Sun Open-Sourcing UltraSPARC Design

AKAImBatman writes "While everyone was busy with the holiday season, Sun Microsystems quietly announced the start of the OpenSPARC project. Unlike previous CPUs that were based on the "Open" SPARC specifications (such as LEON), Sun is releasing the complete Verilog source code to their latest and greatest microprocessor. Their current time frame for releasing the source code to the public is in March of 2006. Given their success with the OpenSolaris project, it seems that this is likely to be more than just vaporware. So get out your Virtex FPGAs and your Verilog compilers, and let's get ready to hack some hardware!"

Comments

Pardon my ignorance...

[Anonymous Coward]

But what are some other examples of open source hardware? How practical is this approach to hardware? I don't mean things like "get iron hot, add carbon, make steel" but more high tech stuff.

Re:Pardon my ignorance...

[Anonymous Coward]

Go to http://www.opencores.org/ for more examples of Open Hardware.

Re:Pardon my ignorance...

[AKAImBatman]

Pardon my threadjack, but I just realized that the editors secretly switched my link for a competing brand. Unlike Folgers, I'm afraid it's much cooler to get processor news straight from the horse's mouth.

Re:Pardon my ignorance...

[Anonymous Coward]

Yes but there is a bigger advert for Windows servers on the other site.

When will alternative chappies get it into their heads that the best way to finance Linux and open source and all that crap, is to allow Microsoft to advertise on their sites?

Re:Pardon my ignorance...

[halleluja]

Go to http://www.opencores.org/ for more examples of Open Hardware.

Interesting. Is this DIY or do you need to rub a plant-supervisor?

I mean, can I solder my own crypto accelerator for about $50 using basic electronics equipment?

Re:Pardon my ignorance...

[harrkev]

Look HERE for more details on how this works.

But the short of it is that you buy an FPGA evaluation board. You compile and place-and-route the compiled code onto the FPGA. Add some memory and peripherals, and you are off and running.

Needless to say, this is not a beginners project by any means. But is it possible for a few hundred dollars and a few hundred hours.

Re:Pardon my ignorance...

[AKAImBatman]

Needless to say, this is not a beginners project by any means. But is it possible for a few hundred dollars and a few hundred hours.

Nonsense! Beginners do it all the time. All you need is $99 to get yourself a Spartan 3 Starter Kit, the free WebPack ISE tools that Xilinx bundles, and a lot of perseverance in learning VHDL, Verilog, JHDL, or some other Hardware Design Language of choice.

As a bonus, the Starter Kit comes with manuals targetted right at newbies to hardware design. It's so easy to understand the hardware layouts of the board that a monkey could do it, and the booklet explaining the history and usage of FGPAs is most informative. Pick up an understanding of Logic Gates from Wikipedia, and you should be ready to get started. (Personally, I think JHDL is the best place to start because it forces you to deal with the actual gates. Once you get that clear in your head VHDL and Verilog become much easier to master.)

Re:Pardon my ignorance...

[harrkev]

Thank you for proving my point. That $99 kit that you linked to has a Spartan 2 XC3S200. That part has the equivalent of 200,000 gates. Do you really think that you could put an UltraSparc in that little space? You would want something with a gate count in the millions. And even if you could fit it in there, that part only has 27KB of block ram, so your cache would be quite small. Performace would, in a word, suck.

So, yes, it would take hundreds of dollars. Look to the higher-end Vertex II parts, or possibly even the Vertex 4.

And when I say hundreds of hours, that includes getting up to speed with the FPGA software, and interfacing the system to other devices. At the very least, you will need an DRAM interface, or some sort of "northbridge" interface to something that can talk to DRAM. And after that, you will need to get an OS running on it in order to really do anything with it. This entire process can take hundreds of hours and will stretch you in hardware and software.

I work with FPGAs for a living, so I know a little about this. I am a lot more fuzzy on what it would take to put Linux on a new processor, but it can't exactly be trivial, especially since the processor would be embedded in what amounts to custom hardware.

Re:Pardon my ignorance...

[AKAImBatman]

Thank you for proving my point. That $99 kit that you linked to has a Spartan 2 XC3S200.

I'm hardly proving your point when you can't even tell the difference between a Spartan 2 and a Spartan 3.

That part has the equivalent of 200,000 gates.

Yep.

Do you really think that you could put an UltraSparc in that little space?

Nope. I was just addressing your point about FPGAs not being for beginners. If your point was intended to be that UltraSPARCs aren't for beginners, then I agree with you. :-)

(Not that it would hurt anything if newbies played with the compiled code in a testing framework a bit. Lotsa stuff you can learn.)

You would want something with a gate count in the millions. And even if you could fit it in there, that part only has 27KB of block ram, so your cache would be quite small. Performace would, in a word, suck.

*shrug* FPGAs are proof of concept devices that work great for development. You've got maybe 100-200 MHz to work with. You can go for some cheap ASICs from Europractice and push it up to ~1GHz. After that, you need to sub-contract a *real* fab.

And when I say hundreds of hours, that includes getting up to speed with the FPGA software, and interfacing the system to other devices. At the very least, you will need an DRAM interface, or some sort of "northbridge" interface to something that can talk to DRAM. And after that, you will need to get an OS running on it in order to really do anything with it. This entire process can take hundreds of hours and will stretch you in hardware and software.

Not everyone is going to be a hardware wizard overnight. But that doesn't mean they can't get a few designs up and running in a minimal amount of time. For example, the Spartan 3 board comes with a well-documented SRAM interface. Perhaps not great for a full UltraSPARC, but you should be able to get a 6502 clone or LEON up and running on it in a short period of time. Xilinx and those at OpenCores have really been making an effort to make it easy for newbies to learn hardware design. :-)

I work with FPGAs for a living, so I know a little about this.

Well I don't work with FPGAs for a living, and I don't know jack. So HAH!

Err... wait a moment... :-P

FP!

[Anonymous Coward]

FP! SP! TP!

Re:FP!

[alman]

FP! SP! TP!
Frame Pointer
Stack Pointer
ummmm...
What's TP??
Oh yeah
Toilet Paper - aka SCO Legal Documentation.

Re:FP!

[Anonymous Coward]

Thread-storage Pointer?

too far?

[ricochet81]

I'm all for their ideas on OpenSolaris, but this may be going a bit too far. Didn't they open Solaris to sell more hardware? I'm pretty sure a company that doesn't make money is like a species that doesnt reproduce... dead.

Re:too far?

[Seumas]

I'm pretty sure a company that doesn't make money is like a species that doesnt reproduce... dead.

Or, one might suggest, a Slashdotter? :)

Re:too far?

[Virak]

I'm pretty sure a company that doesn't make money is like a species that doesnt reproduce... dead.

Obviously not, as Slashdot is still going strong. ;)

Re:too far?

[Anonymous Coward]

It's an unlikely that any FPGA and indeed, any synthesis tool available "off-the-shelf" is going to handle this puppy. This is more of a "get people interested and excited about SPARC" than a genuine path towards producing open hardware.

Re:too far?

[Short Circuit]

They still own the patents on various parts of the implementation.

From what I understand of patent law, if someone else wants to distribute hardware, they'll still need to get patent licenses.

IOW, Sun is becoming an IP company of a rare sort.

Re:too far?

[wetfeetl33t]

Unless...
They have **something else** up their sleeve
And you thought microsoft was into world domination.
Bwahahaha!

Re:too far?

[blastwave]

Personally I enjoy watching Red Hat, Novell/SUSE, Dell and IBM all squirm as Sun undercuts their prices in every product line. I can get Solaris for free, Sun Cluster for free, the tools for free, Java for free, the source code to Solaris for free and a dual core Opteron or multi-core UltraSparc for dirt cheap. The FUD being sprayed by Red Hat/IBM and Novell is just staggering.

Dennis Clarke
http://www.blastwave.org/

Re:too far?

[Anonymous Coward]

Sun mostly sells support, surely?

Re:too far?

[ciroknight]

Yes, but doesn't all of this kind of make sense for Sun? I mean if they die as a company, all of their IP becomes abandonware, and thus not helpful to anyone anymore. But, as a technique to keep themselves from dying, they opensource all of their previous products that they can, which immediately drives up product awareness and puts more eyes on the product, which immediately attracts investors.. and soon enough Sun is out of the spiralling blackhole that they once were in.

The question is will they continue to honor the Open Source way by continuing their commitment, or will they simply abandon what they've open sourced and move on with the new capital? They can still make money off of their open products by supporting it (which is something a very mature company would do anyways), but will they? This is what we have to look forward to in the future. Maybe a few more companies will come onboard with Sun and we'll see a whole new wave of innovation? At least we can hope.

Re:too far?

[Hal_Porter]

Actually I don't think you'd need a patent license

http://www.leox.org/docs/faq_MLleon.html

Q3)Does it exist a patent issue with Sun?

No, you don't have to pay any royalties to Sun, LEON was developed using the SPARC V8 manual from SPARC International and a licence to develop hardware based on the manual.

Back in 1997, SPARC International required a one-time licence fee of $99 to allow you to design a processor according the SPARC manual. Jiri Gaisler did indeed purchase this licence, so LEON was legally developed.
The architecture licence has now been abolished, and designing SPARC processors can be done without any licences what so ever. This is indeed why Jiri Gaisler has selected SPARC, just see how many times Intel, MIPS and ARM
have sued companies that developed processors using their architecture.

So it looks like Sparc has already been cloned. It's an older version of Sparc, but the one time license fee makes it look as if they never really tried to close the architecture, unlike everyone else.

Mind you, an FPGA Sparc will run at a fairly low clock rate compared to a custom chip, so it's not ideal for a desktop application. And for embedded stuff, I'd suspect that the code density will be too low and the core size too high compared to Arm. And if you can afford to make the custom chips, an Arm license is probably not too bad.

If you're doing an FPGA, the vendor has a bunch of royalty free cores, which are hopefully fairly optimised for the design.

Re:too far?

[NineNine]

Yes, but doesn't all of this kind of make sense for Sun? I mean if they die as a company, all of their IP becomes abandonware, and thus not helpful to anyone anymore. But, as a technique to keep themselves from dying, they opensource all of their previous products that they can, which immediately drives up product awareness and puts more eyes on the product, which immediately attracts investors.. and soon enough Sun is out of the spiralling blackhole that they once were in.

You forgot one important thing... a technique for them to actually make money. I think it makes *no* sense to open source their hardware, because who is going to contribute to it? I wouldn't imagine that there are legions on Sun hardware hackers ready to help them re-design their chips. The only people that would benefit from this would be their competitors. I see this as a great way to kill Sun very quickly.

Re:too far?

[ArbitraryConstant]

"I think it makes *no* sense to open source their hardware, because who is going to contribute to it?"

It reduces concern over what will happen if they die. The larger customers can have their own chips fabbed if absoloutely necessary. It also allows people to customize the chips. For example, MIPS is popular because it can be customized in this way.

"The only people that would benefit from this would be their competitors."

How many competitors use SPARC?

Re:too far?

[Hal_Porter]

Sparc is screwed as an architecture I reckon. It lagged behind the other Riscs in terms of performance, and now Sun have started to make Opteron x86-64 boxes. Given time, I think they'll move over to it across the line.

It's a shame really, Sparc is a really elegant architecture on paper. And x86-64 is a pile of hacks. Still, it's a high performance pile of hacks, and I predict that all desktop machines and servers will use it eventually. Embedded is a different world of course.

As to Sun, I can't see what they're trying to do. They used to be Sparc+Solaris+server grade IO. If you take away Sparc and stop charging for Solaris or offering Linux boxes instead, I can't see how they'll make money. Lots of people have Sparc+Solaris servers though and they need support and new hardware, so it won't kill them for a while, but you have to wonder what the long term plan is.

Re:too far?

[WindBourne]

HUmmmm... Kind of curious why you got modded up, so lets talk about your stuff.

RedHat, Suse, Mandrake, etc all offer linux as OSS. This includes not just the compiler but a very wide array of tools. You can download these for free (mandrake only offers a short verion for free, but it is still including a large number of tools). ALL of the source code of anything marked OSS is available.

So, is Solaris now avilable with 100% of source code? Just a little while ago they were not (I no longer stay up with their development, I just talk to a few of their engineers).

Now, you mention DELL and IBM. Well they both sell hardware with services. Neither of them directly deal with Linux (except for IBM with Linux for the mainframe). You can buy just about any size machine from these 2 companies that is both smaller/cheaper to larger/more expensive than what Sun offers. In addition, when I look at the top 500 fastest computers, where is Solaris in there? Does it hold the majority of the top 10, let alone the top 500? Even in hardware, Sun is not there as much as IBM and others. In terms of Market share, Dell and IBM are individually beating Sun.

So, Linux is just about everywhere and has positive growth. Sun is strong on servers, but with flat growth. And you are claiming that Redhat, Novell, Dell, and IBM are spraying FUD? Hummmmmmm

Should I guess where you are from (as well as your modders)?

Re:too far?

[tomhudson]

As others pointed out, most people will just be simulating the chip.

Of course, this means that you have more people trying out different things.

Sun then takes the most interesting stuff, and puts it in their next ge chips.

How is this (getting people to improve your product) a dumb idea?

Re:too far?

[MBGMorden]

As to Sun, I can't see what they're trying to do. They used to be Sparc+Solaris+server grade IO. If you take away Sparc and stop charging for Solaris or offering Linux boxes instead, I can't see how they'll make money. Lots of people have Sparc+Solaris servers though and they need support and new hardware, so it won't kill them for a while, but you have to wonder what the long term plan is.

Their previous stance in the market isn't sustainable anymore. People are buying more commodity hardware over proprietary stuff like Sun's servers/workstations. People are using Linux more over things like Solaris (and as someone who admins servers who run both, Linux, as well as FreeBSD, are both vastly easier to maintain than a Solaris box).

So the solution to this? Become a commodity hardware dealer. They've still got good brand recognition, and have started this with using the Opteron chips. Seriously, if the prices come down I would LOVE to be putting Sun or IBM servers in our data center rather than the Dells that we have been buying. They just need to start cranking out regular X86-64 boxes a la Gateway, Dell, and HP. Put out some simple Sempron server machines with IDE disks. Very cheap servers have a place as backup DNS's or being part of a set of MX routers (and what one company uses for their cheap servers they tend to use for their expensive ones too). Also, by open sourcing so much of their previous stuff, they get on the "good side" of many techies who are in a position to make recommendations on server hardware.

This would of course destroy much of what makes Sun . . . Sun, but they are a business, and they are logically going to go for whatever makes them a sustainable profit.

Re:too far?

[Waffle Iron]

The only people that would benefit from this would be their competitors.

CPU instruction set architectures are very dependent on the network effect: they need to have large numbers of people using them and developing for them, or the CPU's market share shrivels up and dies. (As has happened to MIPS and Alpha.)

The SPARC architecture won't survive in the non-embedded CPU market with just one major vendor selling systems. If Sun wants to continue down the SPARC path, they need competitors in that space.

Having said that, IMO it's already too late for SPARC. Sun should just move everything over to X86-64 like the rest of the world. The value-add in servers is in the CPU packaging and interconnects, not in the particulars of the CPU core. They can use off-the-shelf cores and not waste resources on processor design. There's no harm to Sun in open-sourcing the SPARC and letting others waste their time tinkering with it.

Re:too far?

[Frumious Wombat]

Sun seems to have entered the, "if you strike me down, I'll become more powerful than you can imagine" mode of business. Let's hope (for those of us who ran and like Solaris) that they're right.

It would be interesting if there is enough of this technology running released for either the up and comers (China and India) or smaller firms here, to start making Sparcs, thereby encouraging it to spread as an alternate platform. Since Sun still sells support, consulting, etc, and the tools to drive it are free, it's a great way of creating mind-share.

They may be edging back towards the technical computing market as well, given that their Studio 11 includes Fortran95 for Opteron/Solaris. Nice to see that they're going to go out with a fight, rather than quietly fade away like SGI has.

Re:too far?

[Anonymous Coward]

There is also a significant team in Rochester, MN working with Linux on iSeries in addition to the mainframe zSeries folks.

Oh and to a few other posts while I am here.... I do believe IBM put the POWER architecture out there opensource (at least in part if not all...sorry I've slept since then) before Sun.....trust us, those of us who work on IBM gear are in no way scared of Sun.

GlobalMind.

Re:too far?

[maswan]

Yes, but Sun isn't actually going to be developing that line of processors anymore. The future for Sun's sparc-based computers are either the niagara design or the Fujitsu sparc64 design.

The sparc64 chips are faster than the ultrasparcs anyway, so someone just taking the now open sourced ultrasparc design and doing slight tweaks are not going to be able to compete with sun on performance, and they'd need huge volumes to compete on price/performance.

Re:too far?

[Anonymous Coward]

> Should I guess where you are from

Why guess? Dennis told us who he is and he's not shy about his devotion for Sun/Solaris,
and he is NOT a Sun employee as far as I know.

>> Dennis Clarke
>> http://www.blastwave.org/ [blastwave.org]

How often do you see Slashdotters left their real name?

I actually agree with you on the sorry state of Sun for _the last 5 years_,
however, I think Dennis is commenting based on Sun _today_ and the future.

I also think it's absolutely true that Sun is offering better deal today
on both SW and HW sides than some of the companies he mentioned, at least
in some segments of the market.

I am not a registered user (hence Anonymous Coward), just felt necessary to
support Dennis as he and his blastwave crews have done so much for the Solaris community.

Re:too far?

[sethstorm]

Yes, but doesn't all of this kind of make sense for Sun? I mean if they die as a company, all of their IP becomes abandonware, and thus not helpful to anyone anymore.

They already do this, and ever since Build 23, they've cemented that proof wrt Solaris 10 and the further defense of that in Opensolaris.

The question is will they continue to honor the Open Source way by continuing their commitment, or will they simply abandon what they've open sourced and move on with the new capital?

Given by their treatment of the sun4m platform and hardware past version 6, I'd say they'll abandon it before it's really too slow to work with it.

Re:too far?

[blastwave]

Well I guess its time to look at some facts. I like facts. That are really solid and, well, factual. You know? Tough to argue with.

RedHat, Suse, Mandrake, etc all offer linux as OSS

OpenSolaris has an OSI license. It is called the CDDL. Welcome to open source.

This includes not just the compiler but a very wide array of tools.

Sun offers the Sun ONE Studio tools for free. Vastly superior to GCC in every measurable way. Of course that is my opinion based on years of code crunching. The fact is that these are available for free. Download and go.
I believe that the source is being made open also.

ALL of the source code of anything marked OSS is available

Absolutely. All of the components under the CDDL are open. Have fun.
More on the way.
Heck, Sun just spent FIVE years working on an entirely new filesystem called ZFS and they released it and open sourced it at the same time. How cool is that?
See : http://www.sun.com/smi/Press/sunflash/2005-11/sunf lash.20051130.1.html

Now, you mention DELL and IBM. Well they both sell hardware with services.

I have heard that .. somewhere. I think Sun does that too. So does my corner store.

Neither of them directly deal with Linux

see : http://www.redhat.com/sundown/
Why is there an IBM logo on that page? Why is there an edition RHEL for POWER but not for Sparc ? Why does it say in big BOLD graphics there "Migrate to Linux with IBM + Red Hat"?
Now go look at : http://www.redhat.com/en_us/USA/rhel/compare/serve r/
The absolute cheapest edition is $349 and the top is $2499 !!
I can get Solaris for FREE.
For UltraSparc or for Intel or AMD Opteron.
The cost of an OPTIONAL software support contract is less than 34 cents a day.
I ought to know .. I bought one because it was five times cheaper than my daily coffee intake and I can't live with that either.
See my blog : http://www.blastwave.org/dclarke/blog/pivot/entry. php?id=107
While you're surfing, look at the three guys at :
http://www.novell.com/linux/unixtolinux/
They are all parked on a bench outside the IT Directors office waiting to tell how reiserfs screwed up their data again and they lost the corporate database because of some messed up kernel patch.
But that is just me guessing.

You can buy just about any size machine from these 2 companies that
is both smaller/cheaper to larger/more expensive than what Sun offers.

Sure. I agree with "cheap".
Show me a 64-bit Opteron that is faster, cooler and less costly than a SunFire X2100.
Really. Anyone can make junk that is cheap and monsters that are massively expensive.
Show me a 64-bit machine that has more horsepower than an 8-core 1.2GHz SunFire T1000 or a 64-bit AMD Opteron machine with more horsepower than the SunFire X2100.
For less money.
Oh, and the Opteron gear has to be certified to run Windows as well as Linux as well as a real UNIX.
Good luck.

when I look at the top 500 fastest computers, where is Solaris in there?
Does it hold the majority of the top 10, let alone the top 500?

Take a long hard stare at my blog from a little while ago :
http://www.blastwave.org/dclarke/blog/pivot/entry. php?id=113
I count, what? 16 e

Re:too far?

[AKAImBatman]

Niagara had been released as UltraSPARC T1. UltraSparc T1 is the processor Sun will be releasing as OSS.

Re:too far?

[EmoryBrighton]

Mod Parent Down,
If you follow the sig link -> blastwave you will see "Blastwave is a software service for Solaris x86 and Solaris Sparc users". The author has significant bias toward Sun and the comment shows it. Undercut price? Compared to Dell? Does dell sell Opteron? Can you compare? Java is free? And umm, java does not work for the other platforms? OS X runs Java at almost realtime.

I am a student at Gatech and there seems to be a significant bias here toward sun. Some SUN machines are very slow (you could buy them for less than $100 on ebay), see felix.cc (list of machines from GT CS Dept: http://tinyurl.com/8l23j )

I see Sun posters everywhere in the IT rooms, even some student labs.... How about a fair comparison?

Re:too far?

[georgewilliamherbert]

  Sparc is screwed as an architecture I reckon. It lagged behind the other Riscs in terms of performance

Niagra is a real leapfrog forwards, though, assuming it performs in people's real world applications environments as fast as it does in benchmarks. For workloads which are thread partitionable (large numbers of parallel processes, like apache, a java web applications server, etc) it acts much closer to a SMP multiprocessor server with something like 32 cores than either an Intel hyperthreading or HT/multicore, or AMD multicore CPU.

See the T-1000 benchmarks page.

Paraphrasing from that...
The single CPU 1 GHz 8-core T1000 system hs about 3x faster on SPECweb than dual 3.8 GHz Xeons, 2x as fast on SPECjbb business apps benchmarks than dual Xeons, etc

Your typical FPS game will vary, of course, until Carmack gets around to massively multithreading...

Re:too far?

[Kadin2048]

Uh ... I'm not sure if this is relevant. LEON was a branch of SPARC; what we're talking about here today is UltraSPARC. So while LEON and OpenSPARC might not have any patent landmines in them, it doesn't necessarily mean that the code that Sun is releasing soon won't. Doubtless it includes the implementations of new research that they've done since the original SPARC code was released, research that was probably patented, so I don't think you can say with any certainty that there aren't encumbrances.

Re:too far?

[JonJ]

They are all parked on a bench outside the IT Directors office waiting to tell how reiserfs screwed up their data again and they lost the corporate database because of some messed up kernel patch.
And someone said something about RedHat/Novell using FUD...
Solaris might be open source soon, but the CDDL is not compatible with the GPL, and thus, is not free-as-in-speech software, besides that, Sun regularly has some shady remarks about the GPL, so who's FUDing? I'm not trusting Sun longer then I can throw them, Novell and RedHat on the other side, who fights SCO, open sources Netscape directory server, shows me that I can trust them.

Re:too far?

[boner]

... and maybe someone should get informed....

Solaris might be open source soon it already is, thank you very much. Check out OpenSolaris.

...but the CDDL is not compatible with the GPL, and thus, is not free-as-in-speech software... luckily not everybody subscribes to the dim point of view that only GPL is worthy of the name OpenSource. (and what was your definition of FUD again...?).

...Sun regularly has some shady remarks about the GPL... how would a company, in your worldview, express criticism of the GPL, without being shady?

....Novell and RedHat on the other side, who fights SCO, open sources Netscape directory server, shows me that I can trust them. ... and the meek shall inherit the earth...

C'mon JonJ, the world has moved on, time to move along with them. Sun, Novell and RedHat are in the business of making money, they have to act on what is good for the company.

Re:too far?

[Jah-Wren Ryel]

Sun offers the Sun ONE Studio tools for free. Vastly superior to GCC in every measurable way.

I think it is pretty easy to measure:

1) What ISA's a compiler supports
2) How Free the source code for the compiler is.

Heck, Sun just spent FIVE years working on an entirely new filesystem called ZFS and they released it and open sourced it at the same time. How cool is that?

I dunno since ZFS is not mentioned once in the document on the other end of the link you posted. Even if it is released, it will be under the CDDL which makes it of little value to anyone other than the people who are chained to sun anyway.

I'm a long time user of Solaris all the way back to 2.5.1 and SunOS before that as well as UNIX on Apollo systems, SGI systems and a whole lot of Fortran IV number crunching on IBM 3090 mainframes back in the day.

For a guy who claims to have been around a while you sure make a lot of hyperbolic claims and misleading remarks - if it quacks like a fanboy...

Re:too far?

[Anonymous Coward]

CDDL is not compatible with the GPL, and thus, is not free-as-in-speech software

That's just plain wrong. Free-as-in-speech has absolutely nothing to do with the GPL. It is true that the GPL is generally considered to be free-as-in-speech, but that's not to say that it's the only one, and it's not to say that any license that isn't compatible with it isn't free. Take for example the BSD license. The BSD license gives the user essentially free reign to do whatever they want, including making changes and redistributing binaries without providing the code. You can't do that with the GPL, which would seem to make the GPL less free in that regard. Nevertheless, you can't legally mix GPL code with BSD code because the GPL requires that all code be GPL.

The CDDL is extremely similar to the Mozilla Public License (MPL) -- the primary differences are around patent protection for the user (which you don't get with the GPL, by the way). Are you saying that Mozilla isn't free-as-in-speech? I do realize that the Mozilla foundation recently underwent a major effort to dual-license all their software under both the MPL and the GPL, but that wasn't because the MPL wasn't free enough. Rather, it was to allow them to use GPL code in parts of Mozilla because the GPL wouldn't allow it otherwise (because the GPL requires all code to be GPL).

Have you actually looked at the CDDL license? What specifically about it makes it non-free? It is OSI-approved, which seems to be a pretty strong indicator that the experts do consider it to be free-as-in-speech.

Re:too far?

[Anonymous Coward]

Vastly superior to GCC in every measurable way.

Everyway eh? So I show just one instance of your being wrong, and then you are FUD; How about how many different arch does it run on? How many different companies support it? How portable is Sun's compiler? Not very

Now go look at : http://www.redhat.com/en_us/USA/rhel/compare/serve r/
  The absolute cheapest edition is $349 and the top is $2499 !!
  I can get Solaris for FREE.
  For UltraSparc or for Intel or AMD Opteron.

Oh, I love that. Lets see. What exactly are you getting for the 349? An OS? No. You are getting support. How much does debian, opensuse.org, or fedora cost? Nothing. Which makes it == to OpenSolaris. But why did you compare linux support to sun's code? Well, that is called FUD (BTW, if you go to argue that Fedora is not Redhat, fine, then get centos, which is redhat enterprise sans support).

So lets talk about support. How many companies support Sun's solaris directly? Just one that I know; Sun. And they charge what? And how many companies support just Redhat code? Numerous. Don't believe it? Look at e-bay. Look at Google. You can get good support for any Linux distro at a fraction of the Sun price (However, I will say "caveat emptor"). I would say that a true free market wins out.

They are all parked on a bench outside the IT Directors office waiting to tell how reiserfs screwed up their data again and they lost the corporate database because of some messed up kernel patch.
  But that is just me guessing.

No, that is not you guessing. That is you being an asshole, as well as FUDing. If you are not sure what is the difference, well, then just look at some of the posts around here.

I do not mind that Sun competes. But I hate the lies, and the BS. I really can not tell the difference between you, McNealy, and Bill Gates. All three of you lie. Gates and McNealy do so because they are the CEOs of their companies that are slowly losing to Linux (if Sun was not losing to Linux, then why would a man who has spent 15 years knocking MS, suddenly turn around and work with them; because Linux/BSD is becoming the common enemy). But you, are a different matter. You funded a company supporting solaris. Cool. So now, you think that you as one person can FUD here and will win over other geeks. Well, not likely. In fact, I have been running various SunOS/Solaris over the years, and will do less now. Why? Because I can not stand the lies and FUD coming more and more from Sunbots. IOW, you just lost not one customer, but probably 20 (and that was just from me).

Rather than get into a pissing contest with you over hardware, let me point out something. Sun's sales are still dropping and other companies are accelerating in sales. For all your talk about how good SUN's hardware is, others are eating their lunch (and breakfast and dinner). Finally, Sun every so often gets into the top spot in terms of performance/dollars. It never last long, but Sun's time out of the top tends to be longer and longer. Sadly, I predict that Sun will lose a great deal more sales before the recovery, but it will take getting rid of the fanatics/salesmen who FUD and return to good engineering to carry the day. Speaking of; Good Day.

Re:too far?

[Anonymous Coward]

Or you could measure the quality of the compiler itself: Sun's compiler is a piece of crap. It lets all sorts of awful code through and compiles it without warning.

Re:too far?

[SillyNickName4me]

Doubtless it includes the implementations of new research that they've done since the original SPARC code was released, research that was probably patented, so I don't think you can say with any certainty that there aren't encumbrances.

You have a good point but I think that the parent was pointing at the fact that even when the original SPARC was still 'encimbered', it was not so bad that it was beyond the reach of someoen wanting to toy with it as a hobby. Pasr results don't guarantee anything for the future, but at least it is reasonable to expect we won't see huge submarine patent issues here either.

Re:too far?

[JonAnderson]

Your wasting your time. I can no longer count the number of posts I have seen which are similar, if not identical, to the one you are replying to. The crime is that the CDDL is not compatible with the GPL (when it's really the conditions of the GPL which enforce this). Nothing you can say will make these people understand the reasons behind the CDDL and why it's ok to have a license that isn't the GPL. The irony is that 'they' pillory Sun for it's business decisions but laud companies like Redhat for doing the same. Why Redhat over Sun?

Re:too far?

[JonAnderson]

Oh, I love that. Lets see. What exactly are you getting for the 349? An OS? No. You are getting support. How much does debian, opensuse.org, or fedora cost? Nothing. Which makes it == to OpenSolaris. But why did you compare linux support to sun's code? Well, that is called FUD (BTW, if you go to argue that Fedora is not Redhat, fine, then get centos, which is redhat enterprise sans support).

You're missing the point. The point was that you can't get Redhat EL from Redhat WITHOUT paying for support. You can get Solaris from Sun NOW for FREE. If you want a support contract it's LESS than Redhat. i.e. you don't have to go to another company to get Solaris for free. I would equate Fedora to Solaris Express. I don't know many people who run Solaris Express in a production environment. The original statement was not FUD, just your ignorance made it seem so. You can get an Ultra20 with Solaris and a support contract for the same as what you are paying Redhat at their minimum price.

Re:too far?

[elgatozorbas]

It is not necessarily a dumb idea but isn't it likely that they will give away more info to the competition than they will receive from the community? Whoever can come up with good ideas probably works in the industry, either with Sun (no need to share) or the competition (d'OH).

Re:too far?

[Soruk]

Nevertheless, you can't legally mix GPL code with BSD code because the GPL requires that all code be GPL.

According to GNU's licence list, you can mix GPL code with code licensed under a "GPL-compatible" licence. One of which is the "Modified BSD" licence. The original BSD licence, however, has the advertising clause, and it is that which renders it GPL-incompatible.

Re:too far?

[Anonymous Coward]

Oh, that's right... Sun is desperately trying to grab back all the ground, credibility and customers that it's lost to Red Hat, Novell etc etc after its financials tanked, it did an IP-swap deal with Microsoft, funded SCO's FUD campaign against Linux, and everyone speculated that Sun was going to be bought out... but now, suddenly it's all about Red Hat and Novell squirming.

Nice spin, boyo.

Java for free

Where? Oh, you mean "No cost". Funny... I can get Microsoft's .net for free too.

Re:too far?

[Hal_Porter]

MIPS is patented.

Clones of it need to skip some instructions, LWL and SWL.

http://www.opencores.org/projects.cgi/web/mips/ove rview

People tried to add support for a cloned Mips chip to GCC, but the patch was rejected on the grounds that "it wasn't a Mips part"

http://gcc.gnu.org/ml/gcc/2004-11/msg00371.html

Mind you if these architectures weren't protected by patents, the companies that invented them would be driven out of business by cloners in China and so on.

Re:too far?

[Anonymous Coward]

Sun offers the Sun ONE Studio tools for free. Vastly superior to GCC in every measurable way. Of course that is my opinion based on years of code crunching. The fact is that these are available for free. Download and go.
I believe that the source is being made open also.

Are you kidding?? The Sun compiler is one of the only modern commercial compilers that doesn't build almost all of boost. It's humiliating to use that POS. I'd actually rank it as the worst mainstream C++ compiler in existence.

Meanwhile, GCC builds all of boost as far as I'm aware.

Re:too far?

[Anonymous Coward]

your link there has a number of sun servers that are 5+ years old, some even 10+ - did you expect a sparc 2 or sparc 10 or an sparc ipx to be fast? even the ultra 5s and 10s getting pretty dated, since I've been using a ultra 5 for over 4.5 years and I got it used.

As for comparing to Dell - its easy to say they undercut, because Dell's 1U servers with Intel Celerons cost more than Sun's 1U servers with Opterons. I don't care what OS its running, the Sun 1U boxes are cheaper with nicer hardware

Re:too far?

[tomhudson]

If they gpl it, the competitors can't incorporate it into their chips and SELL the chips unless they give away their modified source as well, which will reveal a LOT about the inner workings of their chips. Sure, they could use it for chips used in-house, but that's not much of a market when a fab costs billions. Unless Intel and AMD are going to do the same thing - which I doubt.

Also, it takes more than a cpu to make a computer (look at all the different buses we've got now as one example).

Plus, the competition isn't willing to just put out a chip that'll cannibalize an existing line (hence the delay in switching from the Itanic).

Re:too far?

[Anonymous Coward]

Sun offers the Sun ONE Studio tools for free.

That's "free" as in no cost at the moment, everybody.

Vastly superior to GCC in every measurable way.

You know, I would never claim *anything* was vastly superior to anything else in every measurable way. Why? It's because I'm not a zealot pulling claims out of my ass.

Of course that is my opinion based on years of code crunching. The fact is that these are available for free. Download and go.

Again, free as in coated in a think creamy patent portfolio, no source, and no cost... at the moment.

Absolutely. All of the components under the CDDL are open. Have fun.

Oh dear... everything under the open source license is open. You really *do* work for Sun marketing, don't you? How about answering his question. "OpenSolaris" doesn't actually work as a fully functional kernel unless you add in Sun special sauce closed bits. Regardless of your lies -- OpenSolaris is nothing of the sort. Sun was calling it OpenSolaris long before it was ever even minimally open.

Re:too far?

[Anonymous Coward]

CentOS, WhiteBox Linux ... just two of the respected and trusted sources of Red Hat EL without the support. Get a clue.

If I get OpenSolaris without support and call Sun about a problem, what will I get?

If I use CentOS and call Red Hat about a problem, what will I get?

Sweet fuck all, in both cases. Stop lying and spinning.

Re:too far?

[rufty_tufty]

Most teenagers don't reproduce though...

Re:too far?

[JonAnderson]

Huh? I don't need to get a clue. I know about CentOS and WBL. Let me try this again as you didn't get it first time around: You said:

Oh, I love that. Lets see. What exactly are you getting for the 349? An OS? No. You are getting support. How much does debian, opensuse.org, or fedora cost? Nothing. Which makes it == to OpenSolaris. But why did you compare linux support to sun's code? Well, that is called FUD (BTW, if you go to argue that Fedora is not Redhat, fine, then get centos, which is redhat enterprise sans support).

This was in response to the OP suggesting that you A) couldn't get RHEL without paying for it and B) that Sun were actually much cheaper when comparing apples to apples. You seem to be missing the point that you can get Solaris 10 (note, NOT opensolaris) from Sun for FREE. Thats right, it's free for domestic AND commercial use. If you want a support contract then that's what you pay for (not the OS). So you don't need to pay a penny if you want to run Solaris and you can get it from Sun (i.e you don't have to get it from somewhere else) So, to be blunt, I haven't got a clue what you are on about as you don't seem to have challenged the OP's points that Solaris is free (as in beer) and that you can't get RHEL from RH without paying for a support contract.

Re:too far?

[Anonymous Coward]

Spin. Red Hat Enterprise Linux is available for nothing -- it's just not called Red Hat Enterprise Linux.... in every other way it is Red Hat Enterprise Linux. OpenSolaris (although critical components are not actually "open" in any way) is also available for nothing, but it's still called "OpenSolaris" because Sun make it available.

In neither case do you get *any* official support unless you pay for it. How many fucking hairs do you want to split in an attempt to big up Sun. Have you *no* shame? At long last, sir, have you no shame?

Re:too far?

[libra-dragon]

I am expecting Niagara's thirst for threads to work really well with Solaris Containers. This will work out great for consolidating servers; since single threaded processes will no longer be fighting for a slice of time, they'll just be handled by 1/4 of a SPARC T1 core.

Re:too far?

[Nevyn]

Absolutely. All of the components under the CDDL are open.

Wow, impressive misdirection. Yes, all the open bits are open. However you still can't build an OS with what Sun have released. Joerg Schilling seems to be comming the closest, by using Linux bits for the missing pieces ... and even he is having problems because he can't even build non-debug kernels (so performance is "not supported").

Some of these issues will probably be worked out in time, but to say the Solaris is completely OSS now just isn't true. And frankly to suggest that a single vendor that has been shipping bits of their own large static-block of code under an OSS license for a few weeks is the same as multiple vendors shipping a modular community OS for years is just Sun PR.

Sun offers the Sun ONE Studio tools for free. Vastly superior to GCC in every measurable way.

By "every measureable" I assume you mean performance of binary output on Sparc. Certainly the "how to compile Solaris code on GCC" articles from Sun lead me to believe that the Sun compilers accept utter crap and have a huge hill to climb to get to gcc -Wall -Wextra levels of usefulness.

Re:too far?

[JonAnderson]

Spin. Red Hat Enterprise Linux is available for nothing -- it's just not called Red Hat Enterprise Linux.... in every other way it is Red Hat Enterprise Linux. OpenSolaris (although critical components are not actually "open" in any way) is also available for nothing, but it's still called "OpenSolaris" because Sun make it available. In neither case do you get *any* official support unless you pay for it. How many fucking hairs do you want to split in an attempt to big up Sun. Have you *no* shame? At long last, sir, have you no shame?

OMFG, this IS a joke right? Or are you REALLY this dumb? This is as simple as I can make it: 1) Solaris 10 (NOT Opensolaris, NOT Opensolaris - there, did you get that?) is FREE to use and is available for free download from Sun where you can use it for free and it's free. Did I mention it is free?. You can also download the source code to Solaris from Opensolaris should you so wish under a fully OSI approved open source license. 2) The OP's 1st POINT was that you CANNOT get RHEL from RH unless you BUY a support contract. RHEL may be available for free from somewhere but it's not from RH. Whatever, the OP's point is true. Just to refresh your memory, this is what you quoted from the OP in your initial reply:

Now go look at : http://www.redhat.com/en_us/USA/rhel/compare/serve [redhat.com] r/ The absolute cheapest edition is $349 and the top is $2499 !! I can get Solaris for FREE. For UltraSparc or for Intel or AMD Opteron.

So, to summarise: You can get Solaris 10 for X86, X64 and SPARC from Sun where you can use it free for commercial or domestic use. You can obtain the source from Opensolaris, make a distro, do whatever the hell you like under the terms of the CDDL. Sun ONLY charge for support. Nearly all of Sun's software stack, including dev tools, is now free to use. You can only get RHEL from RH if you buy a support contract. If you compare support contract prices, Sun are much cheaper than RH. These are just the facts that you tried to refute (piss poorly I might add). I really don't want to get into a pissing contest with your moronic opinions so let's just stick to the facts. btw, I am happy to eat crow if anything I have said isn't true.

Re:too far?

[Anonymous Coward]

Snip titanic amount of bulllshit that misses the point

Sun are much cheaper than RH. These are just the facts that you tried to refute (piss poorly I might add).

You are clearly deranged. Do you understand English? Red Hat Linux Enterprise + support = money. Red Hat Enterprise (in the form of WBL or CENTOS) - support = no cost. Solaris + support = money. Solaries - support = no cost.

Undertandee? Sprechen zie Englisch... el mongo.

RHEL may be available for free from somewhere but it's not from RH.

No, it *is* from RH, it's just not compiled by them. This has no impact. None. Zero. Zip.

If you wish to wank over semantics and hair splitting, go ahead, Sun-fan. You want to argue over support prices, fine... but that wasn't the question. Centos and WBL are compiled up versions of exactly the same source... they are RHL without the name. That's it. The original point about Solaris being free but RHL always costing money was nothing but Sun-fanboy spinning and bullshit -- and your moronic gushing replies are exactly the same. If you wish expend vast quantities of effort re-restating the same fucking thing because you can't read properly, go ahead.

Re:too far?

[blastwave]

> Joerg Schilling seems to be comming the closest,
> by using Linux bits for the missing pieces .

Really? I'll have to ask him about that. I really don't think so though, its not his style and he and I were talking earlier today about the port process and at no point did we discuss the use of GPL'ed code. In fact, his POSIX compliant star is under the CDDL, not the GPL.

No .. I think that SchilliX is original work as well as OpenSolaris code. He even wrote his own libm that was based on the old circa 1992 BSD code.

As for SunONE Studio 10 compiler being better than GCC, well, how about I take a bit of number crunching code and compile it on Opteron with both GCC and SunONE Studio 10 and see which runs faster and produces the same results?

Seems like a reasonable test to me.

Dennis

Re:too far?

[Anonymous Coward]

It lets all sorts of awful code through and compiles it without warning.

Their Java compiler does this too, it's just that all Java code fits that definition.

Re:too far?

[JonAnderson]

Excellent, now we are getting somewhere:

Red Hat Linux Enterprise + support = money. Red Hat Enterprise (in the form of WBL or CENTOS) - support = no cost.

So you admit that you can't get RHEL from RH for free. That's point 1 that the OP made and which you tried with lots of spin and bullshit to refute and failed.

You want to argue over support prices, fine... but that wasn't the question

It was the question. Just because you decided to answer a different question does not alter that fact. The op stated that if he wanted to get and run RHEL from RH then it was $349 minimum. If he wanted to run Solaris from Sun it was free. You seemed to have confused the free (as in speech) Opensolaris with the free (as in beer) Solaris 10. I don't disagree that you can get the equivalent to RHEL from WBL etc. What I did disagree with was your aggressive response to the OP who had actually posted facts and wasn't actually incorrect. I don't object being called Sun-fan or Sun-fanboy but if you look at my posts, I believe I have just stuck to the facts rather than offer any opinion on Sun or RH. Out of interest, would RH support WBL should you subsequently decide to purchase a support contract? Oh, and if you are still laboouring under the impression the RHEL is cheap, take a look at this:

http://hardware.slashdot.org/comments.pl?sid=17037 4&cid=14199464

Re:too far?

[Nevyn]

I thought libm was glibc based, but according to: this it's FreeBSD based ... but whatever, it's certainly very different from the binary blob you get from Sun, which was the main point.

As for SunONE Studio 10 compiler being better than GCC, well, how about I take a bit of number crunching code and compile it on Opteron with both GCC and SunONE Studio 10 and see which runs faster

You did read what I said right? "By every measurable comparison" is not the same thing as x% faster at number crunching code. Sure llnl.gov might like it, but personally I'll stick with decent warnings. If it wasn't for the semi useful first comment I'd assume you were trolling.

Re:too far?

[htd2]

Sadly your view on the free availability of RedHat ES/AS from other sources than RedHat is not shared by any of the commercial software companies that support their products on RedHat.

Unless it is RedHat from RedHat then you will not get support for their products on it. So if you are a SW developer who doesn't need OS platform support but does need support for the DB/Middleware/Appserver layer then your example is of little interest. On the other hand they will be fine with Solaris 10. No of course they could all be using MySQL and JBoss but this is the real world so they probably arn't.

In addition if you have any issues with the OS/HW combination you will have rather serious difficulties getting any of the major platform vendors to honour their commitments to support RedHat/SuSE on their platforms. Of course you could be using a no name vendor but again this is the real world so you probably are not.

Big Machines !!

[iMaple]

So get out your Virtex FPGAs and your Verilog compilers, and let's get ready to hack some hardware!"

Thats going to give us a nice biiig processor :)

Re:Big Machines !!

[Short Circuit]

I imagine it'll allow for GCC support to become vastly improved.

And wasn't UltraSPARC one of the platforms OpenBSD was having difficulty porting to?

Re:Big Machines !!

[AKAImBatman]

And wasn't UltraSPARC one of the platforms OpenBSD was having difficulty porting to?

As I understand it, the processor was never the problem. The SPARC architecture is well documented and easily obtainable. It was all the other fiddley bits of hardware that have made life difficult for OSS developers. :-)

The roadmap is clear

[d41d8cd98f00b204e980]

Sun has a comprehensive roadmap for UltraSPARC going forward and combining forces with Fujutsu on SPARC64.

These new servers absolutely rock, and at superb prices.

I once had the pleasure of a 4-way Opteron v40z with a development version of 64-bit Solaris 10. It was a screamer, especially compared to our 4-way Dell P4 Xeon box, and 64-bit.

It was plenty fast enough to host 4 zones and several developers working on KDE, gcc and all manner of other stuff.

At last, Sun looks like it's turning the corner (despite the best efforts of some of its PHBs - no names mentioned).

Good luck Sun.

Re:The roadmap is clear

[Saint Aardvark]

"going forward"...do you mean "in the future"?

Re:The roadmap is clear

[Anonymous Coward]

Sun execs are still stupid, they should have opened the JVM. We all prefer SPARC to x86 but for most of us compilers make it a non-issue. We didn't need Sun to open their OS, we already have adequate open source unices. What we do need is a decent managed language runtime but Sun is fighting a rear guard action when they should be making an aggressive push into the one market that they can leverage. I predict that when .NET is everywhere, and the game is over, then we'll be seeing Sun open Java.

Re:The roadmap is clear

[Anonymous Coward]

I once had the pleasure of a 4-way

Rrrriiiigghhhttt...

Re:The roadmap is clear

[Jaxoreth]

I once had the pleasure of a 4-way Opteron v40z with a development version of 64-bit Solaris 10. It was a screamer

Stop bragging -- I'm getting jealous. :-)

Re:The roadmap is clear

[Anonymous Coward]

Sun execs are still stupid, they should have opened the JVM.

What do you mean by 'they should of opened the JVM'?

Re:The roadmap is clear

[Wiz]

To be fair, most of the current sucess is due to AMD's Opteron. To say Sun's v40z is quicker than a Dell system isn't really giving Sun any credit, especially as the v40z was designed by Newisys and is just badged by Sun.

Also, Intel's 64-bit x86 implementation is a bag o crap as they only did it to keep up with AMD. They'd have rather you use Itanium still.

Implementability

[Skowronek]

There are some questions. FPGAs aren't that big... a XC2V6000 that costs $4500 is about the right size for four cores of a simple 4-SIMD 24bit fixed-point signal processor - a UltraSPARC will not fit in it, unless it's seriously cut down.

Also speed of FPGAs is a huge let-down, unless a design takes advantage of their structure. There is no reason to believe that the processor will be designed for FPGAs... It is likely to be therefore very slow, even if you can implement it.

Re:Implementability

[Anonymous Coward]

The UltraSPARC-T1 *is* cut down. The cores are single-issue processors. Not even superscalar. Think "64-bit 386 at high clockspeed".

Re:Implementability

[Skowronek]

Forget the high clockspeed :)

Re:Implementability

[Anonymous Coward]

You don't seem to know what Verilog is. It's not just for implementing something on an FPGA but it's for specifying an ASIC too.
That means that if Sun really releases the source under an open license Dell can go to some chip fab with Sun's source and make Sparcs too.

Re:Implementability

[Skowronek]

I was referring to the gung-ho comment about Virtex FPGAs in the story... I mean, I know on Slashdot we don't read articles, but THIS is going too far.

I decline responding to the rest of the comment :)

Re:Implementability

[Anonymous Coward]

I'm also really surprised if the entire SPARC processor is written in synthesizable Verilog. I would think that this processor would contain numerous asynchronous parts (difficult to synthesize properly) and plenty of custom hard macros (designed at the transistor level).

Re:Implementability

[AKAImBatman]

Believe it or not, I was joking about that. I had a whole post typed up giving a few more specifics and inviting others to comment, but the story failed to appear before I needed to catch the train. :-)

I probably wouldn't have mentioned the FPGA except for the fact that Sun makes a really odd comment on the site about "not knowing an FPGA from an RTL". I'm not quite certain what they're getting at, but it's always possible they're going to have a single core, single thread, cut-down version that CAN fit on an FPGA.

Re:Implementability

[Skowronek]

Heh, thought so... I work with FPGAs as a hobby. They are really great for certain tasks, but implementing big general-purpose processors in FPGAs is counter-productive for anything but research/prototyping/ASIC verification. If you are after performance, it is better to use an off-the-shelf CPU and attach a FPGA-based reconfigurable coprocessor.

Of course, if performance is secondary to simplicity or elegance or flexibility, embedding a cut-down CPU in an FPGA is *the* answer.

Re:Implementability

[Jerry Coffin]

Parent should be modded up -- there haven't been many other intelligent comments on this story.

Contrary to TFA's claim, I suspect for a lot of people, Sun's previous attempt at open-sourcing a core (to the microSPARC) would be a lot more interesting if you wanted to put the design on an FPGA (unfortunately, I'm not at all sure this is still available). I'm not sure how well it would work on an FPGA either, but at least it stands a whole lot better chance, and it's probably still plenty of CPU for most typical FPGA-based designs.

Then again, www.opencores.org, www.fpga4fun.com, etc., already have quite a number of CPU cores available, many without the likelihood of patent problems, and such that are likely to accompany using a SPARC core. Better still, quite a few of these have already been tested in various FPGAs and a few have been put into ASICs as well.

--
The universe is a figment of its own imagination.

Re:Implementability

[randyest]

Then again, www.opencores.org, www.fpga4fun.com, etc., already have quite a number of CPU cores available

OK, well -- you got me there -- four (4) is a number. Even "quite" a number, I guess. Of course, the available free verilog and/or VHDL CPU cores only adds up to four if you're very generous with your definition of CPU core, and if you specifically avoid requiring it to be in the same league as a sparc. But if you don't insist on that criteria, you could include all of them and call it 20. But you'd be including a bunch of old, slow, 8- and 16-bit cores that aren't very well tested, don't include consttaints, and are a bitch to implement (I've tried.)

Seriously, the open-source hardware arena is still quite the slim pickins. Of course there are many reasons for that, not the least of which is the fact that you need more than a free copy of gcc and a $300 computer to compile and make it work. What Sun is doing here is good -- near meningless in and of itself, but still good in the long run. They're raising the bar for open source CPU (and other) cores -- and that's sorely needed.

Re:Implementability

[bigberk]

Yes, I imagine there is more to the processor than just the Verilog described parts. It's not like you're going to be able to take this and build your own competing hardware -- plus, imagine all the lower level design involved in actually fabricating (like you said, gate level stuff, etc.). There is a lot of "etc" to this :)

Unlike open source software, with open source hardware you can not just take the source and re-create the product. What you can do, like open source software, is debug, simulate, test, take individual routines/components, and generally admire and be inspired by awesome engineering.

I really like what Sun is doing here. I've already circulated this article among the VLSI people at my university. What's nice about this approach is that Sun is not putting themselves at a competitive disadvantage by showing the schematics. They are facilitating the design of a lot of related hardware and software that will help grow demand for their architecture. They might even invite improvements (or optimizations).

Re:Implementability

[Jerry Coffin]

But you'd be including a bunch of old, slow, 8- and 16-bit cores that aren't very well tested, don't include consttaints, and are a bitch to implement (I've tried.)

Well, I have to admit I've only used a couple of them, but I don't recall having had a lot more trouble using them than using commercial cores. As far as speed goes, I suppose it's largely a question of what you want. Most of what I've built was almost entirely custom hardware, with roughly the smallest CPU I could use to handle a few things where speed didn't matter.

In my case, I used a free clone of a Z80. Having an instruction set I already knew (albeit had partially forgotten) was handy. As far as speed goes, well, it was quite a bit faster than I needed -- not to mention a lot faster than what I lusted after back when I was running CP/M...

In the end, I s'pose it depends on what you're after. If you want a CPU for a general-purpose computer, then you're absolutely right: most of the free cores are close to useless (and FPGAs mostly are too). If you want to build some custom hardware, and want a CPU to handle things that aren't worth the trouble of implementing in hardware, then FPGAs are likely to work pretty well for you -- and so are the free CPU cores.

--
The universe is a figment of its own imagination.

Re:Implementability

[Anonymous Coward]

Wow, ura dumaass, all chip companies use FPGAs to simulate their designs... eg. ATI GPUs. Nobody wants to spend $$$$ for ASIC before the chip passes basic tests.

Re:Implementability

[Anonymous Coward]

I take it for granted that you work for ATI... /me rolls on the floor, laughing...

ok, I'm convinced

[iggymanz]

open source or not, these coolthreads processors are the first thing from Sun that looks exciting in the last six years. Finally, some leadership. Too late?

Re:ok, I'm convinced

[GigsVT]

Advances in CPU and architecture were constantly announced by these bit players like Sun and Apple as if they were the best thing since sliced bread. If and when the product comes out, it's often a let-down, or the major CPU manufacturers have already made something to do the same thing, and do it faster and much cheaper.

I wouldn't get too excited over it.

Re:ok, I'm convinced

[tonyr60]

"If and when the product comes out, it's often a let-down"

The product is already out..... http://blogs.sun.com/roller/page/bmseer#get_ready

Re:ok, I'm convinced

[Anonymous Coward]

Sun has systems based on the UltraSPARC-T1 and they're available right now starting under $3000. Their throughput and performance for multithreaded applications is simply staggering. In fact, the top-of-the-line Sun Fire T2000 server based on this processor is about $6000 less than the comparably-equipped Dell PowerEdge 6850 (which is the best that we could find from them) and our performance tests with real-world applications show the T2000 providing well over twice the performance while drawing well under half the power and taking half the space (2U versus 4U).

Sure it may not be the best machine to use if you want to play Quake 4 or Half Life 2, but for scalable multithreaded apps, it really is quite impressive.

Re:ok, I'm convinced

[Anonymous Coward]

Don't be so negative, I know what you're saying but this is the source code!
Anyone can make changes to improve the design. The changes are likely to be specialized tweaks to make one task run really fast rather than a faster general purpose processor. And depending on the license everyone gets the benefit of the change.

Say someone comes up with an algorithm that would run ten times faster if only there were just a few more registers or if there were a single instruction that would perform some combination of math on all the registers at once.

Certainly not as DIY as hacking the Linux kernel but this is something someone could base a business on.

Re:ok, I'm convinced

[Anonymous Coward]

At Sun's quarterly announcement today, their benchmarks are showing the T1000/T2000 servers at 3 to 4 times the performance/watt of any Xeon or POWER5 server. These new servers ought to be web hosting monsters. I also wonder if they would make good Sun Ray servers (lots of ram and responsiveness via multiple cores might be good for lots of GNOME sessions).

Re:ok, I'm convinced

[Wiz]

It'll be useless for any sort of gaming, as it has 8 integer cores and 1 FPU core between the lot of them. So any heavy FPU stuff is going to be really slow.... it should also be said that FPU cores are complicated which is why Sun was able to do this system.

The integer performances of their cores isn't going to be great either. For web servering, I suspect it'll be a good box (although I'd like to see some non-Sun benchmarks/SPEC results for this). Comparing it to a Dell system isn't fair either, as the Intel processors are junk. How it compares to Sun's own AMD based systems (which are faster and cooler) is more useful - especially as they are already dual core (2 FPUs too....) and are heading 4 cores.

Re:ok, I'm convinced

[htd2]

An IBM P550 with 4 x 1.9 GHz Power 5 processors does 7881 on SPECweb 2005 and costs arround $50,000 (IBM's web pricing system does not cover the config they tested)

The Dell 2850 with 2 x dual core CPU's does 4850 on SPECweb 2005 and costs $22,447

The IBM x346 with 2 x 3.8 GHx CPU's does 4348 on SPECweb and costs $27,874

The Sun T2000 with 1 Chip and 8 cores does 14001 on SPECweb 2005 and costs $26,995

On this basis the T2000 looks pretty compelling its between 1.9x and 3.2x faster than the most current competitors and is cheaper than all of them except the Dell.

Re:ok, I'm convinced

[Wiz]

Two things:

1. I don't trust SPEC figures as they are normally cooked by the companies involved (see Sun and SpecFP).

2. The Xeon systems would be outpaced by Opteron systems, especially the dual core systems.

I'm not saying the T2000 won't be good at web serving, but I would like some 3rd party evidence of it.

Re:ok, I'm convinced

[htd2]

SPEC CPU which you refer to is very dependant on Compiler and Compiler options, that said the indevidual tests that make up Int and Fp are real world code and so any compiler tweeks generally also help real applications as well. I assume you are refering to the Fp optimisations that Sun added to their compilers in the UltraSPARC III 900 Mhz timeframe, in fact at the time the vast majority of people who reviewed what Sun had done concluded that their tuning was fair and also also usefull to other non SPEC codes which is the acid test of any compiler tweak.

I would also assume that Opteron will be faster than Intel particularly with Dual Core Opteron vs Dual Core Xeon, you can see how badly the Dell Scales with 2 dual core CPU's vs the IBM xSeries box with 2 Single Core CPU's.

That said I would expect that a 4 CPU Opteron Box would be about the same speed as the IBM P550 if it is using Dual Core Opterons and about 15% faster if it is using 4 x Single Core Opterons.

In neither case is the Opteron likely to overhaul the T2000 on web serving, single threaded apps on the otherhand would be a very different story and anything with a significant FP element will also perform badly on the T2000.

Why bother?

[Anonymous Coward]

Rat brains are cheaper and self-replicating.

Re:Why bother?

[Anonymous Coward]

Rat brains are cheaper and self-replicating.

Everybody, you will remember the parent's post years from now. This is the beginning of a meme.

Re:Why bother?

[Ruff_ilb]

But are they open source?

Re:Why bother?

[Anonymous Coward]

Sure they are! Here's a code sample: ...AGACTGACTGTCAATGCCCTG...

Re:Why bother?

[Ruff_ilb]

Funny, that lyses my compiler and gives me a "Chromosome not found" error.

Re:Why bother?

[Anonymous Coward]

It sounds like you've got a bug in your RNA assembler. Please contact God, Inc for assistance.

Re:Why bother?

[Richard Dick Head]

Actually, thats part of a cow protein.

Here's a real rat brain sequence. Look it up on BLAST, I'm sure you'll find the results amusing. (After Format, scroll down below the chart for matches).

ttaggtcgtt aggagaactt actgtgaaca ggctctttta tttcttcaaa agatgtctcc
catttcaagc aaggagcacc catggctgag ggttccctcc cggcatctct ttctcagtca
ccttgagtct ggcctaatca aagactgagg ttatgaagtc gatcctagat ggccttgcag
acaccacctt ccgtaccatc accacagacc tcctctacgt gggctcgaat gacattcagt
atgaagatat caaaggagac atggcatcca aattaggata cttcccacag aaattccctc
taacttcctt caggggtagt cccttccaag aaaagatgac cgcaggagac aactccccgt
tggtcccagc aggagacaca acaaacatta cagagttcta taacaagtct ctctcgtcgt
tcaaggagaa tgaggagaac atccagtgtg gggagaactt tatggacatg gagtgcttta
tgattctgaa tcccagccag cagctggcca tcgctgtact gtccctcaca ctgggcacct
tcacggttct ggagaaccta ctggtgctgt gtgtcatcct gcactcccgc agtctccgat
gcaggccttc ctaccacttc atcggcagcc tggcagtggc cgacctcctg ggaagtgtca
tttttgtgta cagctttgtt gacttccatg tattccaccg taaagacagc cccaatgtgt
ttctgttcaa actgggtggg gttacagcct ccttcacagc ttctgtgggc agcctgttcc
tcacagccat cgacaggtac atatccattc acaggcctct ggcctataag aggatcgtca
ccaggcccaa ggccgttgtg gccttttgcc tgatgtggac tatcgcaata gtaatcgctg
tgttgcctct cctgggctgg aactgcaaga agctgcaatc tgtttgctcg gacattttcc
cactcattga cgagacctac ctgatgttct ggattggggt gaccagtgtg ctgctgctgt
tcattgtgta cgcgtacatg tacattctct ggaaggctca cagccacgcg gtccgcatga
ttcagcgtgg gacccagaag agcatcatca tccacacgtc agaagacggc aaggtgcagg
tgacccggcc tgaccaagcc cgcatggaca ttaggctggc caaaaccctg gttctgatcc
tggtggtgtt gatcatctgc tggggccctc tgcttgcgat catggtgtat gacgtcttcg
ggaagatgaa caagcttatc aagacggtgt ttgccttctg cagtatgctc tgcctgctga
actccaccgt gaaccccatc atctatgctc tgaggagcaa ggacctgaga catgctttcc
gaagcatgtt cccttcgtgc gaaggcaccg cacagcctct agacaacagc atgggggact
cagactgcct gcacaagcac gccaacaaca cagccagcat gcacagggcc gcggagagct
gcatcaagag caccgttaag atcgcgaagg tgaccatgtc tgtgtccaca gacacgtccg
ccgaggctct gtgagcctgc tgcttttgtg gc

Re:Why bother?

[Ruff_ilb]

Homo sapiens cannaboid receptor...

Can't get our mind out of the gutter now, can we ;)

Re:Why bother?

[Richard Dick Head]

Actually the first match is the rat cannabinoid receptor, there's a homo sapiens in the list too cause they're all quite similar, but yes.

There's a good study for you bio majors out there: Hypothesis: Rats Can Get Stoned Too.

Re:Why bother?

[Dynedain]

I'm sorry, I have a patent on that particular gene sequence, you'll have to use another.

Re:Why bother?

[boarsai]

I tried coding a test on my rat cluster with ye olde "hello world"... but it kept printing out "kill mmeeeee" .... obviously I still have some bugs to squash out of my code.

Re:Why bother?

[Ruff_ilb]

Hopefully NOT when they're flying our planes.

Re:Why bother?

[Anonymous Coward]

Oh god, I'm sorry.

Re:Why bother?

[dmayle]

They're right! Open source does make it easier to catch bugs, because you have more eyeballs on the code!

In fact, I've noticed a bug in your code already. Your sequence is supposed to be:

AGACTGACTG CT AATGCCCTG

Sun finally "getting it?"

[einhverfr]

It seems that Sun is trying to outdo IBM in terms of the open processor thing. It will be interesting to see which approach works better in the short run and why.

Now all we need is the source code to the standard Java class libraries and we will be good to go :-)

Re:Sun finally "getting it?"

[NewbieProgrammerMan]

Now all we need is the source code to the standard Java class libraries and we will be good to go :-)

So I must have missed the announcement about the JVM being open-sourced, then? ;)

Re:Sun finally "getting it?"

[einhverfr]

So I must have missed the announcement about the JVM being open-sourced, then? ;)

No. But there are several FOSS JVM's. What is missing to make a real FOSS JRE are the class libraries.

Re:Sun finally "getting it?"

[Anonymous Coward]

"Now all we need is the source code to the standard Java class libraries and we will be good to go :-)"

The source code to the standard Java class libraries has been available on the Sun website pretty much since the first release of Java.

Re:Sun finally "getting it?"

[oscartheduck]

I agree that this will be interesting to watch.
 
I'm looking forward to seeing if a company that seems willing to go forward and take chances on open source development succeeds, and if it doesn't what lessons we can learn as a community from this.

Re:Sun finally "getting it?"

[Anonymous Coward]

Umm... the standard Java class libraries ARE open. Download and install the JDK onto your computer and then go into the JDK root directory (C:\Program Files\Java\jdk1.5.0_05\ on my machine) and open "src.zip". It has all of the standard Java class libraries in there.

The only part that isn't open is the actual Java bytecode compiler and executor.

Re:Sun finally "getting it?"

[psykocrime]

Yeah, but that class-lib code is considered (at best) "pseudo-open" by most F/OSS advocates, because the license it's
provided under does not meet the Open Source definition, nor (I believe) is it even close to GPL compatible.

That said, and as others have already pointed out, there are efforts underway (most notably, GNU Classpath) to create Java classlibs that are purely Free / Open Source.


The only part that isn't open is the actual Java bytecode compiler and executor.

Compilers aren't a problem, there are several open-source Java compilers, including the one built into Eclipse. And there are free runtimes as well, including SableVM, Kaffe, JikesRVM, JamVM, etc.

Re:Sun finally "getting it?"

[farble1670]

Now all we need is the source code to the standard Java class libraries and we will be good to go :-)

check out src.zip in your favorite java dist.

Re:Sun finally "getting it?"

[rubycodez]

More useful for us and profitable for them would be for them to put their full weight behind Linux, to make it run well on all their hardware all the way up the the big enterprise boxes. They're already "giving away" Solaris, does it really matter which OS the customer runs (and on the high end partitionable boxes they could run multiple OS). The customer would get freedom and the ability to run even more software

Re:Sun finally "getting it?"

[Anonymous Coward]

My dick isn't compatible with your mom's nose, but that doesn't stop my from poking her up the pooper before she sucks me off.

Seriously, who gives fuck if the libraries are GPL or not? You can look at the source. Every java installation will have the libraries. If you want to extend or modify them, you can create a subclass. Get over it.

Re:Sun finally "getting it?"

[Anonymous Coward]

Shame that it's under a "touch it and we hunt down and kill your first-born son" license. Sun's bullshit about opening their entire server stack is a massive calumny. It's Java-based. Underneath Java you have this "free" SPARC processor, above Java you have "free" desktops. So wedged in the middle and still firmly under the control of Sun Microsystems where it sits ready to let Sun ass-rape future patent violators... you have JAVA in all its slow, memory hogging, designed by academics and committees, glory.

Ahem

[Anonymous Coward]

Given their success with the OpenSolaris project

Is this supposed to be funny?

Looking forward to the possibility of some new linux on SPARC hosts in the racks.

But... eh?

[Boone^]

I'm doubting that Sun synthesizes verilog to get a 2 GHz processor. Their CAD teams must create custom transistor designs and use formal equivalence with the verilog to prove correctness. Synthesizing the entire processor must require more than one Virtex4 or Stratix-II part, so I can't see people really doing anything with this other than proof-of-concept systems. You could possibly cannibilize parts of the design to make it fit in an embedded system (that's the only speed you'd be able to get out of it). I... dunno. There must be a reason.

The best part in my mind? Think of all of the processor design classes in upper-level EE courses that are going to get a whole lot easier! :-)

Re:But... eh?

[exley]

I don't work for Sun and never have, but you're absolutely correct on the point about not using synthesis alone to design a CPU. To get things like the die size and performance that one wants out of a chip like this, synthesized Verilog is not sufficient, and custom design has to come into play. It's somewhat analagous to assemby vs. high-level language -- if you want stremlined code that might take more design time, you go with assembly. If you want something that's quicker but not as efficient (but still good enough), you go high-level. Maybe not the best analogy, but hopefully it will suffice :)

Verilog design and synthesis is commonly used, however, in smaller scale digital designs like ASICs. There is still physical design work that has to be done, but it's not as intensive as laying out custom FETs like in a big-ass CPU.

Jonathan Schwartz's Weblog

[Anonymous Coward]

Jonathan Schwartz's Weblog: Our Most Valuable Intellectual Property

And now you have a pretty good idea of what's in store for tomorrow. (Pay careful attention to the "open market for parts" comment - we're planning on delivering an extraordinary surprise to the industry. No sense in letting the software folks have all the fun...)

Re:Jonathan Schwartz's Weblog

[AKAImBatman]

Any idea what Schwartz means by this:

Based on a 9.6 Ghz 8-core Niagara chip available in volume

9.6 GHz would rock, but the T2000 only comes in 1.2GHz varieties. Is Schwartz practicing new math or somthing?

Re:Jonathan Schwartz's Weblog

[stoborrobots]

Just guessing, but I suppose 1.2 GHz/core * 8 cores/chip = 9.6 GHz/chip in some marketing exec's mind, or somesuch... :-S

Re:Jonathan Schwartz's Weblog

[buysse]

Sadly, new math. 8 cores, at 1.2 Ghz. 8 * 1.2 = 9.6. Weak.

I've seen worse marketese, but not much.

Re:Jonathan Schwartz's Weblog

[kisielk]

Ah, but the clocks run in an 8-phase configuration, giving you an effective clock rate of 9.6 GHz :p

PowerPC and Arm might get cheaper

[putko]

You can license ARM and PowerPC cores -- but they will probably get a bit cheaper if this one is available for free.

Right now Xilinx and Altera make user-configurable FPGA processors. Most of the processor is fixed, but you can encode what happens for special instructions. Here's one: http://www.xilinx.com/products/silicon_solutions/f pgas/virtex/virtex_ii_pro_fpgas/capabilities/power pc.htm

Now if Sun is giving away the processor, there's no reason for you to pay more for a PowerPC-based design -- someone will make a "cheapo" FPGA-extendable UltraSPARC.

Re:PowerPC and Arm might get cheaper

[Anonymous Coward]

The million-dollar question is: will this do any good? (For computer users, not architecture geeks, that is.)

Right now I can go to the store and get a computer (x86 or PPC) and run entirely free software on it. (It's not so easy to get Linux installed on it out-of-the-box, but that's probably possible, too.)

This may have seemed like a weird pipe dream 20 years ago, but it's common today.

Will we get the same kind of drive behind this that we got behind other open-source projects? Will the industry leader in 10 or 20 years be an OpenSPARC processor, as Firefox is now? Will I be able to go to the store in 2015 and buy an OpenSPARC running GNU/Linux, without raising an eyebrow?

And best of all, with Linux already running on SPARC, and Mac OS being relatively portable these days, will this help to make Windows obsolete? (Hey, a guy can dream.)

Sun's brain damage

[mcrbids]

It's become a custom of late to bash Sun. And, given the neurotic, manic image that they've been projecting, It's not hard to see why... One minute they're holding their cards close to their chest, the next moment, they throw their hand out on the table and say "whattaya thinka that?"

Even though Sun has a wonderful history of sharing their sources in many things, including many of the foundations on which Linux is built, it's really hard to give them full credit because their message is so... mixed.

Well, it may be the Sun is finally making a comeback. I came very, very close to buying a Sun last week. The deal-breaker was that I could not buy one with 2x 300 GB SCSI drives, in a 1U config, with 4 front-mount drive bays. These guys could, and did so at a price that rocked, and the server itself is just quality hardware.

I wish Sun well - there's plenty about them we can use! (EG: OpenOffice)

Re:Sun's brain damage

[Thundersnatch]

All those parts from Avadirect look just like the barebones systems I bought from Tyan. I presume they're using Tyan chassis/mobo systems and stuffing them with parts.

You can't compare a "white-box" server like that to something from Sun (or any of the other tier-1 server manufacturers). With tier-1 server boxes, you can get a 7x24 on-site service contract. This is really important when you're hosting machines off-site where you can't easily get to them, or don't want to spend money keeping a bunch of spares around. You also get support for the software bundle and known-good configurations out of the box. I had all sorts of fun issues with the buggy drivers for the on-board RAID controllers on those Tyans.

Re:Sun's brain damage

[Anonymous Coward]

I have one of these and I say.. the chasis sucks and the machine gives off an inordinate amount of heat.

Re:Sun's brain damage

[afidel]

That's some pretty amazing pricing. A server with 2xdual core Opterons, 8GB of ECC ram, 4x146GB 15k RPM SCSI HDD's and a raid controller for $10K. The only thing that's really lacking is a DVD drive, more of my server software is coming on DVD every month.

Re:Sun's brain damage

[P Fayers]

I know what you mean on the disk front. I'm disappointed that Sun haven't got a Niagra system based on the x4100 design so that you can have a 1U box with mirrored disks. But I think Sun might have something else in mind.

There are rumours floating around of a new Sun storage box called Thumper and iSCSI support is starting to appear in Solaris (including boot support). So why not take the model a bit further and have your 1U boxes with no disks at all talking to a storage box which provides access to disks (or virtual disks) over iSCSI? (There's also Honeycomb, another storage server which is rumoured to be 3U high and hold 16 disks which might be pressed into service in the same way.)

Re:Sun's brain damage

[htd2]

It is odd, the T2000 uses the same chassis as the X4200 but the T1000 does not chassis share with the X2100 or X4100.

Shame.

I understand that the upcoming Opteron Blade chassis from Sun will support the Niagra processors.

Great, now I just need to build a fab...

[the_humeister]

...or outsource this to someone.

You know, I'm not sure how much of an impact this will have. There are other sparc manufacturers, but no one really seems to take notice.

I'm only half joking, so don't mod me funny

[appleLaserWriter]

So, how long until we can buy discount chinese-made ultraSPARC chips at Wal-Mart?

Re:I'm only half joking, so don't mod me funny

[evilviper]

So, how long until we can buy discount chinese-made ultraSPARC chips at Wal-Mart?

China doesn't have the technology, just yet, for chips as complex as high-speed CPUs, so I'd say probably another 5 years.

Re:I'm only half joking, so don't mod me funny

[queenb**ch]

Hah! You were tagged. Seriously, I want to know when Sun is going to get off their behinds and release the 128-bit chip....

2 cents,

Queen B

Re:I'm only half joking, so don't mod me funny

[dakirw]

China doesn't have the technology, just yet, for chips as complex as high-speed CPUs, so I'd say probably another 5 years.

I wouldn't be surprised to hear that Taiwanese companies may already have set up foundries in China to take advantage of the lower costs there. If that's the case, then they should have access to all of the necessary technologies to crank out cheap CPUs.

Re:I'm only half joking, so don't mod me funny

[aminorex]

Given that Schwarz is calling T1 a "9.6GHz" chip, and each of the 8 cores is a 64-bit wide unit, I wouldn't be surprised to hear him calling it a "512-bit" cpu, not surprised at all.

woohoo!

[2Bits]

It's a little bit early to say yet, but if all the "design source, verification suite and simulation models" are released as open-source (as TFA said), and if the license would allow design and manufacture of systems based on that chip without paying an arm and a leg (which TFA didn't mention), I'd say: "Woohooo!", and I'd say this for all the developing countries, including China, India, etc.

And I hope this will "sparc" a revival of the sparc acrhictecture!

HP should've done the same with the Alpha architecture instead of letting it die a forgotten death. What a shame!

Re:woohoo!

[hyc]

HP had to bury the Alpha because it put PA-RISC and everything from Intel utterly to shame.

Re:woohoo!

[exley]

Compaq actually sold the Alpha unit to Intel before Compaq and HP merged... I think. I always get confused on what exactly happened when, but the announcement that the transfer was underway was made in June of '01, and the HP-Compaq merger was announced in September of '01. Either way, the sale of Alpha was most likely a factor in the HP-Compaq mess.

FPGAs are key

[grahamsz]

A field programmable gate array is a little (fairly) inexpensive chip with hundreds of thousands of gates that can be programmed into lots of different types of hardware, and reprogrammed at your convience.

I've worked with stuff from Xilinx and it's pretty impressive.

The other bonus to this is that you can take the Verilog or VHDL langauge (used to write hardware) and simulate it with great accuracy.

Re:FPGAs are key

[jumpingfred]

An FPGA that is large engogh for a modern processor to be synthesized into is about $4500. I don't know about the sparc specifically but I would not be surprised at all if it did not fit in any FPGA out there now.

Re:FPGAs are key

[InvalidError]

Try fitting a P4 or UltraSPARC in a Virtex4-1xxLX, you are going to run into several problems.

1) The ASIC runs at 1GHz+ frequency, the V4 implementation would run around 300MHz at best and cost over $10k for the FPGA alone.
2) Most FPGAs block-RAM and LUT-based RAM can be dual-ported at most, this is problematic for register files where a dozen registers may be concurrently accessed during any given cycle. This would require either register duplication or time-multiplexed register access and a corresponding down-clocking of everything else.
3) Logic is expended pretty fast if you do stuff like 64x64 multipliers using logic only. Sure, there are dedicated multipliers in most modern low-cost FPGAs but these are hard-wired to handle DSP-centric MAC operations.
4) People are upset with desktop CPU's power usage but building similar CPUs on FPGAs would require many times more power to achieve the same performance since FPGA's switch fabric and general-purpose programmable elements have way more parasitic capacitance than ASICs' internal hard-wired traces and circuits. With ASIC, 1M logic gates is only ~6M transistors but a ~1M gate-equivalent FPGA with switch fabric and configuration bits goes beyond 50M transistors with much longer routing delays.

FPGAs are not particularly suitable for general-purpose processing where the system has extensive subsystem interdependencies and shared elements. Where they can truly shine is in applications where the data flow is mostly regular and where processing can be broken down into well-defined self-contained stages like telecom, crypto and DSP. Another area where FPGAs can shine is hard-realtime where they can have dedicated logic to handle time-critical events with 100% deterministic deadlines, unlike modern CPUs and OSes where realtime applications have to put up with unpredictable branch mispredicts, cache misses, preemption, out-of-order execution, etc.

That said, the UltraSPARC's verilog source should make for really interesting reading for logic and digital system engineers and academics like myself. This move makes a lot of sense: CPU designers need to hire new talent and this new talent needs to learn about common practice in real-world designs to be of any use or they'll spend most of their first months just catching up. With a real-world design in the wild, CPU-designer job postings could ask people to specify which architectural components they would like to improve and the interviews could steer towards presenting those improvements instead of often irrelevant technicalities.

Re:FPGAs are key

[mvdw]

Agreed with most of what you say, except I'd like to expand on the comment "FPGAs are not particularly suitable for general-purpose processing where the system has extensive subsystem interdependencies and shared elements." FPGA's are also particularly suitable for low-volume applications where the designer can't afford to spin an ASIC, and also to fast-to-market applications where the time to market may be the critical factor in success or failure of a particular project.

I've worked a little with some of the smaller Altera FPGAs (mostly FLEX10K, ACEX and Cyclone series), and I find them a fantastic alternative to implementing stuff in microcontrollers - I have a project at the moment where I use the FPGA as a real-time coprocessor for the microcontroller which does most of the communications processing. In the future I might have the confidence to use one of the many soft processor cores available and remove the micro from the board entirely.

Re:FPGAs are key

[Anonymous Coward]

With ASIC, 1M logic gates is only ~6M transistors
6.32M transistors, to be precise. But unless you've actually cracked one open, you wouldn't know that. It's ok. Fortunately for you, ignorance is not stupidity.

Apart from letting us all know that you couldn't find work with a decent chip maker, what did you add to this discussion exactly? The guy clearly and correctly states that there are ~6M transistors for every 1M logic gates..and you go and agree with him (You see, 6M transistors really is around 6.32M transistors) and then you try to sound clever, or whatever it is you thought you were doing.

Now I'm begining to understand why you couldn't get a job in a decent hardware design team.

Re:FPGAs are key

[InvalidError]

A microcontroller is not exactly in the same complexity league as a superscalar general-purpose 32/64bits CPU. Most microcontrollers are in-order, single-threaded, use internal program and data memory, have very tight per-instruction timing, fairly simple internal architecture and rarely need to access more than 2-3 internal registers simultaneously for any given OP since they only issue one OP every 1-3 cycles, this does not pose a major problem to FPGAs since a triple-ported register file is still quite manageable. Superscalar pipelined CPUs need much beefier register files since general-purpose CPUs have more complex addressing schemes, need to read up to three register for every issued OP and often needs to write a result at the end of the pipeline... if a CPU can issue four instructions per clock, its register files need to have at least octuple read ports and quad write ports to avoid frequent stalls due to register file contention, a construct that would be a major issue on FPGAs since most are designed for dual-ported registers, multi-port reading can be done via register duplication but there is no direct way of doing multi-port writes.

Microcontrollers on FPGA is feasible and practical because most are simple in-order, non-superscalar and non-pipelined, this makes them much more compatible with FPGAs' dual-ported registers. Anything significantly more complex than that will quickly become costly.

Re:FPGAs are key

[rufty_tufty]

Not to nit pick, but block ram/lut ram is intended for small functions needing ram, not for register files. If you want to have what you describe then you still have registers available to do exactly that. i.e. the same elements that are available to you on an asic are also ther in an FPGA.

And FPGAs are more general purpose than ASICs. That's why you can build a CPU on an FPGA - it'll just run slower than an ASIC doing the same thing. But then ask a CPU to simulate a 10 million gate FPGA and it'll do about 1ms per minute.
So the FPGA can be said to do a better job of being a CPU than a CPU does of being an FPGA.

why yes IAAFAAE (I am an fpga and Asic engineer;-))

Re:FPGAs are key

[InvalidError]

I have never tried synthesis of a large multiple-ported register file on FPGA but I suspect it has very limited practical scaling. At the very best, it would require monstrous multiplexers. Guess I'll have to try it and see what happens with a 128x32bits 8xRead+4xWrite (12 addresses) register file - I will not be expecting anything usable. (IIRC, UltraSPARC is actually 512x64.)

As for FPGAs being better CPUs than CPUs can be FPGAs, this is pretty obvious given that FPGAs can do everything in parallel almost exactly like ASICs do while a CPU has to emulate all that parallel logic serially.

BTW, IAAFAAVE. (V = Validation)

Yea! How to combat cpu level DRM

[nurb432]

We just 'make' our own CPU's and run Solaris ( or NetBSD ). So long after AMD and INTEL have sold their souls, practical open computing can live on.

Not that leon ( and other 'open' cpus ) was 'bad' but, this is from the people who brought SPARC to the world..

Re:Yea! How to combat cpu level DRM

[Anonymous Coward]

Where are you going to "make" them? Got a fab plant in your backyard?

Besides, if they really want you to have DRM, they'll just put it in the motherboard instead, then. Or a special decoder card. Or a USB dongle.

Extremely Practical

[nurb432]

Just google for "FPGA development boards".

For 200 bucks you can stuck the code of a CPU ( and perhaps more, if you choose something small ) on it. ( more $ gets you more speed and useable 'space', but 200 for a starter kit is more then enough to answer your question )

And if you got the cash, ASIC is always an option too.

Java Classes Source [Re:Sun finally "getting it?"]

[codergeek42]

You might want to take a look at GNU Classpath: http://www.gnu.org/software/classpath/

According to http://www.kaffe.org/~stuart/japi/htmlout/h-jdk14- classpath.html, it's about ~97% complete with respect to the standard JDK spec (version 1.4).

Big things like OpenOffice.org run just fine on the Free Java stack as well: http://peter.ramshacklestudios.com/images/openoffi ce.org-free-java-stack.png

who(m) does it help?

[Sebastopol]

Excluding the fab, it takes an enormous amount of design and layout effort to go from RTL to masks. SparcT1 is not a purely synthesized design. Even if it were, the tuning required to make synth work is a nontrivial effort requiring a significant tool foundry.

I suppose that once we have open source versions of: schematic capture, synthesis, floorplanning, layout, timing, validation, and mask generation, then we can focus on an open source process and an open source fab. Not bloody likely!!!

I think the biggest benefit here is that now both hackers and Universities now have a REAL architecture to study in their classrooms. I'll definitely be on the prowl for resumes of students who studied real microprocessor Verilog in college, and not simple ISCAS circuits or architectures from the 1980's.

Re:who(m) does it help?

[Anonymous Coward]

It benefits Sun, because they get good PR essentially for free.

Useable Speed?

[nurb432]

Depends on what you consider useable speed.

Personally a few hundred mhz effective speed is useable. Is it 'cutting edge'? No, but more then useable. ( i think they have LEON's up to 300mhz in high end FPGA's ).

Much as do you *need* a 3ghz intel machine? No, A 400mhz PII managed well will get the job done quite well.

Gate useage? Wont know the answer to that until we see the code.. But since you can get a SPARC compliant design into a reasonably sized FPGA, dont count it out, yet.

Re:Useable Speed?

[Boone^]

300 MHz in an FPGA is quite an accomplishment. It's fairly straightforward to get small sections (PHYs and what not) going at 400 MHz in Virtex parts, but across the entire design? Your LUT count will explode as it throws gates at it to solve troublesome paths.

Re:Useable Speed?

[niktesla]

Xilinx Virtex II Pros and the soon-to-be-available Virtex 4's have up to two IBM Power PC hard cores embedded inside them. This cores max out at 300MHz. It's doubtful that a soft processor core would match or beat that speed, especially since the clock rates are limited to 400MHz (except for the SerDes sections).

Re:Useable Speed?

[randyest]

Yeah -- they're embedded -- not implemented in FPGA in the normal method.

Re:Useable Speed?

[Dan Guisinger]

Exactly, they embed 2 low-end PPC cores into the fabric. There are different versions of the Virtex 4, depending on what you need. We are using Virtex 4 SX35s on our projects at ipVolution, which have 192 hardware multiply/accumilator blocks, which are a godsend in doing DSP work. On other Virtex 4 FPGAs, you can get RocketIO & SerDes functionality for multiple Gbps signalling, up to two Ethernet 10/100/1000 MACs, as well as up to two hardcore PPC CPUs.

The Virtex 4 is really marvelous. You may be able to convert parts of the UltraSparc design to use the newer MAC blocks for multiply, add, etc in both floating point and fixed point; as well as use it for parts such as instruction pointer counting, etc since you can use it in accumilate only mode. But they are still really designed for DSP work only. The block memories on the chip are spread around the chip, making large caches nearly impossible to impliment.

There are some things a FPGA is good at; being a general CPU is not one of those things. You'd want to get an ASIC designed; costing several hundred thousand (minimum) to go from Verilog to prototypes.

Yes, there is some hand layout, but not as much as you'd think. Many common functions already have pre-built blocks that companies like IBM substitute, and their tools quite often do all the conversion between Verilog and their pre-designed logic. They are quite good at what they do.

Re:Useable Speed?

[Boone^]

Virtex4 keeps slipping, so it's not quite marvelous yet. It might be relatively big and might seem fast to someone used to working with embedded DSPs, but they're a PITA to engineers used to working with ASICs. The multi-million dollar NREs (for latest 90/65 nm technology) might push more of these engineers to work with FPGAs in the future, however.

Hand layout, of what? Niagara? Hell yes there's hand layout. IIRC Sun uses TI fabs, and they most likely go the foundry route where they present TI with GDSII files and TI starts making masks. Sun eats the yield problems. There's no pre-built blocks to substitute, unless Niagra uses some IP like SerDes or DRAM controllers. Sun's process probably looks like this:
1) Sun logic designers produce verilog which is verified to be functionally correct.
2) human-synthesis groups reduce blocks of verilog to whatever transistor optimized technology they've got (domino logic or what have you), which will have balanced clocks. human-synthesis groups use tools to measure static/dynamic timing, but if software were good enough to create the finished product at the size and clock rate they need they'd simply synthesize the verilog. Synthesis tools use tracks for routing nets/power/ground and sizing blocks. It's kind of like playing SimCity, but realizing that a good portion of the land in your city is reserved for roads. Synthesis tools are far from perfect. If you need speed, you have human-synthesis groups Sun, AMD, and Intell all have them. ASICs != 80%+-custom processors.
3) the transistor and verilog are run through FE tools to determine that they're the same.
4) Final place/route takes the transistor-optimized blocks of logic and adds global wiring. Some EDA tool extracts design to GDSII format.
5) Profit^H^H^H^H^H^H give away the source code for free

Only Parts of the Core are Opened

[cometsupply]

The first sentence states:

"Sun Microsystems Inc. is looking to ramp up interest in its new UltraSPARC T1 processor by open-sourcing parts of the multicore chip."

For those looking to actually burn an UltraSparc onto their favorite FPGA board are going to be out of luck. Sun couldn't release all the code because they probably have some patents or license agreements.

Re:Only Parts of the Core are Opened

[Slashcrap]

For those looking to actually burn an UltraSparc onto their favorite FPGA board are going to be out of luck.

Yeah, but those people are going to be out of luck anyway. There's no way they're going to be able to cram the UltraSparc design onto an FPGA.

I doubt that Sun were even suggesting that. My guess is that the Slashdot editors added that. The problem is that they're always wide-eyed with the possibilities of technology, but don't actually know enough to consider the real-world issues.

They put me in mind of someone that's just starting to learn a bit about processor technology. They read that the 486 only used a few hundred thousand transistors while the P4 uses a few hundred million. And the first thought they have is, "Wow! You could like make a CPU with a thousand 486s on it and have a supercomputer on a chip! Why isn't Intel doing this??? I must post my idea to a newsgroup immediately!". And of course the answer is that in practice the idea sucks. There are minor practical problems like pin count, memory bandwidth, heat dissipation and writing massively parallel code. But they haven't reached those chapters in the textbook yet.

Re:Only Parts of the Core are Opened

[AKAImBatman]

I doubt that Sun were even suggesting that. My guess is that the Slashdot editors added that.

The italics are the poster (i.e. me), not the editors. Read this thread for more info.

They read that the 486 only used a few hundred thousand transistors while the P4 uses a few hundred million. And the first thought they have is, "Wow! You could like make a CPU with a thousand 486s on it and have a supercomputer on a chip! Why isn't Intel doing this??? I must post my idea to a newsgroup immediately!". And of course the answer is that in practice the idea sucks. There are minor practical problems like pin count, memory bandwidth, heat dissipation and writing massively parallel code.

Are those the only problems? Pff.

Pin Count - Use the I2C protocol and route the instructions inside the chip.
Memory Bandwidth - I hope you have one hell of a registry file to use as a cache. ;-)
Heat - Nothing new here
Massively Parallel Code - How about massive numbers of programs running in parallel?

Seriously, though, the real problem is that people tend not to understand that parallel processing does not equal faster processing. Sure, you can cram 32 execution threads into a single chip (which, BTW, is exactly what the UltraSPARC T1 is), but it's only going to be useful for certain cases. It won't make Quake III run any faster, but it sure as heck will allow a loaded down Web Server strut its stuff in style.

How is this something new?

[IGnatius T Foobar]

How is this something new? SPARC has always been, more or less, an open processor design.

Go to http://sparc.org to see.

SPARC already has multiple manufacturers building independent but compatible chips. SPARC was designed to be an open, multi-sourced processor design. Scalable PRocessor ARChitecture.

Re:How is this something new?

[AKAImBatman]

Excuse me, but I didn't I just say the same thing in the *fine* Summary? As I said, this is "new" because Sun is Open Sourcing one of their top-of-the-line processors instead of a knockoff chip.

Re:How is this something new?

[Wesley Felter]

It's the difference between an open spec and open source. Nobody seems to understand that any more.

Re:How is this something new?

[Anonymous Coward]

And the difference between a processor architecure (SPARC v9 / x86 / POWER) and implementations (MicroSPARC, SPARC64, UltraSPARC / Pentium3, Pentium4 / G4, G5).

FPGAs aren't where it's at

[Anonymous Coward]

The point of the open source release is the possibility that you can run Sparc emulation. Then on your emulated CPU, you can run Sparc Linux. Hell, then you can run a 16 way Beowulf cluster under emulation all on your desktop. It will run like a dog, but will be way cool.

Re:FPGAs aren't where it's at

[Wesley Felter]

Er, you could emulate a SPARC years ago; open-sourcing the design of the processor doesn't affect that. But then, I get the feeling that I've been trolled.

Re:FPGAs aren't where it's at

[nurb432]

Doubt it was a troll, just someone trying to be funny, and failing miserably.

Interesting times, no?

[JumpingBull]

So, the uber hackers get their hands/minds on a real processor.
One simple word springs to mind:
Refactoring

My experience has been that the cagey designer (I do hardware, friends do software) wants to find that mystic dotted line that says "cut here". The complexity just falls away after that careful choice.

I'll just watch from the sidelines, but I expect a lot of cross pollenation and hybridization to come from this intersection of hardware and software.

Crossing my fingers here, but if it works....

brain damage?

[Mr2cents]

Sounds more like a sunstroke to me.

No, I dont work for Big Blue. Just wanted sum4m.

[sethstorm]

By some previous moderation, I dont think you're correct about me working for IBM if you were implying that, or if that was even your thought.

Unfortunately, once you've been outed maybe you might want to bring some facts, and some sun4m support for good measure.

free money!

[Flunitrazepam]

With up to 32 simultaneous threads in one incredibly low-power, low-heat processor, you gain the high-volume throughput you need, while saving millions on power and cooling costs.

Wow, you'd have to be an idiot to not want to save millions! Nevermind that the technology (Cool Threads) sounds like the name of a Hot Topic spin off... Sign me up!

Re:free money!

[Anonymous Coward]

Sun ain't got nothin' on my 32-way multithreaded penis! Yup, the whole cheerleading squad...AT ONE TIME!!!

Re:Sun, I know what you're trying to say it...

[cwcpetech]

...but this does not make for an excuse to keep sun4m off OpenSolaris, no matter what mutterings your kernel devs might have to justify it. Now if you made it possible to rig up an ultrasparc over mbus somehow with this, you might get somewhere. However, that still does not justify including sbus, but cutting out the architecture that primarily used it from source code (even if it was an early build). At the very least, it'd be a fitting end to see a source/binary (for drivers if impossible to source) release of OpenSolaris fit to sun4m. That would get people still with Quad Ross SS/20's, Ross SS10's, SS5/170's, and the others out there that definitely could take advantage of some of the features as well as have the ability to fix some of the major offenders (since I guess Bart Smaalders seems to have forgotten about the numerous bugs, and just wants to shove sun4m under the "closed hierarchy" carpet).

It's not about the cost, it's about having the ability to fix the bugs on these machines. Distraction with an open Ultrasparc core isnt a good idea.

The best I can see out of it is Sun trying to follow in the paths of OpenPOWER. At least the company behind OpenPOWER is the same one that at least did one last release (AIX 5L 5.1) that allowed some sort of openness that Sun would drop at the fall of a hat. Add the part that it's probably a partial release of the components, and I doubt it's even that. It might be worth a good look but by no means a complete core. So it's most likely not a distraction.

Yes, there's an error in there.

[Anonymous Coward]

it should have read "Add the part that it's probably a partial release of the components, and I doubt it's even that. It might be worth a good look but by no means a complete core. So it's most likely not a distraction." as a separate paragraph.

Put down the crackpipe

[jmorris42]

> Personally I enjoy watching Red Hat, Novell/SUSE, Dell and IBM all squirm as Sun
> undercuts their prices in every product line.

And how exactly are they doing this?

> I can get Solaris for free, Sun Cluster for free, the tools for free, Java for free, the
> source code to Solaris for free and a dual core Opteron or multi-core UltraSparc for dirt
> cheap.

So? RHEL is a support contract. I doubt Sun is handing out service contracts for free or even price matching RH. If you want the RH software sans support pick your RHEL Rebuild and start installing. Same for the RH GFS, it is Free as in GPL. Java on the other hand is NOT Free. Sun hardware is getting competitive, which is a good thing but 'dirt cheap'? Put down the crackpipe.

Re:Put down the crackpipe

[adrianmonk]

So? RHEL is a support contract. I doubt Sun is handing out service contracts for free or even price matching RH.

Perhaps you should start comparing prices, then:

• Sun Support is available in three levels: Basic, Standard, and Premium. The prices are $120/yr, $240/yr, and $360/yr for a single processor-socket system. $360/yr gets you unlimited live phone support 24/7.
• Meanwhile, Red Hat offers a wider variety of support plans, including separate workstation and server plans. The cheapest server plan is $349/yr and the most expensive is $2499/yr. You'd have to get the $2499 plan to get 24/7 phone support.

So, it would appear that Sun's support prices are actually lower rather than beating Red Hat's. In fact, for one of Sun's cheapest server systems, you can get Platinum support for $2304 for three years. Platinum support includes both 24/7 software support and 24/7 two-hour response time on-site hardware support. That's cheaper then one year of Red Hat's software-only 24/7 support.

Sun hardware is getting competitive, which is a good thing but 'dirt cheap'? Put down the crackpipe.

Again, compare prices:

• You can buy a 1U, Opteron server system from Sun for $745.00. It doesn't have a disk, but you can add one for $150, bringing the price to $895.
• Meanwhile, the cheapest rack mount server of any kind you can get from Dell will cost you $999. It does include a disk, but its processor is a Celeron with 256K cache.

So, the Sun server may not be as cheap as building a system out of spare parts lying around in your basement, but it really is pretty cheap compared to the competition in that space.

Re:Put down the crackpipe

[Hektor_Troy]

So? RHEL is a support contract. I doubt Sun is handing out service contracts for free or even price matching RH. ... Java on the other hand is NOT Free.

Granted, you talk about GPL and hence "Free as in libre", but you're picking and choosing your choice of that word (free) as you go - plus, your choice of capitalizing "NOT" makes it look like a typo, where you forgot to release the shift button; you could have gone with "not" instead.

Pick one, and stay with it.

You start with "free as in no cost", so stick with it. Java is free (as in no cost).

Then you switch to "Free as in libre". Java is not libre.

But then again, the RHEL support contract is neither free nor libre. It's not like you can get it for free from other people and still have it called RHEL support, is it?

Sure, you used a capital "F" to indicate you're talking about "free AND libre", rather than just "free of charge", but that's just patching broken language anyway. Not your fault - RMS or whomever should have picked a better phrase to begin with (like libre or liber), but then again - he came up with the idea quite a while earlier than the European Commision.

Re:Put down the crackpipe

[Anonymous Coward]

> You can buy a 1U, Opteron server system from Sun for $745.00 [sun.com]. It doesn't have a disk, but you can add one for $150, bringing the price to $895.

And before anybody flips out on this point, I would like to point out that you can boot Solaris10 off of pretty much any externally connected drive (FC, UltraSCSI, SAN, USB, etc.) hence the reason that more and more Sun is offering configs without drives, more and more people just don't need internal drives anymore.

Re:Put down the crackpipe

[thesqlizer]

You can buy a 1U, Opteron server system from Sun for $745.00. It doesn't have a disk, but you can add one for $150, bringing the price to $895.

Agreed, and that's indeed a nice price.

My "favorite" part is that the rails--the RAILS! mind you--are an additional $150!

Want cable management? Wave buh-bye to another 95 greenbacks.

Put another way: if you add both, you're paying a STAGGERING 25% of the cost just on racking the server!! You mean to tell me that of the HUGE costs Sun has in developing these machines--to say nothing of actually manufacturing them, 25 cents of every dollar gets eaten up by the folks in the rack-mount department?!

That's like charging $20,000 for a car, but asking for another $5,000 if you actually want umm...err... wheels. :-) [Note: Yes, I know there are wheels out there that cost that much, but they're *really* the exception as are the folks that spend that kinda loot on 'em. Rails for a 1U though are more like oil or gas for the car for most firms. You might try to throw 'em on a shelf once, but you're not likely to do that with more than a couple--that gets old fast.]

All that said, the overall cost of the gear is still pretty derned reasonable--and if it's anywhere near as good as the older Sun gear I personally use regularly, it will be good indeed.

Re:Put down the crackpipe

[Anonymous Coward]

FWIW, the Dell rails cost about the same, and don't come with the cable-management fold-outs like they used to either.

Re:Put down the crackpipe

[Unknown Relic]

Well it should be noted that the Dell server doesn't come with rails either, and they add $99 to the price of an entry level PowerEdge 850 for the "static" rails. These static rails are literally two pieces of metal as shown here, and don't telescope (much less have a cable management arm). If you want rails that actually let you pull the server all the way out of the rack - still no cable management - well, those are $129. It's pretty sad how much any vendor will gouge you on rails. These rail prices from Dell are actually much lower than in the past as well. A few months ago rails for a few PowerEdge 750s I ordered were priced at $200 per server. To their credit, Sun seems to have their 2U server rails priced the same as their 1U offerings. Rails for a Dell 2850 are a painful $250.

Re:Put down the crackpipe

[Anonymous Coward]

To their credit, Sun seems to have their 2U server rails priced the same as their 1U offerings

The rails are the exact same product, hence the same price. Oh, and if you buy in any volume, its fairly easy to get Sun to throw the rails in for free. I've often called up my salesman and asked for sets of rails only to have them arrive at my data center soon after without a quote or PO ever being cut.

Re:Put down the crackpipe

[Anne Thwacks]

That's like charging $20,000 for a car, but asking for another $5,000 if you actually want umm...err... wheels

When did you last buy a new car?

Check the Mercedes catalogue - wheels (steel or alloy), steering wheel (wood, plastic, leather), and even paint (standard finish - $700 or the metallic finish $1400) are extras. The catalogue price is $30,000, but by the time its on the road, its $55,000! And that is still without twin SUN servers in the boot!

Re:Put down the crackpipe

[halleluja]

So, the Sun server may not be as cheap as building a system out of spare parts lying around in your basement, but it really is pretty cheap compared to the competition in that space.

I would go for this one: http://www.dansdata.com/ttransport.htm/ Whether Sun's cheap NOW does not interest me. I'm not willing to strap myself to one company for the coming years. Sure, they're feeling the heat NOW, but what happens to the prices once the heat's over?

Re:Put down the crackpipe

[thesqlizer]

The point was simple math--not literal dollars. Ummm... it's a R-A-T-I-O. ;-)

If it's a $55,000 car, the wheels would then be another $13,750. Think of it that way. Would nearly $14K for wheels on the $55K Merc out of line? THAT's my point.

Re:Put down the crackpipe

[Anonymous Coward]

To be fair, you are giving Redhat a bit of a raw deal with this. All of Redhat's contracts support at least 2 CPUs - Sun's prices do not. If you compare the above with a dual processor system, then it changes. Likewise if it has a large number of processors.

Remember some people like the flexability of using more than one supplier. Your plan makes you very dependent on Sun. Per example, at a previous job I recommended IBM's Opteron servers but we ended up going with Sun's v65x servers. Whilst they weren't bad, when we needed 64-bit we were stuffed and had to wait for Sun to bring out their Opteron server.

Whilst Sun's line of Opteron servers now is impressive, the flexibility of using Linux + any tier 1 vendor does have benefits.

Re:Put down the crackpipe

[Anonymous Coward]

You're indeed getting ripped off. Over here in germany, I've been paying 75 EUR (slightly more than US$ 100) for the rails for PE 1750, 1850 2650 or 2850 in the last few years. The nice rails with calbe management of course.

You dodged their EOL bullet

[cwcpetech]

At least your hardware wont be subject to Sun's "Hardware EOL Games" (e.g. 32bit cut early on in Solaris 10 builds before OpenSolaris), which will certainly hit those Opterons they're selling. If you did build for them, you could even repurpose the machine and have well-supported hardware unlike what Sun has done in the past *cough*sun4m*cough*.

Re:You dodged their EOL bullet

[Anonymous Coward]

Eh? There's no reason to run a 32-bit kernel on an Opteron or UltraSPARC. Solaris was 64-bit on UltraSPARC even back on craptastic 250MHz CPUs. Most of the userland can remain 32 bit; it works fine.

Also, sun4m support was dropped only this year in Solaris 10. That's over 10 years of hardware support in the OS for those old SS10s and SS20s (we're talking 50MHz CPUs! whose gonna miss them?).

Yes, but

[Anonymous Coward]

This article refers to UltraSPARC.

Listening to Open Graphics Project?

[Theovon]

Someone there must have been paying attention to the Open Graphics Project. They're working on a design for an open source graphics card. Naturally, the drivers will be open source, but ultimately, so will the Verilog code to the internal GPU design.

I want one

[nut]

I want one of these processors in my PC running Linux.

Re:I want one

[maxwell demon]

I want one of these processors in my PC running Linux.

Sorry, but a computer using this processor is not usually called PC.

MOD PARENT UP!

[BoneFlower]

+50 balls of steel.

Sun Microsystems - back from the Shark Jump?

[iggymanz]

Sun lost their edge in the dot-com era, too much easy money coming in from people who didn't shop. This T1 is a great first step to recovery, maybe the one other thing they could do with their recent storage acquisition is make SAN technology much easier to manage & scale , it's such a tedious manual affair at the present with draconian licensing (have to pay to activate ports on a fibre switch, for crap's sake, or activate controller features).

mobos?

[user32.ExitWindowsEx]

any matching mobo information out there? it'd be nice to combine this with a properly tuned archive of debian or gentoo for sparc (archive == both source and binaries) and then have a 100% open computing environment

Get your 0-day Microprocessor Ware3z HERE!

[Anonymous Coward]

Sun's microSPARC processor has been available for download for quite some time. It is available as synthesizable verilog source code and I think it comes with a PCI master. If sparc is not your style, download their picoJava processor instead.

Parts are being released, not the entire design!

[jayslambast]

I'd be surprised if they release the codebase for the entire chip. There is alot of industry secrets that go into processors that are not patented. By giving this away, they'd give IBM and HP the abilitity to analyze the performance of the chip with a fine tooth comb. It only provide more FUD for HP and IBM to throw at them.

The interface portions of the chip to be probably be opened up. Mainly to allow other companies to design chipsets for their new system.

I would like to see where they go with this. Software is a great thing to opensource because changes can be make with little effort and it is very cheap to verify your changes worked. Chip design on the other hand is extremely expensive, with slow turn around times and difficulting in debugging. Not only do you have to worry about the code, but how to design it properly for the process that is being used to fabricate it. Opensource is all about turnaround time, and chip design currently can not support that. Now if someone could create a extremely high density reprogrammable chip (500M gates) then all bets are off.

However, this will be a great learning experience to see any code they provide. It will give student and people in developing nations a chance to learn what goes into a 'high' performance chip design.

Now imagine... Apple switches processors *again*

[FFFish]

What if, eh? Is the core OS X codebase portable enough that Apple could change horses in mid-stride?

We'd have computers with an open CPU but closed whatever-their-hand-is extension running an open OS with a closed GUI. Only thing left to deliver would be an open Distribution/Payment Channel with a closed DRM; I'm game for that if it's similar to the Mac-side small-apps economy. (A powerful freebie and a right-priced professional version. The freebie is so good you can easily justify the full-package price, especially knowing that it's going to a small team, so you're directly putting supper on their table.)

This could be a damn fine bit of social revolution.

Warning! Virus!

[Gogo Dodo]

Don't replicate that code sample! It's a virus!

Cat among the pigeons

[tanveer1979]

While from a normal guy point of view, this is not much. I mean even if you have the source code of a processor, you cannot do much with it. But in the hardware world this is nothing short of revolutionary. You see, there are lots of companies which make things like Modems, TV cards, Cell phone chips etc., etc.,, many of them big names like (TI, connexant, broadcom etc.,) who have a licensed processor from ARM or MIPS. For every chip sold they need to pay some royalty, and over a period of time these royalties come to about 10-15% of the profits or even more. Now there companies have talented VLSI engineers who can design a custom core given enough time and resources, but since these companies are not in the business of making CPUs, buying is cheaper. With Open Sparc project, now they will have the actual source code of the CPU. Modifying a CPU to suit a particular application is not a very big deal. this will lead to price drops in MIPS/ARM cores, which means cheaper modems/cell phones etc., etc., But the biggest advantage will be spinning of new CPU design comanies or units whose job will be to modify the core for particular applications (Network switches, comm processors etc.,) and then sell it to VLSI big boys. these modified Cores will cost 1/10th of that of currently avalilable cores. Hats of to you SUN.

Sparc++

[Doc Ruby]

First hacker to optimize the Sparc design in FPGA wins a prize! Or integrate it with a DSP...

Re:Sparc++

[bhima]

From my understanding of the ecosystem of UltraSparc that award would already go to Fujitsu as their offerings are routinely faster.

Re:so...

[Anonymous Coward]

Bill? Is that you?

Re:Now imagine... Apple switches processors *again

[bhima]

While we are having this fantasy...

tack 128 Bit Vector execution unit on it ala AltiVec and I'd be onboard.

Taiwan is part of China. [nt]

[Anonymous Coward]

Taiwan is a part of China.

sha bi

[Anonymous Coward]

Your penis is small.

Uncle Mao told me so.

Open Source Verilog

[d0n quix0te]

I was wondering if any of the slashdotters can throw light on Open Source Verilog compilers. I know of the icarus project but that seems to be quite ... dead? Any active or good open source verilog compilers out there. I would appreciate your inputs.

Re:Open Source Verilog

[Bassman59]

"I was wondering if any of the slashdotters can throw light on Open Source Verilog compilers. I know of the icarus project but that seems to be quite ... dead? Any active or good open source verilog compilers out there. I would appreciate your inputs."

To be blunt about it: they all suck.

Also, the handful of open-source Verilog compilers aren't synthesizers, so you can't make a chip with them.

Your best bet is to download Xilinx and Altera's free (as in beer) tools.

Re:Open Source Verilog

[thechuckbenz]

Icarus operates fairly well. I only occasionally stumble across a legitimate verilog construct that it doesn't like, and a workaround is rarely difficult. I don't think the synthesis side of it works that well, but it's first and foremost a simulator. Another simulator was just released as open source, Veriwell (it's on sourceforge). It's reportedly 4x faster than Icarus, but Windows support is uncertain (it works under cygwin), and it doesn't support newer verilog updates (the 2000 updates, vpi, and such). I haven't tried the free VHDL simulator yet.

Now it makes sense

[krygny]

Oooohhhhhhh! It's open cores - I thought it was open Coors ... then you get iron hot, add carbon, make steel.

Re:Now it makes sense

[aonaran]

I thought it was open Coors ... then you get iron hot, add carbon, make steel.

No no no... that won't work at all, if you open the Coors before attempting work it will never happen. The correct order is get iron hot, add carbon, make steel, repeat as necessary THEN open Coors.

bmc, build 22 would be fine for sun4m, tyvm

[sethstorm]

You must not have known of the 200mhz ROSS modules which arent exactly slow mbus modules in their own right (throw 4 in and see how they stack to an Ultra1 on 32bit operations), and it'd only take a slightly polished version of Build 22 to take care of those who wanted to fill in the spots for themselves. Sure it'd be in all practicalities Solaris 9 repatched, but it would be a source release that's well overdue for these cpu's that would at least be an end point - you'd have somewhere to start building your sbus support, and somewhere to maybe write something to match dtrace.

Not 100% open source

[tendays]

it'd be nice to combine this with a properly tuned archive of debian or gentoo for sparc (archive == both source and binaries) and then have a 100% open computing environment

That won't be 100% open source, you still need the source code of the universe. (That code that physicists are busy reverse engineering)

:-)

Re:Parts are being released, not the entire design

[bbrack]

Now if someone could create a extremely high density reprogrammable chip (500M gates) then all bets are off.

Good luck building an UST-1 in on a 500M gate ASIC - the chip is close to that xtor count as designed...

Re:Sun, I know what you're trying to say it...

[gdamore]

Why would possibly want to have developers working on sun4m hardware? These days you can pick up an UltraSparc based machine with an UltraSparc-IIi for dirt cheap -- just check out ebay. For not much beyond the cost of shipping you can pick up an Ultra 30, 60, 10, or 5 workstation. You can pick up servers for a little more -- e.g. Ultra 450, etc. And any of these machines with a nearly 10 year old processor will blow the doors off of any sun4m machine. S/Bus was used in early UltraSPARC designs (Ultra-1 systems, mainly), and killing S/Bus will indeed happen at some point in the future, but there are still a lot of Ultra/1 systems out there. (And indeed, the UltraSPARC-1 based systems, with processors less than 200MHz, are already not supported due to a CPU bug that made them insecure in 64-bit mode.) Conversely, there is a huge amount of effort required to support both a 32-bit kernel and 64-bit kernel on SPARC, and the distraction really takes resources away from other cool things that we'd like to see Sun continue to improve upon (e.g. ZFS, network performance, etc.) Now if you have some old *sun4d* hardware, I'd understand why you wouldn't want to just toss the machine out. But even that is a very aging arch. I think you folks with Ross modules need to update to the new millenium. :-)

OpenSolaris + OpenSPARC the Freedom platform

[maitas]

Acording to www.gnu.org:
"The GNU Project was launched in 1984 to develop a complete UNIX like operating system which is free software: the GNU system"

  Of course we have GNU/Linux, but that's only the OS part of the system, you still need the hardware part.

  Nowadays, thanks to OpenSolaris + OpenSPARC + GNU we have a mainstream 100% Open Source system (Oerating System + software + hardware).

Thanks Richard, but your service is not longer required ;-)

-- http://maitas.blogspot.com/ --

OpenSPARC + OpenSolaris = 100% Freedom platform

[maitas]

Acording to www.gnu.org:
"The GNU Project was launched in 1984 to develop a complete UNIX like operating system which is free software: the GNU system"

    GNU/Linux is only the OS part of the system, you still need the hardware part.

  Nowadays, thanks to OpenSolaris + OpenSPARC + GNU we have a mainstream 100% Open Source system (Oerating System + software + hardware).

Thanks Richard, but your service is not longer required ;-)

-- http://maitas.blogspot.com/ --

Re:Now imagine... Apple switches processors *again

[Anonymous Coward]

And Dreamcast emulation! ...and a pony! (pony optional)

Quietly ?

[Sarrek]

Not quite .. There was a Huge Webcast and everything from the quarterly meeting.

hehe .. Silly writers.

I have to ask

[WindBourne]

I have been quiet here, but after reading the other responses, I have to ask. Why lie and misdirect? It does not serve you

For example:
I asked So, is Solaris now avilable with 100% of source code? Just a little while ago they were not
You responded with: Sun offers the Sun ONE Studio tools for free. Vastly superior to GCC in every measurable way. Of course that is my opinion based on years of code crunching. The fact is that these are available for free. Download and go. I believe that the source is being made open also. ALL of the source code of anything marked OSS is available Absolutely. All of the components under the CDDL are open. Have fun.
Why do that? I asked a simple legitimate question. That is, is all the source code now available? From reading the other responses, it is still not available. Yes, there are some who are working on patch source code to make it work, but it shows that Sun is not really serious about OSS. Bear in mind, that I have no issue with the CDDL. It does not work with GPL because GPL wants full openness at all time (BSD license is about true freedom; in spite of that, I stay with Linux). In addition, you do the same with the compiler, You state that it is free (as in beer), while we are having a discussion on free (as in open) software. You purposely misdirect.

This gets picked up by people such as myself who then mistrust you, and implicitly Sun. You are actually hurting your cause (keeping solaris alive and maybe even growing it), and from what I have read, your company. If you want people to move to this, you have to be honest. Please keep in mind, that people such as myself have been around for a while and have learned not to trust Sun (I have been coding on *nix since the mid 80's (and coding since '77)). Sun has earned that reputation. I keep hoping that it will stop, but posts such as yours and Sun's actions, perpetuates it.

Re:I have to ask

[blastwave]

I can not believe that you are actually calling me a lier and then hanging your problems with Sun one me.

Amazing.

There is a CDDL license. Everything that is not owned by or has some license issue with a third party is open or in the process of being made open. Those bits that can not be made open for some third party license issue are handed out in a redistributible binary form.

What is there in this that is a lie? Nothing.

Get your emotional problems with Sun sorted out. They have nothing to do with me.

dc

Ever hear of a FPGA?

[nurb432]

You dont actually 'make' them, you 'program' them, and then you make your own boards ( with other FPGA's for support chips, such as USB controllers, RAM, sound, video, etc. ).

Its not as hard as it sounds, as a *low level* ( but useable ) computer, can often fit in one chip. Just google SoC + FPGA.

Embedding DRM at the chip level on FPGAs would be somewhat difficult due to their 'universal' nature. Not impossible, but difficult.

Re:Sun, I know what you're trying to say it...

[sethstorm]

It's that them leaving sun4m and your quoted sun4d outside of the source code release - just due to some major dtrace difficulty is the real issue. Just enough to get a (workable, buildable without extras) source-backed "Solaris 9++" system up would be fine for the sun4m/d platforms. No need to dig up Solaris 6 cd's for the HCL Olympics you have to go through to install relevant support - it's all there to build.

The only thing that keeps the Ultra 1 even close to supported is the Ultra 2 which fixes some serious bugs. Sure, it might require a bit of work, but at least there's something to work with on documentation - whereas they removed support completely after Build 22 for sun4m/d, not even to *think* about doing anything but deflecting questions as they have already.

The most you'd tie up is a few in the legal department.

Re:Sun, I know what you're trying to say it...

[gdamore]

But you never addressed my original question, which is why would you even waste your time trying to *install* a new OS on this ancient hardware? For pennies on the original new cost dollar, you can easily pick up systems with 64-bit processors, that still run all those old 32-bit apps, and several times faster (and more responsive to boot).

The obsolence of the sun4d/sun4m archs has been a well known/planned/documented thing. Sun warned folks about this *years* ago. (Go back to some S9 release notes for proof.)

And supporting the older platforms (just keeping the old code) does cost real time and resources. Everything from the time it takes to pull the extra source files (which on a given pull isn't much, but over the course of thousands of developers and a couple of years, it adds up!), to the time it takes to build it, to whether or not it is going be tested.

And if an interface needs to be updated/changed, you may find yourself going into this old code to "update" it so that it won't break a compile.

This all takes time, and it is not trivial. Believe me, as a former Sun kernel developer, I was quite gleeful whenever I got blessings from management that I didn't need to worry about one form or another of backwards compatibility. It usually meant one less set of #ifdef's I had to sustain.

Actually, in the case of the sun4m and sun4d hardware, it was really a consequence that the kernel memory guys wanted to move away from a 32-bit kernel. There were various problems with larger configurations and 32-bit kernels, and it just made things a lot cleaner/easier if they could assume a 64-bit memory model. The only way to cleanly do this was to remove support for the systems that couldn't run 64-bit kernels (sun4m/sun4d, and some Ultra-1s with early processors with defective 64-bit CPUs.)

The decision to remove 32-bit sparc support was made a long, long time ago -- long before OpenSolaris was ever conceived. So at the time there was no thought that anyone in the open source community would want to maintain support for the older platforms. Even though the source is open, I still suspect a lot of the decisions that drive its development are going to center around what corporate customers want, and support for ancient 32-bit sun4m hardware just isn't likely to rate very high.

SPARC Has Been Open Since Day One

[Anonymous Coward]

"It seems that Sun is trying to outdo IBM in terms of the open processor thing."

IBM's "OpenPower" initiative is just a cheap immitation of the almost twenty year old SPARC International organization (http://sparcinternational.org/).

SPARC has been openly licensable since 1989, and the the 32-bit v8 version of the ISA is an IEEE standard. Also, this is the third processor Sun has released the Verilog of. It also released the MicroSPARC IIep processor and the picoJava core in 1999 under the Sun Community Source License.