Slashdot Mirror
Alpha-Based Samsung Linux Goodness
Peter Dyck writes: "This summer Compaq divested itself of the Alpha technology. The Alpha tech was purchased by Intel who most likely will bury it after grafting its best aspects to their own 64 bit IA-64 system. However, the non-exclusive terms of the deal allowed Samsung
to continue producing and developing the best 64-bit processor architecture there is today. Now, as a happy owner of a four years old DEC AlphaPC164 I was delighted to see this announcement by Samsung Electronics. In short, the upcoming UP1500 motherboard will house a 64 bit 800+ MHz Alpha 21264B CPU, 4 GB DDR memory, 10/100 Mps LAN, USB and yes, it will run Linux."
The UP1500 was developed long before the Compaq/Intel Alphacide... it is not clear whether Samsung has any intention of continuing to support Alpha.
With that said, I feel that Intel makes a superior processor and Alpha's are already a bit outdated. Almost all modern apps require x-86 extensions such as MMX, SSE, and 3dNow, which Alphas do not support. I'd rather be running a hardware platform which supports these innovations and allows software to overcome x86 limitations. Alpha's are 64 bit processors, and they are quite fast, but they do not offer the specialised hardware instructions that x86 supports. Alpha's are like 1960's muscle cars. They're fast, but only because of the brute force under the hood. X86 machines are sleek and smoothe like a Porche because they use brilliant engineering and specialised extensions like SSE. I'll take the Porche over the outdated horsepower any day.
Furthermore, Alphas are limited in the software platforms on which they support. Only certain flavors of Unix will run on an Alpha, while Almost all Unices, Windows, DOS, BSD, OS/2 etc. are supported by x86 based processors.
-atrowe: Card-carrying Mensa member. I have no toleranse for stupidity.
Fact one: what distinguishes Alpha from IPF is not some "pieces" that could be copied over, but a superior design and architecture. In order to take advantage of that, Intel would have to dump IPF and start over, effectively selling Alpha under a different name. That would be an unthinkable about-face.
There is a very nice Alpha-EPIC comparision white paper from Digital, a shame I don't have the URL.
Fact two: the deal just preceded the HP-Compaq one. It's a marchitecture thing.
Leandro Guimarães Faria Corcete DUTRA
DA, DBA, SysAdmin, Data Modeller
GNU Project, Debian GNU/Lin
The present CPU that is employed in Compaq machines like the AlphaServerSC and the Wildfire and in various cluster systems is the Alpha EV67 processor. The previous chip was shipped with a clock speed ranging from 666-833 Mhz. IIRC, the EV67 was able to deliver up to two floating-point results per clock cycle. The load/store units could load 16 B/cycle while the store bandwidth is slightly smaller: 10.6B/cycle. The bandwidth to memory is 5.3B/cycle, however, the type of memory determines the actual bandwidth through the bank cycle time of the memory. We were expecting a scaled up version of this chip named the EV68. It was projected to have an 833Mhz clock speed. I believe that this is perhaps some version of it.
The density used is 0.18 instead of 0.25 which enables the location of a 1.5 MB secondary cache on chip. The largest difference will be that there will be 4 dual channels from the chip to interconnect it with neighouring chips at a bandwidth of 1.6 GB/s per single channel for what Compaq has called "seamless SMP processing". The path to memory is implemented by 4x5 Rambus links as the systems will be fitted with Rambus memory. The direct I/O dual link from the chip also has a bandwidth of 1.6 GB/s. Theoretically the chip could run at speeds of upward 1Ghz.
I know that the Alpha 21264B is based loosely on the EV line of chips (more specifically the 67 and 68), can anybody further verify this with some more details? Thanks.
the byproduct of years of oppression by the white man
Heh heh... I'd like to run FreeBSD on it. IIRC, it supports the Alpha.
I think 21264B is the beefed up version with 0.18 Micron. You should look at the specs: here, while 21264 is here. You can then compare it side by side.
--
Error 500: Internal sig error
Would you care to have a read of here and then explain your car analogy again?
The only Good System is a Sound System
You have to go to the link, and make sure to look at the large image near the bottom.
The image shows the 32bit pci bus only running at 33Mhz! I mean... I own a DIGITAL AlphaStation 4/233, and it has a 33Mhz. THis box is from 97.
Just guessing from what I saw on the page... the kit is a strange malgamation of old, and new technology. The system has 133Mhz, btw nothing new for Alpha, for the memory bus, but not the pci bus.
So... its is 64 bits.... but it isn't that special either.
It isn't a lie if you belive it.
is this part of the specifications:
"2MB of flash ROM
- SRM Console for Linux Install"
This means a REAL setup, with a command prompt. just like a REAL server should have (Think on SUN, PA-RISC, etc) not that crapy menus x86 machines have.
Way to go Samsung. Add 2 or 3 more PCI slots and it'll be even better.
Oh, and did you noticed te AMD 761 North Bridge ? nothing strange here. Athlon shares the same bus with Alpha. AMD licensed it a long time ago, so using an AMD chipset makes perfect sense.
What ? Me, worry ?
What exactly is "REAL WORK???" Last time I checked the people in the GFX dept where I worked did real work and they do most of it in Photoshop...
I remember about 8 years ago, the Pentium was just released with a maximum clock of 100MHz. At the same time, Alpha chips had clock speeds of 275MHz. How come Intel chips have increased clock speed by a factor of 20 while Alpha have increased by a factor of 3?
(Yes, I know that performance depends on much more than just clock speed.)
Quattuor res in hoc mundo sanctae sunt: libri, liberi, libertas et liberalitas.
Other Alpha systems are also not difficult to locate in eBay's Computer section. Just do a search on "alpha". The machines of interest aren't difficult to locate in the results, as there are rarely more than 4 pages' worth.
And the brethren went away edified.
An older board - the UP2000 - is a dual processor SDRAM (not DDR) based Alpha motherboard, which has 6 PCI slots, two of which are 64-bit.
This new board has DDR ram, but only 32-bit PCI, and then only three slots. While nice and all - DDR is good, and of course it's for the Alpha 21264B, not 21264A - this does seem a bit of a step backwards in the IO stakes. Especially when it's noted that the UP2000 has onboard Ultra-2 SCSI as well.
Perhaps this board was originally targetted at the 'lower-end' workstation segment? Does anyone know if a more server-oriented 21264B board is on the way? It seems sadly unlikely given the current circumstances.
If one wants to have 64-bit multiprocessing on a budget, what are the current alternatives?
The Alpha was a good architecture for the time, but with 2+GHz Pentiums I can't see getting excited about a 64 bit workstation. Especially from Samsung, who to the best of my knowledge has never been a player in the workstation market. Workstations are pretty much gone as a market, Sun seems to be the only people staying afloat, SGI is dead, HP has sold thier soul to Intel. The x86 architecture isn't that great but they got the bucks to continue development and beat other better architectures by shear size of thier warchest. I hate to admit it but good engineering often looses to strong marketing (kinda makes you want to cry), but thats the unfortunate truth. I'm not sure if IA-64 will do that well, I think its going to be a tough transition, Intel will probably be forced to make more generations of x86 and AMD seems to be beating them using a lower clock rate, so it may just be a good time to invest in AMD. Its about time that somre revolutionary architecture comes in a shakes things up, things like StrongARM are a step in the right direction, but not really competive for desktop. Transmeta has great technology, but why buy a simulation when you can afford the real thing, Intel has improved their technology by borrowing from Transmeta so Intel in getting ahead and Transmeta without the huge sums of cash is falling further and further behind.
I'm coming to have a lot of respect for Samsung lately, what with their flat panels with integrated TV tuners, HDTV ready flat panels, their nice cheap 770 TFT (of which I have several tied together with a Matrox G200MMS card), their Yepp MP3 player (of which I have one)(it even plays my cdex/lame encoded vbr mp3s), and a host of other cool products, not to mention a nice website. (menu: who we are, what we sell, where we are. Just what we need to know.)
This Alpha board is another in their seemingly endless line of cheap but good products, not cutting edge like IBM or Sony, but taking existing technology and getting it to the masses at a reasonable price and quality.
(/jonbrewer thinks he'll head to etrade and put his money where his mouth is.)
From here: http://www.theinquirer.net/02040103.htm
Samsung Alpha board suffers from DDR famine
And fails to deliver on 1GHz Alpha
By Pete Sherriff , 31 March 2001
THE JOINT VENTURE which produces mobos for the DEC (sorry Compaq) Alpha microprocessor is suffering from a severe shortage of DDR cache memory, according to sources acutely close to the acute famine.
The UP 1500 Alpha, which supports a 21264 Alpha at up to 800MHz speed and comes with 4MB or 8MB of level two DDR cache, is intended to arrive in July, with typical systems costing around $4,500.
But a shortage of cache for the processor is hampering production, leaving system integrators truly "up in arms" and Samsung embarrassed at the short-fall.
Moneyed corporations, non-working 'poor' and criminal prisoners are turning productive citizens into tax-slaves.
>How come Intel chips have increased clock speed by a factor of 20 while Alpha have increased by a factor of 3?
,and some powerusers/scientists, noticed.
Without going too technical, intel designed it's pentium IV to be highly scalable in speed (but look at how poor it performs mhz/mhz-wise compared with AMD), Alpha had a good design from the start and they've built around it, intel went for the marketting hype machine.
Also keep in mind that since over 2 years, not much work or funding has been put on Alpha technology... basically it's the same chip with more cache, reducing die size to increase clock speed and stick yet more cache, nothing much, nothing new, intel did the same with the pentium II/III... but in the same timeframe, intel pushed a lot of R&D and $$$ to pump out it's next generation processors. There's NO DOUBT that with the same energy, you'd probably have a 21464 making the IA64 a bigger joke that it is right now.
The thing that pisses me the most in this story, is I come from an amiga background, I had a lot of respect for both alpha and Mips back then (remember the Raptor Screamernet renderfarm (Mips-based) that you'd stick near you amiga toaster system and it would render 25 to 40 times faster?, or the first lightwave port to alpha, screaming over 40 times the speed or my poor amiga 4000?), I knew that if my platform would eventually die, I'd have a supersweet alternative.
But what happened? Microsoft pulled the plug on Windows2000 on the alpha, ok no problem, there's still some unix alternatives (but kiss goodbye to seeing alpha as a powerfull Windows workstation), and like if that wasn't enough, compaq bought it, waited, left it to die.. just like gateway did with the amiga. Wait till the technology gets too old (funny fact is even 2 years later the alpha CPU is still good and can be compared to current systems...2 years.... think about it).
Anyways, the treatment the Alpha got is so unfair, it went the same way MIPS went, same way amiga went, and it's a proof that it's not the best technologies that wins. When I was still dreaming about seeing Win2K on alpha, and Compaq released it's workstation shortly after buying DEC, I knew there was something wrong because they would NEVER compare to intel, NEVER. but NEVER I thought that one day, the potential INTEL competitor would get bought by.... INTEL.
Here goes my dream of seeing intel shoving 64bits technology into mainstream and normal people and general benchmarks sites noticing "hey, speaking of 64 bits, there's that Alpha processor that is 3 times faster... woah 3times?!? it's worth to check!!! it might be the next AMD!"
It is.. (even if it's pre-amd) only geeks like us
--- Metamoderating abusive downgraders since my 300th post.
If you are hungry for knowledge, slashdot is an all you can eat shmorgasboard...woof!
(Still scanning all the pdfs)
Man 'o man this brings back memories.
I remember a discussion on architecture a while back when I was a newbie about which was better; the invariable "CISC vs RISC" discussion that degenerated into a flame war of mac vs pc.
(being a newbie at the time, that was an introduction to what a flame war was. Glad I had the sense to lurk and listen.)
As the discussion raged on with benchmarks of floating point and integer, FLOPS, expandability, usability and so forth, an Alpha user spoke up.
I forget the exact words but it went something like this:
"I've been reading this thread with great amusement for some time, because *everyone* in it points to a single benchmark run one at a time. On the machine I am posting from I run a NNTP server that transfers about 3G a day, an FTP server that does even more serving internally and externally, I'm a mirror for (I forget who he said) and, keep in mind, before posting to this forum, I was playing Quake @ 50fps. When you can do half of what I am doing on your pc's and mac's or even *touch* my frame rate, then we'll talk."
To say the discussion ended abruptly would be an understatement.
As a point of reference it was about 1994 or so and the pentium was maybe at the 100Mhz mark. 3G of data when 500M was an "increadible" amount of space. Getting Quake up to 30fps on your average pc was darn near impossible to mere mortals (much less a newbie such as myself at the time).
After that, well, Alphas have always been awe inspiring because then, like now (reading the specs) these processors are beasts!
And SMP systems that are becoming common today, well, Alphas and Suns were the only ones I was aware of (at the time) capable of such things...or were more common than their mac/pc counterparts.
Aw, man, I've gone on long enough, sorry about that.
/me wipes away a tear. {sniff}
Thanks to all the posters of the specs, it is going to be a few days until I can wipe this stupid grin off my face.
Cheers,
GISboy
If it is not on fire, it is a software problem.
This is a little thing that people don't talk about much. Of course, it's quite possible that it doesn't deserve to be talked about much.
Memory management is becoming more difficult to do efficiently these days due to the fact that the most commonly used processors (Intel-based) use a memory page size of 4 kilobytes. Each chunk of 4kB must be managed by the operating system. This is the unit of memory used for a great many operations. Swap space is also referred to as the `paging area', where little-used memory pages of running programs get sent.
Of course, 4kB isn't the only page size that Intel CPUs support -- they can also handle 4MB pages (a little large)! 64-bit successors to the Intel x86 platform (both x86-64 and IA64) only support these same page sizes.
Other CPUs can handle different page sizes. I think SPARCs generally have 32kB pages. Alphas apparently do 8kB. Many processors have variable page sizes as well.
While I doubt the page-size issue is going to cause anything to completely keel over anytime soon, I do think that more flexibility could make memory management more efficient and increase performance.
okay, let's review...
The Inquirer has a story posted March 31, 2001 about the UP1500. it says the product is "is intended to arrive in July". it is now November.
these mailing list posts (including some by yours truly), show that the Samsung page in question, has been around since at least April 2001 and so has a page which has listed the UP1500 as "Under Development" ever since.
now, i'm no expert, but i think it is fairly safe to call this vaporware. maybe the motherboard will come out at some point, but for right now, it's silly to treat it as news.
(i will refrain from making commentary about how certain news *cough* organizations should check their sources before posting stories. oops! i just did.)
Incorrect. You can and do overclock the memory. Most often times the FBS and memory speeds are statically linked. In some bios revisions, you may be able to set the FBS as Memory+33mhz. This allows people to use 133mhz FSB processors with slower PC100 memory.
Ever wonder why overclockers are eagerly awaiting the widespread release of PC2700 DDR-SDRAM? Because it can be a bitch to overclock your PC2100 memory past a certain point.
So, your point is basically totally wrong.
Oh, and don't forget that You can also run 64bit/33mHz PCI cards. Nicely enough, most of these cards are backwards compatable with older busses. I have a newer 3Com Gigabit ethernet card that supports 32bit/64bit and 33mhz/66mhz/133mhz. Hell, I don't even know if you can get a motherboard with PCI-X yet, but the damn NIC already supports it.
Anyway, I don't see how this has anything to do with the original poster's point. He may have worded it poorly, but it isn't that hard to figure out his point:
Not having at least 64bit/33mHz PCI in a newer server-oriented board is a major flaw. 32bit/33mhz PCI is quickly becoming stretched thin by the likes of gigabit ethernet and Ultra160 and now U320 SCSI.
Hell, I even stress the PCI bus in my workstation systems at times. Thankfully I now have 64bit/66mhz PCI in my workstations. Thank you Tyan!
I think I will stick with my Tyan Tiger, with 2 x 1.2 Ghz Athlon's. $500 for the board, 2 processors and 256MB of RAM, life does not get better than this.
"Our products just aren't engineered for security,"
-Brian Valentine,VP in charge of MS Windows Development
While there is no doubt that there is lot of cruft in the x86, you have to give Intel credit for getting way more performance out of it than anyone thought they wood. I remember back in the early 90s everyone kept talking about how RISC was going to kick Intel's ass for these very reasons: they would never be able to overcome the limitations of having to support backward compatibility. Yet, they are still standing, and RISC's advantages are very small in real terms.
You should probably doublecheck your sources, as they seem to have misinformed you on a couple of points.
Firstly, the past several generations _are_ RISC chips, with a wrapper around them that translates x86 instructions. This is why Intel chips have more decode stages in the pipeline than any clean architecture would (and why they were so eager to use a trace cache in the Itanium - among other things, it lets them skip the decode stages for instruction batches the processor has seen recently).
Secondly, there is a *huge* performance difference in practice between RISC and CISC architectures, for the simple reason that you can't pipeline CISC processors. You have instructions that do wildly varying amounts of work, taking wildly varying amounts of time to do it, sometimes without the total execution time being known (like the "loop" and "rep [foo]" instructions). Pipelining requires an instruction set with instructions that take roughly the same amount of time and that share many steps in common between instructions. RISC neatly provides all of this.
You can partially pipeline a CISC machine by only pipelining some types of instruction - heck, even a RISC machine will need to special-case things like divide operations - but pipelining is far, far more effective with a RISC architecture.
This was one more nail in the coffin of CISC cores (there are serious hardware and compiler complexity problems too).
It is worth knowing that Microway will continue producing the 264DP motherboard that API dropped a while back. Thus Samsung isn't the only source for Alpha motherboards. And the 264DP rocks:
;-) an 8GB per process vm limit. If you want more virtual memory (don't think swap, just virtual memory), you need to fiddle with your own segment/offset layer or similar.
*) Dual capable
*) Dual memory busses, *each* with 2.6 GB/sec
*) 4GB memory max (I wish this were higher)
*) Dual 64bit PCI busses, don't know the speed
*) Built-in Adaptec SCSI, usb, etc. FWIW, Microway seems to prefer adding an Intraserver PCI SCSI controller (Symbios based) and avoiding the Adaptec controller.
These motherboards can really push data. Systems at 500MHz and 667MHz built around these boards crush x86 cpus at twice or thrice their clock speed. These systems are somewhat expensive, but they're worth every penny. You just can't get similar floating point performance or memory bandwidth from x86 machines, even with the new ServerWorks chipsets.
Because the Alphas are a 64 bit architecture, your per-process memory space is huge. You won't get above 3GB virtual memory per process on x86 under linux, I believe NT has a similar or lower limit and SCO has (had?
For what it is worth, we do in-memory data mining and number crunching in our lab. We regularly have processes with 15GB of virtual memory allocated (of course we're not swapping that much; we may be crazy but we're not stupid =-). For these purposes I love the Alphas. I have no knowledge about web serving, database serving, etc, from Alphas.
-Paul Komarek
In order to be useful most applications, etc. have to be compiled with the least common denominator- period. This translates into something that will not assume that you've got seperate pipelines for execution for the FP and MMX/3DNow type instructions because they want it to run on all those K6-2/3's and Pentium MMX/II machines as well as the Pentium III machines. Backwards compatibility's a double-edged sword.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas