Slashdot Mirror
Phoenix BIOS Software Available for Crusoe
Titten writes "Computerworld reports that Phoenix BIOS software is available for Crusoe. Phoenix worked in secret with Transmeta for more than two years to prepare its PhoenixBIOS software for use with Transmeta's chips. Here's a release from Phoenix. I guess we'll be seeing computers with the Crusoe chip sometime soon now!"
Not for desktop machines my ass. Why do chip manufacturers say this? And why do some of us have ballsy 700mhz Athlons, or dual PIIs, when we were told that kind of setup is "not for desktop machines?"
Oh yeah, and why is the Debian, Transmeta, and Dreamcast logos all look surprisingly similar? Heheh.. Dreamcast II will probably have a Transmeta processor and be running Debian...
Both of the major "broadband" netowrking technologies (cable modem / xDSL) offered in the Austin area require the installation of an Ethernet card to connect to service. Granted, broadband isn't all that widespread now (maybe a couple of million subscribers, I think), but Fast Internet is something you can really sell to the average consumer. It's much much cooler than "fast hard drive" and at least as cool as "fast printer."
Now, what does this have to do with boot code and ethernet? Maybe not a whole lot; but point is that ethernet is probably going to become more standard as time goes on, and I'd certainly like to be able to use it as it's done on most large workstation boxes.
(On the other hand, there is considerable architectural support needed to get this level or interoperability up and running--most implementations seem to have some sort of imbedded Forth in them and a well-defined interface for storage and network controllers. It may well be "too late," at least for this generation of PC's. Let's face it--the new PC specifications from MS have a lot fo problems, but at least they're an attempt to ditch at least some of the 19-year-old baggage of the PC architecture)
-- Have you ever imagined a world with no hypothetical situations?
Phoenix is definitely not everyone's favourite BIOS company in terms of setting things up and user interface but they are the de-facto standard for the BIOS programming interface.
If you take any BIOS programming guide you will find mostly two names: Phoenix and IBM. IBM is actually resposnsible mostly for the PS2 additions. Some of them have been adopted, some not.
So a BIOS design usually looks like - Phoenix starts, defines the interface spec and everyone else follows.
As a user I dislike phoenix as much as you do if not more, but as a programmer I have to admit that if you want your things to work you have to rely on their docs, standards and implementation. Se la vie...
Baker's Law: Misery no longer loves company. Nowadays it insists on it
http://www.sigsegv.cx/
Yes I agree we must allow them some time to let the ball roll so to speak. However, the fact of the matter is that Transmeta does have some good ideas but really no working prototype. I mean where is the laptop that is supposed to be running on this new chip?
I have to admit though that their marketing strategy is a stroke of genius. Bring in one of the most respected techies in the world (Linus Torvalds) and combine that with a cloke of secrecy for a few years and you have a winner. Or at least a very curious public. I have to admit that the hype may be slightly over-inflated and artificual considering that most of it is riding on Linus's laurels, but only time will tell.
As mentioned above, Transmeta has some heavy duty backers, so we can expect them to be around for awhile. It never hurts to have "Big Blue" on your side. What will really be interesting is to see how the other chip manufactures like Intel and AMD will respond. It's almost a given that they will try to create a similar product which will directly compete with the Crusoe processor. Essentially, Transmeta will have to either specialize or diversify to keep afloat in the "crazy" computer world. Anyhow it will be interesting...
Nathaniel P. Wilkerson
NPS Internet Solutions, LLC
www.npsis.com
Nathaniel P. Wilkerson
www.haidacarver.com
The instruction set is x86, yes.
The chip is not pin compatable, and requires it's own unique blend of support chips, I would imagine (as any processor does).
The function of BIOS is like a mini HAL(Hardware Abstraction Layer)to deal with firing up the motherboard components in the right order, setting interrupts, and providing boot code to the CPU. This process is different for every brand of motherboard, and every chip out there. BIOS presents a standard interface to the OS itself.
Yes, the code morphing software loads before the BIOS. IT has to.. the BIOS is written in x86. But then the BIOS has to take care of the rest of the motherboard.
while this may be great for Crusoe - it'll encourage more rapid adoption of their chips.. I can't help but wonder what this will do to Phoenix in the long run. They've basically dropped their pants and mooned Intel in front of the entire industry.. do 'ya think if Crusoe fails they might not be.. umm... irritated?
All I can say is thank you to the computer gods that it has not wormed its putrid way into any system that I have to use. I am getting ready to buy a new system...now there are two things I have to ask NOT be put on my system: Windows and Pheonix!
There is another solution...I could turn off my monitor or even just close my eyes until I hear the hard drives thrashing the lovely tune of the Penguin waking. Only then, when Linux is starting, can I open my eyes and know that everything will be all right.
-- soldack
Consumers don't, of course. But in addition to real firmware being fun for the technically savvy, it's also very useful in embedded systems such as these. Why? Because it's nice to put simple drivers for simple hardware in 1 or 2 MB of ROM rather than wasting scarce RAM on more complicated os-level drivers. Because it's nice to be able to boot from a network. Because it's nice to have hardware diagnostics built in, especially for machines that aren't likely to have lots of flashing lights, beeps, and other traditional failure indicators. In embedded systems, good firmware can make the difference. Now, I realize these aren't strictly embedded systems, but it's easy to see (at least for me) that they would greatly benefit from a nice OF implementation.
Naturally, Transmeta's business plan revolves (for now) around x86 compatibility, and that means top-to-bottom peecee "emulation." I personally think that's a terrible idea that will condemn them to hang in the past while most everyone else rockets into the future, but in context, partnering with Phoenix makes perfect sense. And real firmware really just doesn't.
ObFirmwareAnecdote: The ultralinux project once mocked Sun's motto as "The PROM is the computer" in appreciation of real firmware and its benefits.
I believe you are correct in the areas you cover. However, there are developers, systems administrators, and "average users" out there who don't want "more Intel-compatable." There are people sick of the cruft and crap inherent in being compatable with everything Intel has created since 1980. This crap makes life worse for everyone.
Operating system developers work around old architectural problems--there because to remove them would be to break some DOS application from 1983. Users get poorly-designed, ambiguous, and conflicting peripheral interfaces. Which way does the IDE connector go, pin 1 towards power connector? Which one is pin 1? This cable doesn't have a red stripe. I only get _two_ drives on this channel?
Ever connect a 50-pin Centronics, 50-pin dense, or 68-pin dense SCSI connector to its cable? It only goes one way, if it's made correctly.
Remember when microprocessors were _small_? Intel has doubled the volume of each of their processor housings since the 386, all the while requiring complicated cooling solutions to keep them within operating temperature ranges. This isn't progress, this is regress. They're faster, but they're so sloppy and loud. I'm convinced with the technology we have today (not tomorrow), we should be able to build a machine with performance on-par with that of a Pentium II with no fans. In fact, I'm typing on such a machine, it's a G3 laptop made by Apple, and it runs a multi-user, multi-tasking Unix clone. These machines are expensive because they're expensive to produce. Its components will continue to be produced in small quantities until people demand similar performance from the computers they use every day.
PC board BIOS sets get hacked up every time someone makes a larger IDE disk. I believe that the only reason IDE is popular is because it's cheap, and the only reason IDE is cheap is because profit was the only thing clone-makers were looking at ten years ago. Ever take a look at LILO's source code? Why should anyone have to go through such contortions to boot an operating system? There are decent two-stage bootloaders (I use GRUB, FreeBSD's loader works well, I'm sure there are others), but the PC architecture is just so braindamaged when compared with something like the Alpha, or OpenFirmware on a SPARC or a PowerPC. I don't imply the latter are examples of perfect implementations, but they beat the spotted trousers off the PC. Every time I boot my Alpha, I'm impressed by the sheer usefullness of its SRM firmware. I can boot directly a kernel bootstrap program, initialize the PAL code in the processor, and be off, with none of that 640K silliness. I can boot MILO (or even flash it into firmware!), and "ls" my devices, before choosing what to boot.
Crusoe could have changed this. All of this. Transmeta could have implemented a new, clean instruction set. They could have worked hard to provide top-quality documentation, reference material, and developer resources to a bunch of people looking for something new. They have my favorite "specially-abled alien" (or whatever INS wants to call him) to lead the port of an excellent operating system to this new architecture. They could have implemented a 64-bit memory addressing scheme. I have only read the PDF whitepaper, but it seems to me they could have easily used a 64-bit instruction set. Crusoe runs cool--very cool. I want my desktop to be fanless. That the hard drive makes noise implies that it moves, which it does by design. Only electrons need to move in a microprocessor. Transmeta's Code Morphing (TM) allows an excellent abstraction of the ISA from the hardware below it, and we need this.
But I think I'm too idealistic. Transmeta couldn't have pulled that off. Transmeta is a business, and they want to money by selling a product to people that want it. People think they want Intel because they don't know anything else, and getting funding to fight Intel is probably harder than running a new chip from design to fabrication.
I guess I'm disappointed because Transmeta failed to change the world.
--
Oh well, I presume there will be subnotebooks/whatever we end up calling them based on the 700 mhz crusoe that will actually run Mobile Linux, and that the dichotomy is merely a marketing split to show the two "different" market sectors Transmeta is attacking with their products (handhelds vs. notebooks-of-the-future). I just am not entirely comfortable with the idea that Linux is boxed in as the embedded platform / handheld platform, at least from a marketing perspective.
Is it just me or does anybody think that Transmeta is using Linux+Linus as a way to sell their chips?
/. already has its own ICON for news regarding it -- odd there aren't special icons for say intel or cyrix ?
/. and the rest of the rabid Linux users will think its caviar.
I mean think about it, for a start-up chip company,
Call this Trolling, or Flamebait or whatever, but it seems that if Linus put his 'Thumbs Up' in a box filled with shit,
Yadda yadda yadda, no I'm not pro-MS, and I have nothing against the crusoe cpu, just think at how easily you folks are manipulated.
It's been said a million times: ENOUGH WITH TRANSMETA! Their product is not the super meta-CPU we thought it would be. In fact, it's pretty boring.
BUT, because Linus is in the company, we hear everything about it. The CEO had a flat tire? Good enough! Linus spills his coffee? RobLimo, get right on it!
If Torvalds wasn't working there, would this be news? No, because no one would care. But the Church of Linus keeps everyone interested.
Why don't we just stick a live webcam on his head, so we can track his every move? Make 'LinusCam' a Slashbox, and get it over with.
~~~~~~~~~
auntfloyd
The Transmeta Crusoe processor is supposed to run at much lower power consumption levels than normal CPUs. As a result, the heat generated by one of these processors is significantly lower than its competition. To use the Phoenix BIOS defeats this significant advantage.
I've been to Phoenix, in July mind you, and I can safely say that Phoenix is way too hot to allow adequate operating temperatures for the Transmeta Crusoe processor.
Sorry guys, I think you need to choose your business partners with a lot more forethought.
---
MoooooooooOOOOOOOOOOOOOOOOOoooooooooooooo!!!!!!!
Linus has "bitched" about how stupid most bioses are in regards to Linux and how things get done. What do you think are the chances that we'll see a bios that is almost perfect??? Can you imagine.. the Kernel drivers re-written so that bios work-arounds are removed... the speed increases would be cool!
but then... why hasn't someone written an open source bios?
Do not look at laser with remaining good eye.
OK, so it doesnt understand scsi, but you can configure the RLL hdd, and floppy drive. It also has a debugger in ROM, so you can do debuggy things. Seeing as SCSI cards extend the bios, quite nicely in some cases, it must be possible to make an ISA/PCI card with an extended bios.
If you've got a flash bios you may know that they're modular. Trouble is, if you flash it incorrectly and it breaks, the backup bios generally can't handle PCI vid cards, so you'll have to get an old ISA video card to reflash.
-Yarn - Rio Karma: Excellent
The Mac is proof that consumers don't need to worry about Firmware. Experts can just type a keyboard sequence to enter the firmware prompt.
VIA Press release about Joshua
Socket 370 pinout on P6 bus interface
Integrated 64 Kb L1 cache
Integrated mutually exclusive 256 Kb L2 cache
133 MHz Front Side Bus (FSB) support
3DNow! (TM) Technology
Enhanced dual pipelined MMX (TM) and FPU
Utilized advanced 0.18 fabrication process
I have Pentium II LX mb I wanted to upgrade. I was going to replace the P2 233Mhz with a Celeron 466 or 500 ppga but think I will hold off to see what happens. It's my fourth computer anyway. I use it for R&D.
First off, if the CPU is going to slide up and down on it's voltage, instead of just stepping down once (like the Intel) then new chipset support is needed. If you can put power management on the chip, great, but you've still got the rest of the system sucking juice. If your BIOS can manage power this way for the whole computer, even better.
Then there's integrating wireless communication, support for low and no power storage, maybe firewire... All the things that would be really nice to have now.
Then again, and I'm surprised nobody's mentioned it so far...
SOMEBODY has to burn those mobile-Linux in ROM chips.
-- What you do today will cost you a day of your life.
To say that the BIOS is worthless is to imply fundamental problems in the code that makes up the Basic Input Output System. I think that you are probably talking about the BIOS setup program - the interface that most users see after pressing (pick one: F1, F2, DEL) during POST. It allows you to set the various options and parameters that the BIOS will use.
After working as an OEM hardware tech for numerous years, I'd have to say that what goes in the BIOS setup program is up to the OEM. I have seen prototyped motherboards with really nifty options in the BIOS setup program, only to see those options stripped out for the production boards. Consumers are too stupid to understand all those options, you see, and they might mess things up, so the "unnecessary" options are all stripped out.
I am wondering about the actual price of this thing. I would really like to see a cheap computer that could attract people in my boat and not just high end people with thousands of dollars. Will this just be another toy of the ecconomic elite?
Slashdot social engineering at it's finest
Uh, these are to be consumer devices? What makes you think that consumers are concerned that their BIOS knows what an ethernet card is?
-BrentI would have to agree with you here. The Alpha's was pretty damn cool, too. But I suspect that Transmeta is trying to be as backward compatible as possible. They could avoid that by either not supporting Windows (can't do that, need to make money) or by getting MS to upgrade (won't happen). So the only result could be an upgraded transmeta chip that runs a special firmware and linux combo. This means another different chip and linux to support. Ug.
All this leaves back under the hot Pheonix sun.
-- soldack
So far, the processing power race has only had one thing in mind: more mips from a single processor, nothing else matters, not size, not heat, well, maybe cost matters, but *not that much*. I've always thought that the way forward is to maximize processing through per transistor, and that's exactly where Transmeta is going. This directly affects me in two ways: first, my laptop, which is a bleeding wound as far as battery life is concerned. Second: my desktop 2 years from now. I want it to be 16-way or better, yet I don't want to be able to fry eggs on it, and I don't want it to have enough fans to achieve liftoff. Or a refrigerator. The only way to get there is with more energy-efficient processors. Fewer transistors == less heat, other things being equal. Did I say I hate fans? I hate fans.
One thing about Linus being involved in Transmeta is it suggests the tantalizing possiblity that the code morphing software may ultimately wind up being open-sourced. I couldn't think of anyone who could make a more powerful argument for it. Plus, I'd love to be able to program a machine like this directly in its "microcode" (a relative term as far as crusoe is concerned). Again, with LInus in there, I'd see it as a distinct possibility. Did you ever hear of anyone programming the PPro in microcode? It's possible, but nobody does it because Intel keeps that info locked up tighter than a... well, darn tight. Transmeta might not be so anal about it.
Life's a bitch but somebody's gotta do it.
Yes, Transmeta has been very secretive. There is nothing wrong with that. The extreme level of secrecy actually ended up being a brilliant PR move. The difference between "closed door CPU design" and "dirty closed source software" is that anyone with a compiler and source code can roll thier own software. If everyone on Slashdot got copies of the Crusoe design, then every one of us with access to a .18 micron fab facility could make our own processors and maybe contribute to the design process.
Sarcasm aside, CPU development and software development are completely different. The "open source" model of distributed collaboration just does not work with CPUs.
We also have to remember that people (like them or not) spent upwards of $100M over 4 years, with zero return on investments, to make this new CPU happen. They are entitled to make some money off of a very risky undertaking. They were envisioning 700 MHz mobile CPUs when Intel was selling top of the line 120 MHz chips for workstations.
You shouldn't be paranoid about "another Intel". You should hope they become another Intel. How nice would it be to have Intel, AMD, and Transmeta become the "Big Three" of CPU design and have them push each other to produce better, faster, and cheaper CPUs.
-B
"BIOS" is really a misnomer today. "Boot ROM" would be more appropriate. Not much of what's in ROM is used once any modern OS is running. The legacy underlayer of DOS that's still in the ROMs could be removed without much loss, like the old BASIC interpreter was. (You'd lose the ability to boot DOS, assuming anybody still cares.)
The Chip design *is* revolutionary, and no one *cares* about your opinion.
If you beg to differ, please show us a previous chip that was low powered, for long battery life in mobile markets, and did code morphing. Hehe, that's 2 for me and I didn't think more then 20 seconds.
I'm waiting...
-BrentIntel may be mad about being mooned but Phoenix is the king of the BIOS world and has been working with Intel (and Microsoft) for years. Through APM to ACPI, lots of BIOS changes had to be made. In fact, without a proper ACPI compliant BIOS, Win2k will either not work or have problems with some of the power managment features. New systems can not be certified by MS without a fully ACPI BIOS. Intel needs Win2k to be a success to help build momentum for the coming Itanium push into the server scene.
Besides chips and what not, Intel makes complete systems. They had one of those systems at the MS PlugFest for Win2k/Millenium (I tested my company's fibre channel board with them) and they had several problems with their hardware that they needed their Pheonix buddies to create a work around for. Intel needs Pheonix, just like the whole PC market does.
-- soldack
This may be in there but there doesn't seem to be too many details on what it really does (besides be your ISP and give you a pretty start up screen). Any idea what else it does?
It does seem to connect to the internet before you boot. Why don't they just throw in a web browser, e-mail, news group reader, irc, and AIM? Now THAT would be an Internet-BIOS!
-- soldack
I can't believe how many people are naysaying the future of Transmeta and its processors. Just because OEMs weren't present at the press conference on the 19th and there aren't an 700mhz Crusoe laptops on the shelves doesn't mean Transmeta is about to exit stage left. It's been what, about a week now? Yes, we're all impatient for them to really roll this stuff out, but give it a little time. It's very unrealistic to write off a company after only ten days of inactivity when it has friends like IBM, S3, and a good bios company like Phoenix.
See! From the Phonex press-release, now everyone knows that Transmeta only has ONE processor!!! The big secret is out!
it's in my head
On any Sun, SGI, Apple Mac, you can bring up a command line in firmware at boot time and truly control the boot process.
On my Sun, I can boot off of any partition on any disk. I can also *really* boot from CD-ROMs (not cheesy El Torito CD's that work by emulating a piece of crap floppy disk).
All I have to do is type "boot cdrom" to boot from CD, "boot net" to boot off of a network boot/install server. I can even create aliases so that I can say "boot linux" to boot off of a Linux partition - even if it's on my sixth SCSI disk in the chain.
I don't need to waste time/money installing stupid tools like System Commander to get the same functionality.
This is just the tip of the iceberg. Sun's OpenBoot/OpenFirmware provides many mini-miracles.
SGI (which uses their own firmware) actually built Pentium III computers using a real firmware. And by doing that they were able to defeat the stupid 1024 cylinder limit and other bogus limitations of Wintel PeeCee's.
Phoenix recently recieved a terse E-Mail from Intel, the complete body of which read "Phoenix, you bitch, have you been seeing another hardware company on the side?"
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
The symbol of the Phoenix is perfect for Transmeta, since what other creature could emerge in its full brilliant splendor from the fiery death that is an overheating pentium? I submit to you, the answer is none.
Conspiracy theory for the day: the Kottler Caldera Group of Phoenix was a finalist in the design contest for Phoenix, AZ's official symbol. What could this mean?!?!?!
"If one is really a superior person, the fact is likely to leak out without too much assistance" -- John Andrew Holmes
There's more to this story...
Phoenix has been working on putting some kind of internet service called Phoenixnet built into the BIOS. It seems like the kind of thing you'd want in a dedicated web appliance, but I don't know if the two are related.
http://www.phoenix.com/phoenixnet/phoenixnet.html
Silly. Everyone knows that Robinson Crusoe was written by Stephen King. I mean, "Duh"!
**>>BELCH
After working as a tech for numerous years, I'd have to say that the vast majority of Phoenix BIOSes are really worthless. (Esp those in Packard Bells... but that goes without saying) Fewer controls for PnP/IRQ settings, cache settings, DRAM settings, AGP settings, etc. that can come in handy in a wide variety of instances, especially IRQ conflicts in Linux systems. Oh well, the most popular BIOS in the world can't be bad, right?
Now all we need is a mini-motherboard complete with Phoenix Bios, Crusoe processor, and integrated video and sound. Perfect for a do-it-yourself wearable computer.
Hosting for Creators: http://rpg-works.net
It really suprises me that people still seem to think that coding around the "code-morphing" will actually improve the speed of a program on this chip. Keep in mind, this chip has no OO execution or branch-prediction, and relies on the software above to figure that out as best as possible.
If you even _could_ code (or better yet they _allowed_ your to code) in the native format, it still would suffer, because the things that the silcon doesn't have is exactly what makes stuff run fast.
At least last time I looked.
Provided I don't have to look at a stupid advertisement for transmeta instead of the bios post screen.
:)
And no, it's really a slashdot link.
mcrandello@my-deja.com
rschaar{at}pegasus.cc.ucf.edu if it's important.
If the Crusoe appears X86 compatible to the outside world anyway?
;)
It was said at the Transmeta presentation that the code morphing software is loaded up before the BIOS.
I presume this would mean that the BIOS sees a normal x86 when it loads itself up.
I suspect the BIOS modifications mainly consist in a giant Pinguin replacing the normal POST screen
I strongly believe that trying to be clever is detrimental to your health. -- Linus Torvalds