KVR Audio

maxxxter · Post by **maxxxter** » Mon Feb 07, 2005 1:55 pm

DELETED

DevonB · Post by **DevonB** » Mon Feb 07, 2005 2:19 pm

I doubt intel and amd are finished. Too much invested in code for the x86 platform. I'm sure this chip isn't backwards compatible with x86 code.

Devon

nuffink · Post by **nuffink** » Mon Feb 07, 2005 2:24 pm

There are a couple of articles at The Register...

http://www.theregister.com/2005/02/01/c ... _part_one/
http://www.theregister.co.uk/2005/02/03 ... _part_two/

kahrayzee · Post by **kahrayzee** » Mon Feb 07, 2005 2:24 pm

Plus it isn't like Intel/AMD aren't going to continue evolving their processor line. They've figured out how to put dual cores onto a chip for a reasonable price... certainly they have other technology they can bolt onto their processors as required.

Stupid American Pig · Post by **Stupid American Pig** » Mon Feb 07, 2005 2:26 pm

I hear AMD has plans for 4 cores on thier roadmap as well...

WhiteNoise · Post by **WhiteNoise** » Mon Feb 07, 2005 2:27 pm

It definately looks cool, but until it actually comes out it's hard to say how it will perform. Multiple cores is definately the way to go - it will be great for audio stuff. Each instrument could run in it's own thread very easily.

justin_leapfrog · Post by **justin_leapfrog** » Mon Feb 07, 2005 2:31 pm

Some very interesting reading on the Cell here:
http://www.blachford.info/computer/Cells/Cell0.html

I've read that IBM are expected to make rackmount 64-Cell workstations - Imagine these as a DSP workstation, that's a theoretical 16-terraflops!

Optimised SIMD code on a fast x86 cpu could only do about 13-billion!

VitaminD · Post by **VitaminD** » Mon Feb 07, 2005 2:43 pm

http://www.blachford.info/computer/Cells/Cell4.html wrote: Since there is no need for a fast x86 processor why not just emulate one? Removing the x86 and support chips from a PC will give big cost savings. An x86 computer without an x86 sounds a bit weird but that's never stopped Transmeta who do exactly that, perhaps Transmeta could even provide the x86 emulation technology, they're already thinking of getting out of chip manufacturing [Transmeta].

good luck.

UltraJv · Post by **UltraJv** » Mon Feb 07, 2005 3:13 pm

Transmetas main goal was to save power consumption. The speed of their processors is abysmal. They thought that they could match the speed of current processors but they were way off the mark. Their idea is clever but the code morphing is very slow. If we applied it to something like the Cell, I think we would see speed reduce by 75% Dont get fooled by the numbers game folks, Atari had a 64 bit console (the Jaguar) in the early 90s. The Cell will fall over on its inability to get data in and out as fast as it would like. This is a problem with more than one core, they have to share resources and it leads to bottlenecks. From what Ive heard, the prediction is for 4+ GHZ Cell chips, the first wont be anything like that speed and will probably equal a low to mid range pc at best.

munchkin · Post by **munchkin** » Mon Feb 07, 2005 3:38 pm

I really hope this works out. I have been reading how Intel are struggling to increase the power of their processors. AMD haven't significantly increased processor speed with their 64bit cpu's. There's the problem with higher clock speeds and the heat generated. It feels like we are at a crossroads and people above us are making the decision about which road to take.

Imagine if Wintel lost their strangle hold on the market. We could've been years ahead in terms of computational power and development if they hadn't squeezed better systems out of the market. Their monopoly has meant inflated prices for old fashoined technology. I won't be sorry to see them forced to change.

This may mean Sony etc. will control home computer development and given their reputation for locking people into proprietry hard/software they will probably use the same protectionist strategies as Wintel. But at least we will have jumped further up the development ladder.

UltraJv · Post by **UltraJv** » Mon Feb 07, 2005 3:43 pm

Sun have had 64 bit processors for over 10 years running Unix, if you want to run a Windows app - you need Intel/AMD. No one is forcing you to run Windows, use Linux on 64 bit if want - you can do that today and use Rosegarden or any other Linux audio app. You can also use MACOSX, we dont need another OS as there are plenty to choose from already. The Cell wont run Windows, its not competing with that marketplace. In all honesty, the Windows systems we can have now (Up to 4 processors) are so powerful, no one needs them for audio. What drives software companies to write purely for Windows is its huge user base. If you can convince them to write for the Cell, youve won

Stupid American Pig · Post by **Stupid American Pig** » Mon Feb 07, 2005 3:53 pm

There is no reason that Windows couldnt/wouldnt be ported to the new platform if it turns out a winner. Even NT4 was ported to the Alpha platform 10 years ago(although it was only 32 bit).

Windows doesnt need x86 to run, but it would have to be ported to the new processor. the other porblem would be applications that were written with processor specific instructions like SSE or 3dnow. That is where we would see problems unless code was in place to emulate it on the new chips. As long as the app was written to the Windows SDK it shouldnt care what processor runs it.

munchkin · Post by **munchkin** » Mon Feb 07, 2005 3:58 pm

UltraJv wrote:Transmetas main goal was to save power consumption. The speed of their processors is abysmal. They thought that they could match the speed of current processors but they were way off the mark. Their idea is clever but the code morphing is very slow. If we applied it to something like the Cell, I think we would see speed reduce by 75% Dont get fooled by the numbers game folks, Atari had a 64 bit console (the Jaguar) in the early 90s. The Cell will fall over on its inability to get data in and out as fast as it would like. This is a problem with more than one core, they have to share resources and it leads to bottlenecks. From what Ive heard, the prediction is for 4+ GHZ Cell chips, the first wont be anything like that speed and will probably equal a low to mid range pc at best.

According to the Cell article STI have found a way around the bottle necks. I think they will have learnt from past mistakes and anticipated them in their design.

One of the points the article raised is the old fashioned design of Intel processors. Better systems were squeezed out of the market - not because they weren't technologically superior (which they were) but because of Wintels stranglehold on software/cpu compatibility.

The OS has dictated the type of processor (and visa versa) but what the article is saying is that Cell processors are not dependent on any particular OS architecture. That's its strength and Intels weakness.

You mention the Atari. It failed because it was squeezed out of the market by Mac and then Wintel. At the time it was technologically superior but it wasn't able to establish itself in the market. The numbers game is relevant to those of us who wish for greater computing power and technological development. But it's unimportant to Wintel who force us to buy their substandard and technologically inferior products. I hope this monopoly is broken ASAP.

UltraJv · Post by **UltraJv** » Mon Feb 07, 2005 4:02 pm

The numbers game isnt relevant, its marketing hype - AMD 64 bit chips will at best be 30% faster. Thats if you use XP 64 bit and all the 64 bit drivers etc. Overclocking a P4 gives about the same improvement. AMD are simply using numbers to impress but it dosnt deliver. As for Monopolies, the current one will be replaced with another and so on for ever.

Stupid American Pig · Post by **Stupid American Pig** » Mon Feb 07, 2005 4:09 pm

It is very true that Intel has never had the most advanced chip on the market. Macs were way better until the mid 90s when Intel figured out that clockspeed sells. AMDs generic 486s were faster than Intels, prompting the name pentium in an attempt to squeeze AMD out of the market. Amd hit a ghz 1st, and 1st gen Athlons were faster than PIIIs. Intel didnt really have a faster chip until about 2 years ago when the PIV clock speed was ramped up high enough to outrun AMD.

AMD has suffered revenue problems since Intel scared vendors like Dell out of offering AMD platforms. I cant believe that Dell would not want to support a company that was ALSO based in Austin Texas.

I think that if AMD had the same revenue stream as Intel, processors would be much faster today. Either way, I would like to see a new platform as long as it would support existing hardware and software to at least some degree. Competition only works to our advantage....

Intel & AMD era finished? CELL is here!