Intel gonna be the new gamers choice?

[Acanthus]

Originally posted by: fliguy84
Rahul Sood (Voodoo PC owner) has something interesting to say about this matter

http://voodoopc.blogspot.com/2006/03/if...ad-time-machine.html?www.dailytech.com

They re-ran the tests, the entire blog was debunked, short of the fact that its a very early preview.

The hilarious part was the conroe was running on the wrong memory timings, and gained more from the retest than the FX did from the new bios (the bios update did basically nothing).

 

[ronnn]

Impressive and likely not cheap. Now if Ati showcased a card not to be for sale for at least three months ... the response would be quite different. :beer:

 

[ZimZum]

Originally posted by: Acanthus

Originally posted by: fliguy84
Rahul Sood (Voodoo PC owner) has something interesting to say about this matter

http://voodoopc.blogspot.com/2006/03/if...ad-time-machine.html?www.dailytech.com

They re-ran the tests, the entire blog was debunked, short of the fact that its a very early preview.

The hilarious part was the conroe was running on the wrong memory timings, and gained more from the retest than the FX did from the new bios (the bios update did basically nothing).

Doh, so much for that. People with similar AMD setups on on other forums have run the same benchies Intel did and got similar results. The fact is, in the comparison they are actually painting AMD in the best light possible. Comparing an overclocked FX60 to a Conroe running at stock.

The most shocking part of this is the Conroe Pricing. Which, if it holds true will mean Intel will have not only the highest performing CPU but it will be cheaper than AMDs offerings. Which in turn will assuredly force AMD to drive down prices. Thats the great thing about competition.

edited for spelling

 

[Darkon]

Originally posted by: Acanthus

Originally posted by: DAPUNISHER

Originally posted by: mwmorph

Originally posted by: DAPUNISHER
I believe IBM said there will be Cell blade series servers, but that's about it?

cell is worthless for general computing

Your response is somewhat muddled to me. Are you confirming that specific server type tasking is where Cell will stop?

Cell has a lot of limiting factors sony doesnt talk about. The DSPs suck at everything but fixed functions. It only does in order execution, and all the cores have off die cache...

What on earth are you talking about spe's aren?t DSP's you can do pretty much anything with them except run a sophisticated modern OS like Tiger OS x and no cell doesn't have off die cache

 

[Acanthus]

Originally posted by: Darkon

Originally posted by: Acanthus

Originally posted by: DAPUNISHER

Originally posted by: mwmorph

Originally posted by: DAPUNISHER
I believe IBM said there will be Cell blade series servers, but that's about it?

cell is worthless for general computing

Your response is somewhat muddled to me. Are you confirming that specific server type tasking is where Cell will stop?

Cell has a lot of limiting factors sony doesnt talk about. The DSPs suck at everything but fixed functions. It only does in order execution, and all the cores have off die cache...

What on earth are you talking about spe's aren?t DSP's you can do pretty much anything with them except run a sophisticated modern OS like Tiger OS x and no cell doesn't have off die cache

uhh yeah, it does have off die cache.

in order execution is a massive performance hit.

as for the SPEs, please detail what they can do besides decode indvidual streams of data, being in order processors they cant even use more than one SPE for the same task.

 

[Acanthus]

Originally posted by: ZimZum

Originally posted by: Acanthus

Originally posted by: fliguy84
Rahul Sood (Voodoo PC owner) has something interesting to say about this matter

http://voodoopc.blogspot.com/2006/03/if...ad-time-machine.html?www.dailytech.com

They re-ran the tests, the entire blog was debunked, short of the fact that its a very early preview.

The hilarious part was the conroe was running on the wrong memory timings, and gained more from the retest than the FX did from the new bios (the bios update did basically nothing).

Doh, so much for that. People with similar AMD setups on on other forums have run the same benchies Intel did and got familiar results. The fact is, in the comparison they are actually painting AMD in the best light possible. Comparing an overclocked FX60 to a Conroe running at stock.

The most shocking part of this is the Conroe Pricing. Which, if it holds true will mean Intel will have not only the highest performing CPU but it will be cheaper than AMDs offerings. Which in turn will assuredly force AMD to drive down prices. Thats the great thing about competition.

I am very excited about the pricing myself, especiallly if they overclock well.

 

[otispunkmeyer]

Originally posted by: Acanthus

Originally posted by: Darkon

Originally posted by: Acanthus

Originally posted by: DAPUNISHER

Originally posted by: mwmorph

Originally posted by: DAPUNISHER
I believe IBM said there will be Cell blade series servers, but that's about it?

cell is worthless for general computing

Your response is somewhat muddled to me. Are you confirming that specific server type tasking is where Cell will stop?

Cell has a lot of limiting factors sony doesnt talk about. The DSPs suck at everything but fixed functions. It only does in order execution, and all the cores have off die cache...

What on earth are you talking about spe's aren?t DSP's you can do pretty much anything with them except run a sophisticated modern OS like Tiger OS x and no cell doesn't have off die cache

uhh yeah, it does have off die cache.

in order execution is a massive performance hit.

as for the SPEs, please detail what they can do besides decode indvidual streams of data, being in order processors they cant even use more than one SPE for the same task.

nah it doesnt, the main cpu has access to the cache as a normal cpu would, on die, but the SPE's for some reason arent wired directly to the cache and have to go a longer way round...

 

[DAPUNISHER]

Regardless of the Cell architechture debate going on, IBM thinks it will make a fine server for certain applications

 

[DAPUNISHER]

BTW, Conroe revisited Anand applied the new bios, fixed some stuff he borked in testing and reran some benchies, Conroe still the clear winnar, just 20% in F.E.A.R. instead of 41%. Those questioning the benchies were right, but the bios was a non-factor so STFU about that

 

[Darkon]

Originally posted by: otispunkmeyer

Originally posted by: Acanthus

Originally posted by: Darkon

Originally posted by: Acanthus

Originally posted by: DAPUNISHER

Originally posted by: mwmorph

Originally posted by: DAPUNISHER
I believe IBM said there will be Cell blade series servers, but that's about it?

cell is worthless for general computing

Your response is somewhat muddled to me. Are you confirming that specific server type tasking is where Cell will stop?

Cell has a lot of limiting factors sony doesnt talk about. The DSPs suck at everything but fixed functions. It only does in order execution, and all the cores have off die cache...

What on earth are you talking about spe's aren?t DSP's you can do pretty much anything with them except run a sophisticated modern OS like Tiger OS x and no cell doesn't have off die cache

uhh yeah, it does have off die cache.

in order execution is a massive performance hit.

as for the SPEs, please detail what they can do besides decode indvidual streams of data, being in order processors they cant even use more than one SPE for the same task.

nah it doesnt, the main cpu has access to the cache as a normal cpu would, on die, but the SPE's for some reason arent wired directly to the cache and have to go a longer way round...

They don't have to go longer way spe's have there own 256 kb of sram available which act's as a cache

 

[Acanthus]

Originally posted by: otispunkmeyer

Originally posted by: Acanthus

Originally posted by: Darkon

Originally posted by: Acanthus

Originally posted by: DAPUNISHER

Originally posted by: mwmorph

Originally posted by: DAPUNISHER
I believe IBM said there will be Cell blade series servers, but that's about it?

cell is worthless for general computing

Your response is somewhat muddled to me. Are you confirming that specific server type tasking is where Cell will stop?

Cell has a lot of limiting factors sony doesnt talk about. The DSPs suck at everything but fixed functions. It only does in order execution, and all the cores have off die cache...

What on earth are you talking about spe's aren?t DSP's you can do pretty much anything with them except run a sophisticated modern OS like Tiger OS x and no cell doesn't have off die cache

uhh yeah, it does have off die cache.

in order execution is a massive performance hit.

as for the SPEs, please detail what they can do besides decode indvidual streams of data, being in order processors they cant even use more than one SPE for the same task.

nah it doesnt, the main cpu has access to the cache as a normal cpu would, on die, but the SPE's for some reason arent wired directly to the cache and have to go a longer way round...

Cell article (old)

Cell is basically one SMT enabled power PC core, and the 8 SPEs are for floating point calculations, not integer.

Although i do stand corrected on the cache, 256kb available per core on die. Another old article i read indicated the cache was on package, not on die.

although im laughing pretty hard at this massive core doing 4.5ghz at 30w.

 

[Acanthus]

Originally posted by: Darkon

Originally posted by: otispunkmeyer

Originally posted by: Acanthus

Originally posted by: Darkon

Originally posted by: Acanthus

Originally posted by: DAPUNISHER

Originally posted by: mwmorph

Originally posted by: DAPUNISHER
I believe IBM said there will be Cell blade series servers, but that's about it?

cell is worthless for general computing

Your response is somewhat muddled to me. Are you confirming that specific server type tasking is where Cell will stop?

Cell has a lot of limiting factors sony doesnt talk about. The DSPs suck at everything but fixed functions. It only does in order execution, and all the cores have off die cache...

What on earth are you talking about spe's aren?t DSP's you can do pretty much anything with them except run a sophisticated modern OS like Tiger OS x and no cell doesn't have off die cache

uhh yeah, it does have off die cache.

in order execution is a massive performance hit.

as for the SPEs, please detail what they can do besides decode indvidual streams of data, being in order processors they cant even use more than one SPE for the same task.

nah it doesnt, the main cpu has access to the cache as a normal cpu would, on die, but the SPE's for some reason arent wired directly to the cache and have to go a longer way round...

They don't have to go longer way spe's have there own 256 kb of sram available which act's as a cache

It my understanding that this only acts an L1 cache and the L2 cache is shared among the 8 SPEs and the power PC core.

 

[Keysplayr]

FEAR benches look much more in line to the rest of the gaming benches. And still lands 20% faster with a mainstream Conroe. BIOS update had nil effect.
And a hush falls over the crowd. I'm sure they are hard at work finding other things to pick apart. I'm very glad Anandtech revisited this. They addressed all the complaints of the AMD boyz. The only thing that needs to be done now, is to do the very same testing on the RD580 mobo (with latest BIOS LOL). Somehow, I don't think things will improve anywhere near enough to even get close to the Mainstream Conroe.

Very exciting indeed.

 

[apoppin]

Originally posted by: keysplayr2003
FEAR benches look much more in line to the rest of the gaming benches. And still lands 20% faster with a mainstream Conroe. BIOS update had nil effect.
And a hush falls over the crowd. I'm sure they are hard at work finding other things to pick apart. I'm very glad Anandtech revisited this. They addressed all the complaints of the AMD boyz. The only thing that needs to be done now, is to do the very same testing on the RD580 mobo (with latest BIOS LOL). Somehow, I don't think things will improve anywhere near enough to even get close to the Mainstream Conroe.

Very exciting indeed.

totally agreed

and finally we will see reasonably priced AMD CPUs [for a change]

 

[Acanthus]

Originally posted by: apoppin

Originally posted by: keysplayr2003
FEAR benches look much more in line to the rest of the gaming benches. And still lands 20% faster with a mainstream Conroe. BIOS update had nil effect.
And a hush falls over the crowd. I'm sure they are hard at work finding other things to pick apart. I'm very glad Anandtech revisited this. They addressed all the complaints of the AMD boyz. The only thing that needs to be done now, is to do the very same testing on the RD580 mobo (with latest BIOS LOL). Somehow, I don't think things will improve anywhere near enough to even get close to the Mainstream Conroe.

Very exciting indeed.

totally agreed

and finally we will see reasonably priced AMD CPUs [for a change]

I want to see the days of $200 high end chips again