Official PS3 Thread

[Panajev2001a]

Back to my challenge then

Intel putting good R&D resources behind high-end GPUs and nVIDIA doing the same for high-end CPUs ( x86 market )... I give each company two products ( so they can learn from the mistakes of the previous one )...

I do think Intel would design a better GPU than the CPU nVIDIA would design...

 

[jvd]

Back to my challenge then

Intel putting good R&D resources behind high-end GPUs and nVIDIA doing the same for high-end CPUs ( x86 market )... I give each company two products ( so they can learn from the mistakes of the previous one )...

I do think Intel would design a better GPU than the CPU nVIDIA would design...

I'm not to sure about this . Nvidia is very good at what they do so is ati. Intel already had a 3d card at one point and they make really crappy intergrated video even lagging behind the first nforce board by a huge amount. The only way intell would win is if they used thier deep pockets to buy power vr or something

 

[V3]

Pana,

That challenge is more speculation than "if cell is in PS3"

 

[Panajev2001a]

Back to my challenge then

Intel putting good R&D resources behind high-end GPUs and nVIDIA doing the same for high-end CPUs ( x86 market )... I give each company two products ( so they can learn from the mistakes of the previous one )...

I do think Intel would design a better GPU than the CPU nVIDIA would design...

I'm not to sure about this . Nvidia is very good at what they do so is ati. Intel already had a 3d card at one point and they make really crappy intergrated video even lagging behind the first nforce board by a huge amount. The only way intell would win is if they used thier deep pockets to buy power vr or something

jvd, considering that their solutions have been cheap integrated cards ( and no i740 was not the only one... they have a new one released alongside the Pentium 4 more or less... it is more recent than the i740 and not horrible )and they never put similar R&D resources behind their development and targeted the high-end segment I do not think you can really blame them for what they achieved... they aimed pretty low too...

Also, I said no external help

 

[PC-Engine]

I think the problem with that challenge is that you're expecting Nvidia to build an x86 ISA cpu

How about if we had Nvidia licence a MIPS core and bolting on a bunch of VUs kinda like what Toshiba did? Intel likes high clock rates so I don't see why Nvidia has to follow the same route. I'd bet that Nvidia could make a cpu that does 10 GFLOPS single core running at only 500MHz blowing a P4 out of the water and running a 6 times slower clock.

 

[Panajev2001a]

I think the problem with that challenge is that you're expecting Nvidia to build an x86 ISA cpu

How about if we had Nvidia licence a MIPS core and bolting on a bunch of VUs kinda like what Toshiba did? Intel likes high clock rates so I don't see why Nvidia has to follow the same route. I'd bet that Nvidia could make a cpu that does 10 GFLOPS single core running at only 500MHz blowing a P4 out of the water and running a 6 times slower clock.

I bet nVIDIA would not have a FSB able to feed that puppy nor such an advanced caching system as CPUs like Pentium 4 and Itanium 2 have...

Seiously, x86 is part of the challenge too as it adds a layer of complexity Intel has been carrying around and doing quite well at it too...

I bet that the nVIDIA chip running commercial FP intensive workloads would not perform as well as the Pentium 4 even with higher theretical FP performance as there are lots of situation where the code does not offer much parallelism...

I want to see that ultra wide ideal chip battling in the x86 world in which 1.5-2 ILP is quite a rejoice... their lack of Hz would show up...

nVIDIA would not know where to start looking for in order to maximize parallelism and have their chip run decently as a high-end general purpose CPU...

AMD has not outclassed Intel recently ( let's say after Northwood got its stuff together ) and I do not expect a newcomer like nVIDIA to suddenly master th secrets of CPU designing and deliver a master-piece...

A secret of high speed for Intel CPUs is also because they use a lot of custom designed logic and they do not rely like most GPU desiogners do on automated tools and VHDL models to design their chips...

 

[PC-Engine]

Actually the primary reason why Intel is beating AMD is because of it's monopoly of the market and the amount of money it spends on research etc. Pound for pound AMD is a MUCH more talented group

Give Nvidia some nice funding and I think they'll surprise a lot of people. Oh did I mention that Nvidia sound processor blows all others out of the water on their first try? DICE!

Anyway I doubt Intel can design a GPU that can match Nvidia's even with all their resources.

 

[jvd]

Beats me . No lets talk about something fun. Like um .... women ?

 

[Panajev2001a]

jvd,

you made me think some more about a fiancee of mine ( well my only fiancee... no honey, I was only joking, do not kill meeeeeeeeee !!!! ) who is sleeping and that now is like 2:46 A.M. and I should hit the bed too

Have a good night guys,

Panajev

 

[jvd]

jvd,

you made me think some more about a fiancee of mine ( well my only fiancee... no honey, I was only joking, do not kill meeeeeeeeee !!!! ) who is sleeping and that now is like 2:46 A.M. and I should hit the bed too

Have a good night guys,

Panajev

Well right now i hate all women. Women are the worse things in the world. If i was into guys i'd be gay. Make life so much easier but alas Women are the sex that does it for me. To bad they are mean and heartless. Well I'm going to the party next door to drink. You have a good night ... not like u are really going to go to bed.

 

[Panajev2001a]

I am trying to go to bed sigh sob

 

[Saem]

Actually the primary reason why Intel is beating AMD is because of it's monopoly of the market and the amount of money it spends on research etc. Pound for pound AMD is a MUCH more talented group

I think the opposite. Oh look at that, they both mean nothing. Besides much of the inovation has been coming from Intel while AMD has followed suit, recently with the Hammer and HT has AMD really done anything Intel isn't already doing. BTW, I expect there to be some Hammer 2 type variant rather soon which will have a long pipeline reminicant of the P4. Note, it will NOT be for marketing reasons.

In any case Intel has the cash to have the best and the brightest, AMD does not.

 

[Deepak]

Innovation is almost always directly proportional to R&D Budget!!

 

[PC-Engine]

In any case Intel has the cash to have the best and the brightest, AMD does not.

...and that's why pound for pound AMD is more talented

The most talented groups are the ones that can do a lot with a little...

As a matter of fact I'd really like to see what AMD can come up with if they had the cash...

 

[zurich]

In any case Intel has the cash to have the best and the brightest, AMD does not.

...and that's why pound for pound AMD is more talented

The most talented groups are the ones that can do a lot with a little...

I don't know about that, Intel's Isreal time blew me away with the Pentium-M (especially since i'm a laptop user). Im starting to think AMD is a bit of a one-hit-wonder with their Athlon architecture (A64 seems to have inheireted the marjority of it too - check out the die comparison @ TomsHardware).

 

[V3]

Innovation is almost always directly proportional to R&D Budget!!

That's not always true, Innovation can also comes from desperation .

 

[PC-Engine]

In any case Intel has the cash to have the best and the brightest, AMD does not.

...and that's why pound for pound AMD is more talented

The most talented groups are the ones that can do a lot with a little...

I don't know about that, Intel's Isreal time blew me away with the Pentium-M (especially since i'm a laptop user). Im starting to think AMD is a bit of a one-hit-wonder with their Athlon architecture (A64 seems to have inheireted the marjority of it too - check out the die comparison @ TomsHardware).

Seems you don't really get the point so here it is again. Yes the P4 is better than the Athlon, but with 10 times the resources behind it's design, I would hope so...

 

[jvd]

In any case Intel has the cash to have the best and the brightest, AMD does not.

...and that's why pound for pound AMD is more talented

The most talented groups are the ones that can do a lot with a little...

I don't know about that, Intel's Isreal time blew me away with the Pentium-M (especially since i'm a laptop user). Im starting to think AMD is a bit of a one-hit-wonder with their Athlon architecture (A64 seems to have inheireted the marjority of it too - check out the die comparison @ TomsHardware).

Seems you don't really get the point so here it is again. Yes the P4 is better than the Athlon, but with 10 times the resources behind it's design, I would hope so...

actually clock for clock the athlon is much much faster than the p4. The p4 just scales longer.

 

[PC-Engine]

actually clock for clock the athlon is much much faster than the p4. The p4 just scales longer.

True, which says a lot about AMD's design potential and supports my point well. Anyway I was just comparing the best overall of the best which goes to Intel which is expected with the amount of resources that's devoted.

 

[Saem]

actually clock for clock the athlon is much much faster than the p4. The p4 just scales longer.

So? It doesn't matter how you get the performace, just that you do. This entire obsession with IPC is weird. It's nice it can do more clock for clock, but that some how does't seem relavent when at the end of the the day, the top Intel P4 offering nearly categorically beats the top AMD Athlon offering.

True, which says a lot about AMD's design potential and supports my point well. Anyway I was just comparing the best overall of the best which goes to Intel which is expected with the amount of resources that's devoted.

I don't see AMD competing against the aging well P Pro family as that great. And now AMD isn't really keeping up with the new P4 family as it's starting to show it's legs. Unless of course we see clock rates increase more aggressively with Hammer.

Now all of this stems from "GPUs are more complex than CPUs", which I think has been demonstrated as false -- Pana's post basically says it all.

IBM has some of the best process technology, of course it isn't cheap, but that's not all they possess. One thing TSMC doesn't have is possibly engineering teams that are as talented as the ones IBM has, which it could use to help Nvidia. In the end I think Nvidia is going there to learn from IBM rather than anything else. I wouldn't be suprised if something like this didn't happen with ATI and Intel when they settled their little dispute.