ATI和NV构架效率问题 - POPPUR爱换

原帖由 lzy651 于 2008-6-22 00:16 发表
要说效率，都是二级管，效率都一样，
只是别人对你支持的好和坏了

原帖由 beer966 于 2008-6-22 00:11 发表
另:4楼的图是因为A卡没用到你的物理计算软件:funk:

原帖由 Edison 于 2008-6-22 00:26 发表

AMD是最早提供FAH gpu加速的厂商，你的说法不成立。

原帖由 beer966 于 2008-6-22 00:28 发表
最早提供并不代表研究得更深.......这方面还是承认NV比较投入精力.....:loveliness:

原帖由 cataclyson 于 2008-6-22 00:09 发表
还是绝大多数游戏用不到4D,造成了ATI的SP浪费闲置，当初ATI要能意料到这点估计不至于像今天这么被动。ATI浪费的晶体管太多了:(

原帖由 Ptaltaica 于 2008-6-22 00:17 发表
360开发比PS3容易是因为Cell，和GPU有什么关系？.........

原帖由 china17 于 2008-6-22 00:40 发表
问题是,AMD的努力还是没有使其GPU达到本应能发挥的计算能力

原帖由 feel囝 于 2008-6-22 01:21 AM 发表
还是编译问题吖，800sp是肯定的，但实际怎样利用，可以利用多少，也难说了

原帖由 zuiniubi 于 2008-6-22 01:50 发表
显而易见的问题
这代ati的芯片设计本身就有问题
低能架构
扶不起来的阿斗

Hi All,

I read a debate going on in the forums as to why ATI core is givingless PPD compared to NVIDIA core. And I see many reasons givenincluding the wu's assigned are different (Definitely that also willimpact) as well as MAYBE NVidia core is NOT DOING Science work justCredit work (ONLY possible if someone has goofed up major whileProgramming - LESS LIKELY because I am sure lot of testing has occuredwithin fah stanford group before releasing the GPU beta to the world.)

For a long time (From the time AMD and NVIDIA came out with Unifiedshaders) I have had a thought in my mind, which seems to be playing outnow if the PPD results are valid (i.e no programming error some where).So let me start =>

EVERYONE has to REALIZE the difference in definition of SPs betweenATI/AMD and NVIDIA. To be frank I would say the way ATI named theirsingle SP has SP is a kind of JOKE (in some ways) compared to NVIDIA.

Based on what I have read about the NVIDIA and ATI GPU Architectures onthe Net (Haven't had time nor a ATI card at hand to write code toverify this yet)

IN NVIDIA __EACH SP__ is capable of doing either FP32 or Integer (ALL ops) or Special_functions.
WHILE
IN AMD __FOR EACH Group of 5 SPs ONLY ONE SP__ can do FP32 orInteger_MUL or Special function, while the other 4 SP can only doSIMPLE INTEGER operations.

So if the code is doing lot of FP operations or special functions orInteger_MUL IN EFFECT one gets only 1/5 th the number of SPs in ATI ascompared to what ATI claims as SPs i.e A ATI chip with 320 SPs is ineffect ONLY 64 SPs. So obviously a ATI performance will be only 50%compared to NVIDIA in the WORST CASE.

However in practice one would find that ATI GPUs will give bit moreperformance than the WORST CASE Mentioned above because IF one can mixthe FP32/IntegerMUL/Special_Functions with Simple_Integer ops then theother 4 SPs out of the 5 SP group in ATI can be utilized. Thusimproving over the WORST CASE which I mentioned.

SO BEFORE JUMPING AND CONCLUDING may be WE have to think of the realityinterms of what is ATI 320 SIMPLE SPs worth compared to NVIDIA 128 FULLSPs.

Note: Also one more thing which can aid ATI a bit could be the betterbranch/thread (i.e independent code) granularity in ATI compared toNVidia. But still the 128 Full SPs in NVidia vs 64 Full SPs in AMD/ATImay not help much for ATI wrt this granularity if the Vector sizes onwhich the operations are occuring are large.

Just my thoughts. Happy to get constructive feedback, even to theextent of telling that my thoughts are rubbish, provided it is backedtechnically. I am starting this thread mainly to understand the G80/G92core VS R600 core architecture/programming advantages/disadvantages.

Keep
HanishKVChanishkvc