Overclock.net › Forums › Industry News › Hardware News › [NVIDIA]Microsoft going All-in on GPU computing
New Posts  All Forums:Forum Nav:

[NVIDIA]Microsoft going All-in on GPU computing - Page 7

post #61 of 151
Quote:
Originally Posted by Skylit View Post
Sad thing is.. AMD 58xx/69xx cards have much more theoretical raw computing power than anything nvidia has recently put out.

Like duckie said, the support isn't there :\\
This is so sad for being true, I sure hope AMD will get their chance that they deserve.
post #62 of 151
Quote:
Originally Posted by formula m View Post
mr Duckie comments are about 2009.
AMD is vastly ahead of nvidia of developmental work & working with "other companies".
Can you name any companies they are working with and what products they have specifically in that market? Do you know more than John Carmack?


Quote:
Originally Posted by formula m View Post
Physx isn't successful... it is a marketing gimmick. As cpu physics is far superior implamentation
Havok and Bullet physics do great work but in reality they are nowhere as powerful as PhysX. Not because they're not made by Nvidi but because they're x86 implementations. PhysX works on the CPU as well and guess why it performs so poorly when used in that manner. To fully use PhysX you need a $90 9800gt which does make spending time developing for it seem wasteful compared to the alternatives. Havok and Bullet may be more useful than PhysX but they're not superior. Unless they can somehow be offloading to the GPU through OpenCL or the like, they're never going to be as visually stunning.

Quote:
Originally Posted by formula m View Post
CUDA is wholly irrelevant.
Open standards & the proliferation of muti-core cpus simple overshadow the need for specific hardware.
There's a reason GPU farms are used in Supercomputers. You've also apparently completely missed the reason why AMD and Intel are implementing GPUs into their chips. How much talk lately has been focusing on offloading processes to the iGPU? Lurk more and troll less.

Quote:
Originally Posted by formula m View Post
logic, reasoning & facts will always thwart their pseudo attemps @ marketing on forums.
I use both AMD and Nvidia. You need to stop being personally offended over a piece of electronics. 6xxx series has certainly improved performance but it's still nowhere near as powerful as CUDA. Who cares though? No ones insulting your family. It's just the way it is.
Edited by PoopaScoopa - 6/18/11 at 4:40am
post #63 of 151
Quote:
Originally Posted by PoopaScoopa View Post
Havok and Bullet physics do great work but in reality they are nowhere as powerful as PhysX. Not because they're not made by Nvidi but because they're x86 implementations. PhysX works on the CPU as well and guess why it performs so poorly when used in that manner. To fully use PhysX you need a $90 9800gt which does make spending time developing for it seem wasteful compared to the alternatives. Havok and Bullet may be more useful than PhysX but they're not superior. Unless they can somehow be offloading to the GPU through OpenCL or the like, they're never going to be as visually stunning.
The reason PhysX runs horribly on cpu's is because Nvidia coded it to run horribly on cpu's. Re. Havok and Bullet - they do a lot more than make a banner wave or a little smoke...
Stumpy
(13 items)
 
  
CPUMotherboardGraphicsRAM
A10 5800k gigabyte  7970 mushkin ddr3  
Hard DriveOSPower
intel ssd Win7 64 Corsair TX850W 
  hide details  
Reply
Stumpy
(13 items)
 
  
CPUMotherboardGraphicsRAM
A10 5800k gigabyte  7970 mushkin ddr3  
Hard DriveOSPower
intel ssd Win7 64 Corsair TX850W 
  hide details  
Reply
post #64 of 151
how is it that every discussion about hardware devolves to a red vs. green controversy?

Also why is it that so many people think that companies exist to benefit anyone but themselves? Microsoft will always be trying to increase its market share and shut out any competitors.

nvidia developing CUDA and phsyx to run on its own cards only seems logical yet people think it should just make it for everyone. Plus they know their hardware and how to best utilize it. Having to figure in everyone else's would deplete the usefulness of the venture. That would in no way benefit their bottom line. They have made themselves the number one in graphics acceleration because of this. If AMD/ATI want some of this pie they can develop their own way of doing it not let NVIDIA do it for them. come on now.

I don't think the microsoft language will ever really replace CUDA because it will need to be agnostic and therefore not optimized for any one platform. But I think in the everyday computer market it could be a useful tool.
step up
(12 items)
 
  
CPUMotherboardGraphicsGraphics
4820k EVGA Classified x79 EVGA gtx 980ti SC  EVGA gtx 980ti Classified 
RAMHard DriveCoolingOS
GSkill ripjaws 32 gb samsung 840  custom phase chiller win 10 pro 
MonitorKeyboardCaseMouse
4 of them 3 in surround +1 zboard hawaiian punch rack logitech g502 proteus 
  hide details  
Reply
step up
(12 items)
 
  
CPUMotherboardGraphicsGraphics
4820k EVGA Classified x79 EVGA gtx 980ti SC  EVGA gtx 980ti Classified 
RAMHard DriveCoolingOS
GSkill ripjaws 32 gb samsung 840  custom phase chiller win 10 pro 
MonitorKeyboardCaseMouse
4 of them 3 in surround +1 zboard hawaiian punch rack logitech g502 proteus 
  hide details  
Reply
post #65 of 151
Another idea they stole from Linux
post #66 of 151
then intel and AMD cpu department are screwed
    
CPUGraphicsRAMHard Drive
i5-2410M @ 2.9ghz Nvidia GT 540M 1GB DDR3 4GB DDR3 1333mhz 500GB 5400rpm 
OSMonitorPowerCase
Win 7 Home 64 1366x768 gloss 90w power brick Acer 
  hide details  
Reply
    
CPUGraphicsRAMHard Drive
i5-2410M @ 2.9ghz Nvidia GT 540M 1GB DDR3 4GB DDR3 1333mhz 500GB 5400rpm 
OSMonitorPowerCase
Win 7 Home 64 1366x768 gloss 90w power brick Acer 
  hide details  
Reply
post #67 of 151
Too bad to use it you probably have to pay Microsoft some ridiculous sum of money, making it impossible for low end garage programmers to ever use, without breaking the law.


No?
My First Build
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 940 Biostar 790GX A2+ Powercolor Radeon HD 7870 2x2GB Dominators Stock Clocks 
Hard DriveOSMonitorPower
SAMSUNG 840 SSD 128GB Windows 7 (64 bit) HP w2207 Corsair 750W 
CaseMouseAudio
Antec 300 Logitech G5 On Board 
  hide details  
Reply
My First Build
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 940 Biostar 790GX A2+ Powercolor Radeon HD 7870 2x2GB Dominators Stock Clocks 
Hard DriveOSMonitorPower
SAMSUNG 840 SSD 128GB Windows 7 (64 bit) HP w2207 Corsair 750W 
CaseMouseAudio
Antec 300 Logitech G5 On Board 
  hide details  
Reply
post #68 of 151
Quote:
Originally Posted by Heavy MG View Post
Yet the AMD 58xx and 69xx series have more computing power than any Nvidia with CUDA.
Try bitcoin mining the difference between AMD and Nvidia is insane.
MOLLY
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 955 C3 Biostar TA90GX 128M XFX 5830 Kingston HyperX T1 DDR2 1066 
Hard DriveOptical DriveCoolingOS
Samsung spinpoint f3 1Tb LG DVD-R IDE Hyper 212+ Windows 7 64 ultimate 
MonitorPowerCaseMouse
ASUS VH236H Black 23" 2ms XFX 650W I just don't know anymore... Logitec G500 
  hide details  
Reply
MOLLY
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 955 C3 Biostar TA90GX 128M XFX 5830 Kingston HyperX T1 DDR2 1066 
Hard DriveOptical DriveCoolingOS
Samsung spinpoint f3 1Tb LG DVD-R IDE Hyper 212+ Windows 7 64 ultimate 
MonitorPowerCaseMouse
ASUS VH236H Black 23" 2ms XFX 650W I just don't know anymore... Logitec G500 
  hide details  
Reply
post #69 of 151
Right now i just can't worry about nvidia related stuff, 420Mhash/s is keeping me occupied lol.
mah bebe
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 2600k @ 4.7GHz Asus P8P67 Pro Powercolor PCS+ 6970 CF G.Skill RipjawsX 8GB 1600 
Hard DrivePowerCase
WD 1TB Cav. Black Corsair HX850 HAF 932 Adv. 
  hide details  
Reply
mah bebe
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 2600k @ 4.7GHz Asus P8P67 Pro Powercolor PCS+ 6970 CF G.Skill RipjawsX 8GB 1600 
Hard DrivePowerCase
WD 1TB Cav. Black Corsair HX850 HAF 932 Adv. 
  hide details  
Reply
post #70 of 151
Quote:
Originally Posted by Riou View Post
Yes, C++ is said to be based on DirectCompute. I was just referring to the article in OP, which is just marketing spin. C++ AMP is basically another competing language to CUDA and OpenCL.

Here are the benefits of C++ AMP:
C++ (created in 1979 so I doubt it based on something made later.) is s derivative and enhancement to C and gave birth to C# and Java, OpenCL is a derivative/extension of C not so sure about CUDA
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [NVIDIA]Microsoft going All-in on GPU computing