Overclock.net › Forums › Industry News › Hardware News › [VRZ] Next Generation GPUs: AMD's Cray On a Chip?
New Posts  All Forums:Forum Nav:

[VRZ] Next Generation GPUs: AMD's Cray On a Chip? - Page 5

post #41 of 51
https://en.bitcoin.it/wiki/Why_a_GPU...Nvidia_GPUs.3F
Quote:
Why are AMD GPUs faster than Nvidia GPUs?

Firstly, AMD designs GPUs with many simple ALUs/shaders (VLIW design) that run at a relatively low frequency clock (typically 1120-3200 ALUs at 625-900 MHz), whereas Nvidia's microarchitecture consists of fewer more complex ALUs and tries to compensate with a higher shader clock (typically 448-1024 ALUs at 1150-1544 MHz). Because of this VLIW vs. non-VLIW difference, Nvidia uses up more square millimeters of die space per ALU, hence can pack fewer of them per chip, and they hit the frequency wall sooner than AMD which prevents them from increasing the clock high enough to match or surpass AMD's performance. This translates to a raw ALU performance advantage for AMD:

AMD Radeon HD 6990: 3072 ALUs x 830 MHz = 2550 billion 32-bit instruction per second
Nvidia GTX 590: 1024 ALUs x 1214 MHz = 1243 billion 32-bit instruction per second

This approximate 2x-3x performance difference exists across the entire range of AMD and Nvidia GPUs. It is noticeably visible in all ALU-bound GPGPU workloads such as Bitcoin, password bruteforcers, etc.

Secondly, another difference favoring Bitcoin mining on AMD GPUs instead of Nvidia's is that the mining algorithm is based on SHA-256, which makes heavy use of the 32-bit integer right rotate operation. This operation can be implemented as a single hardware instruction on AMD GPUs, but requires three separate hardware instructions to be emulated on Nvidia GPUs (2 shifts + 1 add). This alone gives AMD another 1.7x performance advantage (~1900 instructions instead of ~3250 to execute the SHA-256 compression function).

Combined together, these 2 factors make AMD GPUs overall 3x-5x faster when mining Bitcoins.
AMD = more weaker sahders
Nvida = fewer stronger shaders

AMD has more total performance when all shaders are in use



Quote:
Originally Posted by 2010rig View Post
I converted a movie that is 1 hour and a half today to watch on my Ipod. The GPU usage was between 30% - 50%, it was going at 320FPS, and the whole thing was done in 6 minutes 30 seconds using Badaboom 2.0.
If you're interested in CUDA vs AMD vs Quick Sync vs x86 encoding:
http://www.anandtech.com/show/4083/t...-2100-tested/9

Quick Sync encoded the fastest with just as good quality as CPU encoding. CUDA encoding had quality issues.
I wouldn't use CUDA for encoding if I were you, I'd get a SB/IB CPU with Z68 chipset instead.
Edited by PoopaScoopa - 8/2/11 at 1:11pm
post #42 of 51
Quote:
Originally Posted by PoopaScoopa View Post
Quick Sync encoded the fastest with just as good quality as CPU encoding. CUDA encoding had quality issues.
I wouldn't use CUDA for encoding if I were you, I'd get a SB/IB CPU with Z68 chipset instead.
But do you really want to spend $210 when you can buy a $140 cpu?

AMD FX ~Seronx
(16 items)
 
  
CPUMotherboardGraphicsRAM
FX-9800P Acer Wasp R7 M440 SK Hynix HMA41GS6AFR8N-TF 
Hard DriveHard DriveOptical DriveCooling
KINGSTON RBU-SNS8152S3128GG2 TOSHIBA MQ01ABD100 HL-DT-ST DVDRAM GUE1N Stock 
OSMonitorKeyboardPower
Microsoft Windows 10 Home Build 14393 Viewsonic XG2401 24 Hz-144 Hz Ducky Channel Shine 3 Stock 65W 
CaseMouseMouse PadAudio
Acer Exoskeleton Steelseries Rival 300 Razer Megasoma AMD-Realtek ALC255 
  hide details  
Reply
AMD FX ~Seronx
(16 items)
 
  
CPUMotherboardGraphicsRAM
FX-9800P Acer Wasp R7 M440 SK Hynix HMA41GS6AFR8N-TF 
Hard DriveHard DriveOptical DriveCooling
KINGSTON RBU-SNS8152S3128GG2 TOSHIBA MQ01ABD100 HL-DT-ST DVDRAM GUE1N Stock 
OSMonitorKeyboardPower
Microsoft Windows 10 Home Build 14393 Viewsonic XG2401 24 Hz-144 Hz Ducky Channel Shine 3 Stock 65W 
CaseMouseMouse PadAudio
Acer Exoskeleton Steelseries Rival 300 Razer Megasoma AMD-Realtek ALC255 
  hide details  
Reply
post #43 of 51
If this were to be made the standard CPU's of the future would be wildly more powerful than what we have today. Even a relatively ancient 8800 or 3800 series is massively more powerful than the most powerful CPU of today, by a factor of three at least.

Currently a 6870 puts out 2 Tflop, if I could have used that as CPU it would be possible to crush the mightiest CRAY with only a handful of cards. The problem is the architecture and software support.

It would become possible to limit motherboard sizes to only accommodate USB, SATA and other ports while leaving RAM and processing power on the GPU. Whether VRAM size could or should be made variable as RAM today is a different discussion but if we take Win 7 as an example 8 GB would be fine for the average user, with more powerful cards or simply cards with more VRAM for the more demanding user. It's very exciting.
Edited by Liranan - 8/3/11 at 6:36am
The girlfriend.
(15 items)
 
The Mistress
(13 items)
 
Media Server
(11 items)
 
CPUMotherboardGraphicsRAM
A8-6410 Lenovo Lancer 4B2 K16.3 R5 128 Shaders/M230 Hynix 8GB DDR3 1600 
Hard DriveHard DriveOSMonitor
Samsung 840 120 GB SSD Seagate Momentus 1TB 5400rmp Win 8.1 CMN1487 TN LED 14" 1366*768 
KeyboardPowerMouseMouse Pad
Lenovo AccuType 2900mAh/41Wh Elan Trackpad/Logitech M90 Super Flower 
Audio
AMD Avalon(Connexant) 
  hide details  
Reply
The girlfriend.
(15 items)
 
The Mistress
(13 items)
 
Media Server
(11 items)
 
CPUMotherboardGraphicsRAM
A8-6410 Lenovo Lancer 4B2 K16.3 R5 128 Shaders/M230 Hynix 8GB DDR3 1600 
Hard DriveHard DriveOSMonitor
Samsung 840 120 GB SSD Seagate Momentus 1TB 5400rmp Win 8.1 CMN1487 TN LED 14" 1366*768 
KeyboardPowerMouseMouse Pad
Lenovo AccuType 2900mAh/41Wh Elan Trackpad/Logitech M90 Super Flower 
Audio
AMD Avalon(Connexant) 
  hide details  
Reply
post #44 of 51
Quote:
Originally Posted by Seronx View Post
If you knew about the differences of the architecture you would should have actually said

AMD -> Integer -> Bitcoin Mining
Nvidia -> Floating Point -> F@H

Nvidia for now has the better Floating Point performance
AMD has the best Integer performance
If I had said what you said... I would have been wrong.


See PoopaScoopa's post below yours.... straight from Bitcoin's site.

NVIDIA cores are more complex so there are fewer is the primary reason for the performance difference. The second reason is integer performance. In addition, my point still stands that CUDA is easier to develop with and NVIDIA is the choice for complex workloads.


Quote:
Originally Posted by Liranan View Post
If this were to be made the standard CPU's of the future would be wildly more powerful than what we have today. Even a relatively ancient 8800 or 3800 series is massively more powerful than the most powerful CPU of today, by a factor of three at least.

Currently a 6870 puts out 2 Tflop, if I could have used that as CPU it would be possible to crush the mightiest CRAY with only a handful of cards. The problem is the architecture and software support.

Also I don't fold, I run BOINC and my 4870 and 6870 love MW@H and DNetC.
No.... CPUs are very good at completing a few general tasks faster. They have path prediction and a host of other features that improve serial performance. GPGPU are good at completing many simplier tasks in parallel. Their uses differ (for now).

The CPU and GPU are converging but you don't want a pure GPU for a CPU.
Edited by DuckieHo - 8/3/11 at 6:35am
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #45 of 51
Quote:
Originally Posted by lordikon View Post
You did this without an Nvidia GPU? Oh...wait...

That's the issue, CUDA isn't open for any hardware manufacturer to use without licensing it. If this is to become a standard then it needs to be an open standard.
Open standard doesn't mean you can alter or contribute at will. "Open standards" are not innately open source.

I would much rather have proprietary software with support than an open standard with far less adaptation.

It's literally like OpenGL v. D3D
Po' Pimpin'
(11 items)
 
  
CPUMotherboardRAMHard Drive
i5 2500k @ stock Biostar TZ68K+ [A3] 4GB  Sandforce 1222 64GB SSD 
Optical DriveCoolingOSMonitor
LG 22x DVD-+RW  Stock Windows 7 x64 Acer S211HL 1080p 
PowerCaseMouse
600w Diablotek Linkworld Electronic Inland 
  hide details  
Reply
Po' Pimpin'
(11 items)
 
  
CPUMotherboardRAMHard Drive
i5 2500k @ stock Biostar TZ68K+ [A3] 4GB  Sandforce 1222 64GB SSD 
Optical DriveCoolingOSMonitor
LG 22x DVD-+RW  Stock Windows 7 x64 Acer S211HL 1080p 
PowerCaseMouse
600w Diablotek Linkworld Electronic Inland 
  hide details  
Reply
post #46 of 51
Quote:
Originally Posted by DuckieHo View Post
The CPU and GPU are converging but you don't want a pure GPU for a CPU.
Not yet, I see no reason to think there won't be a fusion of the two in the future. Intel weren't developing an 80 core Larrabee for nothing, which of course failed.
The girlfriend.
(15 items)
 
The Mistress
(13 items)
 
Media Server
(11 items)
 
CPUMotherboardGraphicsRAM
A8-6410 Lenovo Lancer 4B2 K16.3 R5 128 Shaders/M230 Hynix 8GB DDR3 1600 
Hard DriveHard DriveOSMonitor
Samsung 840 120 GB SSD Seagate Momentus 1TB 5400rmp Win 8.1 CMN1487 TN LED 14" 1366*768 
KeyboardPowerMouseMouse Pad
Lenovo AccuType 2900mAh/41Wh Elan Trackpad/Logitech M90 Super Flower 
Audio
AMD Avalon(Connexant) 
  hide details  
Reply
The girlfriend.
(15 items)
 
The Mistress
(13 items)
 
Media Server
(11 items)
 
CPUMotherboardGraphicsRAM
A8-6410 Lenovo Lancer 4B2 K16.3 R5 128 Shaders/M230 Hynix 8GB DDR3 1600 
Hard DriveHard DriveOSMonitor
Samsung 840 120 GB SSD Seagate Momentus 1TB 5400rmp Win 8.1 CMN1487 TN LED 14" 1366*768 
KeyboardPowerMouseMouse Pad
Lenovo AccuType 2900mAh/41Wh Elan Trackpad/Logitech M90 Super Flower 
Audio
AMD Avalon(Connexant) 
  hide details  
Reply
post #47 of 51
Quote:
Originally Posted by Liranan View Post
Not yet, I see no reason to think there won't be a fusion of the two in the future. Intel weren't developing an 80 core Larrabee for nothing, which of course failed.
AMD, NVIDIA, and Intel have been working on the convergence for the last few years.

AMD has Fusion today.
NVIDIA has Tegra 2 today.
Intel has Sandy Bridge today.

Integeration is only going to get tighter over time. However, a CPU and GPU are still different today.
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #48 of 51
Quote:
Originally Posted by 2010rig View Post
One word: CUDA
AMD Stream.
X79-GCN
(22 items)
 
  
CPUMotherboardGraphicsRAM
Intel 3930K 4.5GHz HT GIGABYTE GA-X79-UP4 AMD R9-290X GEil Evo Potenza DDR3 2400MHz CL10 (4x4GB) 
Hard DriveCoolingCoolingCooling
Samsung 840 Pro 120GB EK Supremacy (CPU) NF F12's P/P (360 Rad)  NF A14's (420 Rad)  
CoolingCoolingCoolingCooling
XSPC Chrome Compression Fittings EK RES X3 150 Primochill PremoFlex Advanced LRT Clear 1/2 ID EK-FC (R9 290X) 
CoolingCoolingCoolingOS
EK D5 Vario Top-X  Phobya G-Changer V2 360mm Phobya G-Changer V2 420mm Win 10 x64 Pro 
MonitorKeyboardPowerCase
BenQ XR3501 35" Curved Corsair Vengeance K90 Seasonic X-1250 Gold (v2) Corsair 900D 
MouseAudio
Logitech G400s Senn HD 598 
  hide details  
Reply
X79-GCN
(22 items)
 
  
CPUMotherboardGraphicsRAM
Intel 3930K 4.5GHz HT GIGABYTE GA-X79-UP4 AMD R9-290X GEil Evo Potenza DDR3 2400MHz CL10 (4x4GB) 
Hard DriveCoolingCoolingCooling
Samsung 840 Pro 120GB EK Supremacy (CPU) NF F12's P/P (360 Rad)  NF A14's (420 Rad)  
CoolingCoolingCoolingCooling
XSPC Chrome Compression Fittings EK RES X3 150 Primochill PremoFlex Advanced LRT Clear 1/2 ID EK-FC (R9 290X) 
CoolingCoolingCoolingOS
EK D5 Vario Top-X  Phobya G-Changer V2 360mm Phobya G-Changer V2 420mm Win 10 x64 Pro 
MonitorKeyboardPowerCase
BenQ XR3501 35" Curved Corsair Vengeance K90 Seasonic X-1250 Gold (v2) Corsair 900D 
MouseAudio
Logitech G400s Senn HD 598 
  hide details  
Reply
post #49 of 51
Quote:
Originally Posted by BradleyW View Post
AMD Stream.
The industry is not using AMD Stream though. Many more are using CUDA.

If you build it, they will not always come. NVIDIA is offering support and investing in CUDA development much more than AMD is.

I did a quick search on Monster.com for CUDA and these are the companies looking for someone with experience. I can't find any looking for "AMD Stream" experience (maybe you can?)

CyberCoders
Intellisis
Aptina Imaging
Citrix
TEKsystems
FARO Technologies, Inc.
Raytheon
Modis, Inc.
Bloomberg
Volt Workforce Solutions
Textron Inc
Visionist Incorp
ConnectTel, Inc.
Pacific Biosciences
Informatic Technologies
BAE Systems
CRAY Inc
Edited by DuckieHo - 8/3/11 at 9:32am
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #50 of 51
Is CUDA relevent anymore..?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [VRZ] Next Generation GPUs: AMD's Cray On a Chip?