Jump to content

Mixing graphics cards

Mike171
By Mike171
in Graphics Cards
 Share
Go to solution Solved by Madgemade,
34 minutes ago, Mike171 said:

AMD graphics card does not support tensorflow-gpu because of opencl

And it never will because AMD have made a tensorflow port that uses HCC instead of OpenCL (which is mostly dead and dying as an API at this point). The AMD Tensorflow can be found here. The installation is not at all easy if you're not experienced in Linux. AMD don't support Windows at all for this stuff. You also need R9 Fury or newer, ideally Vega or RX series (doesn't say in guides but I tried R9 290 and it didn't work).

34 minutes ago, Mike171 said:

mix two different graphics card architectures for GPGPU

There is no hard and fast rule. It depends on the workload, program, etc. For Tensorflow you can do it.

The architecture of the GPU doesn't matter as long as it is supported. Tensorflow currently needs CUDA® Compute Capability 3.5 or higher. This is supported by GTX 700 series and newer (except for 760/770, which are rebranded 600 series).

24 minutes ago, Mike171 said:

cannot take any advantages from sli

SLI is for graphics rendering (such as games) only. It is not used in compute and is not needed. Only NVLink can be used for compute and only on Quadro class cards, not consumer ones. From what I have read NVLink doesn't do much for Compute workloads which can be split easily across GPUs and so it isn't needed for Machine Learning uses.

Posted

With Nvidia, can you mix two different graphics card architectures for GPGPU and tensorflow? //Currently AMD graphics card does not support tensorflow-gpu because of opencl. 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

I know you can mix, but I don't know how it would affect GPGPU.

M.S.C.E. (M.Sc. Computer Engineering), IT specialist in a hospital, 30+ years of gaming, 20+ years of computer enthusiasm, Geek, Trekkie, anime fan

  • Main PC: AMD Ryzen 7 5800X3D - EK AIO 360 D-RGB - Arctic Cooling MX-4 - Asus Prime X570-P - 4x8GB DDR4 3200 HyperX Fury CL16 - Sapphire AMD Radeon 6950XT Nitro+ - 1TB Kingston Fury Renegade - 2TB Kingston Fury Renegade - 512GB ADATA SU800 - 960GB Kingston A400 - Seasonic PX-850 850W  - custom black ATX and EPS cables - Fractal Design Define R5 Blackout - Windows 11 x64 23H2 - 3 Arctic Cooling P14 PWM PST - 5 Arctic Cooling P12 PWM PST
  • Peripherals: LG 32GK650F - Dell P2319h - Logitech G Pro X Superlight with Tiger Ice - HyperX Alloy Origins Core (TKL) - EndGame Gear MPC890 - Genius HF 1250B - Akliam PD4 - Sennheiser HD 560s - Simgot EM6L - Truthear Zero - QKZ x HBB - 7Hz Salnotes Zero - Logitech C270 - Behringer PS400 - BM700  - Colormunki Smile - Speedlink Torid - Jysk Stenderup - LG 24x External DVD writer - Konig smart card reader
  • Laptop: Acer E5–575G-386R 15.6" 1080p (i3 6100U + 12GB DDR4 (4GB+8GB) + GeForce 940MX + 256GB nVME) Win 10 Pro x64 22H2 - Logitech G305 + AAA Lithium battery
  • Networking: Asus TUF Gaming AX6000 - Arcadyan ISP router - 35/5 Mbps vDSL
  • TV and gadgets: TCL 50EP680 50" 4K LED + Sharp HT-SB100 75W RMS soundbar - Samsung Galaxy Tab A8 10.1" - OnePlus 9 256GB - Olymous Cameda C-160 - GameBoy Color 
  • Streaming/Server/Storage PC: AMD Ryzen 5 3600 - LC-Power LC-CC-120 - MSI B450 Tomahawk Max - 2x4GB ADATA 2666 DDR4 - 120GB Kingston V300 - Toshiba DT01ACA100 1TB - Toshiba DT01ACA200 2TB - 2x WD Green 2TB - Sapphire Pulse AMD Radeon R9 380X - 550W EVGA G3 SuperNova - Chieftec Giga DF-01B - White Shark Spartan X keyboard - Roccat Kone Pure Military Desert strike - Logitech S-220 - Philips 226L
  • Livingroom PC (dad uses): AMD FX 8300 - Arctic Freezer 64 - Asus M5A97 R2.0 Evo - 2x4GB DDR3 1833 Kingston - MSI Radeon HD 7770 1GB OC - 120GB Adata SSD - 500W Fractal Design Essence - DVD-RW - Samsung SM 2253BW - Logitech G710+ - wireless vertical mouse - MS 2.0 speakers
Link to comment
Share on other sites
Link to post
Share on other sites
Posted
  • Author
5 minutes ago, 191x7 said:

I know you can mix, but I don't know how it would affect GPGPU.

Did you mean with back propagation, stochastic gradient descent etc, you cannot take any advantages from sli?

Link to comment
Share on other sites
Link to post
Share on other sites
Posted
  • Solution
34 minutes ago, Mike171 said:

AMD graphics card does not support tensorflow-gpu because of opencl

And it never will because AMD have made a tensorflow port that uses HCC instead of OpenCL (which is mostly dead and dying as an API at this point). The AMD Tensorflow can be found here. The installation is not at all easy if you're not experienced in Linux. AMD don't support Windows at all for this stuff. You also need R9 Fury or newer, ideally Vega or RX series (doesn't say in guides but I tried R9 290 and it didn't work).

34 minutes ago, Mike171 said:

mix two different graphics card architectures for GPGPU

There is no hard and fast rule. It depends on the workload, program, etc. For Tensorflow you can do it.

The architecture of the GPU doesn't matter as long as it is supported. Tensorflow currently needs CUDA® Compute Capability 3.5 or higher. This is supported by GTX 700 series and newer (except for 760/770, which are rebranded 600 series).

24 minutes ago, Mike171 said:

cannot take any advantages from sli

SLI is for graphics rendering (such as games) only. It is not used in compute and is not needed. Only NVLink can be used for compute and only on Quadro class cards, not consumer ones. From what I have read NVLink doesn't do much for Compute workloads which can be split easily across GPUs and so it isn't needed for Machine Learning uses.

Gaming Rig:CPU: Xeon E3-1230 v2¦RAM: 16GB DDR3 Balistix 1600Mhz¦MB: MSI Z77A-G43¦HDD: 480GB SSD, 3.5TB HDDs¦GPU: AMD Radeon VII¦PSU: FSP 700W¦Case: Carbide 300R

 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted
  • Author
3 minutes ago, Madgemade said:

And it never will because AMD have made a tensorflow port that uses HCC instead of OpenCL (which is mostly dead and dying as an API at this point). The AMD Tensorflow can be found here. The installation is not at all easy if you're not experienced in Linux. AMD don't support Windows at all for this stuff. You also need R9 Fury or newer, ideally Vega or RX series (doesn't say in guides but I tried R9 290 and it didn't work).

There is no hard and fast rule. It depends on the workload, program, etc. For Tensorflow you can do it.

The architecture of the GPU doesn't matter as long as it is supported. Tensorflow currently needs CUDA® Compute Capability 3.5 or higher. This is supported by GTX 700 series and newer (except for 760/770, which are rebranded 600 series).

SLI is for graphics rendering (such as games) only. It is not used in compute and is not needed. Only NVLink can be used for compute and only on Quadro class cards, not consumer ones. From what I have read NVLink doesn't do much for Compute workloads which can be split easily across GPUs and so it isn't needed for Machine Learning uses.

Thanks

Link to comment
Share on other sites
Link to post
Share on other sites
Posted
Just now, Mike171 said:

Thanks

No probs. If it answered all your questions you can always mark it as solution. Then people will find it easily from Google. ?

Gaming Rig:CPU: Xeon E3-1230 v2¦RAM: 16GB DDR3 Balistix 1600Mhz¦MB: MSI Z77A-G43¦HDD: 480GB SSD, 3.5TB HDDs¦GPU: AMD Radeon VII¦PSU: FSP 700W¦Case: Carbide 300R

 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted
1 minute ago, Mike171 said:

Ok

?

Gaming Rig:CPU: Xeon E3-1230 v2¦RAM: 16GB DDR3 Balistix 1600Mhz¦MB: MSI Z77A-G43¦HDD: 480GB SSD, 3.5TB HDDs¦GPU: AMD Radeon VII¦PSU: FSP 700W¦Case: Carbide 300R

 

Link to comment
Share on other sites
Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share
Go to topic listing Graphics Cards

×