cancel
Showing results for 
Search instead for 
Did you mean: 

Processors

noctis
Adept I

Threadripper 1950x bad performance for tensorflow

Hi, guys,

I recently tested some deep learning applications including inception V3, Resnet implemented with TensorFlow on my machine. The machine has 2 1080ti and 1950x. And the memory speed is 2933 with  64GB capacity.

I saw from some websites that one 1080ti with intel CPU can deal with ~140 pictures per second, while mine can only deal with 80+ pictures. With 2 GPU running together, that figure increases to 150 pictures per second. This means that two 1080ti with 1950x can only reach the performance of 1 1080ti on Intel platform!!!! This is too terrible!

Do you guys have any related experience? I searched the internet but found no solution. There's even no bazel build option for AMD cpu.

Looking for your help. I'm in despair , for spending twice the money to buy such a machine with so disappointing performance.

0 Likes
1 Solution
noctis
Adept I

Sorry guys,

I made a serious mistake. The performance of the benchmark I tested was totally different from that tested by others. The test link is here:

GitHub - tensorflow/benchmarks: Benchmark code

And after I found this and compared with

https://www.leadergpu.com/articles/429-tensorflow-resnet-50-benchmark

1950x is slightly faster!

Sorry for my misjudgment!

Great job, AMD!

View solution in original post

0 Likes
1 Reply
noctis
Adept I

Sorry guys,

I made a serious mistake. The performance of the benchmark I tested was totally different from that tested by others. The test link is here:

GitHub - tensorflow/benchmarks: Benchmark code

And after I found this and compared with

https://www.leadergpu.com/articles/429-tensorflow-resnet-50-benchmark

1950x is slightly faster!

Sorry for my misjudgment!

Great job, AMD!

0 Likes