3 ms·
How is the hardware inferior for current use cases? The compute capabilities are pretty much the same for the latest NVidia/AMD cards in the same price range.
by dharma1 10y ago
How is the hardware inferior for current use cases?
The compute capabilities are pretty much the same for the latest NVidia/AMD cards in the same price range. It's a software thing - lack of optimised DL ops, particularly fast convolution kernels that is hurting the perf of AMD in deep learning
- frozenport 10y agoNo. AMD has worse floating point performance, and has poor support for half precision floating point values. Also gets beat on power consumption per flop.
- dharma1 10y agoFLOPS/$ is better on AMD afaik. double speed fp16 is only on gp100 and I don't know if any framework supports fp16 training yet. But yes AMD need to add 2x fp16 too. The main reason people don't use AMD for deep learning is lack of fast libraries, and AMD's inability to provide a CuDNN equivalent optimised for their hardware
- frozenport 10y agoBut I don't think this is a software issue: clFFT runs faster on NVIDIA. Simple operations like SAXPY run faster on NVIDIA. Etc.
- AlphaSite 10y agoSpeaking of this, AMDs new Vega cards in theory should offer more fp16 than the P100.