Last active
December 9, 2019 14:16
-
-
Save rarecoil/3d0411a01c649d3b99d3ae1e1a792223 to your computer and use it in GitHub Desktop.
AMD Radeon RX 5700 XT (Navi) PlaidML - plaidbench deep learning benchmarks
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Running 1024 examples with mobilenet, batch size 1, on backend plaid | |
| INFO:plaidml:Opening device "opencl_amd_gfx1010.0" | |
| Compiling network... Warming up... Running... | |
| Example finished, elapsed: 6.513s (compile), 7.912s (execution) | |
| ----------------------------------------------------------------------------------------- | |
| Network Name Inference Latency Time / FPS | |
| ----------------------------------------------------------------------------------------- | |
| mobilenet 7.73 ms 1.67 ms / 599.91 fps | |
| Correctness: PASS, max_error: 7.314303729799576e-06, max_abs_error: 6.407499313354492e-07, fail_ratio: 0.0 | |
| Running 1024 examples with resnet50, batch size 1, on backend plaid | |
| INFO:plaidml:Opening device "opencl_amd_gfx1010.0" | |
| Compiling network... Warming up... Running... | |
| Example finished, elapsed: 5.678s (compile), 16.609s (execution) | |
| ----------------------------------------------------------------------------------------- | |
| Network Name Inference Latency Time / FPS | |
| ----------------------------------------------------------------------------------------- | |
| resnet50 16.22 ms 7.96 ms / 125.69 fps | |
| Correctness: PASS, max_error: 3.435083499425673e-06, max_abs_error: 1.7881393432617188e-07, fail_ratio: 0.0 | |
| Running 1024 examples with vgg16, batch size 1, on backend plaid | |
| INFO:plaidml:Opening device "opencl_amd_gfx1010.0" | |
| Compiling network... Warming up... Running... | |
| Example finished, elapsed: 3.232s (compile), 55.677s (execution) | |
| ----------------------------------------------------------------------------------------- | |
| Network Name Inference Latency Time / FPS | |
| ----------------------------------------------------------------------------------------- | |
| vgg16 54.37 ms 50.40 ms / 19.84 fps | |
| Correctness: PASS, max_error: 2.691625468287384e-06, max_abs_error: 4.0046870708465576e-08, fail_ratio: 0.0 | |
| Running 1024 examples with vgg19, batch size 1, on backend plaid | |
| INFO:plaidml:Opening device "opencl_amd_gfx1010.0" | |
| Compiling network... Warming up... Running... | |
| Example finished, elapsed: 3.281s (compile), 92.729s (execution) | |
| ----------------------------------------------------------------------------------------- | |
| Network Name Inference Latency Time / FPS | |
| ----------------------------------------------------------------------------------------- | |
| vgg19 90.56 ms 55.29 ms / 18.09 fps | |
| Correctness: PASS, max_error: 1.7339709756925004e-06, max_abs_error: 1.7136335372924805e-07, fail_ratio: 0.0 | |
| Running 1024 examples with xception, batch size 1, on backend plaid | |
| INFO:plaidml:Opening device "opencl_amd_gfx1010.0" | |
| Downloading data from https://github.com/fchollet/deep-learning-models/releases/download/v0.4/xception_weights_tf_dim_ordering_tf_kernels.h5 | |
| 91889664/91884032 [==============================] - 13s 0us/step | |
| Compiling network... Warming up... Running... | |
| Example finished, elapsed: 8.406s (compile), 19.893s (execution) | |
| ----------------------------------------------------------------------------------------- | |
| Network Name Inference Latency Time / FPS | |
| ----------------------------------------------------------------------------------------- | |
| xception 19.43 ms 9.93 ms / 100.68 fps | |
| Correctness: PASS, max_error: 4.805629032489378e-06, max_abs_error: 4.6193599700927734e-07, fail_ratio: 0.0 | |
| Running 1024 examples with imdb_lstm, batch size 1, on backend plaid | |
| INFO:plaidml:Opening device "opencl_amd_gfx1010.0" | |
| Compiling network... Warming up... Running... | |
| Example finished, elapsed: 6.976s (compile), 74.840s (execution) | |
| ----------------------------------------------------------------------------------------- | |
| Network Name Inference Latency Time / FPS | |
| ----------------------------------------------------------------------------------------- | |
| imdb_lstm 73.09 ms 6.01 ms / 166.47 fps | |
| Correctness: PASS, max_error: 0.0, max_abs_error: 0.0, fail_ratio: 0.0 |
I have not found the RX 5700 XT's architecture to be very useful for anything other than gaming. For DL/ML, I ended up buying more Radeon VII cards, which are officially ROCm supported and the same underlying architecture as the Radeon Instinct MI50.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Damn that's not great :/ I thought 5700 XT was supposed to compete against 2070 ! But that is actually where I expect 1660 Ti to be. Too bad !