OpenCL-Benchmark icon indicating copy to clipboard operation
OpenCL-Benchmark copied to clipboard

Add the ability to submit and compare benchmark result online

Open brokeDude2901 opened this issue 2 years ago • 2 comments

Would be nice if there is a site to view and compare all result online

brokeDude2901 avatar Nov 21 '23 02:11 brokeDude2901

Phoronix has added the FLUIDX3D Benchmark:

https://openbenchmarking.org/test/pts/fluidx3d&eval=d57337adf43caf95155efec209add4dd36b9f344

Maybe they would include this benchmark, too ? (Maybe not, since CLPeak already is present and its similar)

RhynarAI avatar Mar 01 '24 16:03 RhynarAI

This would be a cool feature.

For my contribution, here is an Arc 750 LE GPU result , but with double precision emulation turned on:

|----------------.------------------------------------------------------------| | Device ID | 2 | | Device Name | Intel(R) Arc(TM) A750 Graphics | | Device Vendor | Intel(R) Corporation | | Device Driver | 23.22.26516.34 (Linux) | | OpenCL Version | OpenCL C 1.2 | | Compute Units | 448 at 2400 MHz (3584 cores, 17.203 TFLOPs/s) | | Memory, Cache | 8127 MB, 16384 KB global / 64 KB local | | Buffer Limits | 3860 MB global, 3953458 KB constant | |----------------'------------------------------------------------------------| | Info: OpenCL C code successfully compiled. | | FP64 compute 0.086 TFLOPs/s (1/64) | | FP32 compute 10.462 TFLOPs/s (2/3 ) | | FP16 compute 15.912 TFLOPs/s ( 1x ) | | INT64 compute 1.076 TIOPs/s (1/16) | | INT32 compute 4.087 TIOPs/s (1/4 ) | | INT16 compute 25.986 TIOPs/s ( 2x ) | | INT8 compute 9.787 TIOPs/s (1/2 ) | | Memory Bandwidth ( coalesced read ) 207.97 GB/s | | Memory Bandwidth ( coalesced write) 408.68 GB/s | | Memory Bandwidth (misaligned read ) 395.89 GB/s | | Memory Bandwidth (misaligned write) 240.82 GB/s | | PCIe Bandwidth (send ) 8.81 GB/s | | PCIe Bandwidth ( receive ) 8.49 GB/s |

PCIe Bandwidth ( bidirectional) (Gen4 x16) 8.96 GB/s

sumseq avatar Jul 25 '24 16:07 sumseq