OpenCL ROCm Linux vs. AMDGPU-PRO Benchmarks ROCm 1.4 benchmarks on Ubuntu 16.04 compared to AMDGPU-PRO. OpenCL benchmarks by Michael Larabel for a future article on Phoronix.com. R9 Fury: ROCm 1.4: Processor: Intel Xeon E3-1280 v5 @ 4.00GHz (8 Cores), Motherboard: MSI C236A WORKSTATION (MS-7998) v1.0, Chipset: Intel Sky Lake, Memory: 16384MB, Disk: 256GB TOSHIBA-RD400, Graphics: Sapphire AMD Radeon R9 FURY / NANO 3968MB, Audio: Realtek ALC1150, Monitor: Acer B286HK, Network: Intel Connection OS: Ubuntu 16.04, Kernel: 4.6.0-kfd-compute-rocm-rel-1.4-16 (x86_64), Desktop: Unity 7.4.0, Display Server: X Server 1.18.3, Display Driver: modesetting 1.18.3, OpenGL: 4.1 Mesa 11.2.0 Gallium 0.4, OpenCL: OpenCL 2.0 AMD-APP (2300.5), Compiler: GCC 5.4.0 20160609 + Clang 4.0 + LLVM 4.0.0, File-System: ext4, Screen Resolution: 3840x2160 RX 460: ROCm 1.4: Processor: Intel Xeon E3-1280 v5 @ 4.00GHz (8 Cores), Motherboard: MSI C236A WORKSTATION (MS-7998) v1.0, Chipset: Intel Sky Lake, Memory: 16384MB, Disk: 256GB TOSHIBA-RD400, Graphics: LLVMpipe, Audio: Realtek ALC1150, Monitor: Acer B286HK, Network: Intel Connection OS: Ubuntu 16.04, Kernel: 4.6.0-kfd-compute-rocm-rel-1.4-16 (x86_64), Desktop: Unity 7.4.0, Display Server: X Server 1.18.3, Display Driver: modesetting 1.18.3, OpenGL: 3.3 Mesa 11.2.0 Gallium 0.4, OpenCL: OpenCL 2.0 AMD-APP (2300.5), Compiler: GCC 5.4.0 20160609 + Clang 4.0 + LLVM 4.0.0, File-System: ext4, Screen Resolution: 3840x2160 RX 480: ROCm 1.4: Processor: Intel Xeon E3-1280 v5 @ 4.00GHz (8 Cores), Motherboard: MSI C236A WORKSTATION (MS-7998) v1.0, Chipset: Intel Sky Lake, Memory: 16384MB, Disk: 256GB TOSHIBA-RD400, Graphics: LLVMpipe, Audio: Realtek ALC1150, Monitor: Acer B286HK, Network: Intel Connection OS: Ubuntu 16.04, Kernel: 4.6.0-kfd-compute-rocm-rel-1.4-16 (x86_64), Desktop: Unity 7.4.0, Display Server: X Server 1.18.3, Display Driver: modesetting 1.18.3, OpenGL: 3.3 Mesa 11.2.0 Gallium 0.4, OpenCL: OpenCL 2.0 AMD-APP (2300.5), Compiler: GCC 5.4.0 20160609 + Clang 4.0 + LLVM 4.0.0, File-System: ext4, Screen Resolution: 3840x2160 RX 460: AMDGPU-PRO 16.50: Processor: Intel Xeon E3-1280 v5 @ 4.00GHz (8 Cores), Motherboard: MSI C236A WORKSTATION (MS-7998) v1.0, Chipset: Intel Sky Lake, Memory: 16384MB, Disk: 256GB TOSHIBA-RD400, Graphics: AMD Radeon RX 460 2048MB, Audio: Realtek ALC1150, Monitor: Acer B286HK, Network: Intel Connection OS: Ubuntu 16.04, Kernel: 4.4.0-59-generic (x86_64), Desktop: Unity 7.4.0, Display Server: X Server 1.18.3, Display Driver: amdgpu 1.1.99, OpenGL: 4.5.13462, OpenCL: OpenCL 2.0 AMD-APP (2236.5), Compiler: GCC 5.4.0 20160609, File-System: ext4, Screen Resolution: 3840x2160 RX 480: AMDGPU-PRO 16.50: Processor: Intel Xeon E3-1280 v5 @ 4.00GHz (8 Cores), Motherboard: MSI C236A WORKSTATION (MS-7998) v1.0, Chipset: Intel Sky Lake, Memory: 16384MB, Disk: 256GB TOSHIBA-RD400, Graphics: AMD Radeon RX 480 8192MB, Audio: Realtek ALC1150, Monitor: Acer B286HK, Network: Intel Connection OS: Ubuntu 16.04, Kernel: 4.4.0-59-generic (x86_64), Desktop: Unity 7.4.0, Display Server: X Server 1.18.3, Display Driver: amdgpu 1.1.99, OpenGL: 4.5.13462, OpenCL: OpenCL 2.0 AMD-APP (2236.5), Compiler: GCC 5.4.0 20160609, File-System: ext4, Screen Resolution: 3840x2160 R9 Fury: AMDGPU-PRO 16.50: Processor: Intel Xeon E3-1280 v5 @ 4.00GHz (8 Cores), Motherboard: MSI C236A WORKSTATION (MS-7998) v1.0, Chipset: Intel Sky Lake, Memory: 16384MB, Disk: 256GB TOSHIBA-RD400, Graphics: Sapphire AMD Radeon R9 Fury 4096MB, Audio: Realtek ALC1150, Monitor: Acer B286HK, Network: Intel Connection OS: Ubuntu 16.04, Kernel: 4.4.0-59-generic (x86_64), Desktop: Unity 7.4.0, Display Server: X Server 1.18.3, Display Driver: amdgpu 1.1.99, OpenGL: 4.5.13462, OpenCL: OpenCL 2.0 AMD-APP (2236.5), Compiler: GCC 5.4.0 20160609, File-System: ext4, Screen Resolution: 3840x2160 SHOC Scalable HeterOgeneous Computing 2015-11-10 Target: OpenCL - Benchmark: Triad GB/s > Higher Is Better R9 Fury: ROCm 1.4 ......... 10.59 |============================================ RX 460: ROCm 1.4 .......... 5.21 |====================== RX 480: ROCm 1.4 .......... 7.94 |================================= RX 460: AMDGPU-PRO 16.50 .. 6.25 |========================== RX 480: AMDGPU-PRO 16.50 .. 9.40 |======================================= R9 Fury: AMDGPU-PRO 16.50 . 4.12 |================= SHOC Scalable HeterOgeneous Computing 2015-11-10 Target: OpenCL - Benchmark: FFT SP GFLOPS > Higher Is Better R9 Fury: ROCm 1.4 ......... 399.71 |======================= RX 460: ROCm 1.4 .......... 158.21 |========= RX 480: ROCm 1.4 .......... 403.22 |======================= RX 460: AMDGPU-PRO 16.50 .. 245.13 |============== RX 480: AMDGPU-PRO 16.50 .. 508.20 |============================= R9 Fury: AMDGPU-PRO 16.50 . 751.86 |=========================================== SHOC Scalable HeterOgeneous Computing 2015-11-10 Target: OpenCL - Benchmark: Max SP Flops GFLOPS > Higher Is Better R9 Fury: ROCm 1.4 ......... 5330.67 |=============================== RX 460: ROCm 1.4 .......... 2158.12 |============= RX 480: ROCm 1.4 .......... 5815.52 |================================== RX 460: AMDGPU-PRO 16.50 .. 2066.69 |============ RX 480: AMDGPU-PRO 16.50 .. 5750.69 |================================== R9 Fury: AMDGPU-PRO 16.50 . 7131.18 |========================================== SHOC Scalable HeterOgeneous Computing 2015-11-10 Target: OpenCL - Benchmark: Bus Speed Download GB/s > Higher Is Better R9 Fury: ROCm 1.4 ......... 11.32 |==================================== RX 460: ROCm 1.4 .......... 5.72 |================== RX 480: ROCm 1.4 .......... 8.37 |=========================== RX 460: AMDGPU-PRO 16.50 .. 6.93 |====================== RX 480: AMDGPU-PRO 16.50 .. 13.66 |============================================ R9 Fury: AMDGPU-PRO 16.50 . 13.69 |============================================ SHOC Scalable HeterOgeneous Computing 2015-11-10 Target: OpenCL - Benchmark: Bus Speed Readback GB/s > Higher Is Better R9 Fury: ROCm 1.4 ......... 10.86 |================================== RX 460: ROCm 1.4 .......... 5.27 |================ RX 480: ROCm 1.4 .......... 8.38 |========================== RX 460: AMDGPU-PRO 16.50 .. 7.14 |====================== RX 480: AMDGPU-PRO 16.50 .. 14.20 |============================================ R9 Fury: AMDGPU-PRO 16.50 . 14.21 |============================================ SHOC Scalable HeterOgeneous Computing 2015-11-10 Target: OpenCL - Benchmark: Texture Read Bandwidth GB/s > Higher Is Better R9 Fury: ROCm 1.4 ......... 214.53 |========================================= RX 460: ROCm 1.4 .......... 91.14 |================== RX 480: ROCm 1.4 .......... 193.49 |===================================== RX 460: AMDGPU-PRO 16.50 .. 77.35 |=============== RX 480: AMDGPU-PRO 16.50 .. 160.57 |=============================== R9 Fury: AMDGPU-PRO 16.50 . 223.25 |=========================================== Parboil 2.5 Test: OpenCL BFS Seconds < Lower Is Better R9 Fury: ROCm 1.4 . 1.43 |=================================================== RX 460: ROCm 1.4 .. 1.47 |==================================================== RX 480: ROCm 1.4 .. 1.49 |===================================================== Parboil 2.5 Test: OpenCL LBM Seconds < Lower Is Better R9 Fury: ROCm 1.4 . 36.81 |=============================== RX 460: ROCm 1.4 .. 61.84 |==================================================== RX 480: ROCm 1.4 .. 38.49 |================================ Parboil 2.5 Test: OpenCL TPACF Seconds < Lower Is Better R9 Fury: ROCm 1.4 . 2.24 |============================== RX 460: ROCm 1.4 .. 3.90 |===================================================== RX 480: ROCm 1.4 .. 2.24 |============================== Rodinia 2.4 Test: OpenCL Myocyte Seconds < Lower Is Better R9 Fury: ROCm 1.4 ........ 362.48 |============================================ RX 460: ROCm 1.4 ......... 169.92 |===================== RX 480: ROCm 1.4 ......... 244.31 |============================== RX 460: AMDGPU-PRO 16.50 . 131.59 |================ RX 480: AMDGPU-PRO 16.50 . 88.30 |=========== Rodinia 2.4 Test: OpenCL Heartwall Seconds < Lower Is Better R9 Fury: ROCm 1.4 ......... 6.45 |===================== RX 460: ROCm 1.4 .......... 13.51 |============================================ RX 480: ROCm 1.4 .......... 7.28 |======================== RX 460: AMDGPU-PRO 16.50 .. 7.97 |========================== RX 480: AMDGPU-PRO 16.50 .. 5.35 |================= R9 Fury: AMDGPU-PRO 16.50 . 6.38 |===================== Darktable 2.2.1 Test: Boat - Acceleration: OpenCL Seconds < Lower Is Better R9 Fury: ROCm 1.4 ......... 4.98 |======================= RX 460: ROCm 1.4 .......... 9.57 |============================================= RX 480: ROCm 1.4 .......... 5.72 |=========================== RX 460: AMDGPU-PRO 16.50 .. 9.51 |============================================= RX 480: AMDGPU-PRO 16.50 .. 4.37 |===================== R9 Fury: AMDGPU-PRO 16.50 . 4.22 |==================== Darktable 2.2.1 Test: Masskrug - Acceleration: OpenCL Seconds < Lower Is Better R9 Fury: ROCm 1.4 ......... 6.09 |====================================== RX 460: ROCm 1.4 .......... 7.05 |============================================ RX 480: ROCm 1.4 .......... 5.93 |===================================== RX 460: AMDGPU-PRO 16.50 .. 7.20 |============================================= RX 480: AMDGPU-PRO 16.50 .. 5.76 |==================================== R9 Fury: AMDGPU-PRO 16.50 . 6.30 |======================================= Darktable 2.2.1 Test: Server Room - Acceleration: OpenCL Seconds < Lower Is Better R9 Fury: ROCm 1.4 ......... 1.48 |======================== RX 460: ROCm 1.4 .......... 2.48 |======================================= RX 480: ROCm 1.4 .......... 0.99 |================ RX 460: AMDGPU-PRO 16.50 .. 2.83 |============================================= RX 480: AMDGPU-PRO 16.50 .. 0.99 |================ R9 Fury: AMDGPU-PRO 16.50 . 1.79 |============================ JuliaGPU 1.2pts1 OpenCL Device: GPU Samples/sec > Higher Is Better R9 Fury: ROCm 1.4 ......... 73072755.80 |================================== RX 460: ROCm 1.4 .......... 46101692.27 |===================== RX 480: ROCm 1.4 .......... 70675082.10 |================================= RX 460: AMDGPU-PRO 16.50 .. 50807022.25 |======================== RX 480: AMDGPU-PRO 16.50 .. 81972594.40 |====================================== R9 Fury: AMDGPU-PRO 16.50 . 75992404.70 |=================================== MandelbulbGPU 1.0pts1 OpenCL Device: GPU Samples/sec > Higher Is Better R9 Fury: ROCm 1.4 ......... 44388927.12 |================================== RX 460: ROCm 1.4 .......... 29562658.90 |======================= RX 480: ROCm 1.4 .......... 49050438.67 |====================================== RX 460: AMDGPU-PRO 16.50 .. 32208376.98 |========================= RX 480: AMDGPU-PRO 16.50 .. 48517365.80 |====================================== R9 Fury: AMDGPU-PRO 16.50 . 43447360.40 |================================== MandelGPU 1.3pts1 OpenCL Device: GPU Samples/sec > Higher Is Better R9 Fury: ROCm 1.4 ......... 82051996.27 |============================ RX 460: ROCm 1.4 .......... 28295516.33 |========== RX 480: ROCm 1.4 .......... 59296261.87 |==================== RX 460: AMDGPU-PRO 16.50 .. 35552080.15 |============ RX 480: AMDGPU-PRO 16.50 .. 81101281.90 |============================ R9 Fury: AMDGPU-PRO 16.50 . 107202116.40 |===================================== LuxMark 3.0 OpenCL Device: GPU - Scene: Hotel Score > Higher Is Better R9 Fury: ROCm 1.4 ......... 1201 |======================= RX 460: ROCm 1.4 .......... 381 |======= RX 480: ROCm 1.4 .......... 987 |================== RX 460: AMDGPU-PRO 16.50 .. 897 |================= RX 480: AMDGPU-PRO 16.50 .. 2399 |============================================= R9 Fury: AMDGPU-PRO 16.50 . 2402 |============================================= LuxMark 3.0 OpenCL Device: GPU - Scene: Microphone Score > Higher Is Better R9 Fury: ROCm 1.4 ......... 5695 |================================= RX 460: AMDGPU-PRO 16.50 .. 2623 |=============== RX 480: AMDGPU-PRO 16.50 .. 6924 |========================================= R9 Fury: AMDGPU-PRO 16.50 . 7681 |============================================= LuxMark 3.0 OpenCL Device: GPU - Scene: Luxball HDR Score > Higher Is Better R9 Fury: ROCm 1.4 ......... 11995 |=========================== RX 460: ROCm 1.4 .......... 3664 |======== RX 480: ROCm 1.4 .......... 9196 |===================== RX 460: AMDGPU-PRO 16.50 .. 5547 |============= RX 480: AMDGPU-PRO 16.50 .. 14066 |================================ R9 Fury: AMDGPU-PRO 16.50 . 19394 |============================================