Latest HPC Challenge Numademo Benchmarks Latest numademo system tests by Michael Larabel. ACML 5.3.0: Standard: Processor: 32 x AMD Opteron 6380 @ 2.50GHz (256 Cores), Motherboard: Supermicro H8QGL, Chipset: AMD ATI RD890 + SB7x0/SB8x0/SB9x0, Memory: 1032192MB, Disk: 2 x 256GB Samsung SSD 840, Graphics: Matrox s MGA G200eW WPCM450, Network: Intel 82576 Gigabit Connection OS: CentOS 6.4, Kernel: 3.10.10-advanced+ (x86_64), Compiler: GCC 4.4.7 20120313 + Open64 4.5.2.1 + PGI C-C++ Workstation, File-System: nfs ACML 5.3.0: One Process Per Node: Processor: 32 x AMD Opteron 6380 @ 2.50GHz (256 Cores), Motherboard: Supermicro H8QGL, Chipset: AMD ATI RD890 + SB7x0/SB8x0/SB9x0, Memory: 1032192MB, Disk: 2 x 256GB Samsung SSD 840, Graphics: Matrox s MGA G200eW WPCM450, Network: Intel 82576 Gigabit Connection OS: CentOS 6.4, Kernel: 3.10.10-advanced+ (x86_64), Compiler: GCC 4.4.7 20120313 + Open64 4.5.2.1 + PGI C-C++ Workstation, File-System: nfs CentOS ATLAS: Standard: Processor: 32 x AMD Opteron 6380 @ 2.50GHz (256 Cores), Motherboard: Supermicro H8QGL, Chipset: AMD RD890 + SB7x0/SB8x0/SB9x0, Memory: 1032192MB, Disk: 2 x 256GB Samsung SSD 840, Graphics: Matrox s MGA G200eW WPCM450, Network: Intel 82576 Gigabit Connection OS: CentOS 6.5, Kernel: 3.10.23-advanced+ (x86_64), Compiler: GCC 4.4.7 20120313 + Open64 4.5.2.1 + PGI C-C++ Workstation, File-System: nfs CentOS ATLAS: One Process Per Node: Processor: 32 x AMD Opteron 6380 @ 2.50GHz (256 Cores), Motherboard: Supermicro H8QGL, Chipset: AMD RD890 + SB7x0/SB8x0/SB9x0, Memory: 1032192MB, Disk: 2 x 256GB Samsung SSD 840, Graphics: Matrox s MGA G200eW WPCM450, Network: Intel 82576 Gigabit Connection OS: CentOS 6.5, Kernel: 3.10.23-advanced+ (x86_64), Compiler: GCC 4.4.7 20120313 + Open64 4.5.2.1 + PGI C-C++ Workstation, File-System: nfs HPC Challenge 1.4.3 Test / Class: G-HPL GFLOPS > Higher Is Better ACML 5.3.0: Standard ............... 1485.27000 |============================== ACML 5.3.0: One Process Per Node ... 110.07900 |== CentOS ATLAS: Standard ............. 104.83300 |== CentOS ATLAS: One Process Per Node . 6.44425 | HPC Challenge 1.4.3 Test / Class: G-Ffte GFLOPS > Higher Is Better ACML 5.3.0: Standard ............... 6.94872 |================================= ACML 5.3.0: One Process Per Node ... 3.07791 |=============== CentOS ATLAS: Standard ............. 0.61248 |=== CentOS ATLAS: One Process Per Node . 3.04804 |============== HPC Challenge 1.4.3 Test / Class: EP-DGEMM GFLOPS > Higher Is Better ACML 5.3.0: Standard ............... 8.57251 |================ ACML 5.3.0: One Process Per Node ... 17.26010 |================================ CentOS ATLAS: Standard ............. 0.47812 |= CentOS ATLAS: One Process Per Node . 0.87297 |== HPC Challenge 1.4.3 Test / Class: G-Ptrans GB/s > Higher Is Better ACML 5.3.0: Standard ............... 1.70028 |=============================== ACML 5.3.0: One Process Per Node ... 0.92995 |================= CentOS ATLAS: Standard ............. 1.83521 |================================= CentOS ATLAS: One Process Per Node . 0.94857 |================= HPC Challenge 1.4.3 Test / Class: EP-STREAM Triad GB/s > Higher Is Better ACML 5.3.0: Standard ............... 1.40679 |====== ACML 5.3.0: One Process Per Node ... 7.15146 |================================= CentOS ATLAS: Standard ............. 1.51252 |======= CentOS ATLAS: One Process Per Node . 7.02175 |================================ HPC Challenge 1.4.3 Test / Class: G-Random Access GUP/s > Higher Is Better ACML 5.3.0: Standard ............... 0.08525 |================================= ACML 5.3.0: One Process Per Node ... 0.03849 |=============== CentOS ATLAS: Standard ............. 0.04144 |================ CentOS ATLAS: One Process Per Node . 0.04014 |================ HPC Challenge 1.4.3 Test / Class: Random Ring Latency usecs < Lower Is Better ACML 5.3.0: Standard ............... 7.64532 |================================= ACML 5.3.0: One Process Per Node ... 2.65930 |=========== CentOS ATLAS: Standard ............. 7.70645 |================================= CentOS ATLAS: One Process Per Node . 2.76840 |============ HPC Challenge 1.4.3 Test / Class: Random Ring Bandwidth GB/s > Higher Is Better ACML 5.3.0: Standard ............... 0.00759 |= ACML 5.3.0: One Process Per Node ... 0.20727 |================================= CentOS ATLAS: Standard ............. 0.00666 |= CentOS ATLAS: One Process Per Node . 0.20074 |================================ HPC Challenge 1.4.3 Test / Class: Max Ping Pong Bandwidth MB/s > Higher Is Better ACML 5.3.0: Standard ............... 25003.30 |================================ ACML 5.3.0: One Process Per Node ... 469.27 |= CentOS ATLAS: Standard ............. 1839.00 |== CentOS ATLAS: One Process Per Node . 470.60 |=