5xa5000kw-dl580-3-4-pytorch-batchsize512

KVM testing on Ubuntu 24.04 via the Phoronix Test Suite.

HTML result view exported from: https://openbenchmarking.org/result/2410077-NE-5XA5000KW77.

5xa5000kw-dl580-3-4-pytorch-batchsize512ProcessorMotherboardChipsetMemoryDiskGraphicsAudioNetworkOSKernelDisplay ServerDisplay DriverOpenCLCompilerFile-SystemScreen ResolutionSystem Layer5x a5000kw dl580-3-4 PyTorch Batch size=5124 x Intel Xeon E7-4880 v2 (60 Cores / 120 Threads)QEMU Standard PC (Q35 + ICH9 2009) (edk2-20240813-1.fc40 BIOS)Intel 82G33/G31/P35/P31 + ICH916 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 16 GB + 4 GB RAM21GB VIRTUAL-DISKRed Hat QXL paravirtual graphic card 22GBQEMU Generic2 x Red Hat Virtio 1.0 deviceUbuntu 24.046.8.0-45-generic (x86_64)X ServerNVIDIAOpenCL 3.0 CUDA 12.4.131GCC 13.2.0 + CUDA 12.0ext41024x768KVMOpenBenchmarking.org- Transparent Huge Pages: madvise- CPU Microcode: 0x715- Python 3.12.3- gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Mitigation of PTE Inversion; VMX: flush not necessary SMT vulnerable + mds: Mitigation of Clear buffers; SMT Host state unknown + meltdown: Mitigation of PTI + mmio_stale_data: Unknown: No mitigations + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines; IBPB: conditional; IBRS_FW; STIBP: conditional; RSB filling; PBRSB-eIBRS: Not affected; BHI: Retpoline + srbds: Not affected + tsx_async_abort: Not affected

5xa5000kw-dl580-3-4-pytorch-batchsize512pytorch: CPU - 512 - ResNet-50Standard ErrorStandard Deviation5x a5000kw dl580-3-4 PyTorch Batch size=5129.060.102.13%OpenBenchmarking.org

PyTorch

Device: CPU - Batch Size: 512 - Model: ResNet-50

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 512 - Model: ResNet-505x a5000kw dl580-3-4 PyTorch Batch size=5123691215SE +/- 0.10, N = 49.06MIN: 7.35 / MAX: 10.17

Phoronix Test Suite v10.8.5