NVIDIA Tesla V100

V100 のメモリバンド幅が非常に大きい。
ちなみに P100 & K40m と GTX 1080Ti & V100 では実行しているマシンが異なるので、Host と Device 間のバンド幅に関しては参考程度に。

# ./bandwidthTest --device=1
[CUDA Bandwidth Test] - Starting...
Running on...

Device 1: Tesla K40m
Quick Mode

Host to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 9840.9

Device to Host Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 8315.7

Device to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 182612.1


# ./bandwidthTest -device=1
[CUDA Bandwidth Test] - Starting...
Running on...

Device 1: GeForce GTX 1080 Ti
Quick Mode

Host to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 11685.8

Device to Host Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 12856.6

Device to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 346895.8


# ./bandwidthTest --device=0
[CUDA Bandwidth Test] - Starting...
Running on...

Device 0: Tesla P100-PCIE-16GB
Quick Mode

Host to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 9215.2

Device to Host Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 7586.4

Device to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 402852.2


# ./bandwidthTest -device=0
[CUDA Bandwidth Test] - Starting...
Running on...

Device 0: Tesla V100-PCIE-16GB
Quick Mode

Host to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 11730.0

Device to Host Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 12879.7

Device to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 741333.3