HPCDIY-ERM1GPU4TS(こちら)に GeForce RTX 3090 を4枚実装して、nvidia-smi と deviceQuery を実行してみた。

nvidia-smi

dl@dl-machine:~$ nvidia-smi
Thu Oct 22 17:12:19 2020       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 455.23.05    Driver Version: 455.23.05    CUDA Version: 11.1     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  GeForce RTX 3090    On   | 00000000:81:00.0 Off |                  N/A |
| 63%   57C    P2   194W / 350W |  23629MiB / 24268MiB |     23%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
|   1  GeForce RTX 3090    On   | 00000000:82:00.0 Off |                  N/A |
| 37%   38C    P8    22W / 350W |    373MiB / 24268MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
|   2  GeForce RTX 3090    On   | 00000000:C1:00.0 Off |                  N/A |
| 46%   43C    P8    22W / 350W |    373MiB / 24268MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
|   3  GeForce RTX 3090    On   | 00000000:C2:00.0 Off |                  N/A |
| 47%   44C    P8    22W / 350W |    373MiB / 24268MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A      2009      G   /usr/lib/xorg/Xorg                  4MiB |
|    0   N/A  N/A      7656      C   python                          23621MiB |
|    1   N/A  N/A      2009      G   /usr/lib/xorg/Xorg                  4MiB |
|    1   N/A  N/A      7656      C   python                            365MiB |
|    2   N/A  N/A      2009      G   /usr/lib/xorg/Xorg                  4MiB |
|    2   N/A  N/A      7656      C   python                            365MiB |
|    3   N/A  N/A      2009      G   /usr/lib/xorg/Xorg                  4MiB |
|    3   N/A  N/A      7656      C   python                            365MiB |
+-----------------------------------------------------------------------------+
tf_cnn_benchmarks を実行中です。

deviceQuery
最後に表示される Peer access に注目してください。これが No だと、GPUが多い場合(>4くらい)並列性能が出ないと言われています。
ちなみに GeForce RTX 2080 ti, Titan RTX は No
Quadro, Tesla は Yes

dl@dl-machine:~$ deviceQuery 
deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 4 CUDA Capable device(s)

Device 0: "GeForce RTX 3090"
  CUDA Driver Version / Runtime Version          11.1 / 11.1
  CUDA Capability Major/Minor version number:    8.6
  Total amount of global memory:                 24268 MBytes (25447170048 bytes)
  (82) Multiprocessors, (128) CUDA Cores/MP:     10496 CUDA Cores
  GPU Max Clock rate:                            1695 MHz (1.70 GHz)
  Memory Clock rate:                             9751 Mhz
  Memory Bus Width:                              384-bit
  L2 Cache Size:                                 6291456 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total shared memory per multiprocessor:        102400 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1536
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Managed Memory:                Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 129 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

Device 1: "GeForce RTX 3090"
  CUDA Driver Version / Runtime Version          11.1 / 11.1
  CUDA Capability Major/Minor version number:    8.6
  Total amount of global memory:                 24268 MBytes (25447170048 bytes)
  (82) Multiprocessors, (128) CUDA Cores/MP:     10496 CUDA Cores
  GPU Max Clock rate:                            1695 MHz (1.70 GHz)
  Memory Clock rate:                             9751 Mhz
  Memory Bus Width:                              384-bit
  L2 Cache Size:                                 6291456 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total shared memory per multiprocessor:        102400 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1536
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Managed Memory:                Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 130 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

Device 2: "GeForce RTX 3090"
  CUDA Driver Version / Runtime Version          11.1 / 11.1
  CUDA Capability Major/Minor version number:    8.6
  Total amount of global memory:                 24268 MBytes (25447170048 bytes)
  (82) Multiprocessors, (128) CUDA Cores/MP:     10496 CUDA Cores
  GPU Max Clock rate:                            1695 MHz (1.70 GHz)
  Memory Clock rate:                             9751 Mhz
  Memory Bus Width:                              384-bit
  L2 Cache Size:                                 6291456 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total shared memory per multiprocessor:        102400 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1536
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Managed Memory:                Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 193 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

Device 3: "GeForce RTX 3090"
  CUDA Driver Version / Runtime Version          11.1 / 11.1
  CUDA Capability Major/Minor version number:    8.6
  Total amount of global memory:                 24268 MBytes (25447170048 bytes)
  (82) Multiprocessors, (128) CUDA Cores/MP:     10496 CUDA Cores
  GPU Max Clock rate:                            1695 MHz (1.70 GHz)
  Memory Clock rate:                             9751 Mhz
  Memory Bus Width:                              384-bit
  L2 Cache Size:                                 6291456 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total shared memory per multiprocessor:        102400 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1536
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Managed Memory:                Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 194 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
> Peer access from GeForce RTX 3090 (GPU0) -> GeForce RTX 3090 (GPU1) : No
> Peer access from GeForce RTX 3090 (GPU0) -> GeForce RTX 3090 (GPU2) : No
> Peer access from GeForce RTX 3090 (GPU0) -> GeForce RTX 3090 (GPU3) : No
> Peer access from GeForce RTX 3090 (GPU1) -> GeForce RTX 3090 (GPU0) : No
> Peer access from GeForce RTX 3090 (GPU1) -> GeForce RTX 3090 (GPU2) : No
> Peer access from GeForce RTX 3090 (GPU1) -> GeForce RTX 3090 (GPU3) : No
> Peer access from GeForce RTX 3090 (GPU2) -> GeForce RTX 3090 (GPU0) : No
> Peer access from GeForce RTX 3090 (GPU2) -> GeForce RTX 3090 (GPU1) : No
> Peer access from GeForce RTX 3090 (GPU2) -> GeForce RTX 3090 (GPU3) : No
> Peer access from GeForce RTX 3090 (GPU3) -> GeForce RTX 3090 (GPU0) : No
> Peer access from GeForce RTX 3090 (GPU3) -> GeForce RTX 3090 (GPU1) : No
> Peer access from GeForce RTX 3090 (GPU3) -> GeForce RTX 3090 (GPU2) : No

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 11.1, CUDA Runtime Version = 11.1, NumDevs = 4
Result = PASS