# A100 Benchmark

### host<->device bandwidth

Device 0: A100-SXM4-40GB
 Quick Mode

 Host to Device Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)	Bandwidth(GB/s)
   32000000			24.6

 Device to Host Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)	Bandwidth(GB/s)
   32000000			25.9

 Device to Device Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)	Bandwidth(GB/s)
   32000000			1091.6

### inter-GPU bandwidth

P2P Connectivity Matrix
     D\D     0     1
     0	     1     1
     1	     1     1
Unidirectional P2P=Disabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 1276.55  17.38 
     1  17.51 1279.69 
Unidirectional P2P=Enabled Bandwidth (P2P Writes) Matrix (GB/s)
   D\D     0      1 
     0 1278.64 265.74 
     1 264.77 1288.13 
Bidirectional P2P=Disabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 1288.13  17.78 
     1  18.26 1301.54 
Bidirectional P2P=Enabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 1302.08 515.24 
     1 528.83 1304.80 
P2P=Disabled Latency Matrix (us)
   GPU     0      1 
     0   2.33  24.14 
     1  24.45   2.81 

   CPU     0      1 
     0   3.03   9.05 
     1   9.02   3.00 
P2P=Enabled Latency (P2P Writes) Matrix (us)
   GPU     0      1 
     0   2.33   3.02 
     1   2.98   2.74 

   CPU     0      1 
     0   3.06   2.38 
     1   2.45   3.06