P2P tests with 4090's

base_drivers.txt

[P2P (Peer-to-Peer) GPU Bandwidth Latency Test]
Device: 0, NVIDIA GeForce RTX 4090, pciBusID: 1, pciDeviceID: 0, pciDomainID:0
Device: 1, NVIDIA GeForce RTX 4090, pciBusID: 2, pciDeviceID: 0, pciDomainID:0
Device=0 CANNOT Access Peer Device=1
Device=1 CANNOT Access Peer Device=0

***NOTE: In case a device doesn't have P2P access to other one, it falls back to normal memcopy procedure.
So you can see lesser Bandwidth (GB/s) and unstable Latency (us) in those cases.

P2P Connectivity Matrix
     D\D     0     1
     0       1     0
     1       0     1
Unidirectional P2P=Disabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 913.74  11.97 
     1  12.14 924.01 
Unidirectional P2P=Enabled Bandwidth (P2P Writes) Matrix (GB/s)
   D\D     0      1 
     0 915.35  12.00 
     1  12.15 924.56 
Bidirectional P2P=Disabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 919.68  16.79 
     1  16.80 924.28 
Bidirectional P2P=Enabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 919.62  16.79 
     1  16.83 923.99 
P2P=Disabled Latency Matrix (us)
   GPU     0      1 
     0   1.32  10.94 
     1  10.54   1.29 

   CPU     0      1 
     0   1.40   4.08 
     1   4.14   1.31 
P2P=Enabled Latency (P2P Writes) Matrix (us)
   GPU     0      1 
     0   1.32  10.28 
     1  10.54   1.27 

   CPU     0      1 
     0   1.35   4.16 
     1   4.15   1.30 

NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.

new_drivers.txrt

[P2P (Peer-to-Peer) GPU Bandwidth Latency Test]
Device: 0, NVIDIA GeForce RTX 4090, pciBusID: 1, pciDeviceID: 0, pciDomainID:0
Device: 1, NVIDIA GeForce RTX 4090, pciBusID: 2, pciDeviceID: 0, pciDomainID:0
Device=0 CANNOT Access Peer Device=1
Device=1 CANNOT Access Peer Device=0

***NOTE: In case a device doesn't have P2P access to other one, it falls back to normal memcopy procedure.
So you can see lesser Bandwidth (GB/s) and unstable Latency (us) in those cases.

P2P Connectivity Matrix
     D\D     0     1
     0       1     0
     1       0     1
Unidirectional P2P=Disabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 913.21  12.04 
     1  12.05 923.46 
Unidirectional P2P=Enabled Bandwidth (P2P Writes) Matrix (GB/s)
   D\D     0      1 
     0 915.35  11.99 
     1  12.06 925.10 
Bidirectional P2P=Disabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 919.66  16.82 
     1  16.76 924.82 
Bidirectional P2P=Enabled Bandwidth Matrix (GB/s)
   D\D     0      1 
     0 918.75  16.82 
     1  16.86 924.28 
P2P=Disabled Latency Matrix (us)
   GPU     0      1 
     0   1.39  12.18 
     1  11.54   1.28 

   CPU     0      1 
     0   1.37   4.23 
     1   4.35   1.36 
P2P=Enabled Latency (P2P Writes) Matrix (us)
   GPU     0      1 
     0   1.37  10.28 
     1  11.12   1.27 

   CPU     0      1 
     0   1.50   4.48 
     1   4.30   1.36 

NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.