Skip to content

Commit

Permalink
guess warpRegAllocUnit
Browse files Browse the repository at this point in the history
  • Loading branch information
dpvanbalen committed Nov 20, 2023
1 parent 2edafa6 commit a140e84
Showing 1 changed file with 5 additions and 5 deletions.
10 changes: 5 additions & 5 deletions src/Foreign/CUDA/Analysis/Device.chs
Original file line number Diff line number Diff line change
Expand Up @@ -401,11 +401,11 @@ deviceResources = resources . computeCapability
, maxRegPerThread = 255
, sharedMemAllocUnit = 128
, warpAllocUnit = 4
, warpRegAllocUnit = _
, warpRegAllocUnit = 256
, maxGridsPerDevice = 128
}

Compute 8 9 -> DeviceResources
Compute 8 9 -> DeviceResources
{ threadsPerWarp = 32
, coresPerMP = 64
, warpsPerMP = 48
Expand All @@ -420,11 +420,11 @@ deviceResources = resources . computeCapability
, maxRegPerThread = 255
, sharedMemAllocUnit = 128
, warpAllocUnit = 4
, warpRegAllocUnit = _
, warpRegAllocUnit = 256
, maxGridsPerDevice = 128
}

Compute 9 0 -> trace "*** Warning: Compute Capability 9.0 has Thread Block Clusters, which this occupancy calculation might not support" $ DeviceResources
Compute 9 0 -> trace "*** Warning: Compute Capability 9.0 has Thread Block Clusters, which this occupancy calculation might not support" $ DeviceResources
{ threadsPerWarp = 32
, coresPerMP = 64
, warpsPerMP = 64
Expand All @@ -439,7 +439,7 @@ deviceResources = resources . computeCapability
, maxRegPerThread = 255
, sharedMemAllocUnit = 128
, warpAllocUnit = 4
, warpRegAllocUnit = _
, warpRegAllocUnit = 256
, maxGridsPerDevice = 128
}

Expand Down

0 comments on commit a140e84

Please sign in to comment.