Open kmp5VT opened 2 months ago
The contractions in your example won't work out of the box with cuTensor (see table): https://docs.nvidia.com/cuda/cutensor/latest/api/cutensor.html#contraction-operations
Should we automatically convert F32 to CF32?
Describe the bug
Some mixed element type contractions are failing when trying to contract with cuTENSOR.
To reproduce
The Minimal Working Example (MWE) for this bug:
All of these contractions have no issues when not wrapping the
CuArray
s inCuTensor
sManifest.toml
``` [052768ef] CUDA v5.3.1 [46192b85] GPUArraysCore v0.1.6 [011b41b2] cuTENSOR v2.1.0 ```
Version info
Details on Julia:
Details on CUDA: