Skip to content

Truncate — fwd (cuda, torch.float32, wrapper, default)

← Back to compiled benchmarks

Result

Field Value
Operator Truncate
Category linop
Method fwd
Device cuda
Dtype torch.float32
Compile level wrapper
Compile config default
Status ok
Class compiles_slower
Graph breaks 0
Recompiles —
Eager time 10.72 us
Compiled time 16.84 us
Speedup 0.64x
Max abs diff 0.0
Max rel diff 0.0
Degraded False

Tensors

Tensor Shape Dtype Device Numel Bytes View Contiguous Stride
input [4, 5] torch.float32 cuda:0 20 80.00 B False True [5, 1]
output [4, 3] torch.float32 cuda:0 12 48.00 B True False [5, 1]

Notes

—