MCPcopy Create free account
hub / github.com/evilsocket/cake / make_gpu_tensor

Function make_gpu_tensor

cake-core/benches/bench_utils.rs:17–21  ·  view source on GitHub ↗
(shape: &[usize], seed: u64)

Source from the content-addressed store, hash-verified

15}
16
17fn make_gpu_tensor(shape: &[usize], seed: u64) -> Tensor {
18 super::bench_helpers::make_tensor(shape, seed)
19 .to_device(&gpu_device())
20 .unwrap()
21}
22
23#[divan::bench(args = [256, 1024, 4096])]
24fn rms_norm_forward(bencher: divan::Bencher, hidden_size: usize) {

Callers 15

fused_silu_mul_gpuFunction · 0.85
fused_rms_norm_gated_gpuFunction · 0.85
unfused_silu_mul_gpuFunction · 0.85
fused_exp_mul_gpuFunction · 0.85
unfused_exp_mul_gpuFunction · 0.85
fused_sub_mul_gpuFunction · 0.85
unfused_sub_mul_gpuFunction · 0.85
fused_add3_gpuFunction · 0.85
fused_add_scaled_gpuFunction · 0.85
fused_add_rms_norm_gpuFunction · 0.85

Calls 2

gpu_deviceFunction · 0.85
make_tensorFunction · 0.70

Tested by

no test coverage detected