termoshtt
einsum-derive
BlogDocsChangelog

Call GEMM for `ab,bc->ac` and DOT for `a,a->`

#24
Comparing
call-blas-routines
(
5cb284d
) with
main
(
554c326
)
CodSpeed Performance Gauge
0%
Untouched
12

Benchmarks

Passed

ij_jk[4]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
+1%
5.3 µs5.3 µs
ij_jk_kl[4]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
7 µs6.9 µs
ij_jk[8]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
8.2 µs8.2 µs
ij_jk_kl[8]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
12.4 µs12.4 µs
ij_jk[16]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
24.1 µs24 µs
ij_jk[64]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
884.3 µs883.5 µs
ij_jk_kl[16]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
43.3 µs43.2 µs
ij_jk[32]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
117.5 µs117.5 µs
ij_jk_kl[32]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
226.4 µs226.4 µs
ij_jk_kl[64]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
1.7 ms1.7 ms
ij_jk[128]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
8.5 ms8.5 ms
ij_jk_kl[128]
einsum-derive/benches/einsum.rs::einsum::einsum_bench::einsum
CodSpeed Performance Gauge
0%
16.7 ms16.7 ms

Commits

Click on a commit to change the comparison range
Base
main
554c326
+0.16%
RawSubscripts correspond to GEMM and DOT
5cb284d
3 years ago
by termoshtt
© 2025 CodSpeed Technology
Home Terms Privacy Docs