Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Linear Layer Latency Inference on Llama-3-8B decoder block

153Latency (µs)

CodeGEMM

-211.562,249.224,7107,170.78Dec 19, 2025
Updated 4d ago

Evaluation Results

MethodLinks
2025.12
153
2025.12
163
2025.12
172
2025.12
190
2025.12
250
2025.12
332
2025.12
333
2025.12
336
2025.12
340
2025.12
405
2025.12
445
2025.12
491
2025.12
550
2025.12
646
2025.12
744
2025.12
794
2025.12
818
2025.12
909
2025.12
1,027
2025.12
1,027
2025.12
1,027
2025.12
1,027
2025.12
1,034
2025.12
1,360
2025.12
1,361
2025.12
1,364
2025.12
1,367
2025.12
1,416
2025.12
1,515
2025.12
1,554
2025.12
1,748
2025.12
1,991
2025.12
2,373
2025.12
2,959
2025.12
4,695
2025.12
9,267