ywkang/kernbench2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add SchedulerV2 (pe_accel), DPPolicy overrides, and new benchmarks

Browse Source 
- Add cycle-accurate PE accelerator scheduler (SchedulerV2) with tiled
  GEMM/Math pipelines (DMA_IN → GEMM → MATH → DMA_WB)
- Add DPPolicy num_pes/num_cubes/num_sips overrides for single-PE testing
- Support tuple target_pe for targeting specific PE subsets
- Add gemm_single_pe and gpt3_qkv benchmarks
- Switch default topology to pe_scheduler_v2

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Yangwook
2026-03-26 23:18:49 -07:00
parent 63669f82cb
commit 114510d4b9
22 changed files with 1822 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files
topology.yaml
+1 -1
View File
@@ -61,7 +61,7 @@ cube:
pe_template:
components:
pe_cpu: { kind: pe_cpu, impl: pe_cpu_v1, attrs: { overhead_ns: 2.0 } }
pe_scheduler: { kind: pe_scheduler, impl: pe_scheduler_v1, attrs: { overhead_ns: 1.0 } }
pe_scheduler: { kind: pe_scheduler, impl: pe_scheduler_v2, attrs: { overhead_ns: 1.0 } }
pe_dma: { kind: pe_dma, impl: pe_dma_v1, attrs: { rd_engines: 1, wr_engines: 1 } }
pe_gemm: { kind: pe_gemm, impl: pe_gemm_v1, attrs: { overhead_ns: 0.0, shared_resource: accel_slot, peak_tflops_f16: 8.0 } }
pe_math: { kind: pe_math, impl: pe_math_v1, attrs: { overhead_ns: 0.0, shared_resource: accel_slot } }