Commit Graph

17 Commits

Author SHA1 Message Date
b9e07edc1d Fix #27-#30: dtype感知算力(Cube/AIV速率表) / GM首读下限与整芯片字节列口径+设计文档 / Fixpipe输出落点R4(整case驻留L2否则直写GM) / MergeBatch Cube公式复核注释
- docs/05_L2驻留GM读写与dtype算力口径_设计分析.md: R1-R6公理、S_A/S_B/S_C场景、输出落点R4、dtype速率表(白皮书出处+待标定假设)、逐分支GM/L2归属表 (issue#29/#30 先文档后代码)
- hardware: CUBE_DTYPE_FACTOR(f16/bf16=1, fp8=2x, fp4=4x, fp32=1/2假设) + AIV_DTYPE_FACTOR + q_cube/aiv_elem_rate (issue#28)
- 全分支 t_mmad/t_comp/drain/尾轮主导项/θ_c/R16 语义按输入dtype取算力; 混精度取慢侧; StreamK归约保持fp32(AIV fp32部分和)
- Fixpipe输出落点R4: to_l2 <=> V_in+V_out(+workspace)<=L2; 否则直写GM计入共享总线; ASW场景改S_A整case全驻留(原单batch驻留判定漏计整case输出累积逐出)
- 字节列统一整芯片口径(gm/l2/fix/cube_flops), dma_cmd_count注明单核; GM>=V_in不变量入测试; MergeBatch每步flops=2(b0M)(b0N)K公式注释显式化(#27复核与CSV一致无数值改动)
- tests 39->49 全过; 双压力seed7/6000+seed2024/4000: 0违规/0占位/0NaN/0GM<输入; examples三件套重生成且可复现0diff
2026-09-04 16:26:10 +08:00
9afe6eec02 Fix review issues #17-#22: 恢复 #11/#12/#14 (StreamK fixpipe 单次计账/K=1 AIV单缓冲/advice) + 占位方案不可评估 + 转置 dValue 判据三处同源(form c 双缓冲适配修复) + 恢复 #13/#15 回归测试 + 清理临时 csv/.gitignore + 文档同步 2026-09-03 21:10:05 +08:00
6d5e452fbc Update BMM_Theory: docs/02_分支理论/04_特殊分支.md 2026-09-03 12:44:10 +00:00
88b1e4946c Add BMM_Theory: docs/04_差异对照_bmmv3_vs_BMM_Theory.md 2026-09-03 12:35:49 +00:00
623f20259e Add BMM_Theory: docs/03_测评报告/ 软件复评报告 v2.0 (整改后复评 + issue #11-#16 闭环记录) 2026-09-03 20:06:55 +08:00
4bedf0a109 Fix review issues #11-#16: StreamK fixpipe 单次计账 / K=1 AIV单缓冲方案 / MergeBatch b0 L0A/L0B 上限+路由可行回退 / advice-StreamK / 输入校验 / .gitignore+死代码清理 2026-09-03 20:05:14 +08:00
593ffcf76f Update BMM_Theory: docs/01_软件架构.md (fix review issues #4-#10) 2026-09-03 11:34:29 +00:00
d36e224eb1 Add BMM_Theory: docs/03_测评报告/ 软件测评报告 v1.0 2026-09-03 18:56:03 +08:00
416362dec4 Add BMM_Theory: docs/02_分支理论/07_尾轮处理策略.md 2026-09-03 09:25:51 +00:00
47064c22de Add BMM_Theory: docs/02_分支理论/06_ASW_Basic分支.md 2026-09-03 09:25:50 +00:00
8664dd5e57 Add BMM_Theory: docs/02_分支理论/05_StreamK分支.md 2026-09-03 09:25:49 +00:00
a64fbd72a6 Add BMM_Theory: docs/02_分支理论/04_特殊分支.md 2026-09-03 09:25:48 +00:00
df035598b4 Add BMM_Theory: docs/02_分支理论/03_转Matmul分支.md 2026-09-03 09:25:47 +00:00
6808d5d7fc Add BMM_Theory: docs/02_分支理论/02_IterBatch分支.md 2026-09-03 08:10:33 +00:00
d00faa6502 Add BMM_Theory: docs/02_分支理论/01_MergeBatch分支.md 2026-09-03 08:10:31 +00:00
13c08a52ba Add BMM_Theory: docs/02_分支理论/00_总纲_分支决策树.md 2026-09-03 08:10:30 +00:00
a7b76289e1 Add BMM_Theory: docs/01_软件架构.md 2026-09-03 08:10:30 +00:00