admin/matmul-analysis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

T_cmd=0 仲裁改为策略优先 MergeBatch: K截断即优先(指令/命令次数省b0倍,未量化收益), L1绑定恒劣; 路由层以策略覆盖时延模型差额并显式标注

Browse Source
This commit is contained in:
admin
2026-09-04 10:47:02 +08:00
parent a071d87831
commit 5051e00fa0
3 changed files with 59 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
BMM/BMM_Theory/bmm_theory/router.py
View File

@@ -7,7 +7,9 @@
仲裁规则 (v1.1 §4.5 统一分界):
MergeBatch 最优 <=> K截断(k_L1=K) 且 b_core > b0*(T_comp+T_write)/T_cmd
L1 绑定时 MergeBatch 恒劣于 IterBatch;
两分支同时合法时用端到端时延模型 T_total 仲裁.
两分支同时合法时用端到端时延模型 T_total 仲裁;
例外: T_cmd<=0 (命令时延不可量化/未标定) 时, 指令级收益未建模,
按既定策略: K截断即可优先 MergeBatch (覆盖时延模型仅来自 drain/冗余的差额).
3. StreamK 检查: P <= C/2 且满足切K条件 -> StreamK (B/M/N 买不满时买 K)
4. 兜底: ASW_Basic 切 M/N (含降核模式)
"""
@@ -90,14 +92,28 @@ class BranchRouter:
t_ib = ib.timing.t_total
lat_win = self.merge_batch.name if t_mb <= t_ib else self.iter_batch.name
win = self.merge_batch.name if mb_win else self.iter_batch.name
arbitration = (
f"两分支均合法, 仲裁: "
f"[分界条件] MergeBatch最优={mb_win} ({detail}); "
f"[时延模型] T_MergeBatch={t_mb*1e6:.2f}us vs T_IterBatch={t_ib*1e6:.2f}us -> {lat_win}更优; "
f"[裁决] {win}" + ("" if win == lat_win else f" (分界条件与时延模型不一致, 以时延模型为准: {lat_win})")
)
if win != lat_win:
win = lat_win # 时延模型为最终裁决
# 冲突解决: 默认"时延模型为最终裁决"; 例外是 T_cmd<=0 且分界条件判
# MergeBatch 胜 (K截断) 的情形 —— 此时时延模型不含指令级收益
# (MergeBatch 搬移命令数/主机指令数省 b0 倍, 未量化), 按既定策略
# 优先 MergeBatch (时延模型内的差额只是 drain 惩罚/冗余, 方向已知小量).
policy_merge = (self.spec.t_cmd <= 0 and mb_win and
win != lat_win and win == self.merge_batch.name)
if policy_merge:
arbitration = (
f"两分支均合法, 仲裁: "
f"[分界条件] MergeBatch最优={mb_win} ({detail}); "
f"[时延模型] T_MergeBatch={t_mb*1e6:.2f}us vs T_IterBatch={t_ib*1e6:.2f}us -> {lat_win}更优; "
f"[裁决] {win} (T_cmd<=0 策略: 命令/指令级收益未建模, 时延模型差异仅来自 "
f"drain/冗余, 以 MergeBatch 优先策略裁决)")
else:
arbitration = (
f"两分支均合法, 仲裁: "
f"[分界条件] MergeBatch最优={mb_win} ({detail}); "
f"[时延模型] T_MergeBatch={t_mb*1e6:.2f}us vs T_IterBatch={t_ib*1e6:.2f}us -> {lat_win}更优; "
f"[裁决] {win}" + ("" if win == lat_win else f" (分界条件与时延模型不一致, 以时延模型为准: {lat_win})")
)
if win != lat_win:
win = lat_win # 时延模型为最终裁决
elif any(capable.values()):
win = next(n for n, v in capable.items() if v)
arbitration = f"仅 {win} 条件满足"