HelloSun commited on
Commit
f548785
·
verified ·
1 Parent(s): e9456c7

Clarify which EPQ version (v1/v2) is active for Flint

Browse files
Files changed (1) hide show
  1. results/test-status.md +6 -0
results/test-status.md CHANGED
@@ -29,6 +29,12 @@ llama.cpp base: `970b51ce7c9ba5fff0726c5e9994f1c2f0e4eedc` plus `flint-epq.patch
29
  5. **`tools/runmem_cgroup.py` and the cgroup claim were presented as usable.** Now labelled untested.
30
  6. **Earlier tier throughput was from single runs** and varied by ~±15 % between runs. Replaced with medians of 3 runs.
31
 
 
 
 
 
 
 
32
  ## Known limits and open items
33
 
34
  - EPQ output is not bit-identical to native llama.cpp. Differences come from float summation order and are amplified by 8-bit activation quantization; native llama.cpp itself shows the same magnitude of KLD under a ubatch change (control above). No functional mismatch was found.
 
29
  5. **`tools/runmem_cgroup.py` and the cgroup claim were presented as usable.** Now labelled untested.
30
  6. **Earlier tier throughput was from single runs** and varied by ~±15 % between runs. Replaced with medians of 3 runs.
31
 
32
+ ## EPQ version coverage
33
+
34
+ - v1 (expert paging, mode C): active in all runs.
35
+ - v2 preload: code present, requires an arena of at least 28.6 GiB; never triggered in these tests (the 24 GiB run had 10,082 slots, below the 12,032 needed).
36
+ - v2 row-parallel: Nemotron no-gate only; inactive for Flint.
37
+
38
  ## Known limits and open items
39
 
40
  - EPQ output is not bit-identical to native llama.cpp. Differences come from float summation order and are amplified by 8-bit activation quantization; native llama.cpp itself shows the same magnitude of KLD under a ubatch change (control above). No functional mismatch was found.