mukesh 0ef4fde5d8 gqa(decode-4cases): Case 1 — Cube-SP × PE-TP (inter-CUBE lrab only; PE-TP B=1 waste) (5C.A) ...

Adds the fourth and final 4-cases panel: KV split S_kv-wise across
the 8 cubes (cube=row_wise, S_local = S_kv/C), replicated within
each cube (pe=replicate). PE-TP at B=1 means only PE 0 of each cube
has work; PEs 1-7 early-return (slide-11 PE-TP-at-B=1 waste). No
intra-CUBE comm; inter-CUBE 8-way reduce reuses the lrab-adapted
center-root pattern (root cube 6) — same structural cost as Case 4's
inter-CUBE phase (21 ipcq_copy). 16 tests pass (4 per case).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

2026-06-15 15:30:00 -07:00

..

test_gqa_decode_long_2d_reduce.py

gqa: single-KV-group LLaMA-3.1-70B prefill milestone (Increments 1-5)

2026-06-15 13:42:31 -07:00

test_gqa_decode_mc.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_decode_opt2.py

gqa(adr-0065): N4 — opt2 numeric parity vs opt3 in data mode

2026-06-10 23:13:16 -07:00

test_gqa_decode_sp.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_decode.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_long_context.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_prefill_long_c8_snake.py

gqa: single-KV-group LLaMA-3.1-70B prefill milestone (Increments 1-5)

2026-06-15 13:42:31 -07:00

test_gqa_prefill_long_pe_sp.py

gqa: single-KV-group LLaMA-3.1-70B prefill milestone (Increments 1-5)

2026-06-15 13:42:31 -07:00

test_gqa_prefill_long_tile_ring.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_prefill_ring.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_prefill.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_scoped_writeback.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_short_context.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_gqa_tile_sweep.py

gqa: tile-granular Ring KV (P3c) + rename to gqa_attention_* + ADR-0060/62/63/64 → Accepted

2026-06-10 16:20:00 -07:00

test_milestone_gqa_decode_4cases.py

gqa(decode-4cases): Case 1 — Cube-SP × PE-TP (inter-CUBE lrab only; PE-TP B=1 waste) (5C.A)

2026-06-15 15:30:00 -07:00

test_milestone_gqa_headline.py

bench(milestone-gqa-headline): drop misleading single_user_/multi_user_ panels

2026-06-15 14:50:25 -07:00

test_milestone_gqa_single_kv_group_prefill_panel.py

bench(milestone-gqa-headline): drop misleading single_user_/multi_user_ panels

2026-06-15 14:50:25 -07:00