| # | GEMM | K (in) | N (out) | Shape class | qweight bytes | 2-bit? |
|---|---|---|---|---|---|---|
| G4q | QKV-Q | 6144 | 3072 | M_in | 4.5 MB | YES |
| G4k | QKV-K | 6144 | 3072 | M_in | 4.5 MB | YES |
| G4v | QKV-V | 6144 | 3072 | M_in | 4.5 MB | YES |
| G6 | O proj | 3072 | 3072 | S (square) | 2.25 MB | YES |
| G7g | MLP gate | 3072 | 8192 | M_out | 6.0 MB | YES |
| G7u | MLP up | 3072 | 8192 | M_out | 6.0 MB | YES |
| G9 | MLP down | 8192 | 6144 | L | 12.0 MB | YES |
| G12 | LM head vocab | 256 | 128256 | V (skew) | 8.0 MB | YES |
| G8 | fused gather-dot (fp16 exact rescore, HBM) | 3072 | 512 rows top-K_C | 788 MB fp16 (HBM) | — | SILICON 8/8 (WIDE-256, 0xE4B7110E) |
g8_rescore_p.sv (the WIDE-256 gather+dot, 0xE4B7110E),
kv_journal.sv (per-branch ancestor K/V for depth-N), stage_io_buffer_ff.sv
(FF activation seams — kills the cross-SLR BRAM-strand bug class), and the batch path in
eagle4_drafter.sv (BODY_BATCH_B=2, sibling weight-reuse, building now as
0xE4B7110F). See docs/PDI24_GOLDEN_REFERENCE.md for the full phase→file map.
| Module | Role | Status | xsim | Source |
|---|---|---|---|---|
mac_2bit_shift_add.sv |
One 2-bit MAC cell, multiplier-free (LUT carry only) | ooc-synth ✓ | 1000/1000 | src/primitives/mac_2bit_shift_add.sv |
mac_2bit_mult.sv |
Same I/O, uses *; synth chooses LUT vs DSP58 |
ooc-synth ✓ | 1000/1000 | src/primitives/mac_2bit_mult.sv |
gptq2_dequant16.sv |
16-lane combinational unpack + zero subtract (consumes one int32 of qweight per cycle, produces 16 signed 3-bit weights) | ooc-synth ✓ | 1024/1024 | src/primitives/gptq2_dequant16.sv |
gemv_systolic_ws.sv |
Weight-stationary GEMV (parameterized K, N, LANES, DEPTH) | pending | — | (V5) |
gemv_systolic_os.sv |
Output-stationary GEMV | pending | — | (V6) |
gemv_systolic_is.sv |
Input-stationary GEMV | pending | — | (V7) |
gen_constfold_gemv.py |
Compile-time generator: weights baked into Verilog (no weight memory, no multiplier) | pending | — | (V8) |
The whole drafter array is built by stamping out one lane 128 times in parallel. Each lane owns one output column of every matrix. The diagram below shows what's inside one lane, and where each piece sits on the V-ladder.
What got built so far is two variants of a single 2-bit MAC cell. The two diagrams below trace the actual datapath each Vivado synthesis produced, with measured LUT counts per sub-block. Critical path (1.19–1.29 ns at 7 logic levels) shown as a red dashed overlay.
Why both still exist: at LANES≥16 systolic chains the explicit shift-add structure controls fanout (each lane is independent, no shared multiplier scheduling) and pairs cleanly with the LUT-carry accumulator chain that gives 128-tap inner products a single-cycle II. Whether mult retains its per-cell win at LANES-scale is one of the open questions to confirm at V5/V6.
reports/evidence/wide-g8-silicon/WIDE_G8_SILICON_GAIN.md); and SEL_PIPE full-top
OOC @100 MHz = +0.087 MET [OOC 2026-07-10] (the reclock enabler; module OPT1F −3.452 →
SELPIPE +0.426; verilog/forks/sel_pipe/build/ooc_top_selpipe_100m/timing_100m.rpt). Full
silicon ledger: EAGLE4_MEASURED_RESULTS.md.
| Module | CLB LUT | DSP58 | FF | Data path delay | Logic levels | xsim correctness |
|---|---|---|---|---|---|---|
mac_2bit_mult |
53 | 0 | 21 | 1.19 ns | 7 | 1000/1000 PASS |
mac_2bit_shift_add |
77 | 0 | 21 | 1.29 ns | 7 | 1000/1000 PASS |
gptq2_dequant16 (16-lane comb) |
58 | 0 | 0 | 0.675 ns | 3 | 1024/1024 PASS |
gemv_lane_int_pilot (V5a, K=64, K_PAR=16, no pipeline) |
1703 | 0 | ~280 | 10.21 ns | 54 | 200/200 PASS |
csa_tree16_int20 (SOTA Wallace 3:2 CSA + DSP58 CPA, standalone) |
322 | 1 | ~80 | 0.896 ns | 3 | 500/500 PASS |
gemv_lane_int_pilot_wallace (V5b — full lane with Wallace tree wired in) |
1547 | 1 | — | 2.817 ns | 13 | 200/200 PASS |
gemv_lane_qscale (V5b-scale — + Q1.15 scale apply on DSP58) |
1,556 | 3 | — | 2.848 ns | 13 | 200/200 PASS |
gemv_array_ws (V5c — LANES=16 of gemv_lane_qscale in parallel) |
24,896 | 48 | — | 2.849 ns | 13 | 3200/3200 PASS |
weight_bank_uram (V5d primitive, DEPTH=4096, DATA_W=32) |
1 | 0 | — | 1.051 ns | 1 (URAM288E5) | measured |
gemv_lane_uram (V5d integration: URAM-fed lane) |
1,558 | 3 | 1 URAM | 3.277 ns | 10 | 200/200 PASS |
Surprise: Vivado declined the DSP58 even when I wrote $signed(a) * $signed(w_s) — for a 16×3-bit MAC the DSP58 is too coarse, so synth chose LUT carry chains in both cases. The mult variant is smaller than my explicit shift-add (53 vs 77 LUT) — Vivado's automatic decomposition beat my hand-rolled one at this size. The explicit shift-add still wins at larger fan-out, where forcing structure controls the route — confirm when we get to V5 LANES=16 systolic.
rope_pair_rotate — a cos/sin-vs-rotate
latency skew corrupts attention head-0 for any ctx_len≥2 (masked because every unit gate
ran ctx_len=1 or compared two matched-cadence instances; memory
project_rope_head0_ctxlen_ge2_latency_bug); (2) the attention score-dot truncates
every Q·K to INT20 → overflows pervasively at real σ≈0.93 for ctx_len≥2 → wrong softmax winner
(memory project_int20_attnscore_overflow_ctxlen_ge2). Both are latent in the flashed lineage
and are a measured contributor to accept 1.68-vs-2.13. Root cause of the miss: A/B-twin TBs are blind to
bugs SHARED by both arms (feedback_ab_twin_gate_cancels_shared_bug). Fix + ctx≥2 real-magnitude
gates are queued for the next respin. The COST/xsim columns below are accurate for the isolated primitive.
| Module | Role | Origin | xsim | Cost (where measured) |
|---|---|---|---|---|
rmsnorm_scale_int.sv |
1/sqrt(mean(x²)+ε) scale via isqrt + divide FSM (~57 cyc/call) | TALOS-V2 lift | 200/200 | 438 LUT / 0 DSP / 2.64 ns (closes 300 MHz) |
rope_pair_rotate.sv |
One (i, i+D_HEAD/2) rotation pair: 4 muls + 2 adds, Q4.12 | new (uses cos/sin LUT) | 200/200 | ~2-4 DSP / pair · ~50 LUT |
exp_pwl_q12.sv |
32-entry PWL exp(-x) for softmax | TALOS-V2 lift | 200/200 | ~50 LUT combinational |
sigmoid_pwl_q12.sv |
64-entry PWL sigmoid for SiLU | new (computed from math.exp) | via silu_mul | ~60 LUT combinational |
silu_mul.sv |
gate × sigmoid(gate) × up = SiLU output, 2-stage pipeline | new | 200/200 | 2 DSP / lane · ~150 LUT |
kv_head_uram.sv |
Per-head KV cache: dual-port URAM, append-write + random read | new | 1024/1024 | ~16 URAM/head packed-4 · ~5 LUT control |
attn_score_dot.sv |
Q · K[p] dot product reusing Wallace tree primitive | new (composes csa_tree16_int20) | 200/200 | 16 DSP + ~400 LUT (D_HEAD=16 pilot) |
score_max_tracker.sv + softmax_exp_sum.sv |
Softmax pass-1 (running max) + pass-2 (sum of exp_pwl) | new | 20/20 groups | ~80 LUT / pair |
reports/evidence/wide-g8-silicon/WIDE_G8_SILICON_GAIN.md). That is 10.6× slower
than a GPU forward (658 µs) — the FPGA's win is correctness (M7 8/8 exact rescore) +
perf/W + the offload-appliance role, not raw latency. E2E throughput is 5.46 tok/s
(host fast-path) vs the GPU's ~105–125. Read the pilot table below as history; the live numbers are in
Measurements above and EAGLE4_MEASURED_RESULTS.md / docs/TOKS_ROADMAP_FOR_OPUS.md.
One 2-bit MAC cell at fp16/INT16 activation: 53 LUT, 0 DSP58, 21 FF, 1.19 ns. Multiply this by the parallel-MAC count and add the supporting hardware (weight memory, accumulator chain, scale apply, control, DMA) to get the full picture.
Per drafter step (1 forward through G1..G12), the 8 two-bit GEMVs total ~134 million MACs:
| GEMM | K × N | MACs | qweight bytes |
|---|---|---|---|
| G4_q + G4_k + G4_v | 6144 × 3072 × 3 | 56.6 M | 13.5 MB |
| G6_o | 3072 × 3072 | 9.4 M | 2.25 MB |
| G7g + G7u | 3072 × 8192 × 2 | 50.3 M | 12.0 MB |
| G9_down | 8192 × 6144 | 50.3 M | 12.0 MB |
| G12_lm | 256 × 128256 | 32.8 M | 8.0 MB |
| Total | — | ~199 M | 47.75 MB |
A WS systolic GEMV is fundamentally LANES_N (output cols processed in parallel) × K_PAR (input dim consumed per cycle per col). Each lane owns a slice of qweight in its own URAM bank. The drafter-step total cycles ≈ Σ over GEMVs of (K/K_PAR) × (N/LANES_N).
| Config | LANES_N × K_PAR | Parallel MACs | MAC LUT | Scale-apply DSP | URAM (weights) | Step cycles | Step time @ 300 MHz | vs GPU (0.658 ms) |
|---|---|---|---|---|---|---|---|---|
| Pilot (V5) | 16 × 4 | 64 | ~3.4 K (0.13% LUT) | 16 (0.15%) | ~16 (1.2%) | ~3.1 M | ~10 ms | 15× slower |
| Modest | 64 × 16 | 1024 | ~54 K (2.1%) | 64 (0.6%) | ~64 (5%) | ~194 K | ~0.65 ms | tie |
| Production | 128 × 32 | 4096 | ~217 K (8.4%) | 128 (1.2%) | ~128 (10%) | ~49 K | ~0.16 ms | 4× win |
| Aggressive | 256 × 32 | 8192 | ~434 K (16.9%) | 256 (2.4%) | ~256 (20%) | ~24 K | ~0.08 ms | 8× win |
| Resource | Per-cell | Production (4096 MACs) | V80 budget | Headroom |
|---|---|---|---|---|
| LUT | 53 | 217 K | 2.57 M (~1.8 M post-shell) | 8× still left |
| DSP58 | 0 | 128 (scale-apply only) | 10,848 | 85× still left |
| URAM (qweight) | 0 | ~128 (≈ 4.5 MB per array) | 1,925 (67.7 MB) | drafter total 48 MB → 71% — fits with margin |
| BRAM (scales) | 0 | ~40 | 3,741 RAMB36 (16.84 MB) | also room for G9 here if desired |
| Activation b/w (1 act/cycle, broadcast) | 32 bits/cycle | ~0.6 GB/s | ~32 GB/s PCIe Gen5 | trivial |
| Weight b/w (per-lane URAM bank) | — | ~1.2 TB/s aggregate | URAM ≈ 27 TB/s aggregate | comfortable if banked |
Where systolic puts weights in URAM and streams activations, constfold bakes the weights into the bitstream as combinational adder trees — 0 URAM, 0 weight bandwidth, 0 multiplier. For an output column at K=3072 with sparse 2-bit weights (~50% zero after pruning), the per-output cost is ~K/2 = ~1500 LUT.
| GEMM | N (outputs) | ~LUT per output | Total LUT | Fits V80? |
|---|---|---|---|---|
| G6 (K=3072, N=3072) | 3072 | ~1500 | ~4.6 M | no (2.6× over) |
| G12 (K=256, N=128256) | 128256 | ~128 | ~16.4 M | no (9× over) |
| One G12 column at K=256 | 1 | ~128 | ~128 | demo only |
So constfold doesn't scale to any full drafter-step GEMV. Its useful niche is small constant ops — RoPE/gamma multipliers, RMSNorm constants, fixed permutations. We'll measure ONE constfold column at V8 as a proof-of-concept, then defer it: the systolic path is what scales to a real drafter step.
Corrected V80 budget per Vivado 2026-05-20: the real V80 (xcv80-lsva4737-2MHP-e-S) has 16.84 MB BRAM (3,741 RAMB36) and 67.7 MB URAM (1,925 blocks) — ~80% more than the starter kit's table. The drafter's 48 MB qweight fits cleanly on-chip in multiple ways:
✅ SILICON-CONFIRMED (2026-07-09): Layout B (the winner row) is not just the design choice — it is proven on the card: WIDE PDI 0xE4B7110E runs the whole body with all 2-bit qweight (incl. G9) in URAM, zero HBM compute traffic, M7 8/8, URAM 1,710/88.83% binding. Only the 788 MB fp16 lm-head lives in HBM (G8 gather). The A/C rows below are the rejected alternatives, kept for the rationale.
| Layout | URAM use | BRAM use | HBM compute traffic | Complexity |
|---|---|---|---|---|
| B — All weights in URAM | ~48 MB (71%) | ~3 MB scales (18%) | 0 GB/s | simplest — zero ARVALID during compute |
| A — G9 in BRAM, rest in URAM | ~36 MB (53%) | ~15 MB (90%) | 0 GB/s | BRAM >80% util risks timing |
| C — Block 4 pattern (G9 HBM-streamed) | ~36 MB (53%) | ~3 MB (18%) | ~75 GB/s | silicon-validated but adds m_axi bundle |
Per-tensor placement (Layout B):
| Tensor | Size | Where | Why |
|---|---|---|---|
| G4_q/k/v qweight | 13.5 MB | URAM | Hot, ~370 URAMs at native 72-bit. Read 1×/forward × 3 GEMVs. |
| G6_o qweight | 2.25 MB | URAM | Square, hot. ~60 URAMs. |
| G7g + G7u qweight | 12.0 MB | URAM | Wide-output, big MACs. ~330 URAMs. |
| G12 qweight | 8.0 MB | URAM | Tiny K=256, huge N — bandwidth not capacity. ~220 URAMs. |
| G9 qweight | 12.0 MB | HBM-streamed | Doesn't fit if everything else is in URAM. Same call Block 4 makes today. |
| All scales (fp16) | ~3 MB | BRAM | Tiny, dense, multi-port. |
| KV cache | 5–20 MB | URAM hot + HBM cold spill | Depends on max_ctx. |
| Embedding table | 788 MB | HBM | No alternative — far too big for on-chip. |
| RoPE / gamma constants | <100 KB | LUTROM or BRAM | Truly constant, tiny — constfold-able. |
project_v80_actual_resources.md). The older claim of "4176 × 18 Kbit ≈ 9.4 MB" was the starter-kit number for a smaller SKU. G9 qweight is 12 MB, which DOES fit in 71% of BRAM — re-opens BRAM-resident G9 as a design option (currently still streaming from HBM for the AXI-driver-fan-in safety reasons below).With the corrected V80 budget, the cleanest architecture is everything URAM-resident, no streaming buffer:
| End-state criterion | Target | Realistic at "Production" config |
|---|---|---|
| Per-forward latency | < 0.658 ms (beat GPU drafter) | ~0.20 ms (3× win) if bottlenecks 1–4 close |
| Bit-exact vs C++ ref | YES on every GEMV | YES (already proven on per-cell MAC) |
| Resource fit on V80 | < 80% any single resource | ~10% LUT, ~10% URAM, ~1% DSP |
| Single top-level kernel (avoids integration hang class) | YES | designable but unverified |
| Energy per token (V80 75W vs H200 700W) | 9× lower power-per-token | unmeasured |
What gets confirmed at V5 (K=N=64 pilot, LANES=16, K_PAR=4):
Those four numbers tell us whether the Modest → Production extrapolation holds. If V5 lands LUT × 16 ≈ 850 LUT/lane, accumulator slack > +0.5 ns at 300 MHz, and II=1, we're on track for sub-0.3 ms drafter steps.
The user cloned github.com/y344shi/TALOS-V2-learn into cleanup_EAGLE/TALOS-V2-learn/ — a 1-layer 27-token Karpathy microGPT in SystemVerilog on Cyclone V (Q4.12 fixed-point). It is small, clean, and the synthesis pattern is directly applicable. Full lessons in TALOS_V2_LESSONS.md.
| TALOS file | Lines | Verdict | What we take |
|---|---|---|---|
rtl/src/systolic_matvec16_tile.sv | 67 | port as V5 backbone | WS systolic FSM (start/busy/done), per-lane acc, parameterized LANES/DATA/ACC. Replace flat-weight bus with gptq2_dequant16, widen K_PAR=16. |
rtl/src/include/microgpt_exact_core_math.svh | 117 | lift helpers | sat16, mul_q12, isqrt64, rms_scale_from_sum, and especially the 32-entry exp_neg_q12 PWL LUT for future softmax. |
rtl/src/processing_element.sv | 40 | reference pattern | Clean MAC + pass-through structure for V6/V7 OS/IS systolic forwarding. |
rtl/sim/tb_microgpt_core.sv | — | methodology | Load golden hex, drive RTL, compare — same flow we already use. |
rtl/python/export_weights.py | — | pattern | $readmemh hex flow — adapt for GPTQ-2bit qweight + fp16 scale ROMs. |
INDEX_FOR_EAGLE_PORT.md | 767 | 2nd-opinion port plan | Section-by-section reusability + 6-phase port roadmap from a prior-context agent. Cross-check our plan against it. |
Critical reframe for V5: instead of writing gemv_lane_int_pilot.sv from scratch, port TALOS's systolic_matvec16_tile.sv (60 lines lifted) and add (a) gptq2_dequant16 integration, (b) K_PAR=16 tree-reduce, (c) generalized COLS for our K=64 pilot. Maybe 80 lines new on top of 60 lifted — half the work of the from-scratch path.
| Date | Decision / finding | Source |
|---|---|---|
| 2026-07-10 | Real GPU traces caught a latent depth-N KV-replay bug no synthetic gate could see:
journal replay rd_slot=branch_s never re-keys across survivor commits, but re-parenting is
the COMMON case (5/8 non-identity per commit) → depth≥2 would replay a SIBLING's ancestor KV. Fix
kv_anc_lane.sv gated on real trees (PASS 0 err; RED arm 120/504). RULE: depth-N gates must
drive real/adversarially-re-parented trees, never self-consistent synthetics. |
reports/results/DEPTHN_GPU_TRACE_CAPTURE.md |
| 2026-07-10 | Batch-tree PDI 0xE4B7110F (B=2) built + flashed (route WNS +0.024 MET; SLL cleared; preflash PASS; FLASH_EXIT=0, awaiting cold-AC). Depth-0 expected bit-identical to 0xE4B7110E — an ENABLER proof for depth-N, not a tok/s change. Also: jewel per-SLR purity measured FALSE (E1=1,024 URAM > any SLR's 677/624/624) → E1 spans SLR0-top+SLR1+2×SLR2 rows. | reports/PDI_0xE4B7110F_BATCH_TESTPLAN.md · EAGLE4_MEASURED_RESULTS.md |
| 2026-07-09 | WIDE G8 PDI 0xE4B7110E is the first correct-AND-fast drafter on silicon. M7 all 19 stages golden, ordered 8/8; 6,964.8 µs/forward (N=60 median, ctx=1) = 4.9× over the 33.84 ms scalar baseline. HONEST: 4.9× is below the 28× xsim model (consume-bound fabric); still 10.6× slower than a GPU forward (658 µs). Correctness anchor every speed build must reproduce. | reports/evidence/wide-g8-silicon/WIDE_G8_SILICON_GAIN.md |
| 2026-07-09 | Host fast-path lifts E2E 3.20 → 5.46 tok/s (1.71×), host-only (C MMIO store loop + in-process pread; no RTL/flash). Accept 1.78 unchanged, M7 8/8 intact. CORRECTION: forward is ctx-dependent (6.97 ms @ctx0 → 17.6 ms @ctx620, G5 ctx-scan) — this, not host glue, caps E2E. | reports/results/HOST_FASTPATH_RESULT.md |
| 2026-07-10 | SEL_PIPE breaks the reclock wall: full-top in-context OOC @100 MHz = +0.087 ns MET, 0 failing endpoints (baseline −2.935; module OPT1F −3.452 → SELPIPE +0.426). Bit-exact, cycle-identical, URAM 88.83% preserved. New worst cone = tree→g8 seam, not the selector. Post-route @100 MHz NOT yet proven (OOC optimistic ~3.3 ns derate). URAM-OREG measured a no-op (both +5.291) → deprioritized. | verilog/forks/sel_pipe/build/ooc_top_selpipe_100m/timing_100m.rpt |
| 2026-07-09 | Accept 1.78 → 2.13 gap = DEPTH, not fidelity. Ours today = depth-0 width-8 (accept 1.78); GPU = depth-3 tree width-8 (accept 2.13). Batch-tree PDI 0xE4B7110F (B=2, sibling weight-reuse, depth-0 bit-identical) building to enable affordable depth-N. Prior measurement: batch-B is URAM/BRAM-neutral but LUT-expensive (+233K/unit) → batch-4-on-all-3-engines is LUT-infeasible; B=1 default, opt-in. | docs/TOKS_ROADMAP_FOR_OPUS.md · commit 04016a8 |
| 2026-05-20 | Verilog exploration opened; 8 GEMVs across 5 shape classes catalogued; numerical contract pinned to gptq2_gemv. |
2BIT_MM_INVENTORY.md |
| 2026-05-20 | Two MAC variants land xsim PASS on 1000 stimuli vs Python golden. Both are bit-exact. | build/log/xsim.log |
| 2026-05-20 | ooc-synth: mac_2bit_mult 53 LUT / 0 DSP / 1.19 ns vs mac_2bit_shift_add 77 LUT / 0 DSP / 1.29 ns. Synth automatically infers LUT-carry for 16×3-bit MACs even when written with *. Explicit shift-add deferred until LANES-scale validation. |
build/log/util_mac_2bit_*.rpt |
| 2026-05-20 | Block 4 "0.139 ms / 4.73× faster than GPU" silicon claim confirmed FALSE by user. Real per-forward is csynth ~5.5 ms. Paper headline framing pending revision (see ~/.claude/.../memory/project_block4_retraction_scope.md). |
conversation |
| 2026-05-20 | V80 resource correction: real budget per Vivado is BRAM 16.84 MB / URAM 67.7 MB (1,925 blocks) — ~80% more than starter-kit's smaller-SKU numbers. The "48 MB doesn't fit URAM" claim earlier was wrong. | build/log/util_mac_2bit_*.rpt · memory project_v80_actual_resources.md |
| 2026-05-20 | Layout B locked. All 48 MB qweight in URAM (71%), scales in BRAM, zero HBM compute traffic. Single top-level kernel; integration-hang class (V0/V3/V5) eliminated by construction. V5+ systolic designs assume per-lane URAM banks — drop the streaming-buffer plan. | user decision · memory project_verilog_layout_b_locked.md |
| 2026-05-20 | TALOS-V2 prior-art clone surfaced. Lift systolic_matvec16_tile.sv as V5 backbone; lift exp_neg_q12 PWL LUT for future softmax. gemv_lane_int_pilot.sv from-scratch start absorbed/discarded in favor of TALOS port. |
memory reference_talos_v2_prior_art.md · verilog/docs/TALOS_V2_LESSONS.md |
docs/TOKS_ROADMAP_FOR_OPUS.md §5 (reclock fallback,
depth-vs-denominator order, FF-seam waiver, speed-vs-batch line split) and
docs/DESIGN_KV_ONCHIP_DEPTHN.md §7 (journal mux fan-in, MAX_CTX audit).
Raise end-to-end tokens/s from the 5.46 tok/s silicon anchor per the multi-night lever plan in docs/TOKS_ROADMAP_FOR_OPUS.md (5.46 → ~20–30 tok/s modeled — each [MODELED] rung is replaced by a silicon measurement as it builds). Ordered levers:
0xE4B7110F — ✅ SILICON-VALIDATED (2026-07-10, T1–T5 ALL PASS):
depth-0 regression ordered 8/8 + all 19 CRCs (bit-identical to 0xE4B7110E); batch-test 4/4 —
one URAM weight read fed BOTH sibling MAC sets on silicon; timing 6.965 ms (+0.07%, no LUT
penalty); endurance 100/100 bit-identical; SGLang 5.75 tok/s @ accept 1.78 (no-regression). The
sibling weight-reuse architecture EXISTS in silicon — depth-N builds on it.
reports/PDI_0xE4B7110F_BATCH_TESTPLAN.md §2. Placement note: the batch triggered a
wholesale SLR rebalance — device SLICE 92.66% is the new wall (not LUT 58%); the
+0.024 margin owner is an AVED-shell reset recovery net (congestion canary, watchlisted) — A3:
reports/results/BATCH_LUT_PRESSURE_A3.md.This file is generated and refreshed by the fpga-eagle4-accelerator agent on demand. The maintenance prompt lives at verilog/docs/STATUS_UPDATE_PROCEDURE.md — it tells the agent which source files, synth reports, sim logs, and memory entries to scan, and how to merge them into this HTML.
Trigger an update whenever: (a) a new .sv module lands, (b) make synth_* produces new util/timing reports, (c) make * xsim run flips PASS↔FAIL, (d) a measurement at a new shape class arrives, or (e) the user makes a decision worth recording in the decision log.
EAGLE4 / Llama-3.2-3B / GPTQ-2bit / V80 Versal HBM / Vitis 2025.1.
Numerical contract: reference/include/eagle4_gemm_abstract.h.
Hard rules: reports/VERILOG_STARTER_KIT.md §11.
Memory: ~/.claude/projects/.../memory/MEMORY.md.