FASE #1 hardware freeze for FPGA-Neural V2, N4/P8, single external SDRAM (Alliance Memory AS4C4M16SA-6TIN) serving weights, activations, and results through one physical sdram_controller.v instance. Removes the PSRAM dependency (hardware/v1/rtl/psram_controller.v + memory_interface.v) from the V2 physical path entirely -- V1 itself remains fully unmodified, the golden reference. New RTL: sdram_unified_backend.v (2-way W/AR arbitration over one SDRAM controller, real per-byte DQM write masking added to sdram_controller.v for correct single-byte result writes with no read-modify-write), nms_neural_multiprocessor_sdram_unified.v (the frozen top-level). Two real bugs found and fixed via full-system testing before being accepted (ERR-0023): a deadlock and an off-by-one data-shift bug in the new arbitration logic. Real results: N=4 and N=2 D-Stress bit-exact (256/256 neurons), 40 real AUTO REFRESH events interleaved with zero corruption, real Yosys+nextpnr-ecp5 synthesis/P&R for LFE5U-45F-8CABGA381 (149/245 TRELLIS_IO, a real 45-pin reduction from the prior dual-memory design). Timing is MARGINAL (1/8 P&R seeds >=80MHz), reported honestly rather than masked by the best seed. Real, sourced ball-level pinout for the SDRAM bus + clk/rst (39/149 signals, P&R-verified) using the official Lattice ECP5U-45 pinout CSV found on disk during this step's own pre-commit review -- corrects an earlier draft that wrongly assumed no real pinout data was available. Chip readiness: NO. Real, disclosed blockers remain (no physical host interface exists yet -- the RTL's own reg_* ports are a 110-pin raw test-harness bus; clock source/PLL decision; power/configuration component selection) -- see hardware/v2/docs/{HARDWARE_FREEZE, CHIP_READINESS,OPEN_ITEMS}.md for the complete, itemized status. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013xXuuRUWZScuo1DeYJxs3v
2.5 KiB
2.5 KiB
| 1 | configuration | n_slots | pfd | latency_class | cycles | psram_util_pct | weight_stall_pct | prefetch_effectiveness_pct | sustained_mac_per_cycle | theoretical_mac_per_cycle | utilization_pct_of_theoretical | fmax_mhz | lut4 | ccu2c | ff | dsp | bram | classification |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2 | current_nms_baseline | 2 | n/a | real_v1_psram | 185645 | 90.4 | 92.5 | 0.0 | 0.1765 | 16 | 1.10 | 93.10 | 1948 | 266 | 3522 | 16 | 0 | POST-P&R MEASURED Fmax/resources + RTL SIMULATION cycles (EXP-0022) |
| 3 | nms_plus_prefetch | 1 | 1 | real_v1_psram | 181489 | 59.3 | 87.91 | 0.02 | 0.1806 | 8 | 2.26 | n/a | n/a | n/a | n/a | n/a | n/a | RTL SIMULATION (EXP-0024) |
| 4 | nms_plus_prefetch | 1 | 2 | real_v1_psram | 162876 | 66.4 | 86.44 | 0.39 | 0.2012 | 8 | 2.52 | 132.26 | 1333 | 203 | 2245 | 8 | 0 | POST-P&R MEASURED Fmax/resources + RTL SIMULATION cycles (EXP-0024) |
| 5 | nms_plus_prefetch | 1 | 4 | real_v1_psram | 162876 | 66.4 | 86.44 | 0.39 | 0.2012 | 8 | 2.52 | n/a | n/a | n/a | n/a | n/a | n/a | RTL SIMULATION (EXP-0024) |
| 6 | nms_plus_prefetch | 1 | 8 | real_v1_psram | 162876 | 66.4 | 86.44 | 0.39 | 0.2012 | 8 | 2.52 | 137.76 | 1464 | 200 | 2245 | 8 | 0 | POST-P&R MEASURED Fmax/resources + RTL SIMULATION cycles (EXP-0024) |
| 7 | nms_plus_prefetch | 1 | 16 | real_v1_psram | 162876 | 66.4 | 86.44 | 0.39 | 0.2012 | 8 | 2.52 | n/a | n/a | n/a | n/a | n/a | n/a | RTL SIMULATION (EXP-0024) |
| 8 | nms_plus_prefetch | 2 | 1 | real_v1_psram | 185410 | 90.5 | 93.32 | 0.78 | 0.1767 | 16 | 1.10 | n/a | n/a | n/a | n/a | n/a | n/a | RTL SIMULATION (EXP-0024) |
| 9 | nms_plus_prefetch | 2 | 2 | real_v1_psram | 185408 | 90.5 | 93.32 | 0.78 | 0.1767 | 16 | 1.10 | 97.16 | 1941 | 368 | 3449 | 16 | 0 | POST-P&R MEASURED Fmax/resources + RTL SIMULATION cycles (EXP-0024) |
| 10 | nms_plus_prefetch | 2 | 4 | real_v1_psram | 185404 | 90.5 | 93.32 | 0.78 | 0.1767 | 16 | 1.10 | n/a | n/a | n/a | n/a | n/a | n/a | RTL SIMULATION (EXP-0024) |
| 11 | nms_plus_prefetch | 2 | 8 | real_v1_psram | 185398 | 90.5 | 93.32 | 0.78 | 0.1767 | 16 | 1.10 | 95.25 | 1908 | 362 | 3449 | 16 | 0 | POST-P&R MEASURED Fmax/resources + RTL SIMULATION cycles (EXP-0024) |
| 12 | nms_plus_prefetch | 2 | 16 | real_v1_psram | 185390 | 90.5 | 93.33 | 0.78 | 0.1768 | 16 | 1.11 | n/a | n/a | n/a | n/a | n/a | n/a | RTL SIMULATION (EXP-0024) |
| 13 | ideal_memory_pfd_sweep | 1_or_2 | 1 | latency_0-1_cyc | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | IDEAL-MEMORY SIMULATION (EXP-0023) -- min PFD for >=90% util at latency 0-1 is PFD=4 | |
| 14 | ideal_memory_pfd_sweep | 1_or_2 | 1 | latency_2-4_cyc | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | IDEAL-MEMORY SIMULATION (EXP-0023) -- min PFD for >=90% util at latency 2-4 is PFD=8 | |
| 15 | ideal_memory_pfd_sweep | 1_or_2 | 1 | latency_8_cyc | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | IDEAL-MEMORY SIMULATION (EXP-0023) -- min PFD for >=90% util at latency 8 is PFD=16 | |
| 16 | ideal_memory_pfd_sweep | 1_or_2 | 1 | latency_16_cyc | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | IDEAL-MEMORY SIMULATION (EXP-0023) -- min PFD for >=90% util at latency 16 is PFD=32 |