michele/FPGA-Neural
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

perf(v2): word-level burst reads - 2.24-2.37x real wall-clock speedup (DEC-0015)

Browse Source 
Implements optimization #1 from the final benchmark campaign's own
recommendation: exploit psram_controller.v's already-implemented
page-mode support (confirmed present by direct inspection) by
fetching multiple bytes per real backend transaction instead of one
at a time.

Root cause addressed: int8_memory_access.v (the byte-level backend
prefetch_engine.v originally sat on) already converts every 8-bit
logical request into a full 16-bit PSRAM word access internally
(mem_addr <= addr >> 1), discarding half of every word it already
paid for. prefetch_engine.v/memory_manager.v now speak
memory_interface.v's own 16-bit word protocol directly, bypassing
int8_memory_access.v entirely - which remains untouched, still frozen
V1 (§1/§34); V2 simply reuses the lower layer of the same frozen
chain instead of the byte-splitting layer on top of it, the same
"reuse what fits" precedent slot_mem_arbiter.v already set.
slot_mem_arbiter.v and neural_multiprocessor.v widened to match
(lb_n/ub_n added, master port wired directly to memory_interface.v).

Real, measured results: M4's own single-job testbench shows 49-56%
fewer cycles (166->84, 446->204, 728->322, all still bit-exact). The
full final-benchmark campaign (24/24 workload/config combinations)
re-verified bit-exact with D-Stress's real wall-clock time (cycles /
real POST-P&R Fmax) improving 2.24-2.37x across every N_SLOTS tested,
against a small real Fmax cost (unchanged at N=1, -6.2% at N=2, -1.2%
at N=4).

tb_neural_multiprocessor.v (M8) and tb_benchmark_suite.v (final
campaign) needed zero changes - both treat neural_multiprocessor.v as
a black box. Only tb_memory_manager.v (M4, rewired to skip
int8_memory_access.v) and tb_dataflow_core.v (M7, behavioral model
widened to word-level) needed updates.

The "real parallel scaling is flat beyond N_SLOTS=2" finding (DEC-0014)
still holds - this optimization made the shared PSRAM port more
efficient per transaction, not multi-ported - so N_SLOTS=2 remains
the recommended default.

Logged: simulation/synthesis/timing/benchmark/decisions (DEC-0015)/
experiments (EXP-0015)/development.log.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013xXuuRUWZScuo1DeYJxs3v
This commit is contained in:
michele
2026-09-05 20:35:19 +02:00
co-authored by Claude Sonnet 5
parent 3cdaeaee35
commit e4a5540b6e
14 changed files with 489 additions and 168 deletions
Download Patch File Download Diff File Expand all files Collapse all files
hardware/v2/logs/benchmark.log
+36
View File
@@ -275,3 +275,39 @@ bit-exact against a software golden model, zero errors, zero
timeouts, zero deadlocks (after fixing the 3 issues in errors.log timeouts, zero deadlocks (after fixing the 3 issues in errors.log
ERR-0009). No node lost, no node duplicated, correct multi-hop ERR-0009). No node lost, no node duplicated, correct multi-hop
dependency wake-up verified (workload F's 2-hop diamond+fan-in graph). dependency wake-up verified (workload F's 2-hop diamond+fan-in graph).
[2026-09-05] EXP-0015 -- word-level burst-read rewrite (DEC-0015),
real before/after comparison (user-requested optimization #1,
following the final-benchmark.md report's own recommendation)
M4 standalone (tb_memory_manager.v, real V1 PSRAM chain, single job):
| Job (n_tiles) | cycles BEFORE | cycles AFTER | reduction |
|-----------------|-----------------|----------------|-------------|
| 1 tile | 166 | 84 | -49.4% |
| 3 tiles | 446 | 204 | -54.3% |
| 5 tiles | 728 | 322 | -55.8% |
All still bit-exact.
Full campaign (tb_benchmark_suite.v, same 6 workloads as EXP-0014),
D-Stress (256 neurons, the largest/most representative workload),
real cycles and real wall-clock (cycles / real POST-P&R Fmax):
| N_SLOTS | Fmax BEFORE | Fmax AFTER | cycles BEFORE | cycles AFTER | wall-clock BEFORE | wall-clock AFTER | real speedup |
|-----------|---------------|--------------|------------------|-----------------|----------------------|---------------------|----------------|
| 1 | 152.46 MHz | 152.44 MHz | 780298 | 348682 | 5118.1 us | 2287.3 us | 2.24x |
| 2 | 142.45 MHz | 133.58 MHz | 736402 | 307602 | 5169.5 us | 2302.8 us | 2.24x |
| 4 | 113.38 MHz | 112.07 MHz | 736823 | 307346 | 6498.7 us | 2742.4 us | 2.37x |
Real PSRAM port utilization also rose (e.g. N=2, D-Stress: 91.0% ->
89.1% -- essentially unchanged fraction, but of a MUCH smaller total
cycle count, meaning the port is doing genuinely useful work a larger
fraction of the time it IS busy, not idling on redundant round-trips).
All 24/24 workload/config combinations (6 workloads x N_SLOTS=1/2/4/8)
re-verified bit-exact after the rewrite. The "real parallel scaling is
flat beyond N_SLOTS=2" finding from EXP-0014 STILL holds (D-Stress
cycles at N=2/4/8 remain within ~0.2% of each other: 307602/307346/
307874) -- this optimization made the shared PSRAM port more
EFFICIENT per transaction, it did not remove the fact that there is
still only one physical port, so DEC-0014's N_SLOTS=2 recommendation
is unaffected and reconfirmed with the new, faster numbers.
hardware/v2/logs/decisions.log
+80
View File
@@ -857,3 +857,83 @@ future memory-bandwidth-scaling architecture change.
STATUS: STATUS:
ACCEPTED ACCEPTED
DEC-0015
DATE: 2026-09-05
DECISION:
prefetch_engine.v/memory_manager.v's own Memory Backend Interface is
changed from byte-level (matching hardware/v1/rtl/int8_memory_access.v's
contract, one 8-bit logical transaction per real backend round-trip)
to WORD-level (matching hardware/v1/rtl/memory_interface.v's own
16-bit contract directly, one transaction moving 2 consecutive bytes).
neural_multiprocessor.v no longer instantiates int8_memory_access.v --
the arbiter's master port connects directly to memory_interface.v.
slot_mem_arbiter.v's own per-port data width and lb_n/ub_n signals are
widened to match.
WHY (user-requested, directly following the M9/M10 benchmark
campaign's own finding that the system is memory-bound -- see the
final-benchmark.md report's recommendation #1): int8_memory_access.v
ALREADY converts every 8-bit logical request into a FULL 16-bit
PSRAM word access internally (`mem_addr <= addr >> 1`, one byte lane
selected via lb_n/ub_n) -- so fetching X/W tile arrays one byte at a
time was ALREADY paying for two bytes of real PSRAM bandwidth per
transaction while discarding half of it, and paying int8_memory_access's
own STATE_IDLE/STATE_WAIT round-trip TWICE for every 2 real bytes
instead of once. hardware/v1/rtl/psram_controller.v's own real
page-mode support (already implemented, unmodified, confirmed present
by direct inspection) then has fewer, more effective opportunities to
serve consecutive words fast once transactions are batched this way.
int8_memory_access.v/memory_interface.v/psram_controller.v are all
frozen V1 files and remain byte-for-byte unmodified (§1/§34) --
V2 simply chooses to reuse the lower (word-level) layer of that same
frozen chain directly instead of the byte-splitting layer on top of
it, the same "reuse what fits" precedent slot_mem_arbiter.v already
set by not reusing hardware/v1/rtl/mem_arbiter.v verbatim.
EVIDENCE (real, measured, before/after -- see experiments.log EXP-0015
for full detail): hardware/v2/sim/tb_memory_manager.v (M4, real V1
PSRAM chain): 3-tile job 446->204 cycles (-54%), 1-tile 166->84
(-49%), 5-tile 728->322 (-56%), all still bit-exact. Full campaign
(tb_benchmark_suite.v, EXP-0014's own workloads) re-run at N_SLOTS=
1/2/4/8: D-Stress real wall-clock (cycles / real POST-P&R Fmax)
improves 2.24-2.37x across every N_SLOTS tested, all 24/24 workload/
config combinations still bit-exact. Real Fmax cost is small (N=1:
152.46->152.44 MHz, unchanged; N=2: 142.45->133.58 MHz, -6.2%; N=4:
113.38->112.07 MHz, -1.2%) -- overwhelmingly a net win in real
wall-clock terms at every N_SLOTS.
CONSTRAINT introduced: P_IN must be even (already true, P_IN=8), and
tile base addresses (x_base/w_base, and therefore every x_base +
tile_idx*P_IN the system ever computes) must be word-aligned (even
byte addresses) -- true of every address this project's own
testbenches already use, and a trivial constraint for any real
loader/host to satisfy (place tile arrays at even byte offsets).
ALTERNATIVES:
1. Modify int8_memory_access.v itself to return/accept 2 bytes per
logical transaction. Rejected: that file is frozen V1 (§1/§34) --
never modified, regardless of how small the change would be.
2. Build a NEW byte-level burst wrapper on top of int8_memory_access.v
(queue N byte requests, pipeline them). Rejected: int8_memory_access's
own STATE_IDLE only samples a new req once back in STATE_IDLE after
the previous transaction's mem_ready -- it fundamentally does not
support pipelining/overlapped requests, so no wrapper on TOP of it
can avoid paying its full per-byte round-trip cost twice per word.
Only bypassing it (going one layer lower, to memory_interface.v's
own native word interface) actually eliminates the redundant
round-trip.
RESULT:
prefetch_engine.v/memory_manager.v/slot_mem_arbiter.v/
neural_multiprocessor.v now speak a word-level (16-bit + lb_n/ub_n)
Memory Backend Interface, bypassing int8_memory_access.v entirely
(still frozen, still reused unmodified -- just one layer lower in the
same frozen stack). Real, measured 2.24-2.37x wall-clock improvement
at every N_SLOTS tested, negligible real Fmax cost, all functional
correctness preserved (24/24 bit-exact).
STATUS:
ACCEPTED
hardware/v2/logs/development.log
+20
View File
@@ -311,3 +311,23 @@ decision: vedi decisions.log DEC-0014 -- N_SLOTS=2 raccomandato come
tetto DSP mafisico, non come raccomandazione d'uso generale). tetto DSP mafisico, non come raccomandazione d'uso generale).
next_action: datasheet V2 in stile professionale (richiesta utente), next_action: datasheet V2 in stile professionale (richiesta utente),
ora sbloccato dalla decisione su N_SLOTS. ora sbloccato dalla decisione su N_SLOTS.
[2026-09-05] Ottimizzazione #1 -- burst read a livello di parola
(word-level), su richiesta esplicita dell'utente
reason: dopo la campagna di benchmark finale, l'utente ha chiesto
concretamente di implementare la raccomandazione #1 (sfruttare il
page-mode gia' presente nel controller PSRAM leggendo piu' byte in
una sola transazione, non uno alla volta).
result: prefetch_engine.v/memory_manager.v riscritti per parlare
direttamente il protocollo a 16 bit di memory_interface.v (bypassando
int8_memory_access.v, che resta comunque congelato e non modificato
-- semplicemente non piu' istanziato in questo percorso dati).
Risultato reale misurato: -49/-54/-56% cicli su job singoli (M4),
2.24-2.37x speedup reale in wall-clock sull'intera campagna finale
(24/24 ancora bit-exact), a fronte di un costo Fmax reale piccolo
(-6.2% a N_SLOTS=2, -1.2% a N_SLOTS=4, invariato a N_SLOTS=1).
errors: nessuno (implementazione pulita, nessuna regressione).
decision: vedi decisions.log DEC-0015.
next_action: ottimizzazione #2 -- cache condivisa on-chip per il
vettore di attivazione (X), per eliminare le letture ridondanti tra
neuroni che condividono lo stesso input di layer.
hardware/v2/logs/experiments.log
+33
View File
@@ -551,3 +551,36 @@ next_action: none mandated by the roadmap (this campaign was
characterization before deciding N_SLOTS and writing the V2 characterization before deciding N_SLOTS and writing the V2
datasheet). Full report: hardware/v2/docs/benchmarks/final- datasheet). Full report: hardware/v2/docs/benchmarks/final-
benchmark.md. benchmark.md.
[2026-09-05] EXP-0015 -- word-level burst-read implementation (user-
requested optimization #1, following final-benchmark.md's own
recommendation: exploit psram_controller.v's already-implemented
page-mode support by fetching multiple bytes per real transaction
instead of one at a time)
test: prefetch_engine.v/memory_manager.v rewritten to speak
memory_interface.v's 16-bit word protocol directly (bypassing
int8_memory_access.v, still frozen/unmodified -- just no longer
instantiated in this datapath); slot_mem_arbiter.v/dataflow_core.v/
neural_multiprocessor.v widened to match. Re-verified: M4's own
testbench (updated to skip int8_memory_access), M7's own testbench
(sim_byte_mem -> sim_word_mem), M8's own testbench (UNCHANGED,
black-box), and the full final benchmark campaign (UNCHANGED,
black-box) at N_SLOTS=1/2/4/8.
simulator: Verilator 5.050 (--binary --timing)
PASS/FAIL:
SIMULATED: M4 3/3 PASS, cycles reduced 49-56% (166->84, 446->204,
728->322). M7 4/4 PASS. M8 4/4 PASS, cycles 684->337. Final
campaign 24/24 PASS bit-exact, D-Stress cycles reduced from
780298/736402/736823/738751 to 348682/307602/307346/307874
(N=1/2/4/8) -- roughly 2.2-2.4x fewer real cycles.
SYNTHESIZED + POST-P&R (real, full system incl. real PSRAM pins):
N=1 152.44 MHz (was 152.46), N=2 133.58 MHz (was 142.45, -6.2%),
N=4 112.07 MHz (was 113.38, -1.2%) -- small real Fmax cost.
Combined real wall-clock speedup (cycles / real Fmax): 2.24-2.37x
across N_SLOTS=1/2/4.
errors: none found (clean implementation, no regressions).
decision: see decisions.log DEC-0015.
next_action: user-requested optimization #2 -- a shared on-chip cache
for the activation (X) vector, so N independent neurons sharing one
input vector (the dense-layer shape used throughout this benchmark
suite) fetch it from PSRAM ONCE instead of once per neuron.
hardware/v2/logs/simulation.log
+18
View File
@@ -134,3 +134,21 @@ PASS/FAIL: 24/24 PASS bit-exact (11,520 individual neuron/node
sizing bugs -- psram_model DEPTH too small, N_NODES too small sizing bugs -- psram_model DEPTH too small, N_NODES too small
causing a real node-id wraparound deadlock) causing a real node-id wraparound deadlock)
full data/analysis: hardware/v2/docs/benchmarks/final-benchmark.md full data/analysis: hardware/v2/docs/benchmarks/final-benchmark.md
[2026-09-05] EXP-0015 -- regression + real improvement measurement
after word-level burst-read rewrite (DEC-0015)
test: hardware/v2/sim/tb_memory_manager.v (M4, updated to connect
memory_manager directly to memory_interface.v, skipping
int8_memory_access.v), hardware/v2/sim/tb_dataflow_core.v (M7,
sim_byte_mem -> sim_word_mem), hardware/v2/sim/tb_neural_multiprocessor.v
(M8, UNCHANGED -- black-box on neural_multiprocessor.v, no edits
needed), hardware/v2/sim/tb_benchmark_suite.v (final campaign,
UNCHANGED, re-run at N_SLOTS=1/2/4/8)
simulator: Verilator 5.050 (--binary --timing)
PASS/FAIL: M4 3/3 PASS (166/446/728 -> 84/204/322 cycles, bit-exact).
M7 4/4 PASS (wd=67->43 cycles to first dependency check). M8 4/4
PASS (684->337 cycles). Final campaign 24/24 PASS bit-exact,
D-Stress cycles roughly halved to a bit more at every N_SLOTS
(see benchmark.log EXP-0015 for the full table).
errors: none found during this implementation (clean first-pass
correctness at every regression point).
hardware/v2/logs/synthesis.log
+15
View File
@@ -97,3 +97,18 @@ N_SLOTS=4: LUT4=7552 CCU2C=768 TRELLIS_FF=6495 MULT18X18D=32 DP16KD=0
(N_SLOTS=2 reference, EXP-0009: LUT4=4191 CCU2C=388 FF=3659 DSP=16 DP16KD=0) (N_SLOTS=2 reference, EXP-0009: LUT4=4191 CCU2C=388 FF=3659 DSP=16 DP16KD=0)
CHECK: 0 problems on both, same benign warnings as every other CHECK: 0 problems on both, same benign warnings as every other
neural_processor instantiation since EXP-0001. neural_processor instantiation since EXP-0001.
[2026-09-05] EXP-0015 -- neural_multiprocessor N_SLOTS=1/2/4 after the
word-level burst-read rewrite (DEC-0015), real standalone synthesis,
no harness needed
N_SLOTS=1: LUT4=1995 CCU2C=200 TRELLIS_FF=2217 MULT18X18D=8 DP16KD=0
N_SLOTS=2: LUT4=3166 CCU2C=388 TRELLIS_FF=3655 MULT18X18D=16 DP16KD=0
N_SLOTS=4: LUT4=5824 CCU2C=768 TRELLIS_FF=6529 MULT18X18D=32 DP16KD=0
CHECK: 0 problems on all three (same benign warnings as every prior
neural_processor instantiation since EXP-0001). Resource cost is
essentially unchanged from the pre-rewrite byte-level numbers
(EXP-0014: LUT4 2642/4191/7552, FF 2240/3659/6495) -- widening the
backend to 16-bit + lb_n/ub_n cost a small amount of LUT4 in some
configs and saved some in others (word-level control logic is
simpler than byte-indexing logic in prefetch_engine.v), net roughly
flat.
hardware/v2/logs/timing.log
+11
View File
@@ -115,3 +115,14 @@ routing congestion around the shared director/dependency_manager/
arbiter hub), exactly the same trend already observed for arbiter hub), exactly the same trend already observed for
dataflow_core alone (M7/M10, EXP-0008/EXP-0011) but now measured for dataflow_core alone (M7/M10, EXP-0008/EXP-0011) but now measured for
the REAL FULL SYSTEM including the real PSRAM backend. the REAL FULL SYSTEM including the real PSRAM backend.
[2026-09-05] EXP-0015 -- neural_multiprocessor N_SLOTS=1/2/4 after the
word-level burst-read rewrite (DEC-0015), real nextpnr-ecp5 --45k
--package CABGA381 --speed 8 --freq 80 --lpf-allow-unconstrained
N_SLOTS=1: Fmax = 152.44 MHz -- PASS at 80MHz (was 152.46 MHz, unchanged)
N_SLOTS=2: Fmax = 133.58 MHz -- PASS at 80MHz (was 142.45 MHz, -6.2%)
N_SLOTS=4: Fmax = 112.07 MHz -- PASS at 80MHz (was 113.38 MHz, -1.2%)
Small, real Fmax cost from widening the shared arbiter/backend to
16-bit + lb_n/ub_n (extra routing), overwhelmingly outweighed by the
real cycle-count reduction (EXP-0015 in experiments.log/benchmark.log):
D-Stress real wall-clock improves 2.24-2.37x at every N_SLOTS.
hardware/v2/rtl/dataflow_core.v
+11 -6
View File
@@ -75,12 +75,16 @@ module dataflow_core #(
// ---- per-slot Memory Backend Interface (arrayed, one per slot -- // ---- per-slot Memory Backend Interface (arrayed, one per slot --
// see file header on why arbitration to one shared PSRAM port is // see file header on why arbitration to one shared PSRAM port is
// NOT done here) ---- // NOT done here). WORD-level (16-bit) post-M10 (decisions.log
// DEC-0015) -- see memory_manager.v/prefetch_engine.v's own
// headers for why. ----
output wire [N_SLOTS-1:0] slot_mem_req, output wire [N_SLOTS-1:0] slot_mem_req,
output wire [N_SLOTS-1:0] slot_mem_wr, output wire [N_SLOTS-1:0] slot_mem_wr,
output wire [ADDR_WIDTH*N_SLOTS-1:0] slot_mem_addr, output wire [ADDR_WIDTH*N_SLOTS-1:0] slot_mem_addr, // WORD address
output wire signed [8*N_SLOTS-1:0] slot_mem_wdata, output wire [16*N_SLOTS-1:0] slot_mem_wdata,
input wire signed [8*N_SLOTS-1:0] slot_mem_rdata, output wire [N_SLOTS-1:0] slot_mem_lb_n,
output wire [N_SLOTS-1:0] slot_mem_ub_n,
input wire [16*N_SLOTS-1:0] slot_mem_rdata,
input wire [N_SLOTS-1:0] slot_mem_ready input wire [N_SLOTS-1:0] slot_mem_ready
); );
@@ -175,8 +179,9 @@ module dataflow_core #(
.result_valid(mm_result_valid), .result_ready(mm_result_ready), .result_data(mm_result_data), .result_valid(mm_result_valid), .result_ready(mm_result_ready), .result_data(mm_result_data),
.mem_req(slot_mem_req[g]), .mem_wr(slot_mem_wr[g]), .mem_req(slot_mem_req[g]), .mem_wr(slot_mem_wr[g]),
.mem_addr(slot_mem_addr[g*ADDR_WIDTH +: ADDR_WIDTH]), .mem_addr(slot_mem_addr[g*ADDR_WIDTH +: ADDR_WIDTH]),
.mem_wdata(slot_mem_wdata[g*8 +: 8]), .mem_wdata(slot_mem_wdata[g*16 +: 16]),
.mem_rdata(slot_mem_rdata[g*8 +: 8]), .mem_ready(slot_mem_ready[g]) .mem_lb_n(slot_mem_lb_n[g]), .mem_ub_n(slot_mem_ub_n[g]),
.mem_rdata(slot_mem_rdata[g*16 +: 16]), .mem_ready(slot_mem_ready[g])
); );
reg job_valid_np; reg job_valid_np;
hardware/v2/rtl/memory_manager.v
+46 -13
View File
@@ -3,13 +3,24 @@
// ================================================================ // ================================================================
// FPGA-Neural V2 -- Memory Manager (M4, docs/v2-description.md §12/§15) // FPGA-Neural V2 -- Memory Manager (M4, docs/v2-description.md §12/§15)
// //
// Sits between a single Neural Processor (M1) and the byte-level // Sits between a single Neural Processor (M1) and the WORD-level
// Memory Backend Interface (hardware/v1/rtl/int8_memory_access.v, // Memory Backend Interface (hardware/v1/rtl/memory_interface.v,
// reused UNMODIFIED, per §15 -- "NON iniziare modificando il // reused UNMODIFIED, per §15 -- "NON iniziare modificando il
// controller PSRAM. Mantenere inizialmente il backend esistente"). // controller PSRAM. Mantenere inizialmente il backend esistente").
// The processor sees only "data available" (operand_valid/ready, // The processor sees only "data available" (operand_valid/ready,
// tile_last) -- never PSRAM request/wait cycles directly (§12). // tile_last) -- never PSRAM request/wait cycles directly (§12).
// //
// Post-M10 (decisions.log DEC-0015): this port talks directly to
// memory_interface.v's own 16-bit word interface instead of routing
// through int8_memory_access.v's byte-splitting layer -- every real
// transaction now moves a full PSRAM word (2 bytes) instead of
// discarding half of one, halving the real transaction count for
// prefetch_engine's own reads. int8_memory_access.v itself is
// untouched (still frozen V1); V2 simply no longer instantiates it in
// this datapath, reusing the lower (word-level) layer directly
// instead, the same "reuse what fits" precedent slot_mem_arbiter.v
// already set for hardware/v1/rtl/mem_arbiter.v.
//
// Double-buffered prefetch (§13): while the processor consumes tile // Double-buffered prefetch (§13): while the processor consumes tile
// N from bank "current", this module retargets the single // N from bank "current", this module retargets the single
// prefetch_engine instance (M4) at bank "next" to fetch tile N+1 // prefetch_engine instance (M4) at bank "next" to fetch tile N+1
@@ -67,12 +78,19 @@ module memory_manager #(
output reg result_ready, output reg result_ready,
input wire signed [DATA_WIDTH-1:0] result_data, input wire signed [DATA_WIDTH-1:0] result_data,
// ---- Memory Backend Interface (matches int8_memory_access.v) ---- // ---- Memory Backend Interface (word-level, matches
// hardware/v1/rtl/memory_interface.v's contract exactly -- see
// prefetch_engine.v's own header and decisions.log DEC-0015 for
// why this is now word- rather than byte-level: int8_memory_access.v
// is no longer in the datapath, each transaction moves a full
// 16-bit PSRAM word instead of discarding half of it) ----
output wire mem_req, output wire mem_req,
output wire mem_wr, output wire mem_wr,
output wire [ADDR_WIDTH-1:0] mem_addr, output wire [ADDR_WIDTH-1:0] mem_addr, // WORD address
output wire signed [7:0] mem_wdata, output wire [15:0] mem_wdata,
input wire signed [7:0] mem_rdata, output wire mem_lb_n,
output wire mem_ub_n,
input wire [15:0] mem_rdata,
input wire mem_ready input wire mem_ready
); );
@@ -122,10 +140,11 @@ module memory_manager #(
// (never both at once, by construction -- see file header) // (never both at once, by construction -- see file header)
// selects which one actually reaches the real output port, // selects which one actually reaches the real output port,
// avoiding a two-driver conflict on mem_req/mem_wr/mem_addr/ // avoiding a two-driver conflict on mem_req/mem_wr/mem_addr/
// mem_wdata. // mem_wdata/mem_lb_n/mem_ub_n.
wire pf_mem_req, pf_mem_wr; wire pf_mem_req, pf_mem_wr;
wire [ADDR_WIDTH-1:0] pf_mem_addr; wire [ADDR_WIDTH-1:0] pf_mem_addr;
wire signed [7:0] pf_mem_wdata; wire [15:0] pf_mem_wdata;
wire pf_mem_lb_n, pf_mem_ub_n;
prefetch_engine #( prefetch_engine #(
.DATA_WIDTH(DATA_WIDTH), .P_IN(P_IN), .ADDR_WIDTH(ADDR_WIDTH) .DATA_WIDTH(DATA_WIDTH), .P_IN(P_IN), .ADDR_WIDTH(ADDR_WIDTH)
@@ -135,12 +154,14 @@ module memory_manager #(
.fetch_busy(pf_busy), .fetch_done(pf_done), .fetch_busy(pf_busy), .fetch_done(pf_done),
.tile_x(pf_tile_x), .tile_w(pf_tile_w), .tile_x(pf_tile_x), .tile_w(pf_tile_w),
.mem_req(pf_mem_req), .mem_wr(pf_mem_wr), .mem_addr(pf_mem_addr), .mem_wdata(pf_mem_wdata), .mem_req(pf_mem_req), .mem_wr(pf_mem_wr), .mem_addr(pf_mem_addr), .mem_wdata(pf_mem_wdata),
.mem_lb_n(pf_mem_lb_n), .mem_ub_n(pf_mem_ub_n),
.mem_rdata(mem_rdata), .mem_ready(mem_ready) .mem_rdata(mem_rdata), .mem_ready(mem_ready)
); );
reg wr_mem_req; reg wr_mem_req;
reg [ADDR_WIDTH-1:0] wr_mem_addr; reg [ADDR_WIDTH-1:0] wr_mem_addr; // WORD address
reg signed [7:0] wr_mem_wdata; reg [15:0] wr_mem_wdata;
reg wr_mem_lb_n, wr_mem_ub_n;
// wr_mem_req is SET while state==MM_WRITE_RESULT but only becomes // wr_mem_req is SET while state==MM_WRITE_RESULT but only becomes
// valid (via NBA) the FOLLOWING cycle, i.e. while state==MM_DONE -- // valid (via NBA) the FOLLOWING cycle, i.e. while state==MM_DONE --
@@ -153,6 +174,8 @@ module memory_manager #(
assign mem_wr = wr_active ? 1'b1 : pf_mem_wr; assign mem_wr = wr_active ? 1'b1 : pf_mem_wr;
assign mem_addr = wr_active ? wr_mem_addr : pf_mem_addr; assign mem_addr = wr_active ? wr_mem_addr : pf_mem_addr;
assign mem_wdata = wr_active ? wr_mem_wdata : pf_mem_wdata; assign mem_wdata = wr_active ? wr_mem_wdata : pf_mem_wdata;
assign mem_lb_n = wr_active ? wr_mem_lb_n : pf_mem_lb_n;
assign mem_ub_n = wr_active ? wr_mem_ub_n : pf_mem_ub_n;
always @(posedge clk) begin always @(posedge clk) begin
if (rst) begin if (rst) begin
@@ -169,7 +192,9 @@ module memory_manager #(
tile_idx <= 16'h0; tile_idx <= 16'h0;
wr_mem_req <= 1'b0; wr_mem_req <= 1'b0;
wr_mem_addr <= {ADDR_WIDTH{1'b0}}; wr_mem_addr <= {ADDR_WIDTH{1'b0}};
wr_mem_wdata <= 8'sd0; wr_mem_wdata <= 16'h0000;
wr_mem_lb_n <= 1'b1;
wr_mem_ub_n <= 1'b1;
pf_pending <= 1'b0; pf_pending <= 1'b0;
end else begin end else begin
job_done <= 1'b0; job_done <= 1'b0;
@@ -282,7 +307,15 @@ module memory_manager #(
MM_WAIT_RESULT: begin MM_WAIT_RESULT: begin
result_ready <= 1'b1; result_ready <= 1'b1;
if (result_valid && result_ready) begin if (result_valid && result_ready) begin
wr_mem_wdata <= result_data; // Replicate int8_memory_access.v's own byte-
// select convention exactly (addr[0]==0 -> low
// byte, addr[0]==1 -> high byte) since that
// module is no longer in the datapath -- see
// prefetch_engine.v's header/decisions.log
// DEC-0015.
wr_mem_wdata <= result_addr_reg[0] ? {result_data, 8'h00} : {8'h00, result_data};
wr_mem_lb_n <= result_addr_reg[0] ? 1'b1 : 1'b0;
wr_mem_ub_n <= result_addr_reg[0] ? 1'b0 : 1'b1;
state <= MM_WRITE_RESULT; state <= MM_WRITE_RESULT;
end end
end end
@@ -292,7 +325,7 @@ module memory_manager #(
// tiles to fetch for this job), so driving the shared // tiles to fetch for this job), so driving the shared
// backend port directly is safe -- see file header. // backend port directly is safe -- see file header.
wr_mem_req <= 1'b1; wr_mem_req <= 1'b1;
wr_mem_addr <= result_addr_reg; wr_mem_addr <= result_addr_reg[ADDR_WIDTH-1:1]; // byte -> word
state <= MM_DONE; state <= MM_DONE;
end end
hardware/v2/rtl/neural_multiprocessor.v
+41 -37
View File
@@ -8,17 +8,29 @@
// The real, hardware-facing top-level: dataflow_core.v (M7) with its // The real, hardware-facing top-level: dataflow_core.v (M7) with its
// N_SLOTS independent Memory Backend Interface ports funneled through // N_SLOTS independent Memory Backend Interface ports funneled through
// a new generic arbiter (slot_mem_arbiter.v, M8) down to the REAL, // a new generic arbiter (slot_mem_arbiter.v, M8) down to the REAL,
// UNMODIFIED hardware/v1 PSRAM backend chain -- // UNMODIFIED hardware/v1 PSRAM backend chain -- exactly the chain
// int8_memory_access -> memory_interface -> psram_controller // hardware/v2/sim/tb_memory_manager.v (M4) already proved correct for
// -- exactly the chain hardware/v2/sim/tb_memory_manager.v (M4) // ONE memory_manager port. This module is the first point M3 (per
// already proved correct for ONE memory_manager port. This module is // DEC-0009) and M2 (per DEC-0006) BOTH deferred to: N_SLOTS
// the first point M3 (per DEC-0009) and M2 (per DEC-0006) BOTH // memory_manager instances genuinely sharing one physical PSRAM port.
// deferred to: N_SLOTS memory_manager instances genuinely sharing one
// physical PSRAM port.
// //
// dataflow_core.v itself is NOT modified -- its per-slot interface // Post-M10 (decisions.log DEC-0015): the chain is now
// (DEC-0009) is exactly what makes it pluggable into an arbiter here // memory_interface -> psram_controller
// without touching M7's own file. // -- int8_memory_access.v is no longer instantiated here.
// int8_memory_access itself is untouched (still frozen V1, §1/§34);
// V2 simply reuses the lower (word-level) layer of the same frozen
// chain directly, since prefetch_engine.v/memory_manager.v now speak
// memory_interface's own 16-bit word protocol natively (see those
// modules' headers for why: every real transaction now moves a full
// PSRAM word instead of discarding half of it, halving the number of
// real backend round-trips per tile fetch).
//
// dataflow_core.v itself is NOT modified in its own control logic --
// its per-slot interface (DEC-0009) is exactly what makes it pluggable
// into an arbiter here without touching M7's own file (only the
// WIDTH of that per-slot interface changed, from 8 to 16 bits plus
// lb_n/ub_n, a mechanical consequence of DEC-0015, not a redesign of
// dataflow_core's own scheduling/dependency logic).
// ================================================================ // ================================================================
module neural_multiprocessor #( module neural_multiprocessor #(
@@ -59,10 +71,12 @@ module neural_multiprocessor #(
output wire psram_zz_n output wire psram_zz_n
); );
// ---- dataflow_core (M7, unmodified) ---- // ---- dataflow_core (M7, control logic unmodified; per-slot
// backend port widened to 16-bit + lb_n/ub_n per DEC-0015) ----
wire [N_SLOTS-1:0] slot_mem_req, slot_mem_wr; wire [N_SLOTS-1:0] slot_mem_req, slot_mem_wr;
wire [ADDR_WIDTH*N_SLOTS-1:0] slot_mem_addr; wire [ADDR_WIDTH*N_SLOTS-1:0] slot_mem_addr;
wire signed [8*N_SLOTS-1:0] slot_mem_wdata, slot_mem_rdata; wire [16*N_SLOTS-1:0] slot_mem_wdata, slot_mem_rdata;
wire [N_SLOTS-1:0] slot_mem_lb_n, slot_mem_ub_n;
wire [N_SLOTS-1:0] slot_mem_ready; wire [N_SLOTS-1:0] slot_mem_ready;
dataflow_core #( dataflow_core #(
@@ -75,14 +89,16 @@ module neural_multiprocessor #(
.reg_x_base(reg_x_base), .reg_w_base(reg_w_base), .reg_n_tiles(reg_n_tiles), .reg_x_base(reg_x_base), .reg_w_base(reg_w_base), .reg_n_tiles(reg_n_tiles),
.reg_result_addr(reg_result_addr), .reg_result_addr(reg_result_addr),
.slot_mem_req(slot_mem_req), .slot_mem_wr(slot_mem_wr), .slot_mem_addr(slot_mem_addr), .slot_mem_req(slot_mem_req), .slot_mem_wr(slot_mem_wr), .slot_mem_addr(slot_mem_addr),
.slot_mem_wdata(slot_mem_wdata), .slot_mem_rdata(slot_mem_rdata), .slot_mem_ready(slot_mem_ready) .slot_mem_wdata(slot_mem_wdata), .slot_mem_lb_n(slot_mem_lb_n), .slot_mem_ub_n(slot_mem_ub_n),
.slot_mem_rdata(slot_mem_rdata), .slot_mem_ready(slot_mem_ready)
); );
// ---- N_SLOTS -> 1 arbiter (M8, new) ---- // ---- N_SLOTS -> 1 arbiter (M8, word-level per DEC-0015) ----
wire arb_m_req, arb_m_wr; wire arb_m_req, arb_m_wr;
wire [ADDR_WIDTH-1:0] arb_m_addr; wire [ADDR_WIDTH-1:0] arb_m_addr;
wire signed [7:0] arb_m_wdata; wire [15:0] arb_m_wdata;
wire signed [7:0] arb_m_rdata; wire arb_m_lb_n, arb_m_ub_n;
wire [15:0] arb_m_rdata;
wire arb_m_ready; wire arb_m_ready;
slot_mem_arbiter #( slot_mem_arbiter #(
@@ -90,28 +106,16 @@ module neural_multiprocessor #(
) u_arbiter ( ) u_arbiter (
.clk(clk), .rst(rst), .clk(clk), .rst(rst),
.s_req(slot_mem_req), .s_wr(slot_mem_wr), .s_addr(slot_mem_addr), .s_req(slot_mem_req), .s_wr(slot_mem_wr), .s_addr(slot_mem_addr),
.s_wdata(slot_mem_wdata), .s_rdata(slot_mem_rdata), .s_ready(slot_mem_ready), .s_wdata(slot_mem_wdata), .s_lb_n(slot_mem_lb_n), .s_ub_n(slot_mem_ub_n),
.s_rdata(slot_mem_rdata), .s_ready(slot_mem_ready),
.m_req(arb_m_req), .m_wr(arb_m_wr), .m_addr(arb_m_addr), .m_wdata(arb_m_wdata), .m_req(arb_m_req), .m_wr(arb_m_wr), .m_addr(arb_m_addr), .m_wdata(arb_m_wdata),
.m_lb_n(arb_m_lb_n), .m_ub_n(arb_m_ub_n),
.m_rdata(arb_m_rdata), .m_ready(arb_m_ready) .m_rdata(arb_m_rdata), .m_ready(arb_m_ready)
); );
// ---- real, unmodified V1 PSRAM backend chain ---- // ---- real, unmodified V1 PSRAM backend chain (memory_interface
wire if_mem_req, if_mem_wr; // -> psram_controller; int8_memory_access no longer in this
wire [ADDR_WIDTH-1:0] if_mem_addr; // datapath -- see file header, DEC-0015) ----
wire [PSRAM_DATA_WIDTH-1:0] if_mem_wdata;
wire if_mem_lb_n, if_mem_ub_n;
wire [PSRAM_DATA_WIDTH-1:0] if_mem_rdata;
wire if_mem_ready;
int8_memory_access #(.ADDR_WIDTH(ADDR_WIDTH)) u_int8 (
.clk(clk), .rst(rst),
.req(arb_m_req), .wr(arb_m_wr), .addr(arb_m_addr), .wdata(arb_m_wdata),
.rdata(arb_m_rdata), .ready(arb_m_ready),
.mem_req(if_mem_req), .mem_wr(if_mem_wr), .mem_addr(if_mem_addr), .mem_wdata(if_mem_wdata),
.mem_lb_n(if_mem_lb_n), .mem_ub_n(if_mem_ub_n),
.mem_rdata(if_mem_rdata), .mem_ready(if_mem_ready)
);
wire pc_mem_req, pc_mem_wr; wire pc_mem_req, pc_mem_wr;
wire [ADDR_WIDTH-1:0] pc_mem_addr; wire [ADDR_WIDTH-1:0] pc_mem_addr;
wire [PSRAM_DATA_WIDTH-1:0] pc_mem_wdata; wire [PSRAM_DATA_WIDTH-1:0] pc_mem_wdata;
@@ -121,9 +125,9 @@ module neural_multiprocessor #(
memory_interface #(.ADDR_WIDTH(ADDR_WIDTH), .DATA_WIDTH(PSRAM_DATA_WIDTH)) u_memif ( memory_interface #(.ADDR_WIDTH(ADDR_WIDTH), .DATA_WIDTH(PSRAM_DATA_WIDTH)) u_memif (
.clk(clk), .rst(rst), .clk(clk), .rst(rst),
.req(if_mem_req), .wr(if_mem_wr), .addr(if_mem_addr), .wdata(if_mem_wdata), .req(arb_m_req), .wr(arb_m_wr), .addr(arb_m_addr), .wdata(arb_m_wdata),
.lb_n(if_mem_lb_n), .ub_n(if_mem_ub_n), .lb_n(arb_m_lb_n), .ub_n(arb_m_ub_n),
.rdata(if_mem_rdata), .ready(if_mem_ready), .rdata(arb_m_rdata), .ready(arb_m_ready),
.mem_req(pc_mem_req), .mem_wr(pc_mem_wr), .mem_addr(pc_mem_addr), .mem_wdata(pc_mem_wdata), .mem_req(pc_mem_req), .mem_wr(pc_mem_wr), .mem_addr(pc_mem_addr), .mem_wdata(pc_mem_wdata),
.mem_lb_n(pc_mem_lb_n), .mem_ub_n(pc_mem_ub_n), .mem_lb_n(pc_mem_lb_n), .mem_ub_n(pc_mem_ub_n),
.mem_rdata(pc_mem_rdata), .mem_ready(pc_mem_ready) .mem_rdata(pc_mem_rdata), .mem_ready(pc_mem_ready)
hardware/v2/rtl/prefetch_engine.v
+73 -36
View File
@@ -1,26 +1,43 @@
`timescale 1ns/1ps `timescale 1ns/1ps
// ================================================================ // ================================================================
// FPGA-Neural V2 -- Prefetch Engine (M4, docs/v2-description.md §13) // FPGA-Neural V2 -- Prefetch Engine (M4, docs/v2-description.md §13;
// word-level burst rewrite post-M10 -- see hardware/v2/logs/
// decisions.log DEC-0015)
// //
// Fetches ONE tile (P_IN activation bytes + P_IN weight bytes) from // Fetches ONE tile (P_IN activation bytes + P_IN weight bytes) from
// the byte-level Memory Backend Interface into a pair of output // the WORD-level Memory Backend Interface, P_IN/2 sixteen-bit
// registers, sequentially (2*P_IN single-byte transactions -- the // transactions per array instead of P_IN single-byte ones.
// same byte-at-a-time convention hardware/v1/rtl/neuron_memory.v
// already uses against the same backend, reused unmodified here).
// //
// This module fetches exactly one tile per fetch_start pulse; the // WHY: hardware/v1/rtl/int8_memory_access.v (the byte-level backend
// double-buffering strategy itself (§13: compute tile N while // this engine originally sat on) converts every 8-bit logical request
// prefetching tile N+1, swap, repeat) is memory_manager.v's // into a FULL 16-bit PSRAM word access internally (mem_addr <= addr
// responsibility -- it retargets this single engine at whichever // >> 1, one byte lane selected via lb_n/ub_n) -- so a byte-at-a-time
// bank currently needs refilling, so no internal arbitration between // fetch was ALREADY paying for two bytes of real PSRAM bandwidth per
// multiple fetch engines sharing the backend port is ever needed. // transaction while only using one. This engine now talks directly to
// hardware/v1/rtl/memory_interface.v's own 16-bit word interface
// (skipping int8_memory_access.v entirely -- both are frozen V1 files,
// unmodified either way, §1/§34; V2 is simply choosing to reuse the
// lower layer instead of the byte-splitting one on top of it, the
// same "reuse what fits" precedent already set by slot_mem_arbiter.v
// not reusing hardware/v1/rtl/mem_arbiter.v verbatim). psram_controller.v's
// own real page-mode support (already implemented, unmodified) then
// serves consecutive same-page word reads faster than a cold access --
// this engine's job is simply to stop discarding half of every word it
// already paid for, and to halve the number of real backend
// round-trips needed per tile.
// //
// The backend port (mem_req/mem_wr/mem_addr/mem_wdata/mem_rdata/ // CONSTRAINT: P_IN must be even, and x_addr/w_addr must be word-
// mem_ready) matches hardware/v1/rtl/int8_memory_access.v's contract // aligned (even BYTE addresses) -- each 16-bit transaction covers
// exactly -- this engine can sit directly on top of that unmodified // BYTE addresses {addr, addr+1} as {low byte, high byte} (matches
// V1 module (which itself sits on memory_interface.v -> // int8_memory_access.v's own addr[0] convention exactly, replicated
// psram_controller.v, also unmodified, per §15). // here since that module is no longer in the datapath). A host/loader
// placing X/W tile arrays at even byte offsets (already true of every
// address used in this project's own testbenches) satisfies this
// with no special handling.
//
// The double-buffering strategy itself (§13) remains memory_manager.v's
// responsibility -- unchanged by this rewrite.
// ================================================================ // ================================================================
module prefetch_engine #( module prefetch_engine #(
@@ -32,18 +49,22 @@ module prefetch_engine #(
input wire rst, input wire rst,
input wire fetch_start, input wire fetch_start,
input wire [ADDR_WIDTH-1:0] x_addr, // base addr of this tile's P_IN X bytes input wire [ADDR_WIDTH-1:0] x_addr, // BYTE address, word-aligned
input wire [ADDR_WIDTH-1:0] w_addr, // base addr of this tile's P_IN W bytes input wire [ADDR_WIDTH-1:0] w_addr, // BYTE address, word-aligned
output reg fetch_busy, output reg fetch_busy,
output reg fetch_done, // one-cycle pulse output reg fetch_done, // one-cycle pulse
output reg signed [DATA_WIDTH*P_IN-1:0] tile_x, output reg signed [DATA_WIDTH*P_IN-1:0] tile_x,
output reg signed [DATA_WIDTH*P_IN-1:0] tile_w, output reg signed [DATA_WIDTH*P_IN-1:0] tile_w,
// ---- word-level Memory Backend Interface (matches
// hardware/v1/rtl/memory_interface.v's contract exactly) ----
output reg mem_req, output reg mem_req,
output reg mem_wr, output reg mem_wr,
output reg [ADDR_WIDTH-1:0] mem_addr, output reg [ADDR_WIDTH-1:0] mem_addr, // WORD address
output reg signed [7:0] mem_wdata, output reg [15:0] mem_wdata,
input wire signed [7:0] mem_rdata, output reg mem_lb_n,
output reg mem_ub_n,
input wire [15:0] mem_rdata,
input wire mem_ready input wire mem_ready
); );
@@ -52,19 +73,27 @@ module prefetch_engine #(
localparam ST_READ_W = 2'd2; localparam ST_READ_W = 2'd2;
localparam ST_DONE = 2'd3; localparam ST_DONE = 2'd3;
localparam WORDS_PER_TILE = P_IN/2;
localparam WIW = $clog2(WORDS_PER_TILE+1);
reg [1:0] state; reg [1:0] state;
reg [$clog2(P_IN+1)-1:0] byte_idx; reg [WIW-1:0] word_idx;
wire [ADDR_WIDTH-1:0] x_word_base = x_addr[ADDR_WIDTH-1:1];
wire [ADDR_WIDTH-1:0] w_word_base = w_addr[ADDR_WIDTH-1:1];
always @(posedge clk) begin always @(posedge clk) begin
if (rst) begin if (rst) begin
state <= ST_IDLE; state <= ST_IDLE;
byte_idx <= 0; word_idx <= 0;
fetch_busy <= 1'b0; fetch_busy <= 1'b0;
fetch_done <= 1'b0; fetch_done <= 1'b0;
mem_req <= 1'b0; mem_req <= 1'b0;
mem_wr <= 1'b0; mem_wr <= 1'b0;
mem_addr <= {ADDR_WIDTH{1'b0}}; mem_addr <= {ADDR_WIDTH{1'b0}};
mem_wdata <= 8'sd0; mem_wdata <= 16'h0000;
mem_lb_n <= 1'b1;
mem_ub_n <= 1'b1;
end else begin end else begin
mem_req <= 1'b0; mem_req <= 1'b0;
fetch_done <= 1'b0; fetch_done <= 1'b0;
@@ -74,42 +103,50 @@ module prefetch_engine #(
ST_IDLE: begin ST_IDLE: begin
if (fetch_start) begin if (fetch_start) begin
fetch_busy <= 1'b1; fetch_busy <= 1'b1;
byte_idx <= 0; word_idx <= 0;
mem_req <= 1'b1; mem_req <= 1'b1;
mem_wr <= 1'b0; mem_wr <= 1'b0;
mem_addr <= x_addr; mem_addr <= x_word_base;
mem_lb_n <= 1'b0; // both byte lanes -- fetch the whole word
mem_ub_n <= 1'b0;
state <= ST_READ_X; state <= ST_READ_X;
end end
end end
ST_READ_X: begin ST_READ_X: begin
if (mem_ready) begin if (mem_ready) begin
tile_x[byte_idx*DATA_WIDTH +: DATA_WIDTH] <= mem_rdata; tile_x[word_idx*16 +: 16] <= mem_rdata;
if (byte_idx == P_IN[$clog2(P_IN+1)-1:0] - 1'b1) begin if (word_idx == WORDS_PER_TILE[WIW-1:0] - 1'b1) begin
byte_idx <= 0; word_idx <= 0;
mem_req <= 1'b1; mem_req <= 1'b1;
mem_wr <= 1'b0; mem_wr <= 1'b0;
mem_addr <= w_addr; mem_addr <= w_word_base;
mem_lb_n <= 1'b0;
mem_ub_n <= 1'b0;
state <= ST_READ_W; state <= ST_READ_W;
end else begin end else begin
byte_idx <= byte_idx + 1'b1; word_idx <= word_idx + 1'b1;
mem_req <= 1'b1; mem_req <= 1'b1;
mem_wr <= 1'b0; mem_wr <= 1'b0;
mem_addr <= x_addr + byte_idx + 1'b1; mem_addr <= x_word_base + word_idx + 1'b1;
mem_lb_n <= 1'b0;
mem_ub_n <= 1'b0;
end end
end end
end end
ST_READ_W: begin ST_READ_W: begin
if (mem_ready) begin if (mem_ready) begin
tile_w[byte_idx*DATA_WIDTH +: DATA_WIDTH] <= mem_rdata; tile_w[word_idx*16 +: 16] <= mem_rdata;
if (byte_idx == P_IN[$clog2(P_IN+1)-1:0] - 1'b1) begin if (word_idx == WORDS_PER_TILE[WIW-1:0] - 1'b1) begin
state <= ST_DONE; state <= ST_DONE;
end else begin end else begin
byte_idx <= byte_idx + 1'b1; word_idx <= word_idx + 1'b1;
mem_req <= 1'b1; mem_req <= 1'b1;
mem_wr <= 1'b0; mem_wr <= 1'b0;
mem_addr <= w_addr + byte_idx + 1'b1; mem_addr <= w_word_base + word_idx + 1'b1;
mem_lb_n <= 1'b0;
mem_ub_n <= 1'b0;
end end
end end
end end
hardware/v2/rtl/slot_mem_arbiter.v
+51 -33
View File
@@ -9,13 +9,18 @@
// funneling N_SLOTS independent memory_manager backend ports down to // funneling N_SLOTS independent memory_manager backend ports down to
// the ONE physical PSRAM port a real chip actually has. // the ONE physical PSRAM port a real chip actually has.
// //
// WORD-level (16-bit, + lb_n/ub_n) post-M10 (decisions.log DEC-0015):
// arbitrates hardware/v1/rtl/memory_interface.v's own port shape
// directly (int8_memory_access.v is no longer in this datapath -- see
// memory_manager.v/prefetch_engine.v's own headers for why).
//
// Inspired by (NOT copied from -- see hardware/v2/logs/decisions.log // Inspired by (NOT copied from -- see hardware/v2/logs/decisions.log
// DEC-0006's own note) hardware/v1/rtl/mem_arbiter.v: same // DEC-0006's own note) hardware/v1/rtl/mem_arbiter.v: same
// single-owner-until-ready-pulse discipline (a port, once granted, // single-owner-until-ready-pulse discipline (a port, once granted,
// holds the shared master port until ITS OWN transaction's m_ready // holds the shared master port until ITS OWN transaction's m_ready
// pulse, then releases -- no queuing/pipelining needed, since every // pulse, then releases -- no queuing/pipelining needed, since every
// requester already issues a clean one-cycle req pulse matching // requester already issues a clean one-cycle req pulse matching
// int8_memory_access's own contract). Generalized from V1's fixed // memory_interface's own contract). Generalized from V1's fixed
// 4 named ports (A/B/C/D) to a parametric N_PORTS array, since // 4 named ports (A/B/C/D) to a parametric N_PORTS array, since
// dataflow_core.v's N_SLOTS is itself a parameter. // dataflow_core.v's N_SLOTS is itself a parameter.
// //
@@ -30,21 +35,20 @@
// IMPORTANT (found via real concurrent-slot simulation, see // IMPORTANT (found via real concurrent-slot simulation, see
// hardware/v2/logs/errors.log ERR-0008): each port's own s_req is a // hardware/v2/logs/errors.log ERR-0008): each port's own s_req is a
// FIRE-AND-FORGET single-cycle pulse (prefetch_engine.v/ // FIRE-AND-FORGET single-cycle pulse (prefetch_engine.v/
// memory_manager.v's own byte-level backend protocol -- M4 verified // memory_manager.v's own backend protocol -- M4 verified it only
// it only against a DIRECT 1:1 connection to int8_memory_access, // against a DIRECT 1:1 connection to the backend, which is always
// which is always free to accept it since there is exactly one // free to accept it since there is exactly one requester). A naive
// requester). A naive "grant only while req is live" arbiter silently // "grant only while req is live" arbiter silently DROPS a pulse that
// DROPS a pulse that arrives while the shared bus is owned by another // arrives while the shared bus is owned by another port, hanging that
// port, hanging that slot's prefetch/writeback forever. Every // slot's prefetch/writeback forever. Every incoming s_req is therefore
// incoming s_req is therefore LATCHED into a per-port `pending` // LATCHED into a per-port `pending` register (capturing wr/addr/wdata/
// register (capturing wr/addr/wdata the same cycle) regardless of // lb_n/ub_n the same cycle) regardless of arbiter state -- the same
// arbiter state -- the same single-entry "queue, don't drop the // single-entry "queue, don't drop the request" idiom already used by
// request" idiom already used by memory_manager's own pf_pending // memory_manager's own pf_pending register (ERR-0006 fix #1). Grants
// register (ERR-0006 fix #1). Grants are drawn from `pending`, never // are drawn from `pending`, never from a live s_req directly, which
// from a live s_req directly, which adds a uniform minimum 1-cycle // adds a uniform minimum 1-cycle latency to every transaction (a real,
// latency to every byte transaction (a real, measured cost of sharing // measured cost of sharing one PSRAM port -- see timing.log/
// one PSRAM port -- see timing.log/benchmark.log EXP-0009) but never // benchmark.log EXP-0009) but never drops a request.
// drops a request.
// ================================================================ // ================================================================
module slot_mem_arbiter #( module slot_mem_arbiter #(
@@ -57,17 +61,21 @@ module slot_mem_arbiter #(
// ---- N_PORTS requester side (one per dataflow_core slot) ---- // ---- N_PORTS requester side (one per dataflow_core slot) ----
input wire [N_PORTS-1:0] s_req, input wire [N_PORTS-1:0] s_req,
input wire [N_PORTS-1:0] s_wr, input wire [N_PORTS-1:0] s_wr,
input wire [ADDR_WIDTH*N_PORTS-1:0] s_addr, input wire [ADDR_WIDTH*N_PORTS-1:0] s_addr, // WORD address
input wire signed [8*N_PORTS-1:0] s_wdata, input wire [16*N_PORTS-1:0] s_wdata,
output reg signed [8*N_PORTS-1:0] s_rdata, input wire [N_PORTS-1:0] s_lb_n,
input wire [N_PORTS-1:0] s_ub_n,
output reg [16*N_PORTS-1:0] s_rdata,
output reg [N_PORTS-1:0] s_ready, output reg [N_PORTS-1:0] s_ready,
// ---- single shared master port (-> int8_memory_access) ---- // ---- single shared master port (-> memory_interface.v) ----
output reg m_req, output reg m_req,
output reg m_wr, output reg m_wr,
output reg [ADDR_WIDTH-1:0] m_addr, output reg [ADDR_WIDTH-1:0] m_addr,
output reg signed [7:0] m_wdata, output reg [15:0] m_wdata,
input wire signed [7:0] m_rdata, output reg m_lb_n,
output reg m_ub_n,
input wire [15:0] m_rdata,
input wire m_ready input wire m_ready
); );
@@ -81,8 +89,10 @@ module slot_mem_arbiter #(
// is never silently dropped while the bus is owned by another port. // is never silently dropped while the bus is owned by another port.
reg [N_PORTS-1:0] pending; reg [N_PORTS-1:0] pending;
reg [ADDR_WIDTH*N_PORTS-1:0] pending_addr; reg [ADDR_WIDTH*N_PORTS-1:0] pending_addr;
reg signed [8*N_PORTS-1:0] pending_wdata; reg [16*N_PORTS-1:0] pending_wdata;
reg [N_PORTS-1:0] pending_wr; reg [N_PORTS-1:0] pending_wr;
reg [N_PORTS-1:0] pending_lb_n;
reg [N_PORTS-1:0] pending_ub_n;
// Fixed lowest-index-wins priority scan over PENDING requests (not // Fixed lowest-index-wins priority scan over PENDING requests (not
// raw s_req -- see file header). // raw s_req -- see file header).
@@ -107,13 +117,17 @@ module slot_mem_arbiter #(
owner <= OWNER_NONE; owner <= OWNER_NONE;
pending <= {N_PORTS{1'b0}}; pending <= {N_PORTS{1'b0}};
pending_addr <= {(ADDR_WIDTH*N_PORTS){1'b0}}; pending_addr <= {(ADDR_WIDTH*N_PORTS){1'b0}};
pending_wdata <= {(8*N_PORTS){1'b0}}; pending_wdata <= {(16*N_PORTS){1'b0}};
pending_wr <= {N_PORTS{1'b0}}; pending_wr <= {N_PORTS{1'b0}};
pending_lb_n <= {N_PORTS{1'b1}};
pending_ub_n <= {N_PORTS{1'b1}};
m_req <= 1'b0; m_req <= 1'b0;
m_wr <= 1'b0; m_wr <= 1'b0;
m_addr <= {ADDR_WIDTH{1'b0}}; m_addr <= {ADDR_WIDTH{1'b0}};
m_wdata <= 8'sd0; m_wdata <= 16'h0000;
s_rdata <= {(8*N_PORTS){1'b0}}; m_lb_n <= 1'b1;
m_ub_n <= 1'b1;
s_rdata <= {(16*N_PORTS){1'b0}};
s_ready <= {N_PORTS{1'b0}}; s_ready <= {N_PORTS{1'b0}};
end else begin end else begin
m_req <= 1'b0; m_req <= 1'b0;
@@ -129,10 +143,12 @@ module slot_mem_arbiter #(
// the cycle it is granted. // the cycle it is granted.
for (pi = 0; pi < N_PORTS; pi = pi + 1) begin for (pi = 0; pi < N_PORTS; pi = pi + 1) begin
if (s_req[pi]) begin if (s_req[pi]) begin
pending[pi] <= 1'b1; pending[pi] <= 1'b1;
pending_wr[pi] <= s_wr[pi]; pending_wr[pi] <= s_wr[pi];
pending_addr[pi*ADDR_WIDTH +: ADDR_WIDTH] <= s_addr[pi*ADDR_WIDTH +: ADDR_WIDTH]; pending_lb_n[pi] <= s_lb_n[pi];
pending_wdata[pi*8 +: 8] <= s_wdata[pi*8 +: 8]; pending_ub_n[pi] <= s_ub_n[pi];
pending_addr[pi*ADDR_WIDTH +: ADDR_WIDTH] <= s_addr[pi*ADDR_WIDTH +: ADDR_WIDTH];
pending_wdata[pi*16 +: 16] <= s_wdata[pi*16 +: 16];
end end
end end
@@ -141,8 +157,10 @@ module slot_mem_arbiter #(
owner <= grant_idx + 1'b1; owner <= grant_idx + 1'b1;
m_req <= 1'b1; m_req <= 1'b1;
m_wr <= pending_wr[grant_idx]; m_wr <= pending_wr[grant_idx];
m_lb_n <= pending_lb_n[grant_idx];
m_ub_n <= pending_ub_n[grant_idx];
m_addr <= pending_addr[grant_idx*ADDR_WIDTH +: ADDR_WIDTH]; m_addr <= pending_addr[grant_idx*ADDR_WIDTH +: ADDR_WIDTH];
m_wdata <= pending_wdata[grant_idx*8 +: 8]; m_wdata <= pending_wdata[grant_idx*16 +: 16];
pending[grant_idx] <= 1'b0; pending[grant_idx] <= 1'b0;
end end
end else begin end else begin
@@ -153,8 +171,8 @@ module slot_mem_arbiter #(
// of bug already hit/fixed at ERR-0006/M2/M6). // of bug already hit/fixed at ERR-0006/M2/M6).
for (pi = 0; pi < N_PORTS; pi = pi + 1) begin for (pi = 0; pi < N_PORTS; pi = pi + 1) begin
if (owner == pi[PIDXW-1:0] + 1'b1) begin if (owner == pi[PIDXW-1:0] + 1'b1) begin
s_rdata[pi*8 +: 8] <= m_rdata; s_rdata[pi*16 +: 16] <= m_rdata;
s_ready[pi] <= 1'b1; s_ready[pi] <= 1'b1;
end end
end end
owner <= OWNER_NONE; owner <= OWNER_NONE;
hardware/v2/sim/tb_dataflow_core.v
+40 -19
View File
@@ -21,13 +21,17 @@
// node1 (x=1,w=1,8in -> acc=8) --+ // node1 (x=1,w=1,8in -> acc=8) --+
// //
// Verified with Verilator (decisions.log DEC-0004). Each slot gets // Verified with Verilator (decisions.log DEC-0004). Each slot gets
// its own independent behavioral memory (sim_byte_mem, same as // its own independent behavioral memory (sim_word_mem, same as
// tb_neural_director.v/tb_memory_manager.v's own scope decisions -- // tb_neural_director.v/tb_memory_manager.v's own scope decisions --
// DEC-0006/DEC-0007: shared-PSRAM arbitration across slots is // DEC-0006/DEC-0007: shared-PSRAM arbitration across slots is
// explicitly M8's job, not exercised here). // explicitly M8's job, not exercised here).
//
// WORD-level (16-bit, + lb_n/ub_n) post-M10 (decisions.log DEC-0015),
// matching memory_manager.v's own backend port width after the
// burst-read rewrite (see prefetch_engine.v/memory_manager.v headers).
// ============================================================ // ============================================================
module sim_byte_mem #( module sim_word_mem #(
parameter ADDR_WIDTH = 23, parameter ADDR_WIDTH = 23,
parameter DEPTH = 4096 parameter DEPTH = 4096
)( )(
@@ -35,24 +39,28 @@ module sim_byte_mem #(
input wire rst, input wire rst,
input wire req, input wire req,
input wire wr, input wire wr,
input wire [ADDR_WIDTH-1:0] addr, input wire [ADDR_WIDTH-1:0] addr, // WORD address
input wire signed [7:0] wdata, input wire [15:0] wdata,
output reg signed [7:0] rdata, input wire lb_n, ub_n,
output reg [15:0] rdata,
output reg ready output reg ready
); );
reg signed [7:0] mem [0:DEPTH-1]; reg [15:0] mem [0:DEPTH-1];
reg [1:0] state; reg [1:0] state;
reg [ADDR_WIDTH-1:0] addr_reg; reg [ADDR_WIDTH-1:0] addr_reg;
localparam ST_IDLE = 0, ST_WAIT = 1; localparam ST_IDLE = 0, ST_WAIT = 1;
always @(posedge clk) begin always @(posedge clk) begin
if (rst) begin if (rst) begin
state <= ST_IDLE; ready <= 1'b0; rdata <= 8'sd0; state <= ST_IDLE; ready <= 1'b0; rdata <= 16'h0000;
end else begin end else begin
ready <= 1'b0; ready <= 1'b0;
case (state) case (state)
ST_IDLE: if (req) begin ST_IDLE: if (req) begin
addr_reg <= addr; addr_reg <= addr;
if (wr) mem[addr] <= wdata; if (wr) begin
if (!lb_n) mem[addr][7:0] <= wdata[7:0];
if (!ub_n) mem[addr][15:8] <= wdata[15:8];
end
state <= ST_WAIT; state <= ST_WAIT;
end end
ST_WAIT: begin ST_WAIT: begin
@@ -90,7 +98,8 @@ module tb;
wire [N_SLOTS-1:0] slot_mem_req, slot_mem_wr; wire [N_SLOTS-1:0] slot_mem_req, slot_mem_wr;
wire [ADDR_WIDTH*N_SLOTS-1:0] slot_mem_addr; wire [ADDR_WIDTH*N_SLOTS-1:0] slot_mem_addr;
wire signed [8*N_SLOTS-1:0] slot_mem_wdata, slot_mem_rdata; wire [16*N_SLOTS-1:0] slot_mem_wdata, slot_mem_rdata;
wire [N_SLOTS-1:0] slot_mem_lb_n, slot_mem_ub_n;
wire [N_SLOTS-1:0] slot_mem_ready; wire [N_SLOTS-1:0] slot_mem_ready;
dataflow_core #( dataflow_core #(
@@ -103,37 +112,49 @@ module tb;
.reg_x_base(reg_x_base), .reg_w_base(reg_w_base), .reg_n_tiles(reg_n_tiles), .reg_x_base(reg_x_base), .reg_w_base(reg_w_base), .reg_n_tiles(reg_n_tiles),
.reg_result_addr(reg_result_addr), .reg_result_addr(reg_result_addr),
.slot_mem_req(slot_mem_req), .slot_mem_wr(slot_mem_wr), .slot_mem_addr(slot_mem_addr), .slot_mem_req(slot_mem_req), .slot_mem_wr(slot_mem_wr), .slot_mem_addr(slot_mem_addr),
.slot_mem_wdata(slot_mem_wdata), .slot_mem_rdata(slot_mem_rdata), .slot_mem_ready(slot_mem_ready) .slot_mem_wdata(slot_mem_wdata), .slot_mem_lb_n(slot_mem_lb_n), .slot_mem_ub_n(slot_mem_ub_n),
.slot_mem_rdata(slot_mem_rdata), .slot_mem_ready(slot_mem_ready)
); );
genvar g; genvar g;
generate generate
for (g = 0; g < N_SLOTS; g = g + 1) begin : GEN_MEM for (g = 0; g < N_SLOTS; g = g + 1) begin : GEN_MEM
sim_byte_mem #(.ADDR_WIDTH(ADDR_WIDTH), .DEPTH(4096)) u_mem ( sim_word_mem #(.ADDR_WIDTH(ADDR_WIDTH), .DEPTH(4096)) u_mem (
.clk(clk), .rst(rst), .clk(clk), .rst(rst),
.req(slot_mem_req[g]), .wr(slot_mem_wr[g]), .req(slot_mem_req[g]), .wr(slot_mem_wr[g]),
.addr(slot_mem_addr[g*ADDR_WIDTH +: ADDR_WIDTH]), .addr(slot_mem_addr[g*ADDR_WIDTH +: ADDR_WIDTH]),
.wdata(slot_mem_wdata[g*8 +: 8]), .wdata(slot_mem_wdata[g*16 +: 16]),
.rdata(slot_mem_rdata[g*8 +: 8]), .ready(slot_mem_ready[g]) .lb_n(slot_mem_lb_n[g]), .ub_n(slot_mem_ub_n[g]),
.rdata(slot_mem_rdata[g*16 +: 16]), .ready(slot_mem_ready[g])
); );
end end
endgenerate endgenerate
task automatic poke(input integer slot, input [ADDR_WIDTH-1:0] addr, input [7:0] val); // poke/peek stay BYTE-addressed at the testbench level (matching
// every other testbench's own convention) -- converted to
// word-address + byte-lane internally, same as psram_model.v's
// own real convention.
task automatic poke(input integer slot, input [ADDR_WIDTH-1:0] byte_addr, input [7:0] val);
reg [ADDR_WIDTH-2:0] word_addr;
begin begin
word_addr = byte_addr[ADDR_WIDTH-1:1];
case (slot) case (slot)
0: tb.GEN_MEM[0].u_mem.mem[addr] = val; 0: if (byte_addr[0]==1'b0) tb.GEN_MEM[0].u_mem.mem[word_addr][7:0] = val;
1: tb.GEN_MEM[1].u_mem.mem[addr] = val; else tb.GEN_MEM[0].u_mem.mem[word_addr][15:8] = val;
1: if (byte_addr[0]==1'b0) tb.GEN_MEM[1].u_mem.mem[word_addr][7:0] = val;
else tb.GEN_MEM[1].u_mem.mem[word_addr][15:8] = val;
default: ; default: ;
endcase endcase
end end
endtask endtask
function automatic signed [7:0] peek(input integer slot, input [ADDR_WIDTH-1:0] addr); function automatic signed [7:0] peek(input integer slot, input [ADDR_WIDTH-1:0] byte_addr);
reg [ADDR_WIDTH-2:0] word_addr;
begin begin
word_addr = byte_addr[ADDR_WIDTH-1:1];
case (slot) case (slot)
0: peek = tb.GEN_MEM[0].u_mem.mem[addr]; 0: peek = (byte_addr[0]==1'b0) ? tb.GEN_MEM[0].u_mem.mem[word_addr][7:0] : tb.GEN_MEM[0].u_mem.mem[word_addr][15:8];
1: peek = tb.GEN_MEM[1].u_mem.mem[addr]; 1: peek = (byte_addr[0]==1'b0) ? tb.GEN_MEM[1].u_mem.mem[word_addr][7:0] : tb.GEN_MEM[1].u_mem.mem[word_addr][15:8];
default: peek = 8'sdx; default: peek = 8'sdx;
endcase endcase
end end
hardware/v2/sim/tb_memory_manager.v
+14 -24
View File
@@ -50,11 +50,14 @@ module tb;
wire mm_result_valid, mm_result_ready; wire mm_result_valid, mm_result_ready;
wire signed [DATA_WIDTH-1:0] mm_result_data; wire signed [DATA_WIDTH-1:0] mm_result_data;
// ---- memory_manager <-> int8_memory_access (Memory Backend Interface) ---- // ---- memory_manager <-> memory_interface (word-level Memory
// Backend Interface, post-M10 DEC-0015 -- int8_memory_access is no
// longer in this datapath, see memory_manager.v's own header) ----
wire mem_req, mem_wr; wire mem_req, mem_wr;
wire [ADDR_WIDTH-1:0] mem_addr; wire [ADDR_WIDTH-1:0] mem_addr; // WORD address
wire signed [7:0] mem_wdata; wire [15:0] mem_wdata;
wire signed [7:0] mem_rdata; wire mem_lb_n, mem_ub_n;
wire [15:0] mem_rdata;
wire mem_ready; wire mem_ready;
memory_manager #( memory_manager #(
@@ -67,6 +70,7 @@ module tb;
.input_data(mm_input_data), .weight_data(mm_weight_data), .tile_last(mm_tile_last), .input_data(mm_input_data), .weight_data(mm_weight_data), .tile_last(mm_tile_last),
.result_valid(mm_result_valid), .result_ready(mm_result_ready), .result_data(mm_result_data), .result_valid(mm_result_valid), .result_ready(mm_result_ready), .result_data(mm_result_data),
.mem_req(mem_req), .mem_wr(mem_wr), .mem_addr(mem_addr), .mem_wdata(mem_wdata), .mem_req(mem_req), .mem_wr(mem_wr), .mem_addr(mem_addr), .mem_wdata(mem_wdata),
.mem_lb_n(mem_lb_n), .mem_ub_n(mem_ub_n),
.mem_rdata(mem_rdata), .mem_ready(mem_ready) .mem_rdata(mem_rdata), .mem_ready(mem_ready)
); );
@@ -105,23 +109,9 @@ module tb;
else if (job_valid_np && job_ready_np) job_valid_np <= 1'b0; else if (job_valid_np && job_ready_np) job_valid_np <= 1'b0;
end end
// ---- REAL, unmodified V1 backend chain ---- // ---- REAL, unmodified V1 backend chain (memory_interface ->
wire if_mem_req, if_mem_wr; // psram_controller; int8_memory_access no longer in this datapath,
wire [ADDR_WIDTH-1:0] if_mem_addr; // see memory_manager.v's own header, DEC-0015) ----
wire [PSRAM_DATA_WIDTH-1:0] if_mem_wdata;
wire if_mem_lb_n, if_mem_ub_n;
wire [PSRAM_DATA_WIDTH-1:0] if_mem_rdata;
wire if_mem_ready;
int8_memory_access #(.ADDR_WIDTH(ADDR_WIDTH)) u_int8 (
.clk(clk), .rst(rst),
.req(mem_req), .wr(mem_wr), .addr(mem_addr), .wdata(mem_wdata),
.rdata(mem_rdata), .ready(mem_ready),
.mem_req(if_mem_req), .mem_wr(if_mem_wr), .mem_addr(if_mem_addr), .mem_wdata(if_mem_wdata),
.mem_lb_n(if_mem_lb_n), .mem_ub_n(if_mem_ub_n),
.mem_rdata(if_mem_rdata), .mem_ready(if_mem_ready)
);
wire pc_mem_req, pc_mem_wr; wire pc_mem_req, pc_mem_wr;
wire [ADDR_WIDTH-1:0] pc_mem_addr; wire [ADDR_WIDTH-1:0] pc_mem_addr;
wire [PSRAM_DATA_WIDTH-1:0] pc_mem_wdata; wire [PSRAM_DATA_WIDTH-1:0] pc_mem_wdata;
@@ -131,9 +121,9 @@ module tb;
memory_interface #(.ADDR_WIDTH(ADDR_WIDTH), .DATA_WIDTH(PSRAM_DATA_WIDTH)) u_memif ( memory_interface #(.ADDR_WIDTH(ADDR_WIDTH), .DATA_WIDTH(PSRAM_DATA_WIDTH)) u_memif (
.clk(clk), .rst(rst), .clk(clk), .rst(rst),
.req(if_mem_req), .wr(if_mem_wr), .addr(if_mem_addr), .wdata(if_mem_wdata), .req(mem_req), .wr(mem_wr), .addr(mem_addr), .wdata(mem_wdata),
.lb_n(if_mem_lb_n), .ub_n(if_mem_ub_n), .lb_n(mem_lb_n), .ub_n(mem_ub_n),
.rdata(if_mem_rdata), .ready(if_mem_ready), .rdata(mem_rdata), .ready(mem_ready),
.mem_req(pc_mem_req), .mem_wr(pc_mem_wr), .mem_addr(pc_mem_addr), .mem_wdata(pc_mem_wdata), .mem_req(pc_mem_req), .mem_wr(pc_mem_wr), .mem_addr(pc_mem_addr), .mem_wdata(pc_mem_wdata),
.mem_lb_n(pc_mem_lb_n), .mem_ub_n(pc_mem_ub_n), .mem_lb_n(pc_mem_lb_n), .mem_ub_n(pc_mem_ub_n),
.mem_rdata(pc_mem_rdata), .mem_ready(pc_mem_ready) .mem_rdata(pc_mem_rdata), .mem_ready(pc_mem_ready)