test(realdata): PR-K — synthetic regen of doppler/fullchain realdata fixtures

Replaces the legacy ADI CN0566 .npy capture flow with a synthetic radar
scene generated by tb/cosim/real_data/gen_realdata_hex.py via the
existing radar_scene + fpga_model bit-accurate Python models.

Dimensions now match production radar_params.vh:
  RP_FFT_SIZE=2048, RP_DECIMATION_FACTOR=4, RP_NUM_RANGE_BINS=512,
  CHIRPS_PER_FRAME=48, NUM_DOPPLER_BINS=48 (3 sub-frames x 16-pt FFT).

Previously both TBs were pinned to legacy 32-chirp / 2-subframe / 1024->64
DECIM=16 dimensions. range_bin_decimator.v's 2-bit comparisons against
DECIMATION_FACTOR/2 only behave correctly for small DECIM, so the old
DECIM=16 path no longer worked even though the TBs compiled — that is
why Full-Chain Real-Data was reporting pass=0/fail=3.

Changes:
  tb/cosim/real_data/gen_realdata_hex.py  (new) - synthesises 6 fixture
    files from a 2-target scene via DopplerProcessor (3-subframe) and
    RangeBinDecimator (peak, 2048->512). Reproducible (fixed seed 42).

  tb/cosim/real_data/golden_reference.py  (deleted, 1436 lines) - the
    legacy generator depended on out-of-tree ADI .npy captures and
    modelled only the 2-subframe / 32-chirp path.

  tb/cosim/real_data/hex/  - 43 orphan artifacts deleted (CFAR / MTI /
    notched / detection / range-FFT debug dumps that nothing in the
    active TB or regression was loading); 6 fixtures regenerated at
    production dimensions:
      doppler_input_realdata.hex     24576 packed lines (was 2048)
      doppler_ref_{i,q}.hex          24576 lines each   (was 2048)
      fullchain_range_input.hex      98304 packed lines (was 32768)
      fullchain_doppler_ref_{i,q}.hex 24576 lines each  (was 2048)

  tb/tb_doppler_realdata.v          - CHIRPS 32->48, RANGE_BINS 64->512,
                                       DOPPLER_FFT 32->48, MAX_CYCLES bumped.
  tb/tb_fullchain_realdata.v        - same + INPUT_BINS 1024->2048,
                                       DECIM_FACTOR 16->4, fixed
                                       decim_bin_index width to
                                       RP_RANGE_BIN_WIDTH_MAX, fixed
                                       start_bin width 10->11.

  run_regression.sh                 - "Doppler Real-Data" label updated
                                       (no longer "ADI CN0566"); both
                                       realdata tests get explicit
                                       --timeout values (300 / 600 s).

Standalone results:
  tb_doppler_realdata    - 24584/24584 PASS (3.36 s sim, ~50 s wall)
  tb_fullchain_realdata  - 24585/24585 PASS (4.10 s sim, ~5 min wall)

Full regression now: 41 passed / 1 failed (only remaining FAIL is
FFT Engine, pre-existing pre-PR-K regex-reveal — unrelated).

9_Firmware/9_2_FPGA/tb/cosim/real_data/gen_realdata_hex.py

@@ -0,0 +1,182 @@
+#!/usr/bin/env python3
+"""
+gen_realdata_hex.py — Synthetic stimulus + bit-exact golden for the realdata
+co-simulation testbenches (tb_doppler_realdata, tb_fullchain_realdata).
+
+Replaces the legacy ADI CN0566 hardware captures (32-chirp / 2-subframe /
+32-bin Doppler) with a synthetic radar scene at production dimensions
+(48-chirp / 3-subframe / 48-bin Doppler) so the regression no longer
+depends on out-of-tree .npy files.
+
+Outputs (six files, all under tb/cosim/real_data/hex/):
+  doppler_input_realdata.hex      48 chirps x 512 range bins, packed {Q,I}
+  doppler_ref_i.hex / _q.hex      512 range bins x 48 Doppler bins (signed 16-bit)
+  fullchain_range_input.hex       48 chirps x 2048 range bins, packed {Q,I}
+  fullchain_doppler_ref_i.hex
+  fullchain_doppler_ref_q.hex     same shape as doppler_ref_*
+
+Dimensions match production (radar_params.vh: RP_FFT_SIZE=2048,
+RP_DECIMATION_FACTOR=4, RP_NUM_RANGE_BINS=512, RP_NUM_DOPPLER_BINS=48).
+
+Pipeline modeled (bit-exact to RTL):
+  doppler-only:  scene -> doppler_processor (3-subframe, 16-pt FFT, Hamming)
+  fullchain:     scene -> range_bin_decimator (2048->512 peak, DECIM=4)
+                       -> doppler_processor
+
+Usage:  python3 gen_realdata_hex.py
+"""
+
+import os
+import sys
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from fpga_model import DopplerProcessor, RangeBinDecimator
+from radar_scene import Target, generate_doppler_frame
+
+
+# ----------------------------------------------------------------------------
+# Production dimensions (radar_params.vh: PR-F + RP_FFT_SIZE/RP_DECIMATION_FACTOR)
+# ----------------------------------------------------------------------------
+NUM_SUBFRAMES        = 3
+DOPPLER_FFT_SIZE     = 16
+DOPPLER_TOTAL_BINS   = NUM_SUBFRAMES * DOPPLER_FFT_SIZE   # 48
+CHIRPS_PER_SUBFRAME  = 16
+CHIRPS_PER_FRAME     = NUM_SUBFRAMES * CHIRPS_PER_SUBFRAME  # 48
+
+# Doppler-only TB: post-decim range bins fed straight into doppler.
+# Matches production RP_NUM_RANGE_BINS = RP_FFT_SIZE / RP_DECIMATION_FACTOR.
+DOPPLER_RANGE_BINS   = 512
+
+# Fullchain TB: pre-decim 2048-bin range FFT -> range_bin_decimator (DECIM=4) -> doppler.
+FULLCHAIN_INPUT_BINS    = 2048
+FULLCHAIN_OUTPUT_BINS   = DOPPLER_RANGE_BINS              # 512
+FULLCHAIN_DECIM_FACTOR  = FULLCHAIN_INPUT_BINS // FULLCHAIN_OUTPUT_BINS   # 4
+
+OUT_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "hex")
+
+# ----------------------------------------------------------------------------
+# Scene — two distinct targets so both the FFT bin layout and the slow-time
+# Doppler axis carry information across all three sub-frames.
+# ----------------------------------------------------------------------------
+SCENE_TARGETS = [
+    Target(range_m=300,  velocity_mps= 10.0, rcs_dbsm=20.0),
+    Target(range_m=800,  velocity_mps=-20.0, rcs_dbsm=15.0),
+]
+SCENE_SEED = 42
+
+
+def write_hex_32(path, samples_iq):
+    """Packed 32-bit {Q[31:16], I[15:0]} per line for $readmemh."""
+    with open(path, 'w') as f:
+        for (i_val, q_val) in samples_iq:
+            packed = ((q_val & 0xFFFF) << 16) | (i_val & 0xFFFF)
+            f.write(f"{packed:08X}\n")
+
+
+def write_hex_16(path, values):
+    """One signed 16-bit value per line, two's-complement hex."""
+    with open(path, 'w') as f:
+        for v in values:
+            f.write(f"{v & 0xFFFF:04X}\n")
+
+
+def make_doppler_processor():
+    """DopplerProcessor at production dimensions (3 sub-frames, 512 range bins)."""
+    return DopplerProcessor()                        # defaults: NUM_SUBFRAMES=3, RANGE_BINS=512
+
+
+def flatten_doppler_map(doppler_i, doppler_q):
+    """RTL stream order: rbin 0 [dbin 0..47], rbin 1 [...], ..., rbin 63 [...]."""
+    flat_i, flat_q = [], []
+    for rb in range(DOPPLER_RANGE_BINS):
+        for db in range(DOPPLER_TOTAL_BINS):
+            flat_i.append(doppler_i[rb][db])
+            flat_q.append(doppler_q[rb][db])
+    return flat_i, flat_q
+
+
+def gen_doppler_realdata():
+    """tb_doppler_realdata: post-decim 48 x 512 -> doppler 512 x 48."""
+    print("[doppler_realdata] generating ...")
+    frame_i, frame_q = generate_doppler_frame(
+        SCENE_TARGETS,
+        n_chirps=CHIRPS_PER_FRAME,
+        n_range_bins=DOPPLER_RANGE_BINS,
+        seed=SCENE_SEED,
+    )
+
+    stim = []
+    for c in range(CHIRPS_PER_FRAME):
+        for rb in range(DOPPLER_RANGE_BINS):
+            stim.append((frame_i[c][rb], frame_q[c][rb]))
+    write_hex_32(os.path.join(OUT_DIR, "doppler_input_realdata.hex"), stim)
+
+    dp = make_doppler_processor()
+    doppler_i, doppler_q = dp.process_frame(frame_i, frame_q)
+    flat_i, flat_q = flatten_doppler_map(doppler_i, doppler_q)
+    write_hex_16(os.path.join(OUT_DIR, "doppler_ref_i.hex"), flat_i)
+    write_hex_16(os.path.join(OUT_DIR, "doppler_ref_q.hex"), flat_q)
+
+    print(f"  stimulus: {len(stim)} packed lines (expected {CHIRPS_PER_FRAME * DOPPLER_RANGE_BINS})")
+    print(f"  golden:   {len(flat_i)} lines i / {len(flat_q)} lines q "
+          f"(expected {DOPPLER_RANGE_BINS * DOPPLER_TOTAL_BINS})")
+
+
+def gen_fullchain_realdata():
+    """tb_fullchain_realdata: 48 x 2048 -> RangeBinDecimator (DECIM=4 peak) -> doppler 512 x 48."""
+    print("[fullchain_realdata] generating ...")
+    frame_i, frame_q = generate_doppler_frame(
+        SCENE_TARGETS,
+        n_chirps=CHIRPS_PER_FRAME,
+        n_range_bins=FULLCHAIN_INPUT_BINS,
+        seed=SCENE_SEED,
+    )
+
+    stim = []
+    for c in range(CHIRPS_PER_FRAME):
+        for rb in range(FULLCHAIN_INPUT_BINS):
+            stim.append((frame_i[c][rb], frame_q[c][rb]))
+    write_hex_32(os.path.join(OUT_DIR, "fullchain_range_input.hex"), stim)
+
+    # fpga_model.RangeBinDecimator is hard-coded to 2048->512, DECIM=4 — production.
+    decim_i_2d, decim_q_2d = [], []
+    for c in range(CHIRPS_PER_FRAME):
+        di, dq = RangeBinDecimator.decimate(frame_i[c], frame_q[c], mode=1, start_bin=0)
+        decim_i_2d.append(di)
+        decim_q_2d.append(dq)
+
+    dp = make_doppler_processor()
+    doppler_i, doppler_q = dp.process_frame(decim_i_2d, decim_q_2d)
+    flat_i, flat_q = flatten_doppler_map(doppler_i, doppler_q)
+    write_hex_16(os.path.join(OUT_DIR, "fullchain_doppler_ref_i.hex"), flat_i)
+    write_hex_16(os.path.join(OUT_DIR, "fullchain_doppler_ref_q.hex"), flat_q)
+
+    print(f"  stimulus: {len(stim)} packed lines "
+          f"(expected {CHIRPS_PER_FRAME * FULLCHAIN_INPUT_BINS})")
+    print(f"  golden:   {len(flat_i)} lines i / {len(flat_q)} lines q "
+          f"(expected {DOPPLER_RANGE_BINS * DOPPLER_TOTAL_BINS})")
+
+
+def main():
+    os.makedirs(OUT_DIR, exist_ok=True)
+    gen_doppler_realdata()
+    gen_fullchain_realdata()
+
+    print("\nGenerated files:")
+    for f in (
+        "doppler_input_realdata.hex",
+        "doppler_ref_i.hex",
+        "doppler_ref_q.hex",
+        "fullchain_range_input.hex",
+        "fullchain_doppler_ref_i.hex",
+        "fullchain_doppler_ref_q.hex",
+    ):
+        path = os.path.join(OUT_DIR, f)
+        with open(path) as fp:
+            n_lines = sum(1 for _ in fp)
+        print(f"  {f:40s}  {n_lines:7d} lines  ({os.path.getsize(path):7d} bytes)")
+
+
+if __name__ == '__main__':
+    main()

9_Firmware/9_2_FPGA/tb/cosim/real_data/hex/STALE_NOTICE.md

@@ -1,42 +0,0 @@
-# Golden Reference Hex Files
-
-These hex files are **committed golden references** for strict bit-exact
-real-data regression tests (`tb_doppler_realdata.v`, `tb_fullchain_realdata.v`).
-
-## When to regenerate
-
-Regenerate whenever the Doppler processing pipeline changes:
-
-- `doppler_processor.v` (FFT size, window, sub-frame structure)
-- `xfft_16.v` / `fft_engine.v` (butterfly arithmetic, twiddle lookup)
-- `range_bin_decimator.v` (decimation mode, peak detection logic)
-- `fft_twiddle_16.mem` (twiddle factor ROM)
-
-## How to regenerate
-
-```bash
-cd 9_Firmware/9_2_FPGA
-python3 tb/cosim/real_data/golden_reference.py
-# Then copy the Doppler-specific files:
-python3 -c "
-import numpy as np, os, shutil
-h = 'tb/cosim/real_data/hex'
-# Regenerate packed stimulus from range FFT npy
-ri = np.load(f'{h}/range_fft_all_i.npy')
-rq = np.load(f'{h}/range_fft_all_q.npy')
-with open(f'{h}/doppler_input_realdata.hex','w') as f:
-    for c in range(32):
-        for r in range(64):
-            i=int(ri[c,r])&0xFFFF; q=int(rq[c,r])&0xFFFF
-            f.write(f'{(q<<16)|i:08X}\n')
-shutil.copy2(f'{h}/doppler_map_i.hex', f'{h}/doppler_ref_i.hex')
-shutil.copy2(f'{h}/doppler_map_q.hex', f'{h}/doppler_ref_q.hex')
-"
-```
-
-## Architecture
-
-Generated against the **dual 16-point FFT** Doppler architecture
-(2 staggered-PRI sub-frames x 16-point Hamming-windowed FFT).
-
-Source data: ADI CN0566 Phaser radar (10.525 GHz X-band FMCW, 4 MSPS).

9_Firmware/9_2_FPGA/tb/cosim/real_data/hex/detections.txt

@@ -1,361 +0,0 @@
-# AERIS-10 Golden Reference Detections
-# Threshold: 10000
-# Format: range_bin doppler_bin magnitude
-0 0 35364
-0 1 16147
-0 15 11821
-0 16 24536
-0 17 11208
-0 31 10122
-1 0 25697
-1 1 12174
-1 15 13421
-1 16 20002
-1 17 11568
-1 31 11299
-2 0 16788
-2 16 20207
-2 31 10711
-3 0 29174
-3 1 13965
-3 15 13305
-3 16 31517
-3 17 13478
-3 31 14101
-4 0 41986
-4 1 19241
-4 15 21030
-4 16 39714
-4 17 17538
-4 31 20394
-5 0 23766
-5 1 11599
-5 15 14843
-5 16 18211
-5 31 12009
-6 0 42015
-6 1 21423
-6 15 21018
-6 16 47402
-6 17 22815
-6 31 22736
-7 0 32152
-7 1 15393
-7 15 14318
-7 16 28911
-7 17 13876
-7 31 17156
-8 0 11067
-10 0 18848
-10 15 10020
-10 16 20027
-10 31 10048
-11 0 65534
-11 1 37617
-11 2 14940
-11 15 43078
-11 16 65534
-11 17 39344
-11 31 45926
-12 0 58975
-12 1 22078
-12 15 34440
-12 16 59096
-12 17 22512
-12 31 28677
-13 0 38442
-13 1 29490
-13 15 37679
-13 16 44951
-13 17 27726
-13 31 39144
-14 0 52660
-14 1 27797
-14 2 13534
-14 15 39671
-14 16 57929
-14 17 24160
-14 31 31478
-15 0 30021
-15 1 12219
-15 15 17232
-15 16 29524
-15 17 13424
-15 31 17850
-16 0 17593
-16 16 24710
-16 31 13046
-17 0 17606
-17 1 11119
-17 16 13182
-18 16 15914
-19 0 55785
-19 1 29069
-19 15 29418
-19 16 55308
-19 17 27886
-19 31 30649
-20 0 49230
-20 1 24486
-20 15 21233
-20 16 45472
-20 17 21749
-20 31 21614
-21 0 26167
-21 1 13823
-21 15 10487
-21 16 29034
-21 17 15861
-21 31 10454
-22 0 12791
-22 16 22520
-22 17 11384
-22 31 11790
-23 0 29337
-23 1 17065
-23 15 14174
-23 16 39414
-23 17 23310
-23 31 17173
-24 0 16889
-24 15 15710
-24 16 22395
-24 31 15003
-25 0 65535
-25 1 40375
-25 15 54011
-25 16 61127
-25 17 38944
-25 31 48889
-26 0 46367
-26 1 39852
-26 15 29630
-26 16 53587
-26 17 40655
-26 31 34936
-27 0 65535
-27 1 65535
-27 15 64456
-27 16 65535
-27 17 65535
-27 31 58334
-28 0 65535
-28 1 57641
-28 15 65535
-28 16 65535
-28 17 54928
-28 31 65535
-29 0 65535
-29 1 44117
-29 2 13478
-29 14 11179
-29 15 65535
-29 16 65535
-29 17 45898
-29 18 10817
-29 31 60442
-30 0 44530
-30 1 36909
-30 2 14573
-30 15 43430
-30 16 51839
-30 17 37271
-30 31 47866
-31 0 40957
-31 1 52081
-31 2 12755
-31 15 42794
-31 16 41071
-31 17 50472
-31 18 11556
-31 31 43866
-32 0 35747
-32 1 19597
-32 15 25173
-32 16 39213
-32 17 21782
-32 31 29106
-33 0 34216
-33 1 41661
-33 15 42368
-33 16 38638
-33 17 40522
-33 31 45908
-34 0 36589
-34 1 17165
-34 15 16488
-34 16 26972
-34 17 12089
-34 31 13576
-35 0 65536
-35 1 42536
-35 15 61612
-35 16 65536
-35 17 43084
-35 31 60807
-36 0 55831
-36 1 26499
-36 15 28393
-36 16 50059
-36 17 24420
-36 31 23905
-38 0 52721
-38 1 33692
-38 15 32463
-38 16 53145
-38 17 37178
-38 31 30632
-39 0 32288
-39 1 19461
-39 15 20183
-39 16 27198
-39 17 16723
-39 31 14041
-40 0 47793
-40 1 29861
-40 15 23082
-40 16 43109
-40 17 30298
-40 31 31219
-41 0 65536
-41 1 57642
-41 15 52984
-41 16 65536
-41 17 57420
-41 31 57035
-42 0 46393
-42 1 24862
-42 15 27123
-42 16 44734
-42 17 25836
-42 31 33316
-43 0 65535
-43 1 43056
-43 13 10481
-43 14 22074
-43 15 24792
-43 16 39506
-43 17 36481
-43 30 24870
-43 31 39062
-44 0 65535
-44 1 65535
-44 2 19166
-44 3 21321
-44 13 32864
-44 14 41461
-44 15 65535
-44 16 65535
-44 17 58493
-44 19 13967
-44 29 29756
-44 30 54069
-44 31 65535
-45 0 65535
-45 1 58886
-45 14 22013
-45 15 65116
-45 16 65535
-45 17 65535
-45 29 13068
-45 30 25759
-45 31 61393
-46 0 53411
-46 1 44267
-46 15 30631
-46 16 58196
-46 17 36338
-46 31 28840
-47 0 54574
-47 1 35574
-47 15 38960
-47 16 46623
-47 17 32070
-47 31 40091
-48 0 22302
-48 1 10865
-48 15 13917
-48 16 18042
-48 31 10930
-49 0 49013
-49 1 28270
-49 15 25242
-49 16 41969
-49 17 24924
-49 31 26704
-50 0 43858
-50 1 35227
-50 15 39737
-50 16 39085
-50 17 36353
-50 31 38658
-51 0 33868
-51 1 23364
-51 15 23348
-51 16 33246
-51 17 24398
-51 31 27415
-52 0 24500
-52 1 10467
-52 15 13661
-52 16 21073
-52 17 10681
-52 31 10164
-53 0 46806
-53 1 31121
-53 15 34423
-53 16 44568
-53 17 28497
-53 31 35229
-54 0 49713
-54 1 32292
-54 15 40878
-54 16 54060
-54 17 34252
-54 31 43616
-55 0 25597
-55 1 13662
-55 15 13184
-55 16 20525
-55 17 10363
-55 31 12295
-56 0 53620
-56 1 23575
-56 14 12578
-56 15 34783
-56 16 64499
-56 17 32442
-56 31 32139
-58 0 65535
-58 1 35008
-58 15 43324
-58 16 64959
-58 17 35348
-58 31 39154
-59 0 13238
-59 1 11374
-59 14 16623
-59 16 22346
-59 17 12583
-60 0 31259
-60 1 17900
-60 15 19638
-60 16 26331
-60 17 12543
-60 31 18056
-61 0 14596
-61 15 13600
-61 16 23597
-61 17 10681
-61 31 14915
-62 0 22096
-62 1 10515
-62 16 23642
-62 17 11146
-62 31 10180
-63 0 58324
-63 1 25269
-63 15 32920
-63 16 54165
-63 17 27625
-63 31 29816

9_Firmware/9_2_FPGA/tb/cosim/real_data/hex/fullchain_cfar_detections.txt

@@ -1,8 +0,0 @@
-# AERIS-10 Full-Chain CFAR Detection List
-# Chain: decim -> MTI -> Doppler -> DC notch(w=2) -> CA-CFAR
-# CFAR: guard=2, train=8, alpha=0x30, mode=CA
-# Format: range_bin doppler_bin magnitude threshold
-2 14 57128 48153
-2 29 20281 15318
-2 30 44783 22389
-3 26 19423 19422

9_Firmware/9_2_FPGA/tb/cosim/real_data/hex/fullchain_detections.txt

@@ -1,211 +0,0 @@
-# AERIS-10 Full-Chain Golden Reference Detections
-# Threshold: 10000
-# Format: range_bin doppler_bin magnitude
-0 0 65534
-0 1 28729
-0 2 18427
-0 3 14971
-0 14 11972
-0 15 53110
-0 16 65534
-0 17 39344
-0 31 45926
-1 0 65535
-1 1 65535
-1 2 16977
-1 4 13219
-1 10 10505
-1 12 12931
-1 14 23806
-1 15 65535
-1 16 65535
-1 17 65535
-1 31 55887
-2 0 65535
-2 1 65535
-2 2 19166
-2 3 21321
-2 13 32864
-2 14 41461
-2 15 65535
-2 16 65535
-2 17 58493
-2 19 13967
-2 29 29756
-2 30 54069
-2 31 65535
-3 0 53605
-3 1 48935
-3 2 24589
-3 4 12240
-3 5 14117
-3 8 12950
-3 11 13274
-3 12 12930
-3 14 23437
-3 15 29552
-3 16 65433
-3 17 52933
-3 18 24719
-3 19 17650
-3 20 15096
-3 21 13518
-3 27 13184
-3 28 15554
-3 29 23742
-3 30 17775
-3 31 19771
-4 0 46125
-4 1 38769
-4 14 10924
-4 15 33844
-4 16 38130
-4 17 36763
-4 31 34850
-5 0 29649
-5 1 15983
-5 16 27615
-5 17 13683
-5 31 11456
-6 0 29881
-6 1 14520
-6 15 14126
-6 16 33068
-6 17 14679
-6 31 16608
-7 0 15305
-7 16 16195
-8 0 22601
-8 16 32614
-8 17 16367
-8 31 16217
-9 0 33316
-9 1 14993
-9 15 19928
-9 16 38915
-9 17 20159
-9 31 17219
-10 0 31772
-10 1 16438
-10 15 13737
-10 16 29059
-10 17 15821
-10 31 13104
-11 0 30448
-11 1 15802
-11 15 12669
-11 16 30129
-11 17 14340
-11 31 12767
-12 0 12892
-12 16 15956
-13 16 10903
-17 0 11395
-17 16 10440
-19 0 11910
-19 16 12017
-20 0 42212
-20 1 17994
-20 15 23540
-20 16 42519
-20 17 15949
-20 31 23792
-21 0 19822
-21 2 13933
-21 3 12130
-21 13 11590
-21 14 14794
-21 15 14160
-21 16 36543
-21 17 19530
-21 31 13754
-22 0 12654
-26 16 10634
-30 0 11396
-38 0 12032
-38 1 11693
-38 16 18530
-39 0 13327
-39 1 12416
-39 2 10940
-39 15 10351
-39 16 25217
-39 17 11785
-39 31 12383
-40 16 14316
-45 0 16350
-45 16 16146
-46 0 11710
-46 16 12568
-46 31 12206
-47 15 12053
-47 16 17267
-49 0 22212
-49 15 11409
-49 16 20817
-50 0 14287
-50 16 13382
-51 0 15578
-51 1 11129
-51 16 12819
-53 16 10532
-55 0 24789
-55 15 10455
-55 16 25212
-55 17 10510
-55 31 12207
-56 0 45192
-56 1 16083
-56 15 22284
-56 16 40302
-56 17 17318
-56 31 19185
-57 0 41535
-57 1 27102
-57 2 50048
-57 3 30784
-57 6 10595
-57 8 12880
-57 12 16303
-57 13 21792
-57 14 36737
-57 15 51757
-57 16 53641
-57 17 33656
-57 18 11096
-57 31 50828
-58 0 49157
-58 1 40063
-58 15 33919
-58 16 44578
-58 17 46214
-58 31 35365
-59 0 40306
-59 1 20804
-59 15 16849
-59 16 25943
-59 18 11684
-59 30 15453
-59 31 26587
-60 0 19637
-60 15 12275
-60 16 18527
-60 30 10157
-60 31 17278
-61 0 15147
-61 14 10732
-61 15 12364
-61 16 25489
-61 17 13371
-61 31 14278
-62 1 14922
-62 2 13924
-62 16 29381
-62 17 20524
-62 31 12898
-63 0 48946
-63 1 23135
-63 15 21609
-63 16 44144
-63 17 21410
-63 31 19856