Skip to content

progress_log.md

Latest commit

 

History

History
1391 lines (973 loc) · 35 KB

progress_log.md

File metadata and controls

1391 lines (973 loc) · 35 KB
Raw

cpathgen Delta Optimization — Progress Log

Autonomous search for the optimal adaptive delta strategy in _spur_mlq.

Lever: delta (step size / threshold increment) — static or dynamically adjusted per-step based on paths generated.

Adaptation rule: f(paths_this_step, target_pps) → new_delta

  • paths_this_step < target_pps → delta *= scale_up (sparse step, grow window)
  • paths_this_step > target_pps → delta *= scale_down (dense step, shrink window)
  • target_pps == 0 → static delta (no adaptation)

Metric: cpathgen pfxt_time (ms), averaged over leon2 / leon3mp / netcard at K=1M. Correctness gate: avg_err == 0.0 and max_err == 0.0 — required before any timing result counts.

Thread count: 16 (optimal for this 20-thread machine; see thread sweep in session log)

Baseline — static delta=2.5 (pre-autoloop)

Recorded before search began. Hardware: Linux x86-64, 20-thread machine.

benchmark time_ms avg_err max_err steps avg_pps
leon2 71.2 0.0 0.0 1 1,635,380
leon3mp — — — — —
netcard — — — — —

Note: leon3mp and netcard baseline times not recorded individually before autoloop; will be established in Round 1 (static delta=2.5 is config ID 6 in Phase 1).

Observation: With delta=2.5 and 10 queues, leon2 K=1M finishes in a single step (all 1M paths fit in the initial active cost bands — no bounds bump needed). This is the ideal case. Dynamic delta is most impactful when the distribution forces multiple steps; Phase 1 static sweep will map where the 1-step boundary lies.

Rounds below are appended automatically by scripts/autoloop.py.

Round 1 — static delta=0.1

Phase 1 | Config ID 1

Parameters

param value
delta_init 0.1
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 126.7 0.0000 0.0000 4 254294 0.100 0.100
leon3mp 122.7 0.0000 0.0000 5 200259 0.100 0.100
netcard 120.9 0.0000 0.0000 4 251647 0.100 0.100

Avg time across benchmarks: 123.4 ms

Verdict: ➖ plateau (123.4 ms avg)

Best so far: static delta=0.1 @ 123.43333333333334 ms

Round 2 — static delta=0.5

Phase 1 | Config ID 2

Parameters

param value
delta_init 0.5
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 54.6 0.0000 0.0000 1 1124444 0.500 0.500
leon3mp 49.7 0.0000 0.0000 1 1001296 0.500 0.500
netcard 50.1 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 51.5 ms

Verdict: ➖ plateau (51.5 ms avg)

Best so far: static delta=0.5 @ 51.46666666666667 ms

Round 3 — static delta=1.0

Phase 1 | Config ID 3

Parameters

param value
delta_init 1.0
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 60.1 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 45.9 0.0000 0.0000 1 1001296 1.000 1.000
netcard 55.6 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 53.9 ms

Verdict: 🔻 regression (51.5 → 53.9 ms avg)

Best so far: static delta=0.5 @ 51.46666666666667 ms

Round 4 — static delta=1.5

Phase 1 | Config ID 4

Parameters

param value
delta_init 1.5
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 65.1 0.0000 0.0000 1 1468888 1.500 1.500
leon3mp 50.1 0.0000 0.0000 1 1132948 1.500 1.500
netcard 61.9 0.0000 0.0000 1 1355772 1.500 1.500

Avg time across benchmarks: 59.0 ms

Verdict: 🔻 regression (51.5 → 59.0 ms avg)

Best so far: static delta=0.5 @ 51.46666666666667 ms

Round 5 — static delta=2.0

Phase 1 | Config ID 5

Parameters

param value
delta_init 2.0
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 55.3 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.1 0.0000 0.0000 1 1001296 2.000 2.000
netcard 52.2 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 50.2 ms

Verdict: ➖ plateau (50.2 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 6 — static delta=2.5

Phase 1 | Config ID 6

Parameters

param value
delta_init 2.5
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 69.2 0.0000 0.0000 1 1635380 2.500 2.500
leon3mp 54.0 0.0000 0.0000 1 1260400 2.500 2.500
netcard 63.6 0.0000 0.0000 1 1506344 2.500 2.500

Avg time across benchmarks: 62.3 ms

Verdict: 🔻 regression (50.2 → 62.3 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 7 — static delta=3.5

Phase 1 | Config ID 7

Parameters

param value
delta_init 3.5
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 44.3 0.0000 0.0000 1 1124444 3.500 3.500
leon3mp 74.4 0.0000 0.0000 1 1753152 3.500 3.500
netcard 39.9 0.0000 0.0000 1 1039476 3.500 3.500

Avg time across benchmarks: 52.9 ms

Verdict: 🔻 regression (50.2 → 52.9 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 8 — static delta=5.0

Phase 1 | Config ID 8

Parameters

param value
delta_init 5.0
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 63.9 0.0000 0.0000 1 1635380 5.000 5.000
leon3mp 49.3 0.0000 0.0000 1 1260400 5.000 5.000
netcard 58.3 0.0000 0.0000 1 1506344 5.000 5.000

Avg time across benchmarks: 57.2 ms

Verdict: 🔻 regression (50.2 → 57.2 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 9 — static delta=10.0

Phase 1 | Config ID 9

Parameters

param value
delta_init 10.0
target_pps 0.0
scale_up 1.0
scale_down 1.0
delta_min 0.1
delta_max 100.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 130.3 0.0000 0.0000 1 3142768 10.000 10.000
leon3mp 96.9 0.0000 0.0000 1 2428584 10.000 10.000
netcard 113.9 0.0000 0.0000 1 2855388 10.000 10.000

Avg time across benchmarks: 113.7 ms

Verdict: 🔻 regression (50.2 → 113.7 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 10 — adaptive d0=0.5 tpps=50000 up=1.5 dn=0.7

Phase 2 | Config ID 10

Parameters

param value
delta_init 0.5
target_pps 50000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 54.5 0.0000 0.0000 1 1124444 0.300 0.300
leon3mp 48.3 0.0000 0.0000 1 1001296 0.300 0.300
netcard 51.0 0.0000 0.0000 1 1039476 0.300 0.300

Avg time across benchmarks: 51.3 ms

Verdict: 🔻 regression (50.2 → 51.3 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 11 — adaptive d0=0.5 tpps=50000 up=1.5 dn=0.9

Phase 2 | Config ID 11

Parameters

param value
delta_init 0.5
target_pps 50000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 55.6 0.0000 0.0000 1 1124444 0.400 0.400
leon3mp 49.4 0.0000 0.0000 1 1001296 0.400 0.400
netcard 50.9 0.0000 0.0000 1 1039476 0.400 0.400

Avg time across benchmarks: 52.0 ms

Verdict: 🔻 regression (50.2 → 52.0 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 12 — adaptive d0=0.5 tpps=50000 up=2.5 dn=0.7

Phase 2 | Config ID 12

Parameters

param value
delta_init 0.5
target_pps 50000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 56.1 0.0000 0.0000 1 1124444 0.300 0.300
leon3mp 49.8 0.0000 0.0000 1 1001296 0.300 0.300
netcard 50.7 0.0000 0.0000 1 1039476 0.300 0.300

Avg time across benchmarks: 52.2 ms

Verdict: 🔻 regression (50.2 → 52.2 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 13 — adaptive d0=0.5 tpps=50000 up=2.5 dn=0.9

Phase 2 | Config ID 13

Parameters

param value
delta_init 0.5
target_pps 50000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 54.4 0.0000 0.0000 1 1124444 0.400 0.400
leon3mp 50.5 0.0000 0.0000 1 1001296 0.400 0.400
netcard 50.5 0.0000 0.0000 1 1039476 0.400 0.400

Avg time across benchmarks: 51.8 ms

Verdict: 🔻 regression (50.2 → 51.8 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 14 — adaptive d0=0.5 tpps=200000 up=1.5 dn=0.7

Phase 2 | Config ID 14

Parameters

param value
delta_init 0.5
target_pps 200000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 55.8 0.0000 0.0000 1 1124444 0.300 0.300
leon3mp 48.1 0.0000 0.0000 1 1001296 0.300 0.300
netcard 55.6 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 53.2 ms

Verdict: 🔻 regression (50.2 → 53.2 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 15 — adaptive d0=0.5 tpps=200000 up=1.5 dn=0.9

Phase 2 | Config ID 15

Parameters

param value
delta_init 0.5
target_pps 200000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 56.7 0.0000 0.0000 1 1124444 0.500 0.500
leon3mp 49.5 0.0000 0.0000 1 1001296 0.500 0.500
netcard 52.1 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 52.8 ms

Verdict: 🔻 regression (50.2 → 52.8 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 16 — adaptive d0=0.5 tpps=200000 up=2.5 dn=0.7

Phase 2 | Config ID 16

Parameters

param value
delta_init 0.5
target_pps 200000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 53.8 0.0000 0.0000 1 1124444 0.500 0.500
leon3mp 48.7 0.0000 0.0000 1 1001296 0.500 0.500
netcard 50.1 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 50.9 ms

Verdict: 🔻 regression (50.2 → 50.9 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 17 — adaptive d0=0.5 tpps=200000 up=2.5 dn=0.9

Phase 2 | Config ID 17

Parameters

param value
delta_init 0.5
target_pps 200000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 55.6 0.0000 0.0000 1 1124444 0.500 0.500
leon3mp 49.6 0.0000 0.0000 1 1001296 0.500 0.500
netcard 51.6 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 52.3 ms

Verdict: 🔻 regression (50.2 → 52.3 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 18 — adaptive d0=0.5 tpps=500000 up=1.5 dn=0.7

Phase 2 | Config ID 18

Parameters

param value
delta_init 0.5
target_pps 500000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 54.9 0.0000 0.0000 1 1124444 0.500 0.500
leon3mp 49.6 0.0000 0.0000 1 1001296 0.500 0.500
netcard 49.3 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 51.3 ms

Verdict: 🔻 regression (50.2 → 51.3 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 19 — adaptive d0=0.5 tpps=500000 up=1.5 dn=0.9

Phase 2 | Config ID 19

Parameters

param value
delta_init 0.5
target_pps 500000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 55.2 0.0000 0.0000 1 1124444 0.500 0.500
leon3mp 49.8 0.0000 0.0000 1 1001296 0.500 0.500
netcard 50.3 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 51.8 ms

Verdict: 🔻 regression (50.2 → 51.8 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 20 — adaptive d0=0.5 tpps=500000 up=2.5 dn=0.7

Phase 2 | Config ID 20

Parameters

param value
delta_init 0.5
target_pps 500000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 54.9 0.0000 0.0000 1 1124444 0.500 0.500
leon3mp 48.1 0.0000 0.0000 1 1001296 0.500 0.500
netcard 51.5 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 51.5 ms

Verdict: 🔻 regression (50.2 → 51.5 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 21 — adaptive d0=0.5 tpps=500000 up=2.5 dn=0.9

Phase 2 | Config ID 21

Parameters

param value
delta_init 0.5
target_pps 500000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 55.4 0.0000 0.0000 1 1124444 0.500 0.500
leon3mp 48.5 0.0000 0.0000 1 1001296 0.500 0.500
netcard 50.4 0.0000 0.0000 1 1039476 0.500 0.500

Avg time across benchmarks: 51.4 ms

Verdict: 🔻 regression (50.2 → 51.4 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 22 — adaptive d0=1.0 tpps=50000 up=1.5 dn=0.7

Phase 2 | Config ID 22

Parameters

param value
delta_init 1.0
target_pps 50000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 59.9 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 45.2 0.0000 0.0000 1 1001296 1.000 1.000
netcard 56.3 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 53.8 ms

Verdict: 🔻 regression (50.2 → 53.8 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 23 — adaptive d0=1.0 tpps=50000 up=1.5 dn=0.9

Phase 2 | Config ID 23

Parameters

param value
delta_init 1.0
target_pps 50000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 60.1 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 46.5 0.0000 0.0000 1 1001296 1.000 1.000
netcard 56.2 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 54.3 ms

Verdict: 🔻 regression (50.2 → 54.3 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 24 — adaptive d0=1.0 tpps=50000 up=2.5 dn=0.7

Phase 2 | Config ID 24

Parameters

param value
delta_init 1.0
target_pps 50000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 59.5 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 46.1 0.0000 0.0000 1 1001296 1.000 1.000
netcard 54.6 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 53.4 ms

Verdict: 🔻 regression (50.2 → 53.4 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 25 — adaptive d0=1.0 tpps=50000 up=2.5 dn=0.9

Phase 2 | Config ID 25

Parameters

param value
delta_init 1.0
target_pps 50000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 60.8 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 46.4 0.0000 0.0000 1 1001296 1.000 1.000
netcard 56.1 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 54.4 ms

Verdict: 🔻 regression (50.2 → 54.4 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 26 — adaptive d0=1.0 tpps=200000 up=1.5 dn=0.7

Phase 2 | Config ID 26

Parameters

param value
delta_init 1.0
target_pps 200000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 59.5 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 45.9 0.0000 0.0000 1 1001296 1.000 1.000
netcard 54.7 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 53.4 ms

Verdict: 🔻 regression (50.2 → 53.4 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 27 — adaptive d0=1.0 tpps=200000 up=1.5 dn=0.9

Phase 2 | Config ID 27

Parameters

param value
delta_init 1.0
target_pps 200000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 59.6 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 46.2 0.0000 0.0000 1 1001296 1.000 1.000
netcard 56.0 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 53.9 ms

Verdict: 🔻 regression (50.2 → 53.9 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 28 — adaptive d0=1.0 tpps=200000 up=2.5 dn=0.7

Phase 2 | Config ID 28

Parameters

param value
delta_init 1.0
target_pps 200000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 60.5 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 46.2 0.0000 0.0000 1 1001296 1.000 1.000
netcard 56.4 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 54.4 ms

Verdict: 🔻 regression (50.2 → 54.4 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 29 — adaptive d0=1.0 tpps=200000 up=2.5 dn=0.9

Phase 2 | Config ID 29

Parameters

param value
delta_init 1.0
target_pps 200000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 60.8 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 45.5 0.0000 0.0000 1 1001296 1.000 1.000
netcard 55.9 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 54.1 ms

Verdict: 🔻 regression (50.2 → 54.1 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 30 — adaptive d0=1.0 tpps=500000 up=1.5 dn=0.7

Phase 2 | Config ID 30

Parameters

param value
delta_init 1.0
target_pps 500000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 61.1 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 45.3 0.0000 0.0000 1 1001296 1.000 1.000
netcard 55.9 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 54.1 ms

Verdict: 🔻 regression (50.2 → 54.1 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 31 — adaptive d0=1.0 tpps=500000 up=1.5 dn=0.9

Phase 2 | Config ID 31

Parameters

param value
delta_init 1.0
target_pps 500000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 59.8 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 46.1 0.0000 0.0000 1 1001296 1.000 1.000
netcard 55.8 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 53.9 ms

Verdict: 🔻 regression (50.2 → 53.9 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 32 — adaptive d0=1.0 tpps=500000 up=2.5 dn=0.7

Phase 2 | Config ID 32

Parameters

param value
delta_init 1.0
target_pps 500000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 59.5 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 46.4 0.0000 0.0000 1 1001296 1.000 1.000
netcard 55.0 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 53.6 ms

Verdict: 🔻 regression (50.2 → 53.6 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 33 — adaptive d0=1.0 tpps=500000 up=2.5 dn=0.9

Phase 2 | Config ID 33

Parameters

param value
delta_init 1.0
target_pps 500000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 61.0 0.0000 0.0000 1 1299420 1.000 1.000
leon3mp 45.3 0.0000 0.0000 1 1001296 1.000 1.000
netcard 56.0 0.0000 0.0000 1 1200332 1.000 1.000

Avg time across benchmarks: 54.1 ms

Verdict: 🔻 regression (50.2 → 54.1 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 34 — adaptive d0=2.0 tpps=50000 up=1.5 dn=0.7

Phase 2 | Config ID 34

Parameters

param value
delta_init 2.0
target_pps 50000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 57.3 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.4 0.0000 0.0000 1 1001296 2.000 2.000
netcard 53.5 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.4 ms

Verdict: 🔻 regression (50.2 → 51.4 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 35 — adaptive d0=2.0 tpps=50000 up=1.5 dn=0.9

Phase 2 | Config ID 35

Parameters

param value
delta_init 2.0
target_pps 50000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 56.7 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 44.3 0.0000 0.0000 1 1001296 2.000 2.000
netcard 53.6 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.5 ms

Verdict: 🔻 regression (50.2 → 51.5 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 36 — adaptive d0=2.0 tpps=50000 up=2.5 dn=0.7

Phase 2 | Config ID 36

Parameters

param value
delta_init 2.0
target_pps 50000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 58.8 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 42.3 0.0000 0.0000 1 1001296 2.000 2.000
netcard 52.8 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.3 ms

Verdict: 🔻 regression (50.2 → 51.3 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 37 — adaptive d0=2.0 tpps=50000 up=2.5 dn=0.9

Phase 2 | Config ID 37

Parameters

param value
delta_init 2.0
target_pps 50000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 58.1 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.4 0.0000 0.0000 1 1001296 2.000 2.000
netcard 52.9 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.5 ms

Verdict: 🔻 regression (50.2 → 51.5 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 38 — adaptive d0=2.0 tpps=200000 up=1.5 dn=0.7

Phase 2 | Config ID 38

Parameters

param value
delta_init 2.0
target_pps 200000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 55.6 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.6 0.0000 0.0000 1 1001296 2.000 2.000
netcard 55.0 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.4 ms

Verdict: 🔻 regression (50.2 → 51.4 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 39 — adaptive d0=2.0 tpps=200000 up=1.5 dn=0.9

Phase 2 | Config ID 39

Parameters

param value
delta_init 2.0
target_pps 200000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 57.6 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 44.5 0.0000 0.0000 1 1001296 2.000 2.000
netcard 52.6 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.6 ms

Verdict: 🔻 regression (50.2 → 51.6 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 40 — adaptive d0=2.0 tpps=200000 up=2.5 dn=0.7

Phase 2 | Config ID 40

Parameters

param value
delta_init 2.0
target_pps 200000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 57.8 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.8 0.0000 0.0000 1 1001296 2.000 2.000
netcard 52.5 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.4 ms

Verdict: 🔻 regression (50.2 → 51.4 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 41 — adaptive d0=2.0 tpps=200000 up=2.5 dn=0.9

Phase 2 | Config ID 41

Parameters

param value
delta_init 2.0
target_pps 200000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 55.7 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.4 0.0000 0.0000 1 1001296 2.000 2.000
netcard 51.1 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 50.1 ms

Verdict: ➖ plateau (50.1 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 42 — adaptive d0=2.0 tpps=500000 up=1.5 dn=0.7

Phase 2 | Config ID 42

Parameters

param value
delta_init 2.0
target_pps 500000.0
scale_up 1.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 58.2 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.9 0.0000 0.0000 1 1001296 2.000 2.000
netcard 52.9 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.7 ms

Verdict: 🔻 regression (50.2 → 51.7 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 43 — adaptive d0=2.0 tpps=500000 up=1.5 dn=0.9

Phase 2 | Config ID 43

Parameters

param value
delta_init 2.0
target_pps 500000.0
scale_up 1.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 58.2 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.9 0.0000 0.0000 1 1001296 2.000 2.000
netcard 52.3 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.5 ms

Verdict: 🔻 regression (50.2 → 51.5 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 44 — adaptive d0=2.0 tpps=500000 up=2.5 dn=0.7

Phase 2 | Config ID 44

Parameters

param value
delta_init 2.0
target_pps 500000.0
scale_up 2.5
scale_down 0.7
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 57.5 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.6 0.0000 0.0000 1 1001296 2.000 2.000
netcard 51.2 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 50.8 ms

Verdict: 🔻 regression (50.2 → 50.8 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms

Round 45 — adaptive d0=2.0 tpps=500000 up=2.5 dn=0.9

Phase 2 | Config ID 45

Parameters

param value
delta_init 2.0
target_pps 500000.0
scale_up 2.5
scale_down 0.9
delta_min 0.1
delta_max 50.0

Results (K=1M, 16 threads)

benchmark time_ms avg_err max_err steps avg_pps d_min d_max
leon2 56.8 0.0000 0.0000 1 1299420 2.000 2.000
leon3mp 43.9 0.0000 0.0000 1 1001296 2.000 2.000
netcard 52.7 0.0000 0.0000 1 1200332 2.000 2.000

Avg time across benchmarks: 51.1 ms

Verdict: 🔻 regression (50.2 → 51.1 ms avg)

Best so far: static delta=2.0 @ 50.20000000000001 ms