extern/whisper.cpp
1
0
Fork 1
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-07-14 14:55:18 +02:00
Code Issues Packages Projects Releases Wiki Activity
Files
130044f2285555c666a5e0975a50d8b51fc74fa2
whisper.cpp/scripts/bench-all-gg.txt
Georgi Gerganov 06bdaa6c0c bench : update benches
2025-06-25 16:45:19 +03:00

287 lines
19 KiB
Plaintext
Raw Blame History

## M1 Pro (old 22c96b4)
make -j && ./scripts/bench-all.sh 8
Running memcpy benchmark
memcpy: 39.10 GB/s (heat-up)
memcpy: 44.75 GB/s ( 1 thread)
memcpy: 44.78 GB/s ( 1 thread)
memcpy: 44.97 GB/s ( 2 thread)
memcpy: 48.04 GB/s ( 3 thread)
memcpy: 50.55 GB/s ( 4 thread)
memcpy: 55.20 GB/s ( 5 thread)
memcpy: 65.60 GB/s ( 6 thread)
memcpy: 70.64 GB/s ( 7 thread)
memcpy: 73.34 GB/s ( 8 thread)
sum: -5120002535.000000
make -j && ./scripts/bench-all.sh 1 0 0
Running ggml_mul_mat benchmark with 1 threads
64 x 64: Q4_0 237.1 GFLOPS (128 runs) | Q4_1 168.6 GFLOPS (128 runs)
64 x 64: Q5_0 136.4 GFLOPS (128 runs) | Q5_1 135.6 GFLOPS (128 runs) | Q8_0 243.1 GFLOPS (128 runs)
64 x 64: F16 140.4 GFLOPS (128 runs) | F32 316.6 GFLOPS (128 runs)
128 x 128: Q4_0 496.6 GFLOPS (128 runs) | Q4_1 348.6 GFLOPS (128 runs)
128 x 128: Q5_0 273.2 GFLOPS (128 runs) | Q5_1 274.1 GFLOPS (128 runs) | Q8_0 505.1 GFLOPS (128 runs)
128 x 128: F16 300.4 GFLOPS (128 runs) | F32 653.9 GFLOPS (128 runs)
256 x 256: Q4_0 791.7 GFLOPS (128 runs) | Q4_1 615.3 GFLOPS (128 runs)
256 x 256: Q5_0 651.0 GFLOPS (128 runs) | Q5_1 674.7 GFLOPS (128 runs) | Q8_0 803.1 GFLOPS (128 runs)
256 x 256: F16 869.6 GFLOPS (128 runs) | F32 957.2 GFLOPS (128 runs)
512 x 512: Q4_0 973.3 GFLOPS (128 runs) | Q4_1 897.9 GFLOPS (128 runs)
512 x 512: Q5_0 1078.8 GFLOPS (128 runs) | Q5_1 998.4 GFLOPS (128 runs) | Q8_0 752.4 GFLOPS (128 runs)
512 x 512: F16 892.5 GFLOPS (128 runs) | F32 1399.6 GFLOPS (128 runs)
1024 x 1024: Q4_0 1402.7 GFLOPS (128 runs) | Q4_1 1218.5 GFLOPS (128 runs)
1024 x 1024: Q5_0 1444.8 GFLOPS (128 runs) | Q5_1 1444.7 GFLOPS (128 runs) | Q8_0 1395.7 GFLOPS (128 runs)
1024 x 1024: F16 1524.1 GFLOPS (128 runs) | F32 1726.6 GFLOPS (128 runs)
2048 x 2048: Q4_0 1479.4 GFLOPS ( 87 runs) | Q4_1 1378.5 GFLOPS ( 81 runs)
2048 x 2048: Q5_0 1454.6 GFLOPS ( 85 runs) | Q5_1 1462.9 GFLOPS ( 86 runs) | Q8_0 1483.2 GFLOPS ( 87 runs)
2048 x 2048: F16 1488.0 GFLOPS ( 87 runs) | F32 1538.2 GFLOPS ( 90 runs)
4096 x 4096: Q4_0 1509.7 GFLOPS ( 11 runs) | Q4_1 1433.0 GFLOPS ( 11 runs)
4096 x 4096: Q5_0 1422.4 GFLOPS ( 11 runs) | Q5_1 1437.0 GFLOPS ( 11 runs) | Q8_0 1523.0 GFLOPS ( 12 runs)
4096 x 4096: F16 1551.3 GFLOPS ( 12 runs) | F32 1451.0 GFLOPS ( 11 runs)
| CPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| M1 Pro | METAL | tiny | 1 | 0 | 39.21 | 1.74 | 0.61 | 0.04 | 22c96b4 |
| M1 Pro | METAL | base | 1 | 0 | 70.76 | 2.60 | 0.93 | 0.06 | 22c96b4 |
| M1 Pro | METAL | small | 1 | 0 | 217.28 | 6.42 | 2.14 | 0.17 | 22c96b4 |
| M1 Pro | METAL | medium | 1 | 0 | 596.74 | 14.43 | 4.75 | 0.45 | 22c96b4 |
make -j && ./scripts/bench-all.sh 1 1 1
| CPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| M1 Pro | METAL | tiny | 1 | 1 | 30.77 | 1.59 | 0.54 | 0.03 | 22c96b4 |
| M1 Pro | METAL | base | 1 | 1 | 60.42 | 2.29 | 0.81 | 0.05 | 22c96b4 |
| M1 Pro | METAL | small | 1 | 1 | 183.82 | 5.12 | 1.81 | 0.14 | 22c96b4 |
| M1 Pro | METAL | medium | 1 | 1 | 517.92 | 11.60 | 4.01 | 0.38 | 22c96b4 |
## M2 Ultra
make -j && ./scripts/bench-all.sh 8
Running memcpy benchmark
memcpy: 48.01 GB/s (heat-up)
memcpy: 56.00 GB/s ( 1 thread)
memcpy: 56.20 GB/s ( 1 thread)
memcpy: 102.69 GB/s ( 2 thread)
memcpy: 140.32 GB/s ( 3 thread)
memcpy: 179.04 GB/s ( 4 thread)
memcpy: 159.61 GB/s ( 5 thread)
memcpy: 159.02 GB/s ( 6 thread)
memcpy: 180.29 GB/s ( 7 thread)
memcpy: 198.10 GB/s ( 8 thread)
sum: -5119999345.000000
make -j && ./scripts/bench-all.sh 1
Running ggml_mul_mat benchmark with 1 threads
64 x 64: Q4_0 37.7 GFLOPS (128 runs) | Q4_1 36.0 GFLOPS (128 runs)
64 x 64: Q5_0 20.1 GFLOPS (128 runs) | Q5_1 19.8 GFLOPS (128 runs) | Q8_0 39.5 GFLOPS (128 runs)
64 x 64: F16 29.9 GFLOPS (128 runs) | F32 22.6 GFLOPS (128 runs)
128 x 128: Q4_0 71.0 GFLOPS (128 runs) | Q4_1 62.2 GFLOPS (128 runs)
128 x 128: Q5_0 33.4 GFLOPS (128 runs) | Q5_1 31.6 GFLOPS (128 runs) | Q8_0 79.8 GFLOPS (128 runs)
128 x 128: F16 52.4 GFLOPS (128 runs) | F32 32.7 GFLOPS (128 runs)
256 x 256: Q4_0 88.6 GFLOPS (128 runs) | Q4_1 77.2 GFLOPS (128 runs)
256 x 256: Q5_0 40.3 GFLOPS (128 runs) | Q5_1 36.8 GFLOPS (128 runs) | Q8_0 102.5 GFLOPS (128 runs)
256 x 256: F16 64.6 GFLOPS (128 runs) | F32 36.4 GFLOPS (128 runs)
512 x 512: Q4_0 94.7 GFLOPS (128 runs) | Q4_1 83.6 GFLOPS (128 runs)
512 x 512: Q5_0 45.9 GFLOPS (128 runs) | Q5_1 41.3 GFLOPS (128 runs) | Q8_0 112.8 GFLOPS (128 runs)
512 x 512: F16 72.3 GFLOPS (128 runs) | F32 37.7 GFLOPS (128 runs)
1024 x 1024: Q4_0 98.9 GFLOPS ( 47 runs) | Q4_1 88.2 GFLOPS ( 42 runs)
1024 x 1024: Q5_0 49.0 GFLOPS ( 23 runs) | Q5_1 43.9 GFLOPS ( 21 runs) | Q8_0 121.0 GFLOPS ( 57 runs)
1024 x 1024: F16 72.6 GFLOPS ( 34 runs) | F32 36.0 GFLOPS ( 17 runs)
2048 x 2048: Q4_0 101.3 GFLOPS ( 6 runs) | Q4_1 90.0 GFLOPS ( 6 runs)
2048 x 2048: Q5_0 50.8 GFLOPS ( 3 runs) | Q5_1 45.3 GFLOPS ( 3 runs) | Q8_0 124.1 GFLOPS ( 8 runs)
2048 x 2048: F16 70.7 GFLOPS ( 5 runs) | F32 30.4 GFLOPS ( 3 runs)
4096 x 4096: Q4_0 101.7 GFLOPS ( 3 runs) | Q4_1 90.3 GFLOPS ( 3 runs)
4096 x 4096: Q5_0 52.2 GFLOPS ( 3 runs) | Q5_1 45.7 GFLOPS ( 3 runs) | Q8_0 123.0 GFLOPS ( 3 runs)
4096 x 4096: F16 60.3 GFLOPS ( 3 runs) | F32 29.8 GFLOPS ( 3 runs)
make -j && ./scripts/bench-all.sh 1 1 0
| CPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| M2 ULTRA | METAL | tiny | 1 | 0 | 10.15 | 1.20 | 0.36 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | tiny-q5_0 | 1 | 0 | 10.21 | 1.15 | 0.39 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | tiny-q5_1 | 1 | 0 | 9.26 | 1.15 | 0.38 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | tiny-q8_0 | 1 | 0 | 9.00 | 1.12 | 0.37 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | base | 1 | 0 | 15.77 | 1.73 | 0.45 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | base-q5_0 | 1 | 0 | 16.90 | 1.63 | 0.44 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | base-q5_1 | 1 | 0 | 16.93 | 1.64 | 0.44 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | base-q8_0 | 1 | 0 | 16.13 | 1.63 | 0.43 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | small | 1 | 0 | 45.15 | 3.45 | 0.92 | 0.05 | dc8dda60 |
| M2 ULTRA | METAL | small-q5_0 | 1 | 0 | 50.63 | 3.36 | 0.94 | 0.06 | dc8dda60 |
| M2 ULTRA | METAL | small-q5_1 | 1 | 0 | 50.56 | 3.36 | 0.94 | 0.06 | dc8dda60 |
| M2 ULTRA | METAL | small-q8_0 | 1 | 0 | 47.52 | 3.20 | 0.92 | 0.05 | dc8dda60 |
| M2 ULTRA | METAL | medium | 1 | 0 | 122.55 | 7.38 | 1.95 | 0.12 | dc8dda60 |
| M2 ULTRA | METAL | medium-q5_0 | 1 | 0 | 140.61 | 6.73 | 2.02 | 0.14 | dc8dda60 |
| M2 ULTRA | METAL | medium-q5_1 | 1 | 0 | 140.48 | 6.76 | 2.04 | 0.14 | dc8dda60 |
| M2 ULTRA | METAL | medium-q8_0 | 1 | 0 | 131.00 | 6.57 | 1.96 | 0.13 | dc8dda60 |
| M2 ULTRA | METAL | medium-dis | 1 | 0 | 110.85 | 1.00 | 0.24 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | large-v2 | 1 | 0 | 222.28 | 10.96 | 3.03 | 0.21 | dc8dda60 |
| M2 ULTRA | METAL | large-v2-q5_0 | 1 | 0 | 258.64 | 9.79 | 3.04 | 0.25 | dc8dda60 |
| M2 ULTRA | METAL | large-v2-q5_1 | 1 | 0 | 258.32 | 9.87 | 3.05 | 0.24 | dc8dda60 |
| M2 ULTRA | METAL | large-v2-q8_0 | 1 | 0 | 236.55 | 9.61 | 2.87 | 0.23 | dc8dda60 |
| M2 ULTRA | METAL | large-v2-dis | 1 | 0 | 199.84 | 1.14 | 0.27 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | large-v3-turbo | 1 | 0 | 201.52 | 1.77 | 0.45 | 0.03 | dc8dda60 |
| M2 ULTRA | METAL | large-v3-turbo-q5_0 | 1 | 0 | 233.14 | 1.56 | 0.47 | 0.04 | dc8dda60 |
| M2 ULTRA | METAL | large-v3-turbo-q8_0 | 1 | 0 | 214.23 | 1.53 | 0.44 | 0.04 | dc8dda60 |
make -j && ./scripts/bench-all.sh 1 1 1
| CPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| M2 ULTRA | METAL | tiny | 1 | 1 | 7.72 | 1.05 | 0.32 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | tiny-q5_0 | 1 | 1 | 8.20 | 0.98 | 0.31 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | tiny-q5_1 | 1 | 1 | 8.13 | 0.99 | 0.31 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | tiny-q8_0 | 1 | 1 | 7.96 | 0.93 | 0.30 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | base | 1 | 1 | 13.52 | 1.39 | 0.35 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | base-q5_0 | 1 | 1 | 14.88 | 1.31 | 0.34 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | base-q5_1 | 1 | 1 | 14.76 | 1.33 | 0.34 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | base-q8_0 | 1 | 1 | 14.04 | 1.28 | 0.34 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | small | 1 | 1 | 38.78 | 2.72 | 0.67 | 0.04 | dc8dda60 |
| M2 ULTRA | METAL | small-q5_0 | 1 | 1 | 44.01 | 2.64 | 0.69 | 0.05 | dc8dda60 |
| M2 ULTRA | METAL | small-q5_1 | 1 | 1 | 44.02 | 2.66 | 0.69 | 0.05 | dc8dda60 |
| M2 ULTRA | METAL | small-q8_0 | 1 | 1 | 40.79 | 2.49 | 0.67 | 0.05 | dc8dda60 |
| M2 ULTRA | METAL | medium | 1 | 1 | 104.48 | 5.57 | 1.61 | 0.10 | dc8dda60 |
| M2 ULTRA | METAL | medium-q5_0 | 1 | 1 | 122.24 | 5.00 | 1.58 | 0.12 | dc8dda60 |
| M2 ULTRA | METAL | medium-q5_1 | 1 | 1 | 121.99 | 5.02 | 1.59 | 0.12 | dc8dda60 |
| M2 ULTRA | METAL | medium-q8_0 | 1 | 1 | 111.68 | 4.99 | 1.52 | 0.11 | dc8dda60 |
| M2 ULTRA | METAL | medium-dis | 1 | 1 | 93.23 | 0.87 | 0.21 | 0.01 | dc8dda60 |
| M2 ULTRA | METAL | large-v2 | 1 | 1 | 189.82 | 8.36 | 2.35 | 0.19 | dc8dda60 |
| M2 ULTRA | METAL | large-v2-q5_0 | 1 | 1 | 225.73 | 7.34 | 2.40 | 0.22 | dc8dda60 |
| M2 ULTRA | METAL | large-v2-q5_1 | 1 | 1 | 225.88 | 7.60 | 2.40 | 0.22 | dc8dda60 |
| M2 ULTRA | METAL | large-v2-q8_0 | 1 | 1 | 203.55 | 7.32 | 2.26 | 0.20 | dc8dda60 |
| M2 ULTRA | METAL | large-v2-dis | 1 | 1 | 168.20 | 0.98 | 0.24 | 0.02 | dc8dda60 |
| M2 ULTRA | METAL | large-v3-turbo | 1 | 1 | 170.22 | 1.46 | 0.37 | 0.03 | dc8dda60 |
| M2 ULTRA | METAL | large-v3-turbo-q5_0 | 1 | 1 | 201.88 | 1.27 | 0.38 | 0.04 | dc8dda60 |
| M2 ULTRA | METAL | large-v3-turbo-q8_0 | 1 | 1 | 182.37 | 1.24 | 0.36 | 0.03 | dc8dda60 |
## M4 Max
make -j && ./scripts/bench-all.sh 8
Running memcpy benchmark
memcpy: 57.23 GB/s (heat-up)
memcpy: 68.85 GB/s ( 1 thread)
memcpy: 70.00 GB/s ( 1 thread)
memcpy: 104.83 GB/s ( 2 thread)
memcpy: 124.54 GB/s ( 3 thread)
memcpy: 144.30 GB/s ( 4 thread)
memcpy: 141.24 GB/s ( 5 thread)
memcpy: 147.03 GB/s ( 6 thread)
memcpy: 147.18 GB/s ( 7 thread)
memcpy: 149.83 GB/s ( 8 thread)
sum: -5120001475.000000
make -j && ./scripts/bench-all.sh 1
Running ggml_mul_mat benchmark with 1 threads
64 x 64: Q4_0 49.6 GFLOPS (128 runs) | Q4_1 46.8 GFLOPS (128 runs)
64 x 64: Q5_0 28.1 GFLOPS (128 runs) | Q5_1 26.8 GFLOPS (128 runs) | Q8_0 52.3 GFLOPS (128 runs)
64 x 64: F16 38.1 GFLOPS (128 runs) | F32 26.0 GFLOPS (128 runs)
128 x 128: Q4_0 87.6 GFLOPS (128 runs) | Q4_1 79.9 GFLOPS (128 runs)
128 x 128: Q5_0 44.7 GFLOPS (128 runs) | Q5_1 41.6 GFLOPS (128 runs) | Q8_0 98.9 GFLOPS (128 runs)
128 x 128: F16 64.1 GFLOPS (128 runs) | F32 35.4 GFLOPS (128 runs)
256 x 256: Q4_0 104.2 GFLOPS (128 runs) | Q4_1 92.3 GFLOPS (128 runs)
256 x 256: Q5_0 57.3 GFLOPS (128 runs) | Q5_1 51.5 GFLOPS (128 runs) | Q8_0 127.7 GFLOPS (128 runs)
256 x 256: F16 71.4 GFLOPS (128 runs) | F32 40.6 GFLOPS (128 runs)
512 x 512: Q4_0 109.5 GFLOPS (128 runs) | Q4_1 98.0 GFLOPS (128 runs)
512 x 512: Q5_0 62.4 GFLOPS (128 runs) | Q5_1 54.6 GFLOPS (128 runs) | Q8_0 135.0 GFLOPS (128 runs)
512 x 512: F16 82.6 GFLOPS (128 runs) | F32 44.6 GFLOPS (128 runs)
1024 x 1024: Q4_0 112.1 GFLOPS ( 53 runs) | Q4_1 100.9 GFLOPS ( 47 runs)
1024 x 1024: Q5_0 65.4 GFLOPS ( 31 runs) | Q5_1 56.7 GFLOPS ( 27 runs) | Q8_0 140.9 GFLOPS ( 66 runs)
1024 x 1024: F16 88.0 GFLOPS ( 41 runs) | F32 43.4 GFLOPS ( 21 runs)
2048 x 2048: Q4_0 113.4 GFLOPS ( 7 runs) | Q4_1 102.0 GFLOPS ( 6 runs)
2048 x 2048: Q5_0 67.1 GFLOPS ( 4 runs) | Q5_1 57.7 GFLOPS ( 4 runs) | Q8_0 142.7 GFLOPS ( 9 runs)
2048 x 2048: F16 84.6 GFLOPS ( 5 runs) | F32 37.5 GFLOPS ( 3 runs)
4096 x 4096: Q4_0 113.8 GFLOPS ( 3 runs) | Q4_1 102.0 GFLOPS ( 3 runs)
4096 x 4096: Q5_0 67.7 GFLOPS ( 3 runs) | Q5_1 58.0 GFLOPS ( 3 runs) | Q8_0 142.9 GFLOPS ( 3 runs)
4096 x 4096: F16 73.7 GFLOPS ( 3 runs) | F32 36.1 GFLOPS ( 3 runs)
make -j && ./scripts/bench-all.sh 1 1 0
| CPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| M4 Max | METAL | tiny | 1 | 0 | 12.83 | 0.94 | 0.30 | 0.01 | dc8dda60 |
| M4 Max | METAL | tiny-q8_0 | 1 | 0 | 12.95 | 0.80 | 0.31 | 0.01 | dc8dda60 |
| M4 Max | METAL | base | 1 | 0 | 23.54 | 1.37 | 0.33 | 0.02 | dc8dda60 |
| M4 Max | METAL | base-q8_0 | 1 | 0 | 24.14 | 1.24 | 0.33 | 0.02 | dc8dda60 |
| M4 Max | METAL | small | 1 | 0 | 71.59 | 3.02 | 0.71 | 0.06 | dc8dda60 |
| M4 Max | METAL | small-q8_0 | 1 | 0 | 73.34 | 2.65 | 0.72 | 0.06 | dc8dda60 |
| M4 Max | METAL | medium | 1 | 0 | 208.53 | 7.02 | 1.58 | 0.16 | dc8dda60 |
| M4 Max | METAL | medium-q8_0 | 1 | 0 | 212.87 | 6.00 | 1.58 | 0.17 | dc8dda60 |
| M4 Max | METAL | large-v2 | 1 | 0 | 379.84 | 11.47 | 2.52 | 0.29 | dc8dda60 |
| M4 Max | METAL | large-v2-q8_0 | 1 | 0 | 390.45 | 9.19 | 2.48 | 0.29 | dc8dda60 |
| M4 Max | METAL | large-v3-turbo | 1 | 0 | 345.74 | 1.99 | 0.44 | 0.05 | dc8dda60 |
make -j && ./scripts/bench-all.sh 1 1 1
| CPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| M4 Max | METAL | tiny | 1 | 1 | 11.70 | 0.74 | 0.23 | 0.01 | dc8dda60 |
| M4 Max | METAL | tiny-q8_0 | 1 | 1 | 12.36 | 0.67 | 0.23 | 0.01 | dc8dda60 |
| M4 Max | METAL | base | 1 | 1 | 21.76 | 1.12 | 0.25 | 0.02 | dc8dda60 |
| M4 Max | METAL | base-q8_0 | 1 | 1 | 22.60 | 0.94 | 0.26 | 0.02 | dc8dda60 |
| M4 Max | METAL | small | 1 | 1 | 67.26 | 2.27 | 0.50 | 0.06 | dc8dda60 |
| M4 Max | METAL | small-q8_0 | 1 | 1 | 68.67 | 1.93 | 0.53 | 0.06 | dc8dda60 |
| M4 Max | METAL | medium | 1 | 1 | 193.58 | 5.31 | 1.20 | 0.16 | dc8dda60 |
| M4 Max | METAL | medium-q8_0 | 1 | 1 | 198.60 | 4.31 | 1.21 | 0.16 | dc8dda60 |
| M4 Max | METAL | large-v2 | 1 | 1 | 357.54 | 8.73 | 1.99 | 0.27 | dc8dda60 |
| M4 Max | METAL | large-v2-q8_0 | 1 | 1 | 363.98 | 6.43 | 1.99 | 0.28 | dc8dda60 |
| M4 Max | METAL | large-v3-turbo | 1 | 1 | 322.32 | 1.66 | 0.37 | 0.05 | dc8dda60 |
# V100
GGML_CUDA=1 make -j && ./scripts/bench-all.sh 8 1 0
| GPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| V100 | AVX2 CUDA | tiny | 8 | 0 | 5.99 | 1.01 | 0.30 | 0.01 | dc8dda60 |
| V100 | AVX2 CUDA | tiny-q5_1 | 8 | 0 | 6.07 | 1.00 | 0.26 | 0.01 | dc8dda60 |
| V100 | AVX2 CUDA | base | 8 | 0 | 10.96 | 1.44 | 0.43 | 0.02 | dc8dda60 |
| V100 | AVX2 CUDA | base-q5_1 | 8 | 0 | 11.11 | 1.41 | 0.37 | 0.02 | dc8dda60 |
| V100 | AVX2 CUDA | small | 8 | 0 | 31.04 | 2.84 | 0.86 | 0.04 | dc8dda60 |
| V100 | AVX2 CUDA | small-q5_1 | 8 | 0 | 31.69 | 2.82 | 0.71 | 0.04 | dc8dda60 |
| V100 | AVX2 CUDA | medium | 8 | 0 | 83.95 | 6.05 | 1.82 | 0.09 | dc8dda60 |
| V100 | AVX2 CUDA | medium-q5_0 | 8 | 0 | 85.86 | 5.58 | 1.45 | 0.10 | dc8dda60 |
| V100 | AVX2 CUDA | large-v2 | 8 | 0 | 138.50 | 8.70 | 2.71 | 0.15 | dc8dda60 |
| V100 | AVX2 CUDA | large-v2-q5_0 | 8 | 0 | 142.31 | 7.82 | 2.03 | 0.16 | dc8dda60 |
| V100 | AVX2 CUDA | large-v3-turbo | 8 | 0 | 128.39 | 1.42 | 0.44 | 0.02 | dc8dda60 |
| V100 | AVX2 CUDA | large-v3-turbo-q5_0 | 8 | 0 | 131.24 | 1.17 | 0.33 | 0.03 | dc8dda60 |
GGML_CUDA=1 make -j && ./scripts/bench-all.sh 8 1 1
| GPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| V100 | AVX2 CUDA | tiny | 8 | 1 | 4.85 | 0.97 | 0.26 | 0.01 | dc8dda60 |
| V100 | AVX2 CUDA | tiny-q5_1 | 8 | 1 | 4.97 | 0.89 | 0.19 | 0.01 | dc8dda60 |
| V100 | AVX2 CUDA | base | 8 | 1 | 7.23 | 1.28 | 0.35 | 0.02 | dc8dda60 |
| V100 | AVX2 CUDA | base-q5_1 | 8 | 1 | 7.38 | 1.24 | 0.26 | 0.02 | dc8dda60 |
| V100 | AVX2 CUDA | small | 8 | 1 | 20.87 | 2.44 | 0.71 | 0.03 | dc8dda60 |
| V100 | AVX2 CUDA | small-q5_1 | 8 | 1 | 19.80 | 2.35 | 0.51 | 0.03 | dc8dda60 |
| V100 | AVX2 CUDA | medium | 8 | 1 | 54.56 | 5.31 | 1.46 | 0.06 | dc8dda60 |
| V100 | AVX2 CUDA | medium-q5_0 | 8 | 1 | 56.09 | 4.67 | 1.05 | 0.07 | dc8dda60 |
| V100 | AVX2 CUDA | large-v2 | 8 | 1 | 87.05 | 7.65 | 2.16 | 0.10 | dc8dda60 |
| V100 | AVX2 CUDA | large-v2-q5_0 | 8 | 1 | 94.65 | 6.60 | 1.47 | 0.11 | dc8dda60 |
| V100 | AVX2 CUDA | large-v3-turbo | 8 | 1 | 76.46 | 1.29 | 0.37 | 0.02 | dc8dda60 |
| V100 | AVX2 CUDA | large-v3-turbo-q5_0 | 8 | 1 | 79.62 | 1.03 | 0.23 | 0.02 | dc8dda60 |
Reference in New Issue View Git Blame Copy Permalink