Alexander Neumann
|
b2307cafa2
|
Seed chunker with random per-repository polynomial
|
2015-04-06 00:22:19 +02:00 |
|
Alexander Neumann
|
24b14e21cc
|
Remove pools for nodes and IDs
This removes the allocation pools for nodes and IDs. I feel they aren't
really needed. Benchmarks:
benchmark old ns/op new ns/op delta
BenchmarkChunkEncrypt 197890867 198616293 +0.37%
BenchmarkChunkEncryptParallel 196127004 198819818 +1.37%
BenchmarkArchiveDirectory 1098848419 1087237723 -1.06%
BenchmarkPreload 30464455 29910239 -1.82%
BenchmarkLoadTree 3265092 3088543 -5.41%
BenchmarkEncryptWriter 37213511 37134683 -0.21%
BenchmarkEncrypt 36037879 36166546 +0.36%
BenchmarkDecryptReader 38165659 38556734 +1.02%
BenchmarkEncryptDecryptReader 77027044 77194987 +0.22%
BenchmarkDecrypt 36017602 35937888 -0.22%
BenchmarkSaveJSON 47906 50270 +4.93%
BenchmarkSaveFrom 49775973 50520969 +1.50%
BenchmarkLoadJSONID 105290245 107281849 +1.89%
BenchmarkChunkerWithSHA256 151501430 148264078 -2.14%
BenchmarkChunkerWithMD5 93606346 94036392 +0.46%
BenchmarkChunker 74285431 75933882 +2.22%
BenchmarkPipelineWalker 387689 346467 -10.63%
benchmark old MB/s new MB/s speedup
BenchmarkChunkEncrypt 52.99 52.79 1.00x
BenchmarkChunkEncryptParallel 53.46 52.74 0.99x
BenchmarkEncryptWriter 225.42 225.90 1.00x
BenchmarkEncrypt 232.77 231.94 1.00x
BenchmarkDecryptReader 219.79 217.57 0.99x
BenchmarkEncryptDecryptReader 108.90 108.67 1.00x
BenchmarkDecrypt 232.90 233.42 1.00x
BenchmarkSaveFrom 84.26 83.02 0.99x
BenchmarkChunkerWithSHA256 69.21 70.72 1.02x
BenchmarkChunkerWithMD5 112.02 111.51 1.00x
BenchmarkChunker 141.15 138.09 0.98x
benchmark old allocs new allocs delta
BenchmarkChunkEncrypt 110 110 +0.00%
BenchmarkChunkEncryptParallel 100 100 +0.00%
BenchmarkArchiveDirectory 475591 476635 +0.22%
BenchmarkPreload 28059 24182 -13.82%
BenchmarkLoadTree 3124 2889 -7.52%
BenchmarkEncryptWriter 19 19 +0.00%
BenchmarkEncrypt 13 13 +0.00%
BenchmarkDecryptReader 16 15 -6.25%
BenchmarkEncryptDecryptReader 39 39 +0.00%
BenchmarkDecrypt 11 11 +0.00%
BenchmarkSaveJSON 74 74 +0.00%
BenchmarkSaveFrom 109 112 +2.75%
BenchmarkLoadJSONID 103630 97849 -5.58%
BenchmarkChunkerWithSHA256 13 13 +0.00%
BenchmarkChunkerWithMD5 12 12 +0.00%
BenchmarkChunker 6 6 +0.00%
BenchmarkPipelineWalker 212 165 -22.17%
benchmark old bytes new bytes delta
BenchmarkChunkEncrypt 64697 64697 +0.00%
BenchmarkChunkEncryptParallel 64681 64681 +0.00%
BenchmarkArchiveDirectory 193385504 193790864 +0.21%
BenchmarkPreload 4064701 3942000 -3.02%
BenchmarkLoadTree 344954 325396 -5.67%
BenchmarkEncryptWriter 12793 12793 +0.00%
BenchmarkEncrypt 1950 1950 +0.00%
BenchmarkDecryptReader 3120 2774 -11.09%
BenchmarkEncryptDecryptReader 1528036 1528036 +0.00%
BenchmarkDecrypt 1919 1919 +0.00%
BenchmarkSaveJSON 5524 5524 +0.00%
BenchmarkSaveFrom 31353 40804 +30.14%
BenchmarkLoadJSONID 12872020 16010968 +24.39%
BenchmarkChunkerWithSHA256 26821 26821 +0.00%
BenchmarkChunkerWithMD5 13554 13554 +0.00%
BenchmarkChunker 13458 13458 +0.00%
BenchmarkPipelineWalker 58584 55560 -5.16%
|
2015-03-28 16:51:37 +01:00 |
|
Alexander Neumann
|
662e07d17a
|
Refactor crypto layer, switch HMAC for Poyl1305-AES
HMAC-SHA256 calls SHA256() twice which is very expensive. Therefore,
this commit uses Poly1305-AES instead of HMAC-SHA256.
benchcmp:
benchmark old ns/op new ns/op delta
BenchmarkChunkEncrypt 261033772 195114818 -25.25%
BenchmarkChunkEncryptParallel 260973195 195787368 -24.98%
BenchmarkArchiveDirectory 1050500651 1002615884 -4.56%
BenchmarkPreload 23544286 24994508 +6.16%
BenchmarkLoadTree 350065 427665 +22.17%
BenchmarkEncryptWriter 87789753 31069126 -64.61%
BenchmarkEncrypt 88283197 38259043 -56.66%
BenchmarkDecryptReader 90478843 40714818 -55.00%
BenchmarkEncryptDecryptReader 179917626 81231730 -54.85%
BenchmarkDecrypt 87871591 37784207 -57.00%
BenchmarkSaveJSON 52481 56861 +8.35%
BenchmarkSaveFrom 75404085 51108596 -32.22%
BenchmarkLoadJSONID 90545437 82696805 -8.67%
benchmark old MB/s new MB/s speedup
BenchmarkChunkEncrypt 40.17 53.74 1.34x
BenchmarkChunkEncryptParallel 40.18 53.56 1.33x
BenchmarkEncryptWriter 95.55 270.00 2.83x
BenchmarkEncrypt 95.02 219.26 2.31x
BenchmarkDecryptReader 92.71 206.03 2.22x
BenchmarkEncryptDecryptReader 46.62 103.27 2.22x
BenchmarkDecrypt 95.46 222.01 2.33x
BenchmarkSaveFrom 55.62 82.07 1.48x
benchmark old allocs new allocs delta
BenchmarkChunkEncrypt 112 110 -1.79%
BenchmarkChunkEncryptParallel 103 100 -2.91%
BenchmarkArchiveDirectory 383704 392083 +2.18%
BenchmarkPreload 21765 21874 +0.50%
BenchmarkLoadTree 341 436 +27.86%
BenchmarkEncryptWriter 20 17 -15.00%
BenchmarkEncrypt 14 13 -7.14%
BenchmarkDecryptReader 18 15 -16.67%
BenchmarkEncryptDecryptReader 46 39 -15.22%
BenchmarkDecrypt 16 12 -25.00%
BenchmarkSaveJSON 81 86 +6.17%
BenchmarkSaveFrom 117 121 +3.42%
BenchmarkLoadJSONID 80525 80264 -0.32%
benchmark old bytes new bytes delta
BenchmarkChunkEncrypt 118956 64697 -45.61%
BenchmarkChunkEncryptParallel 118972 64681 -45.63%
BenchmarkArchiveDirectory 160236600 177498232 +10.77%
BenchmarkPreload 2772488 3302992 +19.13%
BenchmarkLoadTree 49102 46484 -5.33%
BenchmarkEncryptWriter 28927 8388146 +28897.64%
BenchmarkEncrypt 2473 1950 -21.15%
BenchmarkDecryptReader 527827 2774 -99.47%
BenchmarkEncryptDecryptReader 4100875 1528036 -62.74%
BenchmarkDecrypt 2509 2154 -14.15%
BenchmarkSaveJSON 4971 5892 +18.53%
BenchmarkSaveFrom 40117 31742 -20.88%
BenchmarkLoadJSONID 9444217 9442106 -0.02%
This closes #102.
|
2015-03-14 20:00:41 +01:00 |
|
Alexander Neumann
|
3bf66d24fa
|
chunker: Add test, only pass hash.Hash to New()
|
2015-02-11 13:12:36 +01:00 |
|
Alexander Neumann
|
bbb3145f11
|
Add chunker pool
|
2015-02-09 23:38:50 +01:00 |
|
Alexander Neumann
|
c884704bce
|
Refactor pool stats
|
2015-02-08 13:46:51 +01:00 |
|
Alexander Neumann
|
e2fea0d088
|
Rename khepri -> restic
|
2014-12-05 21:45:49 +01:00 |
|
Alexander Neumann
|
719e121c74
|
Use large, dynamic buffer for encrypting maps
|
2014-11-23 22:58:41 +01:00 |
|
Alexander Neumann
|
f95788ed90
|
Limit memory usage, add several sync.Pool
|
2014-11-23 21:26:20 +01:00 |
|