Influxdata
/
influxdb
mirror of https://github.com/influxdata/influxdb.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
28,853 Commits
205 Branches
453 Tags
382 MiB
Commit Graph

8 Commits (9c33541275e3268b4e35b08d9129ff2f60f8128c)

Author SHA1 Message Date
Stuart Carnie a0300064df feat(tsm1): Improve performance of Gorilla float block decoding 
```
name                        old time/op   new time/op    delta
FloatArrayDecodeAll/1-8      45.9ns ± 1%    13.8ns ± 1%   -70.00%  (p=0.000 n=9+9)
FloatArrayDecodeAll/55-8      686ns ± 0%     232ns ± 1%   -66.10%  (p=0.000 n=9+8)
FloatArrayDecodeAll/550-8    5.78µs ± 0%    2.22µs ± 1%   -61.61%  (p=0.000 n=9+9)
FloatArrayDecodeAll/1000-8   10.2µs ± 2%     4.0µs ± 5%   -60.47%  (p=0.000 n=10+10)

name                        old speed     new speed      delta
FloatArrayDecodeAll/1-8     414MB/s ± 1%  1383MB/s ± 1%  +233.76%  (p=0.000 n=9+9)
FloatArrayDecodeAll/55-8    144MB/s ± 0%   424MB/s ± 1%  +194.19%  (p=0.000 n=9+9)
FloatArrayDecodeAll/550-8   133MB/s ± 0%   346MB/s ± 1%  +160.09%  (p=0.000 n=9+10)
FloatArrayDecodeAll/1000-8  135MB/s ± 2%   340MB/s ± 5%  +153.03%  (p=0.000 n=10+10)
```
 2018-11-01 18:59:20 +00:00
Edd Robinson e282d012c8 Address PR feedback 2018-11-01 18:59:20 +00:00
Edd Robinson 5e7b2cb273 Fix index bug in float encoder 2018-11-01 18:59:20 +00:00
Jeff Wendling 5376530392 Improvements to batch float encoder 
- Inlined the closure to avoid a function call.
- Changed append(b, make([]byte, 8)...) to inline the make call.
- Check for NaN once at the end assuming NaN is infrequent.

New performance delta comparing the current iterators to the new batch
function:

name                   old time/op    new time/op    delta
EncodeFloats/10_seq      1.32µs ± 2%    0.17µs ± 2%  -87.39%  (p=0.000 n=10+10)
EncodeFloats/10_ran      2.09µs ± 1%    0.15µs ± 0%  -92.97%  (p=0.000 n=10+9)
EncodeFloats/100_seq     8.37µs ± 2%    1.28µs ± 2%  -84.74%  (p=0.000 n=10+10)
EncodeFloats/100_ran     19.1µs ± 1%     1.3µs ± 1%  -93.08%  (p=0.000 n=9+9)
EncodeFloats/1000_seq    60.4µs ± 1%    12.6µs ± 0%  -79.13%  (p=0.000 n=9+7)
EncodeFloats/1000_ran     212µs ± 1%      12µs ± 1%  -94.53%  (p=0.000 n=9+8)

name                   old alloc/op   new alloc/op   delta
EncodeFloats/10_seq       0.00B          0.00B          ~     (all equal)
EncodeFloats/10_ran       0.00B          0.00B          ~     (all equal)
EncodeFloats/100_seq      0.00B          0.00B          ~     (all equal)
EncodeFloats/100_ran      0.00B          0.00B          ~     (all equal)
EncodeFloats/1000_seq     0.00B          0.00B          ~     (all equal)
EncodeFloats/1000_ran     0.00B          0.00B          ~     (all equal)

name                   old allocs/op  new allocs/op  delta
EncodeFloats/10_seq        0.00           0.00          ~     (all equal)
EncodeFloats/10_ran        0.00           0.00          ~     (all equal)
EncodeFloats/100_seq       0.00           0.00          ~     (all equal)
EncodeFloats/100_ran       0.00           0.00          ~     (all equal)
EncodeFloats/1000_seq      0.00           0.00          ~     (all equal)
EncodeFloats/1000_ran      0.00           0.00          ~     (all equal)
 2018-11-01 18:59:19 +00:00
Edd Robinson 8190edbf14 Batch oriented float encoders 
This commit adds a tsm1 function for encoding a batch of floats into a
buffer. Further, it replaces the `bitstream` library used in the
existing encoders (and all the current decoders) with inlined bit
expressions within the encoder, significantly reducing the function call
overhead for larger batches.

The following benchmarks compare the performance of the existing
iterator based encoders, and the new batch oriented encoders. They look
at a sequential input slice and a randomly generated input slice.

name                   old time/op    new time/op    delta
EncodeFloats/10_seq      1.14µs ± 3%    0.24µs ± 3%  -78.94%  (p=0.000 n=10+10)
EncodeFloats/10_ran      1.69µs ± 2%    0.21µs ± 3%  -87.43%  (p=0.000 n=10+10)
EncodeFloats/100_seq     7.07µs ± 1%    1.72µs ± 1%  -75.62%  (p=0.000 n=7+9)
EncodeFloats/100_ran     15.8µs ± 4%     1.8µs ± 1%  -88.60%  (p=0.000 n=10+9)
EncodeFloats/1000_seq    50.2µs ± 3%    16.2µs ± 2%  -67.66%  (p=0.000 n=10+10)
EncodeFloats/1000_ran     174µs ± 2%      16µs ± 2%  -90.77%  (p=0.000 n=10+10)

name                   old alloc/op   new alloc/op   delta
EncodeFloats/10_seq       0.00B          0.00B          ~     (all equal)
EncodeFloats/10_ran       0.00B          0.00B          ~     (all equal)
EncodeFloats/100_seq      0.00B          0.00B          ~     (all equal)
EncodeFloats/100_ran      0.00B          0.00B          ~     (all equal)
EncodeFloats/1000_seq     0.00B          0.00B          ~     (all equal)
EncodeFloats/1000_ran     0.00B          0.00B          ~     (all equal)

name                   old allocs/op  new allocs/op  delta
EncodeFloats/10_seq        0.00           0.00          ~     (all equal)
EncodeFloats/10_ran        0.00           0.00          ~     (all equal)
EncodeFloats/100_seq       0.00           0.00          ~     (all equal)
EncodeFloats/100_ran       0.00           0.00          ~     (all equal)
EncodeFloats/1000_seq      0.00           0.00          ~     (all equal)
EncodeFloats/1000_ran      0.00           0.00          ~     (all equal)
 2018-11-01 18:59:19 +00:00
Edd Robinson 2e00954703 Rename float batch decoders 2018-11-01 18:59:19 +00:00
Edd Robinson 7de29bfe01 Update tsi1 package from OSS 2018-10-01 12:08:37 +01:00
Edd Robinson 074f263e08 Initial import of tsm1.Engine 2018-10-01 12:08:37 +01:00