dolysis
/
rocksdb
mirror of https://github.com/facebook/rocksdb.git
3
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Java FFI blog post - Post-publication issues with images (2) (#12372) 

Summary:
Replace unreliable-in-chrome PDF w/PNG of same graph

jmh-result-pinnable-vs-output-plot.pdf is showing as thumbnail on Chrome, rendering OK on Safari for some; I have converted it to PNG in the hope that will display correctly in all environments.

Pull Request resolved: https://github.com/facebook/rocksdb/pull/12372

Reviewed By: cbi42

Differential Revision: D54076718

Pulled By: jowlyzhang

fbshipit-source-id: 2eff995f0239ab7850a40063d841380738953533
This commit is contained in:
Alan Paxton 2024-02-22 15:01:55 -08:00 committed by Facebook GitHub Bot
parent e09b9d0cb9
commit d1386de632
3 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
docs/_posts/2024-02-20-foreign-function-interface.markdown
View File

@ -190,7 +190,7 @@ q "select Benchmark,Score from ./plot/jmh-result-fixed.csv where \"Param: keyCou
The second method call over the FFI boundary to release a pinnable slice has a cost. We compared the `ffiGetOutputSlice()` and `ffiGetPinnableSlice()` benchmarks in order to examine this cost. We ran it with a fixed ky size (128 bytes); the key size is likely to be pretty much irrelevant anyway; we varied the value size read from 16 bytes to 16k, and we found a crossover point between 1k and 4k for performance: The second method call over the FFI boundary to release a pinnable slice has a cost. We compared the `ffiGetOutputSlice()` and `ffiGetPinnableSlice()` benchmarks in order to examine this cost. We ran it with a fixed ky size (128 bytes); the key size is likely to be pretty much irrelevant anyway; we varied the value size read from 16 bytes to 16k, and we found a crossover point between 1k and 4k for performance:
![Plot](/static/images/jni-ffi/jmh-result-pinnable-vs-output-plot.pdf) ![Plot](/static/images/jni-ffi/jmh-result-pinnable-vs-output-plot.png)
- `ffiGetOutputSlice()` is faster when values read are 1k in size or smaller. The cost of an extra copy in the C++ side from the pinnable slice buffer into the supplied buffer allocated by Java Foreign Memory API is less than the cost of the extra call to release a pinnable slice. - `ffiGetOutputSlice()` is faster when values read are 1k in size or smaller. The cost of an extra copy in the C++ side from the pinnable slice buffer into the supplied buffer allocated by Java Foreign Memory API is less than the cost of the extra call to release a pinnable slice.
- `ffiGetPinnableSlice()` is faster when values read are 4k in size, or larger. Consistent with intuition, the advantage grows with larger read values. - `ffiGetPinnableSlice()` is faster when values read are 4k in size, or larger. Consistent with intuition, the advantage grows with larger read values.

BIN
docs/static/images/jni-ffi/jmh-result-pinnable-vs-output-plot.pdf vendored
View File

Binary file not shown.

BIN
docs/static/images/jni-ffi/jmh-result-pinnable-vs-output-plot.png vendored Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 11 KiB