-
We currently use the BabelStream kernels as memory bound workloads and a fibonacci kernel (per thread: read random index [0;42], compute fibonacci number, store result). The latter kernel is compute a…
-
ASV sometimes adds a `~` to ratios if there are many sample (with high variance, I'm guessing), but there's a bug somewhere how these results are evaluated when the significant changes are given out.
…
-
We should ensure our tools (analyzer, analysis server, dart2js, cfe) have very good benchmark coverage for all metrics we care about and continuously track those metrics on golem.
* [ ] Ensure we h…
-
@chrisohaver how hard/easy is it to do a new k8s performance test with the current CoreDNS master branch (or the v008 release)? Because I think we should redo it: new cache, singleinflight gone from t…
miekg updated
7 years ago
-
As we care performance a lot, it may be necessary to run benchmarks for each PR submission.
-
## 🎛 Description
**Choose one**: 🙋 feature request
Something we will look at and continue to look at is performance. Not only is it an important element for the development team, but is a critic…
-
## webpack performance test harness
For a while, we had setup a performance benchmark system on CI which would compare the performance of any fix/feature at PR time made to webpack/webpack. However…
-
# Progress
- [x] Implement TPU executor that works on a single TPU chip (without tensor parallelism) #5292
- [x] Support single-host tensor parallel inference #5871
- [x] Support multi-host ten…
-
Background: In (DREAM) Challenges organized using Synapse, participants are required to accept the challenge / platform rules and Data User Agreement. Now the implementation of this process has not fo…
-
I'm appreciate a redditor, jeromegn, post an unpublished TFB and found H2O (C) is surprising slower than H2O.cr (Crystal+H2O) is probably due to monkey patches for Crystal.
https://www.techempower.co…