diff options
| author | Pjotr Prins | 2025-11-26 12:01:37 +0100 |
|---|---|---|
| committer | Pjotr Prins | 2025-11-26 12:01:37 +0100 |
| commit | 806a01c27a1fd366c120820de0c38caea1d4cdcd (patch) | |
| tree | fae0f2e33453c77f4fdc58c952bda397892c984a /test | |
| parent | c9824193dd3fd6ff398942885f110b27d8b646a0 (diff) | |
| download | pangemma-806a01c27a1fd366c120820de0c38caea1d4cdcd.tar.gz | |
openblas speed on par
Diffstat (limited to 'test')
| -rw-r--r-- | test/performance/releases.org | 21 |
1 files changed, 19 insertions, 2 deletions
diff --git a/test/performance/releases.org b/test/performance/releases.org index 4cc92f1..792cb2e 100644 --- a/test/performance/releases.org +++ b/test/performance/releases.org @@ -1,6 +1,7 @@ * GEMMA performance stats -** GEMMA 0.98.5 +** GEMMA 1.00-pre1 + Measurements taken on a recent AMD Ryzen 7 3700X 8-Core Processor @2.195GHz. @@ -8,7 +9,7 @@ We are facing a time regression. premake5 gmake2 && make verbose=1 config=release -j 8 gemma && time LD_LIBRARY_PATH=$GUIX_ENVIRONMENT/lib ./build/bin/Release/gemma -g ./example/mouse_hs1940.geno.txt.gz -p ./example/mouse_hs1940.pheno.txt -n 1 -a ./example/mouse_hs1940.anno.txt -k ./output/result.cXX.txt -lmm -no-check -debug -With openblas 0.3.21 we go a bit faster. Still behind though, there is room for tweaking. But I want to run some bigger files first. +With openblas 0.3.21 we go a bit faster. Still 10% behind though, there is room for tweaking. But I want to run some bigger files first. #+begin_src sh Pangemma --- GEMMA 0.98.5 compatible executable 1.0.0 (2025-11-22) with guile 3.0.9 by Xiang Zhou, Pjotr Prins and team (C) 2012-2025 @@ -28,6 +29,7 @@ user 0m13.168s sys 0m5.919s #+end_src sh +Before it was #+begin_src sh Pangemma --- GEMMA 0.98.5 compatible executable 1.0.0 (2025-11-22) with guile 3.0.9 by Xiang Zhou, Pjotr Prins and team (C) 2012-2025 @@ -67,6 +69,21 @@ this led me to try the newer openblas on the older gemma - and indeed, the regre Well, at least I found the problem. Time for a special openblas build like I used to do. + +*** Bigger run + +We translate this 10Gb (gzip compressed) job from our pangenome precompute + +``` +/bin/gemma -loco 3 -k /export2/data/wrk/services/gemma-wrapper/tmp/tmp/panlmm/93f6b39ec06c09fb9ba9ca628b5fb990921b6c60.3.cXX.txt.cXX.txt -o a3248cec40b3fe6b9e8672352b3ab2d7280c426c.3.assoc.txt -p pheno.json.txt -g pangenome-13M-genotypes.txt -a snps-matched.txt -lmm 9 -maf 0.1 -n 2 -outdir /export2/data/wrk/services/gemma-wrapper/tmp/tmp/panlmm/d20251126-4190721-c8bbo8 +``` + +to + +``` +time LD_LIBRARY_PATH=$GUIX_ENVIRONMENT/lib ./build/bin/Release/gemma -g tmp/pangenome-13M-genotypes.txt -p tmp/pheno.json.txt -n 1 -a tmp/snps-matched.txt -k tmp/93f6b39ec06c09fb9ba9ca628b5fb990921b6c60.3.cXX.txt.cXX.txt -lmm 9 -no-check +``` + ** GEMMA 0.98.5-pre1 Measurements taken on a recent AMD Ryzen 7 3700X 8-Core Processor @2.195GHz. |