tvl-depot/tvix/cli/default.nix

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

96 lines
3.7 KiB
Nix
Raw Normal View History

{ depot, pkgs, lib, ... }:
(depot.tvix.crates.workspaceMembers.tvix-cli.build.override {
runTests = true;
}).overrideAttrs (finalAttrs: previousAttrs:
let
tvix-cli = finalAttrs.finalPackage;
feat(tvix/cli): add macrobenchmark This commit adds a simple MVP benchmark, built on our nix infrastructure instead of cargo. It simply runs `tvix-eval` inside of GNU time, and prints the three essential statistics in a short JSON blob. You can run the benchmark with a simple `nix run`, like: nix run -f . tvix.cli.benchmark-hello nix run -f . tvix.cli.benchmark-firefox nix run -f . tvix.cli.benchmark-cross-firefox Currently these blobs are stored only in the CI logs, which I'm sure get garbage-collected at some point. We should be putting them in the git trailers, but that can wait for a future CL. I tried using `cargo bench` for this but found it incredibly frustrating. Maybe I'm doing it wrong. It seems to be designed for microbenchmarks only, and very hard to control. It kept building all sorts of unnecessary stuff (like the tests), and unlike crate2nix it was doing all the builds on only a single machine instead of using more than one machine. Worse, for that single machine it kept picking my laptop instead of my fast servers! It seems excessively cargo-flavored for such a straightforward task. Benchmark: {"hello.outPath":{"kbytes":"244736","system":"0.36","user":"2.76"}} Benchmark: {"firefox.outPath":{"kbytes":"1506736","system":"2.38","user":"32.01"}} Benchmark: {"pkgsCross.aarch64-multiplatform.firefox.outPath":{"kbytes":"11334548","system":"10.70","user":"107.07"}} Change-Id: I85bc046ec551360284d7ecfc81a03914f0085909 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10216 Autosubmit: Adam Joseph <adam@westernsemico.com> Reviewed-by: grfn <grfn@gws.fyi> Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
2023-12-09 05:15:31 +01:00
benchmark-gnutime-format-string =
description:
"Benchmark: " +
(builtins.toJSON {
"${description}" = {
kbytes = "%M";
system = "%S";
user = "%U";
};
});
# You can run the benchmark with a simple `nix run`, like:
#
# nix-build -A tvix.cli.meta.ci.extraSteps.benchmark-nixpkgs-cross-hello-outpath
feat(tvix/cli): add macrobenchmark This commit adds a simple MVP benchmark, built on our nix infrastructure instead of cargo. It simply runs `tvix-eval` inside of GNU time, and prints the three essential statistics in a short JSON blob. You can run the benchmark with a simple `nix run`, like: nix run -f . tvix.cli.benchmark-hello nix run -f . tvix.cli.benchmark-firefox nix run -f . tvix.cli.benchmark-cross-firefox Currently these blobs are stored only in the CI logs, which I'm sure get garbage-collected at some point. We should be putting them in the git trailers, but that can wait for a future CL. I tried using `cargo bench` for this but found it incredibly frustrating. Maybe I'm doing it wrong. It seems to be designed for microbenchmarks only, and very hard to control. It kept building all sorts of unnecessary stuff (like the tests), and unlike crate2nix it was doing all the builds on only a single machine instead of using more than one machine. Worse, for that single machine it kept picking my laptop instead of my fast servers! It seems excessively cargo-flavored for such a straightforward task. Benchmark: {"hello.outPath":{"kbytes":"244736","system":"0.36","user":"2.76"}} Benchmark: {"firefox.outPath":{"kbytes":"1506736","system":"2.38","user":"32.01"}} Benchmark: {"pkgsCross.aarch64-multiplatform.firefox.outPath":{"kbytes":"11334548","system":"10.70","user":"107.07"}} Change-Id: I85bc046ec551360284d7ecfc81a03914f0085909 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10216 Autosubmit: Adam Joseph <adam@westernsemico.com> Reviewed-by: grfn <grfn@gws.fyi> Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
2023-12-09 05:15:31 +01:00
#
# TODO(amjoseph): store these results someplace more durable, like git trailers
#
mkExprBenchmark = { expr, description }:
let name = "tvix-cli-benchmark-${description}"; in
(pkgs.runCommand name { } ''
export SSL_CERT_FILE=${pkgs.cacert.out}/etc/ssl/certs/ca-bundle.crt
feat(tvix/cli): add macrobenchmark This commit adds a simple MVP benchmark, built on our nix infrastructure instead of cargo. It simply runs `tvix-eval` inside of GNU time, and prints the three essential statistics in a short JSON blob. You can run the benchmark with a simple `nix run`, like: nix run -f . tvix.cli.benchmark-hello nix run -f . tvix.cli.benchmark-firefox nix run -f . tvix.cli.benchmark-cross-firefox Currently these blobs are stored only in the CI logs, which I'm sure get garbage-collected at some point. We should be putting them in the git trailers, but that can wait for a future CL. I tried using `cargo bench` for this but found it incredibly frustrating. Maybe I'm doing it wrong. It seems to be designed for microbenchmarks only, and very hard to control. It kept building all sorts of unnecessary stuff (like the tests), and unlike crate2nix it was doing all the builds on only a single machine instead of using more than one machine. Worse, for that single machine it kept picking my laptop instead of my fast servers! It seems excessively cargo-flavored for such a straightforward task. Benchmark: {"hello.outPath":{"kbytes":"244736","system":"0.36","user":"2.76"}} Benchmark: {"firefox.outPath":{"kbytes":"1506736","system":"2.38","user":"32.01"}} Benchmark: {"pkgsCross.aarch64-multiplatform.firefox.outPath":{"kbytes":"11334548","system":"10.70","user":"107.07"}} Change-Id: I85bc046ec551360284d7ecfc81a03914f0085909 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10216 Autosubmit: Adam Joseph <adam@westernsemico.com> Reviewed-by: grfn <grfn@gws.fyi> Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
2023-12-09 05:15:31 +01:00
${lib.escapeShellArgs [
"${pkgs.time}/bin/time"
"--format" "${benchmark-gnutime-format-string description}"
"${tvix-cli}/bin/tvix"
"--no-warnings"
"-E" expr
]}
touch $out
'');
feat(tvix/cli): add macrobenchmark This commit adds a simple MVP benchmark, built on our nix infrastructure instead of cargo. It simply runs `tvix-eval` inside of GNU time, and prints the three essential statistics in a short JSON blob. You can run the benchmark with a simple `nix run`, like: nix run -f . tvix.cli.benchmark-hello nix run -f . tvix.cli.benchmark-firefox nix run -f . tvix.cli.benchmark-cross-firefox Currently these blobs are stored only in the CI logs, which I'm sure get garbage-collected at some point. We should be putting them in the git trailers, but that can wait for a future CL. I tried using `cargo bench` for this but found it incredibly frustrating. Maybe I'm doing it wrong. It seems to be designed for microbenchmarks only, and very hard to control. It kept building all sorts of unnecessary stuff (like the tests), and unlike crate2nix it was doing all the builds on only a single machine instead of using more than one machine. Worse, for that single machine it kept picking my laptop instead of my fast servers! It seems excessively cargo-flavored for such a straightforward task. Benchmark: {"hello.outPath":{"kbytes":"244736","system":"0.36","user":"2.76"}} Benchmark: {"firefox.outPath":{"kbytes":"1506736","system":"2.38","user":"32.01"}} Benchmark: {"pkgsCross.aarch64-multiplatform.firefox.outPath":{"kbytes":"11334548","system":"10.70","user":"107.07"}} Change-Id: I85bc046ec551360284d7ecfc81a03914f0085909 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10216 Autosubmit: Adam Joseph <adam@westernsemico.com> Reviewed-by: grfn <grfn@gws.fyi> Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
2023-12-09 05:15:31 +01:00
mkNixpkgsBenchmark = attrpath:
mkExprBenchmark {
description = builtins.replaceStrings [ ".drv" ] [ "-drv" ] attrpath;
expr = "(import ${pkgs.path} {}).${attrpath}";
};
# Constructs a Derivation invoking tvix-cli inside a build, ensures the
# calculated tvix output path matches what's passed in externally.
mkNixpkgsEvalTest = attrpath: expectedPath:
let
name = "tvix-eval-test-${builtins.replaceStrings [".drv"] ["-drv"] attrpath}";
in
(pkgs.runCommand name { } ''
export SSL_CERT_FILE=${pkgs.cacert.out}/etc/ssl/certs/ca-bundle.crt
TVIX_OUTPUT=$(${tvix-cli}/bin/tvix -E '(import ${pkgs.path} {}).${attrpath}')
EXPECTED='${/* the verbatim expected Tvix output: */ "=> \"${builtins.unsafeDiscardStringContext expectedPath}\" :: string"}'
echo "Tvix output: ''${TVIX_OUTPUT}"
if [ "$TVIX_OUTPUT" != "$EXPECTED" ]; then
echo "Correct would have been ''${EXPECTED}"
exit 1
fi
echo "Output was correct."
touch $out
'');
feat(tvix/cli): add macrobenchmark This commit adds a simple MVP benchmark, built on our nix infrastructure instead of cargo. It simply runs `tvix-eval` inside of GNU time, and prints the three essential statistics in a short JSON blob. You can run the benchmark with a simple `nix run`, like: nix run -f . tvix.cli.benchmark-hello nix run -f . tvix.cli.benchmark-firefox nix run -f . tvix.cli.benchmark-cross-firefox Currently these blobs are stored only in the CI logs, which I'm sure get garbage-collected at some point. We should be putting them in the git trailers, but that can wait for a future CL. I tried using `cargo bench` for this but found it incredibly frustrating. Maybe I'm doing it wrong. It seems to be designed for microbenchmarks only, and very hard to control. It kept building all sorts of unnecessary stuff (like the tests), and unlike crate2nix it was doing all the builds on only a single machine instead of using more than one machine. Worse, for that single machine it kept picking my laptop instead of my fast servers! It seems excessively cargo-flavored for such a straightforward task. Benchmark: {"hello.outPath":{"kbytes":"244736","system":"0.36","user":"2.76"}} Benchmark: {"firefox.outPath":{"kbytes":"1506736","system":"2.38","user":"32.01"}} Benchmark: {"pkgsCross.aarch64-multiplatform.firefox.outPath":{"kbytes":"11334548","system":"10.70","user":"107.07"}} Change-Id: I85bc046ec551360284d7ecfc81a03914f0085909 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10216 Autosubmit: Adam Joseph <adam@westernsemico.com> Reviewed-by: grfn <grfn@gws.fyi> Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
2023-12-09 05:15:31 +01:00
benchmarks = {
benchmark-hello = (mkNixpkgsBenchmark "hello.outPath");
benchmark-cross-hello = (mkNixpkgsBenchmark "pkgsCross.aarch64-multiplatform.hello.outPath");
benchmark-firefox = (mkNixpkgsBenchmark "firefox.outPath");
benchmark-cross-firefox = (mkNixpkgsBenchmark "pkgsCross.aarch64-multiplatform.firefox.outPath");
# Example used for benchmarking LightSpan::Delayed in commit bf286a54bc2ac5eeb78c3d5c5ae66e9af24d74d4
benchmark-nixpkgs-attrnames = (mkExprBenchmark { expr = "builtins.length (builtins.attrNames (import ${pkgs.path} {}))"; description = "nixpkgs-attrnames"; });
feat(tvix/cli): add macrobenchmark This commit adds a simple MVP benchmark, built on our nix infrastructure instead of cargo. It simply runs `tvix-eval` inside of GNU time, and prints the three essential statistics in a short JSON blob. You can run the benchmark with a simple `nix run`, like: nix run -f . tvix.cli.benchmark-hello nix run -f . tvix.cli.benchmark-firefox nix run -f . tvix.cli.benchmark-cross-firefox Currently these blobs are stored only in the CI logs, which I'm sure get garbage-collected at some point. We should be putting them in the git trailers, but that can wait for a future CL. I tried using `cargo bench` for this but found it incredibly frustrating. Maybe I'm doing it wrong. It seems to be designed for microbenchmarks only, and very hard to control. It kept building all sorts of unnecessary stuff (like the tests), and unlike crate2nix it was doing all the builds on only a single machine instead of using more than one machine. Worse, for that single machine it kept picking my laptop instead of my fast servers! It seems excessively cargo-flavored for such a straightforward task. Benchmark: {"hello.outPath":{"kbytes":"244736","system":"0.36","user":"2.76"}} Benchmark: {"firefox.outPath":{"kbytes":"1506736","system":"2.38","user":"32.01"}} Benchmark: {"pkgsCross.aarch64-multiplatform.firefox.outPath":{"kbytes":"11334548","system":"10.70","user":"107.07"}} Change-Id: I85bc046ec551360284d7ecfc81a03914f0085909 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10216 Autosubmit: Adam Joseph <adam@westernsemico.com> Reviewed-by: grfn <grfn@gws.fyi> Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
2023-12-09 05:15:31 +01:00
};
evalTests = {
eval-nixpkgs-stdenv-drvpath = (mkNixpkgsEvalTest "stdenv.drvPath" pkgs.stdenv.drvPath);
eval-nixpkgs-stdenv-outpath = (mkNixpkgsEvalTest "stdenv.outPath" pkgs.stdenv.outPath);
eval-nixpkgs-hello-outpath = (mkNixpkgsEvalTest "hello.outPath" pkgs.hello.outPath);
eval-nixpkgs-firefox-outpath = (mkNixpkgsEvalTest "firefox.outPath" pkgs.firefox.outPath);
eval-nixpkgs-firefox-drvpath = (mkNixpkgsEvalTest "firefox.drvPath" pkgs.firefox.drvPath);
eval-nixpkgs-cross-stdenv-outpath = (mkNixpkgsEvalTest "pkgsCross.aarch64-multiplatform.stdenv.outPath" pkgs.pkgsCross.aarch64-multiplatform.stdenv.outPath);
eval-nixpkgs-cross-hello-outpath = (mkNixpkgsEvalTest "pkgsCross.aarch64-multiplatform.hello.outPath" pkgs.pkgsCross.aarch64-multiplatform.hello.outPath);
};
feat(tvix/cli): add macrobenchmark This commit adds a simple MVP benchmark, built on our nix infrastructure instead of cargo. It simply runs `tvix-eval` inside of GNU time, and prints the three essential statistics in a short JSON blob. You can run the benchmark with a simple `nix run`, like: nix run -f . tvix.cli.benchmark-hello nix run -f . tvix.cli.benchmark-firefox nix run -f . tvix.cli.benchmark-cross-firefox Currently these blobs are stored only in the CI logs, which I'm sure get garbage-collected at some point. We should be putting them in the git trailers, but that can wait for a future CL. I tried using `cargo bench` for this but found it incredibly frustrating. Maybe I'm doing it wrong. It seems to be designed for microbenchmarks only, and very hard to control. It kept building all sorts of unnecessary stuff (like the tests), and unlike crate2nix it was doing all the builds on only a single machine instead of using more than one machine. Worse, for that single machine it kept picking my laptop instead of my fast servers! It seems excessively cargo-flavored for such a straightforward task. Benchmark: {"hello.outPath":{"kbytes":"244736","system":"0.36","user":"2.76"}} Benchmark: {"firefox.outPath":{"kbytes":"1506736","system":"2.38","user":"32.01"}} Benchmark: {"pkgsCross.aarch64-multiplatform.firefox.outPath":{"kbytes":"11334548","system":"10.70","user":"107.07"}} Change-Id: I85bc046ec551360284d7ecfc81a03914f0085909 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10216 Autosubmit: Adam Joseph <adam@westernsemico.com> Reviewed-by: grfn <grfn@gws.fyi> Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
2023-12-09 05:15:31 +01:00
in
{
meta = {
ci.targets = (builtins.attrNames benchmarks) ++ (builtins.attrNames evalTests);
};
feat(tvix/cli): add macrobenchmark This commit adds a simple MVP benchmark, built on our nix infrastructure instead of cargo. It simply runs `tvix-eval` inside of GNU time, and prints the three essential statistics in a short JSON blob. You can run the benchmark with a simple `nix run`, like: nix run -f . tvix.cli.benchmark-hello nix run -f . tvix.cli.benchmark-firefox nix run -f . tvix.cli.benchmark-cross-firefox Currently these blobs are stored only in the CI logs, which I'm sure get garbage-collected at some point. We should be putting them in the git trailers, but that can wait for a future CL. I tried using `cargo bench` for this but found it incredibly frustrating. Maybe I'm doing it wrong. It seems to be designed for microbenchmarks only, and very hard to control. It kept building all sorts of unnecessary stuff (like the tests), and unlike crate2nix it was doing all the builds on only a single machine instead of using more than one machine. Worse, for that single machine it kept picking my laptop instead of my fast servers! It seems excessively cargo-flavored for such a straightforward task. Benchmark: {"hello.outPath":{"kbytes":"244736","system":"0.36","user":"2.76"}} Benchmark: {"firefox.outPath":{"kbytes":"1506736","system":"2.38","user":"32.01"}} Benchmark: {"pkgsCross.aarch64-multiplatform.firefox.outPath":{"kbytes":"11334548","system":"10.70","user":"107.07"}} Change-Id: I85bc046ec551360284d7ecfc81a03914f0085909 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10216 Autosubmit: Adam Joseph <adam@westernsemico.com> Reviewed-by: grfn <grfn@gws.fyi> Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
2023-12-09 05:15:31 +01:00
# Expose benchmarks and evalTests as standard CI targets.
passthru = benchmarks // evalTests;
})