diff options
author | Eelco Dolstra <eelco.dolstra@logicblox.com> | 2013-08-07 11:51:55 +0000 |
---|---|---|
committer | Eelco Dolstra <eelco.dolstra@logicblox.com> | 2013-08-07 14:02:04 +0200 |
commit | a583a2bc59a4ee2b067e5520f6c5bc0c61852c32 (patch) | |
tree | 137be66f58b664dbf0aa21138000d711d5e00162 /src/libstore/local-store.cc | |
parent | 263d6682224f516aed74286453c5e2e097a38aa6 (diff) |
Run the daemon worker on the same CPU as the client
On a system with multiple CPUs, running Nix operations through the
daemon is significantly slower than "direct" mode:
$ NIX_REMOTE= nix-instantiate '<nixos>' -A system
real 0m0.974s
user 0m0.875s
sys 0m0.088s
$ NIX_REMOTE=daemon nix-instantiate '<nixos>' -A system
real 0m2.118s
user 0m1.463s
sys 0m0.218s
The main reason seems to be that the client and the worker get moved
to a different CPU after every call to the worker. This patch adds a
hack to lock them to the same CPU. With this, the overhead of going
through the daemon is very small:
$ NIX_REMOTE=daemon nix-instantiate '<nixos>' -A system
real 0m1.074s
user 0m0.809s
sys 0m0.098s
Diffstat (limited to 'src/libstore/local-store.cc')
-rw-r--r-- | src/libstore/local-store.cc | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/src/libstore/local-store.cc b/src/libstore/local-store.cc index 29095e1ea..34b4a5158 100644 --- a/src/libstore/local-store.cc +++ b/src/libstore/local-store.cc @@ -5,6 +5,7 @@ #include "pathlocks.hh" #include "worker-protocol.hh" #include "derivations.hh" +#include "affinity.hh" #include <iostream> #include <algorithm> @@ -1021,6 +1022,7 @@ void LocalStore::startSubstituter(const Path & substituter, RunningSubstituter & case 0: /* child */ try { + restoreAffinity(); if (dup2(toPipe.readSide, STDIN_FILENO) == -1) throw SysError("dupping stdin"); if (dup2(fromPipe.writeSide, STDOUT_FILENO) == -1) |