about summary refs log tree commit diff
path: root/src/libstore/worker-protocol.hh
diff options
context:
space:
mode:
authorEelco Dolstra <eelco.dolstra@logicblox.com>2013-08-07T11·51+0000
committerEelco Dolstra <eelco.dolstra@logicblox.com>2013-08-07T12·02+0200
commita583a2bc59a4ee2b067e5520f6c5bc0c61852c32 (patch)
tree137be66f58b664dbf0aa21138000d711d5e00162 /src/libstore/worker-protocol.hh
parent263d6682224f516aed74286453c5e2e097a38aa6 (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/worker-protocol.hh')
-rw-r--r--src/libstore/worker-protocol.hh2
1 files changed, 1 insertions, 1 deletions
diff --git a/src/libstore/worker-protocol.hh b/src/libstore/worker-protocol.hh
index 07f825b9273f..9317f89c3763 100644
--- a/src/libstore/worker-protocol.hh
+++ b/src/libstore/worker-protocol.hh
@@ -6,7 +6,7 @@ namespace nix {
 #define WORKER_MAGIC_1 0x6e697863
 #define WORKER_MAGIC_2 0x6478696f
 
-#define PROTOCOL_VERSION 0x10d
+#define PROTOCOL_VERSION 0x10e
 #define GET_PROTOCOL_MAJOR(x) ((x) & 0xff00)
 #define GET_PROTOCOL_MINOR(x) ((x) & 0x00ff)