On x86, use thread-local storage instead of stealing a reg for gct

Benchmarks show that using TLS instead of stealing a register is
better by a few percent on x86, due to the lack of registers.

This only affects -threaded; without -threaded we're (now) using
static storage for the GC data.

1 files changed, 6 insertions, 1 deletions

diff --git a/rts/sm/GCThread.h b/rts/sm/GCThread.h
index c563d95efd..5646eddb51 100644
--- a/rts/sm/GCThread.h
+++ b/rts/sm/GCThread.h
@@ -209,10 +209,15 @@ extern gc_thread **gc_threads;
 
 #define SET_GCT(to) gct = (to)
 
-#if defined(sparc_HOST_ARCH)
+#if defined(sparc_HOST_ARCH) || defined(i386_HOST_ARCH)
 // Don't use REG_base or R1 for gct on SPARC because they're getting clobbered 
 //	by something else. Not sure what yet. -- BL 2009/01/03
 
+// Using __thread is better than stealing a register on x86, because
+// we have too few registers available.  In my tests it was worth
+// about 5% in GC performance, but of course that might change as gcc
+// improves. -- SDM 2009/04/03
+
 extern __thread gc_thread* gct;
 #define DECLARE_GCT __thread gc_thread* gct;