linux/0162-extra-optmization-flags.patch at main · clearlinux-pkgs/linux · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Colin Ian King <colin.king@intel.com>
Date: Thu, 3 Aug 2023 16:52:11 +0100
Subject: [PATCH] extra optmization flags

Enable 2 extra optimimization flags:
  -fivopts
     Perform induction variable optimizations (strength reduction,
     induction variable merging and induction variable elimination)
     on trees.
  -fmodulo-sched
     Perform swing modulo scheduling immediately before the first
     scheduling pass. This pass looks at innermost loops and reorders
     their instructions by overlapping different iterations.

stress-ng microbenchmark improvements (average from 10 x 1min runs) on
Alderlake with gcc 13.2.0:
  af-alg:  1.0% (kernel AF_ALG crypto)
  pipe:    1.5% (pipe + context switch)
  fork:    4.4% (process fork/exit)
  mmap:    3.0% (memory mapping)
  switch:  4.9%	(context switching)

KVM QEMU bootspeed improvements (based on average of 100 boots): 0.5%

---
 arch/x86/Makefile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/Makefile b/arch/x86/Makefile
index 415a5d138de4..6b91d1306a7c 100644
--- a/arch/x86/Makefile
+++ b/arch/x86/Makefile
@@ -67,7 +67,7 @@ export BITS
 #
 #    https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53383
 #
-KBUILD_CFLAGS += -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -mno-avx -mno-avx2 -O3 -fno-tree-vectorize -march=westmere -mpopcnt
+KBUILD_CFLAGS += -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -mno-avx -mno-avx2 -O3 -fno-tree-vectorize -march=westmere -mpopcnt -fivopts -fmodulo-sched
 KBUILD_RUSTFLAGS += -Ctarget-feature=-sse,-sse2,-sse3,-ssse3,-sse4.1,-sse4.2,-avx,-avx2

 ifeq ($(CONFIG_X86_KERNEL_IBT),y)
--
https://clearlinux.org