commit 32e5d44be2906e7bb73fe3bdcb64c115d187c3c2
Author: Mattias Andrée <[email protected]>
AuthorDate: Fri May 6 01:17:27 2016 +0200
Commit: Mattias Andrée <[email protected]>
CommitDate: Fri May 6 01:17:27 2016 +0200
Update STATUS
Signed-off-by: Mattias Andrée <[email protected]>
diff --git a/STATUS b/STATUS
index d2c3bf2..fc7a519 100644
--- a/STATUS
+++ b/STATUS
@@ -6,7 +6,7 @@ following combinations of cc and libc:
gcc + musl
clang + glibc
-All benchmarks are done on a x86-64 (specifically an Intel
+All benchmarks are done on an x86-64 (specifically an Intel
Core 2 Quad CPU Q9300), without any extensions turned on
during compilation, and without any use of extensions in
assembly code. The benchmarks are performed with Linux as
@@ -14,8 +14,11 @@ the OS's kernel with 50 µs timer slack, and the benchmarking
processes are fixed to one CPU.
- The following functions are probably implemented optimally:
+ The following functions are probably implemented optimally:
+zseti(a, +) ............. tomsfastmath is faster
+zseti(a, -) ............. tomsfastmath is faster
+zsetu ................... tomsfastmath is faster
zswap ................... always fastest
zzero ................... always fastest (shared with gmp)
zsignum ................. always fastest (shared with gmp)
@@ -31,6 +34,10 @@ zbtest .................. always fastest
zadd_unsigned ........... fastest after ~70 compared against zadd too (x86-64)
ztrunc(a, a, b) ......... fastest until ~100, then 77 % (gcc) or 68 % (clang)
of tomsfastmath
+zbset(a, a, 1) .......... always fastest (93 % of gmp (clang))
+zbset(a, a, 0) .......... always fastest
+zbset(a, a, -1) ......... always fastest
+zlsb .................... always fastest <<suspicious>>
The following functions are probably implemented optimally, but
@@ -39,6 +46,10 @@ ztrunc(a, a, b) ......... fastest until ~100, then 77 %
(gcc) or 68 % (clang) of
zneg(a, b) .............. always fastest
zabs(a, b) .............. always fastest
ztrunc(a, b, c) ......... always fastest (alternating with gmp between
1400~3000 (clang+glibc))
+zbset(a, b, 1) .......... always fastest
+zbset(a, b, 0) .......... always fastest
+zbset(a, b, -1) ......... always fastest
+zsplit .................. alternating with gmp for fastest
The following functions require structural changes for
@@ -53,6 +64,15 @@ zxor .................... fastest until ~700, alternating
with gmp (gcc+glibc)
znot .................... always fastest
+ The following functions are probably implemented optimally
+ or close to optimally, except it contains some code that
+ should not be necessary after some bugs have been fixed:
+
+zbits ................... always fastest
+zcmpi(a, +) ............. always fastest
+zcmpi(a, -) ............. always fastest
+zcmpu ................... always fastest
+
@@ -64,27 +84,13 @@ left column. Double-parenthesis means there may be a better
way
to do it. Inside square-brackets, there are some comments on
multi-bit comparisons.
-zseti ................... tomsfastmath is faster [always]
-zsetu ................... tomsfastmath is faster [always]
zsub_unsigned ........... fastest [always] (compared against zsub too)
zadd .................... fastest [after ~110, tomsfastmath before] (x86-64)
zsub .................... fastest [always]
-zbits ................... fastest [always]
-zlsb .................... fastest [always]
zlsh .................... fastest [until ~1000, then gmp]
zrsh .................... fastest [almost never]
-zsplit .................. fastest [alternating with gmp and slightly slow than
gmp]
-zcmpmag ................. fastest [always]
+zcmpmag ................. fastest [always] (suspicious)
zcmp .................... fastest [almost never]
-zcmpi(a, +) ............. fastest [always]
-zcmpi(a, -) ............. fastest [always]
-zcmpu ................... fastest [always]
-zbset(a, b, 1) .......... fastest [always]
-zbset(a, a, 1) .......... fastest [always]
-zbset(a, b, 0) .......... fastest [always]
-zbset(a, a, 0) .......... fastest [always]
-zbset(a, b, -1) ......... fastest [always]
-zbset(a, a, -1) ......... fastest [always]
zgcd .................... 21 % of gmp (zcmpmag)
zmul .................... slowest
zsqr .................... slowest (zmul)
