diff options
| author | tege <tege@gmplib.org> | 2002-02-19 02:55:50 +0100 |
|---|---|---|
| committer | tege <tege@gmplib.org> | 2002-02-19 02:55:50 +0100 |
| commit | 10771eb65077071674fa4fb188179db21c4cf844 (patch) | |
| tree | 58ebf2ae654b47bc9ee4e632c03eed4b1f4a76dd /tune/README | |
| parent | bb6a085e894aa3644bc7f3d6d16302063eaced2f (diff) | |
| download | gmp-10771eb65077071674fa4fb188179db21c4cf844.tar.gz | |
Clean up *_THRESHOLD names.
Diffstat (limited to 'tune/README')
| -rw-r--r-- | tune/README | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/tune/README b/tune/README index 3830910e0..a6cda1240 100644 --- a/tune/README +++ b/tune/README @@ -1,4 +1,4 @@ -Copyright 2000, 2001 Free Software Foundation, Inc. +Copyright 2000, 2001, 2002 Free Software Foundation, Inc. This file is part of the GNU MP Library. @@ -87,7 +87,7 @@ SCO OpenUNIX 8 /etc/hw PARAMETER TUNING The "tuneup" program runs some tests designed to find the best settings for -various thresholds, like KARATSUBA_MUL_THRESHOLD. Its output can be put +various thresholds, like MUL_KARATSUBA_THRESHOLD. Its output can be put into gmp-mparam.h. The program is built and run with make tune @@ -280,7 +280,7 @@ up to twice as fast as mul_basecase. In practice loop overheads and the products on the diagonal mean it falls short of this. Here's an example running the two and showing by what factor an NxN mul_basecase is slower than an NxN sqr_basecase. (Some versions of sqr_basecase only allow sizes -below KARATSUBA_SQR_THRESHOLD, so if it crashes at that point don't worry.) +below SQR_KARATSUBA_THRESHOLD, so if it crashes at that point don't worry.) ./speed -s 1-20 -r mpn_sqr_basecase mpn_mul_basecase @@ -437,12 +437,12 @@ normal libgmp.la. Note further that the various routines may recurse into themselves on sizes far enough above applicable thresholds. For example, mpn_kara_mul_n will recurse into itself on sizes greater than twice the compiled-in -KARATSUBA_MUL_THRESHOLD. +MUL_KARATSUBA_THRESHOLD. When doing the above comparison between mul_basecase and kara_mul_n what's probably of interest is mul_basecase versus a kara_mul_n that does one level of Karatsuba then calls to mul_basecase, but this only happens on sizes less -than twice the compiled KARATSUBA_MUL_THRESHOLD. A larger value for that +than twice the compiled MUL_KARATSUBA_THRESHOLD. A larger value for that setting can be compiled-in to avoid the problem if necessary. The same applies to toom3 and DC, though in a trickier fashion. |