summaryrefslogtreecommitdiff
path: root/doc
diff options
context:
space:
mode:
authortege <tege@gmplib.org>2000-12-02 22:58:41 +0100
committertege <tege@gmplib.org>2000-12-02 22:58:41 +0100
commit6d397f91353b39cec43d198f94136bab68e11e1e (patch)
tree32c155bc7834ac8df29b0ad482ed11233fcf09ff /doc
parentabb7fd4699fcf1409be81efc21dc2ba466d6fd48 (diff)
downloadgmp-6d397f91353b39cec43d198f94136bab68e11e1e.tar.gz
Add some Cray tasks.
Diffstat (limited to 'doc')
-rw-r--r--doc/tasks.html5
1 files changed, 5 insertions, 0 deletions
diff --git a/doc/tasks.html b/doc/tasks.html
index 7601ccd94..46a33e687 100644
--- a/doc/tasks.html
+++ b/doc/tasks.html
@@ -226,6 +226,11 @@
<li> Cray: Write <code>mpn_mul_basecase</code> and
<code>mpn_sqr_basecase</code>. Same comment applies to this as to the
same functions for Fujitsu VPP.
+<li> Cray T90: In IEEE environment, use _int_mult_upper(x,y).
+ Documented online in "Migration to CRAY T90 IEEE Floating Point".
+<li> Cray with CFP (all vector systems except T90 IEEE): i24mult no longer
+ works. It is "outmoded". What else can we use for reasonable speed?
+ What makes current assembly mulww.s fail?
<li> Improve <code>count_leading_zeros</code> for 64-bit machines:
<pre>
if ((x &gt&gt 32) == 0) { x &lt&lt= 32; cnt += 32; }