[EMAIL PROTECTED] wrote:
I was therefore not surprised to see the P4 do better
than the Athlon for the LL tests. But was a little
surprised to see the Athlon show considerably faster
times over every bit-depth of factoring, with the
higher the bit-depth the more the Athlon blew the P4
away.
v24 has many enhancements for factoing by 64-bit machines. Therefore the Athlon64
now >outshines the P4 in factoring.
The comparison was using v24.13, so shouldn't the P4, using EM64T (64-bit extensions),
"shine" in factoring too? Just a thought.
I tried to explain that behavior already in one of my postings in the "cpu affinity
..." thread:
"But anyway the AMD64 CPUs would fare better in regards of throughput doing TF in 64
bit mode, where they are between ~45 to ~125% faster than in 32 bit mode thanks to
the new code (see http://mersenneforum.org/showpost.php?p=52715&postcount=21). The
same is not true for EM64T CPUs, very likely because of the way, how the integer
division is (had to be) implemented."
Just a few days ago a document appeared, where someone made a thorough comparison of
the way, how instructions are executed in 32 bit and 64 bit modes on several x86 CPUs
- especially throughput and latency for nearly all important instructions.
Unfortunately the integer division instructions are not covered, but alone from the
principle of how the double pumped ALUs work, one could conclude, that the integer
divisions (which is emulated via code from microcode ROM) suffer from an
architecture, which is optimized more towards throughput (e.g. with higher latencies
and lower throughput even for usually simple instructions like "shift right").
The document is here:
http://swox.com/doc/x86-timing.pdf
An rather informative thread (as usual on RWT) discussing it, can be found here:
http://realworldtech.com/forums/index.cfm?action=detail&PostNum=3608&Thread=1&entryID=54448&roomID=11
Regards,
Matthias
_______________________________________________
Prime mailing list
[email protected]
http://hogranch.com/mailman/listinfo/prime
