Author: Sune Fischer
Date: 01:53:32 03/25/02
Go up one level in this thread
On March 25, 2002 at 01:33:46, Tom Kerrigan wrote: >Yes, rah rah, go Alpha. But how are we supposed to use these NPS numbers to >guess how much faster the Hammer will be than the AXP? (Which, BTW, is the issue >at hand.) Actually the issue was how much could be gained in going from 32 to 64 bit. But talking about the Hammer, I read this on Anandtech and was surprised to read about having to turn on 64-bit data oprands with a REX instruction? If we have to go assembler everytime we want to do a 64-bit operation, the Hammer is almost useless, at least to me. "AMD says compiling for x86-64 increases the code size by around 10%, so the miss-rate increase in caches due to larger code size is minimal (perhaps 5%). On the other hand, 64-bit data types would cause the same ~40% increase to data cache miss-rate and an increase in data bandwidth requirements. As such, AMD actually sets the default data operand size to 32-bits in the 64-bit addressing mode. The motivation is that 64-bit data operands are not likely to be needed and could hurt performance; in those situations where 64-bit data operands are desired, they can be activated using the new REX prefix (woohoo, yet another x86 instruction prefix :))." -S. >-Tom
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.