Author: Matt Taylor
Date: 21:18:42 02/08/03
Go up one level in this thread
On February 07, 2003 at 23:29:42, Robert Hyatt wrote: >On February 06, 2003 at 15:47:02, Tom Kerrigan wrote: > >>On February 06, 2003 at 00:29:46, Russell Reagan wrote: >> >>>A few questions about 64-bit machines... >>> >>>1. For chess programming (mainly for bitboards), do either AMD or Intel's 64-bit >>>chips have any features that would make it more desirable over the other? Number >>>of registers, cache size, special instructions (bsf, popcnt, etc.), or whatever >>>else. >> >>IA-64 pros: >>* Lots of registers >>* Lots of cache >> >>x86-64 pros: >>* High clock speed >>* Out of order execution >>* Twice as many registers as x86 >>* Runs x86 software fast >>* Will be available in cheap PCs (imagine, a 64-bit PC chip for $50...) >> >>x86-64 wins hands down IMO. >> >>>2. How much will one be able to take advantage of the hardware using a C/C++ >>>compiler and no assembly programming? The reason I ask this question is because >> >>Programs written in C/C++ will get a performance gain from just recompiling for >>x86-64 because they'll be able to use the extra registers and all the bitboard >>operations will become 64-bit operations. If the programs have assembly, the >>assembly will have to be updated for the program to run in 64-bit mode. >> >>Assembly on IA-64 is a moot point because it's nearly impossible to write >>assembly for the chip. >> >>-Tom > >Why would you think that? > >Go back 30 years to the HP 2100 micro-programmable machine. The >micro-instructions were similar in concept with the IA64 in a gross way, >yet we had students programming that machine with no problems. It is different. >It requires a bit of "new thinking". But multiple instruction parcels in a >single long word is not _that_ bad... > >I think that trying to keep up with all the optimization details for even >IA32 is a very complex undertaking... Haha, that's an understatement. Optimization for IA-32 is an insanely complex problem. I like optimizing for the original Pentium and Athlon because they are relatively simple. The Pentium 2/3 are more complex, and Pentium 4 is annoyingly complex to optimize for. There are optimizations that help everyone, and then optimizations that help specific chips. Then there are size optimizations... The only thing that bothers me about Itanium assembly is the syntax. Masm/Intel syntax for IA-32 is ugly. IA-64 syntax is worse. Perhaps it's changed... -Matt
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.