Replies to post #28913 on Advanced Micro Devices Inc (AMD)

[HailMary]

if not most programs the advantage of more registers is completely counterbalanced by the fact that there is code bloat

BTW - that past presentation by Kevin McGrath (sp?) from AMD showed the average instruction size only increased around 10-15% when going from 32-bit to 64-bit. I don't remember the exact number. Anyone?

therefore QSORT would get no benefit from the extra registers

Judging by the c-code I used, I think there may in fact be some benefit to using extra registers. How else do we account for the fact that the optimized compile was significantly faster? I think wbmw may have suggested the 32-bit compiler is not as good as the 64-bit compiler at optimizations. This could be a factor. It is too bad we don't have a compiler that is identical other than enabling 64-bit and extra registers.

Anyone working on a thesis? This might make a good paper. I just don't have the time to pursue this much.

All I know is I can get significant gains in performance by using an Opteron running 64-bit Linux, even if I don't recompile the 32-bit app. That says something loud and clear. It says my company is going to buy a lot more Opterons.

[Dan3]

Re: Alas, I guess it would be considered unfair to compare quicksort on 64-bit integers in legacy and in long mode?

Those guys are all in shock, and we can forgive a little babbling on their part.

Sun, whose market share was a nest full of unhatched eggs they'd already counted, is now shipping a killer line of Opteron systems while Itanium sales wilt.

IBM, after their success with a small Opteron server line has just moved Opteron to a full line of workstations and is expanding their server line.

HP, the company that many think did more of Itanium's design than Intel, has announced Opteron workstations and servers.

Intel has basically announced they're frantically working on a clone of AMD's chip, and will ship it if and when they can figure out how to make it.

Cray now has multiple lines of high end Opteron systems, Red Storm and Octiga Bay.

One of Intel's former advantages was that they could supply engineered solutions to white box makers. There are now multiple, fully engineered suppliers of Opteron systems to the white box market, Sanmina/Newisys and Celestica.

Microsoft may be late with a full production version of Windows for AMD64, but they've managed to put NX into the next service pack, giving AMD an easily marketed, obvious advantage over Xeon, Pentium, and Centrino - particularly for the lucrative corporate and government markets.

Intel's process technology, which formerly let them ship chips that used less power, making AMD systems the ones pushing the limits of power supplies, heat sinks, and noise, has flat out collapsed in the move to 90nm, while AMD's "risky" SOI process has been having no such troubles.

Least cost, industry standard system platforms will boot but be unreliable using Pentium (due to just barely adequate inexpensive power supplies) but will be rock solid if the more efficient Athlon 64 is used.

The poor guys are reeling, they don't know what hit them!

[yourbankruptcy]

The most important math benchmark for business use would be a typical hash value computation, as part of hashmaps implementation.

I should say some of applications perform 5-6 HashTable lookups for every market data quota they get, i.e. 400*5 times per second.

Which platform will do me that faster?