Author: Aaron Gordon
Date: 21:43:52 06/26/03
Go up one level in this thread
On June 27, 2003 at 00:25:31, Peter Stayne wrote: >AMD's dominance of rc5 and ogr is easily explained, there is a variant of the >ROT assembly command available on the amd's that are not on the Intel chips, >Intels have to do the same thing in software. I've been far less ferverent with >d.net stuff lately, so my ranking may have slipped a bit :) I'm >peter_milkman@hotmail.com on there. > >I bought this sytem based off of benchmarks of quite a few different >applications. Firstly, germaine to this forum, I had heard that AMD's were >faster in chess, but not much. And now, I feel pretty secure that my system is >at least on par with the Athlons. > >Secondly, my website (which is only in an ugly, functional, non marketed state >at http://www.abstractvoice.com ) is now accepting music. I've written a VB app >to automatically download posted MP3's and encode low bitrate versions. I >checked various bench sites and the Intel chips beat the Athlons in the industry >standard LAME encoder, which is what I use. In my tests my Athlon XP 2.5GHz beats a P4-3.38GHz, it's close but it does. >Thirdly, games. The games I play are based off the Quake 3 engine, which favor >the intel chips. This is because Quake3 doesn't use any SIMD instructions on the Athlon. The 3DNow! code is non-existant (or broken) and since it detects an AMD chip, it doesn't use SSE. I however compiled some AMD optimized dlls for Quake3, and with those dlls it allows an Athlon XP 2600+ to beat a P4-3.32GHz, and the XP 2.5GHz (even 2GHz/200fsb/400DDR) blows it away... None of the review pages care (I sent it to them, they won't test), they'd rather use the biased dlls that don't support the Athlon worth a crap. Sisoft sandra does the same thing. They test the SIMD SSE2 instructions vs the raw Athlon FPU.. and people use that to compare FPU power. This is ridiculous, as if 3DNow! or SSE were used it would get *MUCH* higher scores. Infact I have two such programs that test 3dnow/sse on my Athlon, and it's over 2x faster than what Sisoft (raw fpu) gets. If they really want to be retarded, they should just test the Athlon with SDRAM set to CAS 5, underclocked to 50MHz, 100fsb, disable the L1 and L2 cache, then compare it to a P4 overclocked fully optimized for SSE2... They're doing the same thing, only not as extreme. >Fourthly, my 3d Studio Max runs much faster on Xeons than MP's. I'll try to test this when I get the chance. In all of my rendering tests the Athlon blows the P4 away, in both Povray and YASRT. Also there is a program called "Realstorm", which is software realtime raytracing/rendering. The Athlon blows the P4 away there too (www.realstorm.com). From seeing some P4-3GHz C (200/800fsb) results, and comparing it to my results, the P4 would have to run over 4GHz and 263/1050fsb to equal my score. I submitted my results and you will be able to compare them (or I can send them to you directly) when realstorm.com updates their database. >Don't get me wrong, I'm an AMD fan, my last 3 boxes were all oc'd Athlons, and >my next box will be as well (Opterons). But for what I need one for right now, >this was my best choice. > >Pete > > > >On June 26, 2003 at 23:47:46, Aaron Gordon wrote: > >>On June 26, 2003 at 23:21:34, Peter Stayne wrote: >> >>>Ya like I said, hard to compare since our evals are very different. look at the >>>node counts at each ply for one thing. >>> >>>the 11 second score I consider superfluous, since we don't have millisecond >>>resolution. If Fritz rounds up, then that could be 11.0 or 11.999 seconds. or >>>normal rounding could be 10.5 to 11.49999, which would give us sizable >>>differences in kn/s. as we get into the >50 seconds, that difference becomes >>>much less. >> >>Indeed, but you also posted a result < 50 seconds, so I posted my closest >>result. >> >>>Really, in either of our tests, there is no comparable point for a good >>>comparison, since it's obvious both of ours are examining different lines to >>>different depths of varying ease of evaluation on the cpu. Both of ours waver >>>between the same numbers. >>> >>>You win price/performance by a long shot! (that is, if we're purely talking >>>about chess :) ). >> >>The Dual Athlon 2.4 isn't "only" fast at chess.. I've done video encoding, mp3 >>encoding, chess, kernel compiles, all retardedly fast.. :) Any benchmark you'd >>like to do, let me know.. I'll run it for ya. I think you'd be surprised. :) >>Especially in RC5, where I believe this 2x2.4 is roughly 3x faster than a dual >>Xeon 3.06. If you're interested in trying the RC5 (72bit decryption) benchmark, >>I'll put up a copy on my FTP. >> >>>Then again, we'll both be owned by the Opterons if early results are to be >>>believed.
This page took 0.01 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.