Author: Vincent Diepeveen
Date: 14:21:53 05/27/02
Go up one level in this thread
On May 27, 2002 at 15:43:31, Aaron Gordon wrote: Whatever gets said, one thing is for sure. AMDs lack of a compiler is going to have a major impact onto performance of their machines. If even Slate who is not exactly worlds dumbest tester, makes major mistakes with sandar, then that is because sandra doesn't run with a K7 optimized version for the k7. And why doesn't it run with a k7 optimized version? Right: there is no easy way to use the K7 stuff on windows! In short the mistakes are made quickly. With regard to the non-specint things (the graphics tests) it is clear that CCC is not the place to get extensive discussion. All it shows how hard objective testing is. Note that there is a fairly easy way to find out what is the fastest cpu around for the different programs. Just watch the hardware they use at the world championships :) If they take themselves a K7 from their home you know what is faster for'em :) >Of course. Without a proper setup you can make the nicest system run like >absolute crap. The via 4in1 drivers for example contain and AGP driver required >for proper operation. Tons of people were complaining, "Why is my AMD box only >getting 1,300 in 3DMark 2000 and blah blah blah"... or something similar (low >results). After they installed the latest 4in1's their score jumped up to >~5,000-8,000. Just goes to show what a little bit of knowledge can go. I've seen >SO many pages that have done this. Not even bothering installing the 4in1's off >the CD even. > >I'm sure you have done a lot of testing and I know you're only trying to do us a >favor and we do appreciate that. Unfortunately when this testing isn't done >properly it is a complete waste. The CPU, Motherboard, RAM, Videocard, etc all >play a huge part in the testing obviously. You can have a 1.73GHz AthlonXP, the >best ram, the best video card but a horrible board in terms of performance and >actually end up having a much slower system run circles around it. Chipsets make >a huge difference. Some aren't nearly as effecient as others (KT266 @ 50%, >KT266a @ 95-97%. KT133a is also 95-97%, KT133 is only %50 or so). > >Even if you buy a $1000 board like my friend did way back in the day does not >mean you're getting a 'fast' board or one with more options. A friend of mine >(not too knowledgable about computers at the time) decided to get a $1000 tyan >board and two 400MHz Xeon CPU's for $1200 each. He figured with it being so >expensive how can it NOT beat my box? Well, was it fast? Yes. Could it even come >close to my $200 (board+cpu) Celeron 300 @ 644Mhz on an Abit BH6? Nope. It was >even faster at only 450-504mhz. That was the last Intel box he got actually.. I >kept up with the Celerons for a while. They aren't exactly fast but are fun to >overclock. Now, for the non-overclockers. Getting such boards ($80-100 >overclockers boards, Epox 8KHA+, 8K3A, Abit KT7a, KR7, etc) will provide the >best performance even when not overclocking due to all the tweak options & >whatnot. > >Also Slate, I HAVE done testing against a P4's. I haven't tested against any of >the 2.2GHz+ P4's or the northwoods directly but I did pit my box and some of my >older box's against a P4-2GHz. I made sure both computers were running idental >OS's, benchmark settings, drivers, etc. I was totally fair in all tests >optimizing both computers as much as possible. Yes, my two main AMD machines did >kick the hell out of it quite easily. When I get the money up I'll try to "rent" >a P4-2.53 and two of the fastest boards (one for DDR, one RDRAM) and test it >again. As you have seen in some of my previous tests I did run some tests >against some northwoods (crafty benchmark for example) where even the old >overclocked Thunderbird had a good edge over a 2465MHz Northwood P4 with a bus >of almost 550MHz. Also you forget to mention that the "mflops" in sisoft Sandra >for the P4 are with the P4 using SSE2. The AthlonXP was NOT using 3DNow/SSE. Try >comparing the results of the straight P4 fpu to the straight AthlonXP FPU. >Doesn't seem like you're even trying to play fair... > > > >On May 27, 2002 at 09:46:12, Slater Wold wrote: > >>Come on man. This was MY comparison, and for sure, the last comparison I ever >>do for CCC. >> >>1.) I wasn't comparing money. Who cares about money. If I wanted to compare >>money, would I have bothered actually BUYING the equipment? Of course not! I >>would have just made a list of what costs how much, and that would have been it. >> MY test was to see which CPU was faster. >> >>2.) I did everything I could to make sure this test was as "equal" as possible. >>I used the same exact programs, I used the same exact drivers, hell, I even used >>the same video card in some tests. But that's not good enough. I didn't use >>the correct EXE, or I didn't use the right BIOS settings. >> >>I am sitting here, in front of an AMD 1.73Ghz, beside me, an Intel P4 2.53Ghz. >>I have run application after application, and I am here to tell you, that P4 is >>faster. PERIOD. >> >>Send me the EXEs, settings, driver, whatever you'd like. But I am telling you, >>I've been up for 23 hours now, doing this shit. I've seen it all with my own 2 >>eyes. When you have a P4 sitting next to you, then come talk to me. >> >> >>Good nite.
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.