Author: Ron Langeveld
Date: 09:27:43 05/27/02
Go up one level in this thread
Hi Slater, I couldn't believe the conclusion, so I repeated the tests which are easy to repeat; the engine tests. It shows that "something" is different or your machine. Like for instance using a Hiarcs 8 beta or non-standard settings. I get different lines, also with Shredder 6.02. I did not repeat Fritz 7 test because I don't know which version you ran, but my version 7.0.0.6 got 186442kN in 3:09 for depth 14/40 which is 986,4 nodes/sec. The Intel got 897,1 nodes/sec and that is about 10% slower. Like you, I used the Nolot 3 position. 32MB hash was used in all cases and the remaining settings were out-of-the-box (default). See below for the Hiarcs and Shredder tests. Regards, Ron On May 27, 2002 at 05:39:40, Slater Wold wrote: >Intel. > >Please read the tests, and then read the bottom. PLEASE do *NOT* respond to >this post without reading the bottom. > > >Sandra (latest version, downloaded from http://www.downloads.com): > >Arthmetic: > >AMD 1.73Ghz: > >4787 MIPS >2392 MFLOPS > >Intel 2.53Ghz: > >4863 MIPS >3093 MFLOPS > > >Multimedia: > >AMD 1.73Ghz > >9520 it/s >11084 it/s > >Intel 2.53Ghz: > >9999 it/s >12207 it/s > > >Crafty 18.15 (downloaded STRAIGHT from Bob's FTP): > >AMD 1.73Ghz: > >902662 nps @ 108 > >Intel 2.53Ghz: > >965223 nps @ 101 > >**The P4 version is slower than the standard version.** > >Quake 3 Arena Demo 1 - with sound - using the same exact video card & drivers: > >AMD 1.73Ghz: > >1260 frames at 9.1s = 139.0 FPS > >Intel 2.53Ghz: > >1260 frames at 4.5s = 281.3 FPS > > > >Various Chess Engines (test with Nolot #3): > >Fritz 7 (NO MMX): > >AMD 1.73Ghz: > >1.Bxg5! > ² (0.50) Depth: 15/43 00:02:57 219235kN > >Intel 2.53Ghz: > >1.Bxg5! > ² (0.50) Depth: 15/43 00:02:41 219235kN > > >Fritz 7: > >AMD 1.73Ghz: > >1.Bxg5 Bxg5 2.h3 Bh4 3.hxg4 Bxg3+ 4.Kd2 Qd7 5.Rxh7 Rxh7 6.Bxh7 Bf4+ 7.Kc2 0-0-0 >8.Bd3 > ² (0.50) Depth: 14/40 00:02:15 121119kN > >Intel 2.53Ghz: > >1.Bxg5 Bxg5 2.h3 Bh4 3.hxg4 Bxg3+ 4.Kd2 Qd7 5.Rxh7 Rxh7 6.Bxh7 Bf4+ 7.Kc2 0-0-0 >8.Bd3 > ² (0.50) Depth: 14/40 00:02:05 121119kN > > >Hiarcs 8: > >AMD 1.73Ghz: > >1.Bxg5 Bxg5 2.0-0 Bf4 3.Qc2 Kd7 > ± (0.80) Depth: 12/34 00:01:23 15206kN I get 1.Bxg5 Bxg5 2.Rf1 Bf4 3.Kf2 Kd7 4.Kg1 Qe7 5.Qc2 Bxg3 6.hxg3 h5 (0.28) Depth 12/36 00:02:42 33282kN that is 33282/162 = 205,4 nodes/sec whereas the Intel got 15206/74 = 205,4 which idicates equal performance > >Intel 2.53Ghz: > >1.Bxg5 Bxg5 2.0-0 Bf4 3.Qc2 Kd7 > ± (0.80) Depth: 12/34 00:01:14 15206kN > > >Shredder Paderborn: > >AMD 1.73Ghz: > >1.Bxg5 Bxg5 2.0-0 Qd7 3.h3 Bxf3 4.Qxf3 0-0-0 5.Qf7 a6 6.Qxd7+ Rxd7 7.Kh2 g6 >8.Ne2 Ne7 9.Rf7 > ² (0.59) Depth: 13/26 00:01:24 27302kN I get 1.h3 Bxf3 2.gxf3 g6 3.Kd2 Kd7 4.Kc2 Rf8 5.f4 Nf7 6.Be3 (0.81) Depth 13/26 00:01:12 26621kN that is 26621/72 = 369,7 nodes/sec whereas the Intel gets 27302/80 = 341,2 my AMD 2100+ is more than 8% faster than the Pentium IV > >Intel 2.53Ghz: > >1.Bxg5 Bxg5 2.0-0 Qd7 3.h3 Bxf3 4.Qxf3 0-0-0 5.Qf7 a6 6.Qxd7+ Rxd7 7.Kh2 g6 >8.Ne2 Ne7 9.Rf7 > ² (0.59) Depth: 13/26 00:01:20 27302kN > > >Conclusion: > >This test wasn't based on price, or value, or anything like that. So please >don't post anything about "value". It's not worth much to me. (As anyone who >knows me will tell you.) This was about one thing, and one thing alone: >matching Intels top CPU to AMDs top CPU. Sorry to say, but Intel won EACH and >EVERY test I ran. > >However, to most people, the 10% gain, just isn't enough to warrant $1k. I >completly agree. I would *never* recommend the P4 over the AMD. Because, come >on, who wants to spend $1k more for 10%!? > >And before you start telling me that you can overclock, take a look at Aarons >website. He has screenshots of a 1.812Mhz AMD, and it's only faster in _1_ >benchmark (the arthmetic benchmark). > >AMD 1.812Ghz vs Intel 2.53Ghz: > >AMD > >4977 MIPS \ >2493 MFLOPS / Arthmetic > >9768 it/s \ >11457 it/s / Multimedia > > >Intel > >4863 MIPS \ >3093 MFLOPS / Arthmetic > >9999 it/s \ >12207 it/s / Multimedia > > >And of course the PC1066 memory is almost 2x the speed of his memory in these >tests. > >Every test I ran, every application I ran, was faster on the P4. From encoding >movies, to encoding MP3's or RA. There is no "official" benchmark for these >programs, but the ending times (for MPEG video) were about 30% faster on the P4 >compared to the AMD 1.73Ghz. MP3's were about 15% faster on the P4. > >So, if you want the fastest CPU on the market, well, that's going to be the P4 >2.53Ghz. Because it is, just that, the fastest CPU on the market.
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.