harware vs software advances

[Antonio Torrecillas]

I have a running Pentium MMX 100 Mhz with 32Mb RAM running a _be_patient_98.

Let me know if you find the hardware adecuate for this testing.
I promise not to touch the turbo button

[Janzert]

Here's a timeline directly from Intel for the release dates of their processors

http://www.intel.com/pressroom/kits/quickrefyr.htm

btw, I have a 266mhz P2 that should be coming out of active use in the next few months I could probably dedicate to running games if wanted. Also I think I may have a pentium 90mhz sitting around that just might still boot up.

Janzert

[Don]

I'm pretty sure the Pentium 90 was the fastest processor at that time. Intel introduced the Pentium Pro family (to which the P200 belongs) in late 1995.

Yes, I'll assume that circa 1994 CG3 was running on the P90.

Any idea of the speed? I don't suppose anybody has a Pentium 90 laying around do they?

Don

This could be fun... I got REBEL 10 from this page and I'm now trying to get it to run in VMWare in order to compare with these benchmarks.

When did Rebel 10 come out?

I'm running CG3 in dosemu which is said to be native speed, but I'm going to check that out for sure.

[bob]

Here's a timeline directly from Intel for the release dates of their processors

http://www.intel.com/pressroom/kits/quickrefyr.htm

btw, I have a 266mhz P2 that should be coming out of active use in the next few months I could probably dedicate to running games if wanted. Also I think I may have a pentium 90mhz sitting around that just might still boot up.

Janzert

Those are beyond screwed up. They have different release dates for the same processor. Search for "Pentium Pro". was it 1995 or 1997? Couldn't be both. I know we got our first one in 1996, here. Fairly early in the year but I am not sure exactly when, now. I used one in the 1996 WMCCC event for sure.

[Don]

I have a running Pentium MMX 100 Mhz with 32Mb RAM running a _be_patient_98.

Let me know if you find the hardware adecuate for this testing.
I promise not to touch the turbo button

I just want a CPU reference time. Can you run chess genius 3 on it at some fixed depth level that will take a few minutes? Then I can run the same test on my i7 and we can get an adjustment.

The only way it makes sense to run this test is on modern hardware giving odds in favor of Genius. It would be silly to run for instance 100 games at equal time and try to compute a rating from that because a single draw out of 100 games would make a huge difference in the rating when there is such disparity.

The most ideal way to run this test is with source code from an ancient program from around that time - recompiling it and making it work on either winboard or UCI. I have a tester that allows any kind of time odds games. But I am not aware of such a program except for Crafty.

[bob]

I have a running Pentium MMX 100 Mhz with 32Mb RAM running a _be_patient_98.

Let me know if you find the hardware adecuate for this testing.
I promise not to touch the turbo button

I just want a CPU reference time. Can you run chess genius 3 on it at some fixed depth level that will take a few minutes? Then I can run the same test on my i7 and we can get an adjustment.

The only way it makes sense to run this test is on modern hardware giving odds in favor of Genius. It would be silly to run for instance 100 games at equal time and try to compute a rating from that because a single draw out of 100 games would make a huge difference in the rating when there is such disparity.


Note that your test is not a good benchmark, unless you are going to use genius for all your testing. Different programs scale differently, depending on how they were optimized for a particular piece of hardware.

BTW if the disparity is large, the draw is the least of your troubles. Elo is not good for measuring the difference between two opponents that are hundreds of Elo apart. You need a common set of players with very weak ones to calibrate the low-rated program and very strong ones to calibrate the stronger engine. That entails its own set of interesting problems since the weak and strong opponents need calibration and they can't play inter-pool since they are too far apart. So you need some "connective tissue" in the middle. I ran into this on my "skill" testing stuff and said "hell with it..."

The most ideal way to run this test is with source code from an ancient program from around that time - recompiling it and making it work on either winboard or UCI. I have a tester that allows any kind of time odds games. But I am not aware of such a program except for Crafty.

[Don]

According to cpu-world the Pentium 200 only came out in June of 1996:

http://www.cpu-world.com/CPUs/Pentium/I ... 02200.html

That was just a normal pentium, using the new manufacturing facility they had built. The P5/200 was way cheaper to build than the P6 architecture, and they were still addressing the el-cheapo end of the market as well as the server market where the P6/200 was king.

I read on the web somewhere that someone can do 1 million superpi digits in about 12 seconds. The same person said that the same calculation on a P90 Satellite pro 410cs takes 32 minutes. That is a speedup of only about 160.

Usually on a laptop things are 20 or 30% slower for the laptop version of the same chip. For example my laptop core 2 duo is about 30% slower than my desktop core 2 duo (but the laptop is clocked slower and this is not the case for the p90 notebook. But there are still usually some bus issues and such than makes laptops a little bit slower.

Also, superpi is not chess so chess could be faster or slower. But it is some indication that the P90 is not 1000x slower. That's why I want to find out for sure with a real chess benchmark.

[rbarreira]

Here's a timeline directly from Intel for the release dates of their processors

http://www.intel.com/pressroom/kits/quickrefyr.htm

btw, I have a 266mhz P2 that should be coming out of active use in the next few months I could probably dedicate to running games if wanted. Also I think I may have a pentium 90mhz sitting around that just might still boot up.

Janzert

Those are beyond screwed up. They have different release dates for the same processor. Search for "Pentium Pro". was it 1995 or 1997? Couldn't be both. I know we got our first one in 1996, here. Fairly early in the year but I am not sure exactly when, now. I used one in the 1996 WMCCC event for sure.

Probably a die shrink, the second one.

[Don]

According to cpu-world the Pentium 200 only came out in June of 1996:

http://www.cpu-world.com/CPUs/Pentium/I ... 02200.html

That was just a normal pentium, using the new manufacturing facility they had built. The P5/200 was way cheaper to build than the P6 architecture, and they were still addressing the el-cheapo end of the market as well as the server market where the P6/200 was king.

The 12 seconds was on an i7-620m, forgot to mention that.

I read on the web somewhere that someone can do 1 million superpi digits in about 12 seconds. The same person said that the same calculation on a P90 Satellite pro 410cs takes 32 minutes. That is a speedup of only about 160.

Usually on a laptop things are 20 or 30% slower for the laptop version of the same chip. For example my laptop core 2 duo is about 30% slower than my desktop core 2 duo (but the laptop is clocked slower and this is not the case for the p90 notebook. But there are still usually some bus issues and such than makes laptops a little bit slower.

Also, superpi is not chess so chess could be faster or slower. But it is some indication that the P90 is not 1000x slower. That's why I want to find out for sure with a real chess benchmark.

[Don]

I have a running Pentium MMX 100 Mhz with 32Mb RAM running a _be_patient_98.

Let me know if you find the hardware adecuate for this testing.
I promise not to touch the turbo button

Can you run Chess Genius 3 on this machine? We could get a reference time if so ...

I think there is a legal version of CG3 on the web if you go here:

http://www.gambitchess.com/progr.htm

I suggest setting the depth to something that takes at least a few minutes to search and play the moves 1. h3 a76 and let it do a deep search.

Don