harware vs software advances

Discussion of chess software programming and technical issues.

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
mhull
Posts: 13192
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: harware vs software advances

Post by mhull » Sat Sep 11, 2010 9:37 pm

Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
Matthew Hull

User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 7:30 pm
Location: Chicago, Illinois, USA
Contact:

Re: harware vs software advances

Post by michiguel » Sat Sep 11, 2010 10:10 pm

mhull wrote:
Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
That is why Crafty is a bad example. Bob has been saying that programs in the past were not optimized for current hardware. Well, Crafty _was_ being optimized for hardware of the future, in detriment to the current hardware of the time. He was saying this all the time in the 90's and there were countless threads about it. "64 bits will become standard" etc. etc.

Miguel

User avatar
mhull
Posts: 13192
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: harware vs software advances

Post by mhull » Sat Sep 11, 2010 10:39 pm

michiguel wrote:
mhull wrote:
Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
That is why Crafty is a bad example. Bob has been saying that programs in the past were not optimized for current hardware. Well, Crafty _was_ being optimized for hardware of the future, in detriment to the current hardware of the time. He was saying this all the time in the 90's and there were countless threads about it. "64 bits will become standard" etc. etc.

Miguel
None of that matters when comparing the same program on old and new hardware. All that matters is what was the HW speedup and what was the ELO gain for the 1995 program at that speedup.
Matthew Hull

User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 2:27 pm

Re: harware vs software advances

Post by Don » Sat Sep 11, 2010 10:45 pm

michiguel wrote:
mhull wrote:
Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
That is why Crafty is a bad example. Bob has been saying that programs in the past were not optimized for current hardware. Well, Crafty _was_ being optimized for hardware of the future, in detriment to the current hardware of the time. He was saying this all the time in the 90's and there were countless threads about it. "64 bits will become standard" etc. etc.

Miguel
Thank you Miguel. You said it better than I could and this is really clear.

I was getting frustrated trying to make Bob understand this concept and it was just because I am not very good at explaining myself.

User avatar
mhull
Posts: 13192
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: harware vs software advances

Post by mhull » Sat Sep 11, 2010 11:15 pm

Don wrote:
michiguel wrote:
mhull wrote:
Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
That is why Crafty is a bad example. Bob has been saying that programs in the past were not optimized for current hardware. Well, Crafty _was_ being optimized for hardware of the future, in detriment to the current hardware of the time. He was saying this all the time in the 90's and there were countless threads about it. "64 bits will become standard" etc. etc.

Miguel
Thank you Miguel. You said it better than I could and this is really clear.

I was getting frustrated trying to make Bob understand this concept and it was just because I am not very good at explaining myself.
It's not relevant to the question being asked. All you need to know is how much ELO to attribute to HW speedup. How something was or was not optimized to HW has nothing to do with the answer. Only the speedup numbers matter. Whether a particular project sped up 10x or 1000x is the only answer you need. Then figure how how much ELO that translates to. That gives you the HW ELO increase (for that chess project). Any increase above that in a newer version (of that project) would be a candidate for software ELO improvement. This seems to be the general idea.

I mean, can the HW speedup for 1995 Crafty be determined or not? If it can, where is the problem?
Matthew Hull

User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 2:27 pm

Re: harware vs software advances

Post by Don » Sat Sep 11, 2010 11:21 pm

mhull wrote:
Don wrote:
michiguel wrote:
mhull wrote:
Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
That is why Crafty is a bad example. Bob has been saying that programs in the past were not optimized for current hardware. Well, Crafty _was_ being optimized for hardware of the future, in detriment to the current hardware of the time. He was saying this all the time in the 90's and there were countless threads about it. "64 bits will become standard" etc. etc.

Miguel
Thank you Miguel. You said it better than I could and this is really clear.

I was getting frustrated trying to make Bob understand this concept and it was just because I am not very good at explaining myself.
It's not relevant to the question being asked. All you need to know is how much ELO to attribute to HW speedup. How something was or was not optimized to HW has nothing to do with the answer. Only the speedup numbers matter. Whether a particular project sped up 10x or 1000x is the only answer you need. Then figure how how much ELO that translates to. That gives you the HW ELO increase (for that chess project). Any increase above that in a newer version (of that project) would be a candidate for software ELO improvement. This seems to be the general idea.

I mean, can the HW speedup for 1995 Crafty be determined or not? If it can, where is the problem?
Ok, if we go along with your premise, then why not use the Rebel numbers? In your words, "only the speedup numbers matter." And the numbers are 100 to 1 for Rebel. And I'm actually using an IDENTICAL program, you are not.

User avatar
mhull
Posts: 13192
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: harware vs software advances

Post by mhull » Sat Sep 11, 2010 11:42 pm

Don wrote:Ok, if we go along with your premise, then why not use the Rebel numbers? In your words, "only the speedup numbers matter." And the numbers are 100 to 1 for Rebel. And I'm actually using an IDENTICAL program, you are not.
Ok, Rebel it is. 100 to 1. No problem. What's the ELO increase for Rebel with 100x speedup? I can go with that for Rebel.

For 1995 crafty, the speedup will be more than 100x, for the well-understood reason that it's more capable of it by design. But that's not important. What matters is what is 1995 Crafty's speedup, and what does that particular speedup translate to in ELO terms for 1995 crafty?

It's the same measurement for both programs, differing only by degree of speedup. But the function curve (in theory) should be similar, it's just that the two examples will touch the curve at different points along the speedup-ELO function.
Matthew Hull

bob
Posts: 20918
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: harware vs software advances

Post by bob » Sun Sep 12, 2010 3:21 am

michiguel wrote:
mhull wrote:
Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
That is why Crafty is a bad example. Bob has been saying that programs in the past were not optimized for current hardware. Well, Crafty _was_ being optimized for hardware of the future, in detriment to the current hardware of the time. He was saying this all the time in the 90's and there were countless threads about it. "64 bits will become standard" etc. etc.

Miguel
And your point would be? I did the same for parallel search if you recall. But I _designed_ Crafty to work on 32 bit machines. The original rotated stuff had a ton of 32 bit optimizations, with the idea that one day they would not be needed.

Somehow you seem to imply that planning for the future, having the foresight to predict that 64 bit processors would become standard, and such, is somehow a bad idea? I give raises to people that thing progressively, not fire 'em. That's a good characteristic to have, not a bad one. And for me, it worked. I had a competitive program in 1995. It gained more from 64 bit hardware than those that chose to stick with status quo. That is a problem?

bob
Posts: 20918
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: harware vs software advances

Post by bob » Sun Sep 12, 2010 3:25 am

Don wrote:
michiguel wrote:
mhull wrote:
Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
That is why Crafty is a bad example. Bob has been saying that programs in the past were not optimized for current hardware. Well, Crafty _was_ being optimized for hardware of the future, in detriment to the current hardware of the time. He was saying this all the time in the 90's and there were countless threads about it. "64 bits will become standard" etc. etc.

Miguel
Thank you Miguel. You said it better than I could and this is really clear.

I was getting frustrated trying to make Bob understand this concept and it was just because I am not very good at explaining myself.
There is no "concept" to explain here. I chose a data structure that was forward-thinking, based on a lot of hardware experience, and somehow that makes this an invalid discussion? Shoot, let's all use regressive software design techniques, never ask "what might we have next year so that we can factor that into our design" and, as a result, never see the advantages the new things offer. makes sense to me.

I'd reverse this. Any program that only gets a 100x speedup from P5/90 to i7 is a piece of crap. That's the right way to think. Not to make it the gold standard just because it is the worst example of progress you can find. This sounds more like a debate technique where you first decide on the answer you want, and then start searching for data to support that, whether the data is sensible or not is irrelevant, it just needs to exist.

bob
Posts: 20918
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: harware vs software advances

Post by bob » Sun Sep 12, 2010 3:27 am

Don wrote:
mhull wrote:
Don wrote:
michiguel wrote:
mhull wrote:
Don wrote:
mhull wrote:
Don wrote:
bob wrote:I do not quite see the point for all the tangents. It would seem to me we have a pretty good idea of what/how to test.
Says you.

Rebel is only 100 times faster on modern hardware. You add a zero to this and consider it a reasonable test.
I don't think it persuasive to introduce an orange in a comparison of apples. Crafty is apples to apples.
Rebel vs Rebel - apples to apples.

Crafty 1995 vs Crafty 2010 - apples vs oranges.
It's still going to be crafty 1995 vs crafty 1995 (new hardware), just like your Rebel test. The only difference is that crafty 1995 has more speedup potential than Reble because it (1995 version) can also benefit from 64-bit. Rebel can't.
That is why Crafty is a bad example. Bob has been saying that programs in the past were not optimized for current hardware. Well, Crafty _was_ being optimized for hardware of the future, in detriment to the current hardware of the time. He was saying this all the time in the 90's and there were countless threads about it. "64 bits will become standard" etc. etc.

Miguel
Thank you Miguel. You said it better than I could and this is really clear.

I was getting frustrated trying to make Bob understand this concept and it was just because I am not very good at explaining myself.
It's not relevant to the question being asked. All you need to know is how much ELO to attribute to HW speedup. How something was or was not optimized to HW has nothing to do with the answer. Only the speedup numbers matter. Whether a particular project sped up 10x or 1000x is the only answer you need. Then figure how how much ELO that translates to. That gives you the HW ELO increase (for that chess project). Any increase above that in a newer version (of that project) would be a candidate for software ELO improvement. This seems to be the general idea.

I mean, can the HW speedup for 1995 Crafty be determined or not? If it can, where is the problem?
Ok, if we go along with your premise, then why not use the Rebel numbers? In your words, "only the speedup numbers matter." And the numbers are 100 to 1 for Rebel. And I'm actually using an IDENTICAL program, you are not.
First, he _is_ using the identical program also. Unless you want to argue that Crafty 10.18 is not the same as Crafty 10.18? If you want to use the Rebel numbers, that's fine. And you can report what Rebel gained. I'm going to report gaining more from the hardware, because I did. I designed Crafty for the future at the time, not for status quo. Some consider that a good approach. But apparently not everyone. Microsoft needs you guys...

Post Reply