what is the sense of unlimited hardware tournaments ?
thats show you no improved of the program it self.
if i run Tornado with 8 core vs Tornado 1 core, sure the 8 core will win !
51st CSVN programmers tournament at Leiden - Hiarcs wins
Moderators: hgm, Rebel, chrisw
-
- Posts: 918
- Joined: Mon Jan 05, 2009 7:40 pm
- Location: Germany
- Full name: Engin Üstün
-
- Posts: 1167
- Joined: Thu Dec 25, 2008 9:07 pm
- Full name: Herbert L
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
Yeah, that would be nice.
Maybe Harvey can confirm that.
Maybe Harvey can confirm that.
-
- Posts: 4889
- Joined: Thu Mar 09, 2006 6:34 am
- Location: Pen Argyl, Pennsylvania
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
Ok we get it - you only like equal hardware events, you can put it to bed now . I have enjoyed unequal hardware events - it doesn't always work out the way you think it will - check out Fritz Vz Deep Thought in 1995.Engin wrote:what is the sense of unlimited hardware tournaments ?
thats show you no improved of the program it self.
if i run Tornado with 8 core vs Tornado 1 core, sure the 8 core will win !
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
That is such a load of BS.mehmet karaman wrote:28 core Intel Xeon E5-2697 (2.6 GHz) is 3.9x faster than Intel I7 6700K ( 4 Ghz)hgm wrote:I doubt that. It is not easy to use 1024 cores in a beneficial way.
3.9x speed is equal to ~ 80-100 elo
Komodo 10.4 is is 327 elo stronger than Hiarcs 14 (4cpu)
I think Hiarcs 14.x is at least ~200 elo points stronger than Hiarcs 14
Speedwise difference between Haswell and Skylake for cores running on the same frequency is around 5% for chess.
All core turbo for E5-2697 v3 is 3.1GHz. Total nps loss on 28 cores compared to 4 cores is also around only 5%.
So nps-wise dual E5-2697 v3 is like 5 times faster than i7 6700K.
Strength-wise that is certainly more than 100Elo, but almost certainly not more than 150Elo.
So the only valid conclusion is that Hiaracs book is at least 150Elo stronger in a single game sample than Komodo book. In other words there is absolutely no meaningful conclusion possible after single game of 2 engines that are less than 200 Elo apart taking hardware into account and using own books.
These kind of tournaments are total BS anyway and pathetic commercial authors that would otherwise have zero possibility to advertise their products try their best to exploit it.
-
- Posts: 4889
- Joined: Thu Mar 09, 2006 6:34 am
- Location: Pen Argyl, Pennsylvania
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
Would it hurt to at least try to be a little bit friendlier ? Are we not here for the same reasons - we enjoy computer chess true? So why do we have post so much negatively about something we enjoy? If something bothers you , just ignore it , no need to get your panties in a twitch about it.
-
- Posts: 27808
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
Unfortunately some people are only here to grind their axes, and display their naturally evil mood. The less they contribute, the greater their hatred for those who do. To see that some people have fun with computer chess makes their blood boil.
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
Some ppl are here to enjoy computer chess, some other to earn money. The second kind I personally can't stand, while you obviously don't see the difference.MikeB wrote:Would it hurt to at least try to be a little bit friendlier ? Are we not here for the same reasons - we enjoy computer chess true? So why do we have post so much negatively about something we enjoy? If something bothers you , just ignore it , no need to get your panties in a twitch about it.
It is understandable for ppl to pay for the best or even one of top 3 best chess programs. However, it is also natural that commercial chess engines that are much weaker struggle to sell their products and are trying in every way to advertise them. Commercial advertising is strictly forbidden by the charter of this forum. Then you have a grey zone when some ppl obviously with financial interest come and brag about their random and pretty unfair wins in some tournaments over top programs using this as obvious advertising strategy. The fact that you see this through the rose-coloured glasses doesn't make it less true.
-
- Posts: 2011
- Joined: Sun May 25, 2008 11:12 pm
- Location: Whitchurch. Shropshire, UK.
- Full name: Harvey Williamson
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
If you mean me I have never had any financial interest in Hiarcs or Computer Chess.Milos wrote:Then you have a grey zone when some ppl obviously with financial interest
-
- Posts: 4889
- Joined: Thu Mar 09, 2006 6:34 am
- Location: Pen Argyl, Pennsylvania
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
I have never seen the behavior you described from Harvey, you are sadly mistaken I believe. Anyway, you made your point. Let's move on.
-
- Posts: 142
- Joined: Tue Jan 28, 2014 8:37 am
- Location: TURKEY
Re: 51st CSVN programmers tournament at Leiden - Hiarcs wins
The Stockfish (28082015) benchmark of i7-6700k is 9726Milos wrote:That is such a load of BS.mehmet karaman wrote:28 core Intel Xeon E5-2697 (2.6 GHz) is 3.9x faster than Intel I7 6700K ( 4 Ghz)hgm wrote:I doubt that. It is not easy to use 1024 cores in a beneficial way.
3.9x speed is equal to ~ 80-100 elo
Komodo 10.4 is is 327 elo stronger than Hiarcs 14 (4cpu)
I think Hiarcs 14.x is at least ~200 elo points stronger than Hiarcs 14
Speedwise difference between Haswell and Skylake for cores running on the same frequency is around 5% for chess.
All core turbo for E5-2697 v3 is 3.1GHz. Total nps loss on 28 cores compared to 4 cores is also around only 5%.
So nps-wise dual E5-2697 v3 is like 5 times faster than i7 6700K.
Strength-wise that is certainly more than 100Elo, but almost certainly not more than 150Elo.
So the only valid conclusion is that Hiaracs book is at least 150Elo stronger in a single game sample than Komodo book. In other words there is absolutely no meaningful conclusion possible after single game of 2 engines that are less than 200 Elo apart taking hardware into account and using own books.
These kind of tournaments are total BS anyway and pathetic commercial authors that would otherwise have zero possibility to advertise their products try their best to exploit it.
The Stockfish (28082015) benchmark 2x Xeon E5-2695 v3 (28 core) is 33219
2x Xeon E5-2697 v3 is ~%15 faster than 2x Xeon E5-2695 v3
https://sites.google.com/site/computers ... benchmarks
Hiarcs 14t book is ~70 elo points weaker than some free books like Saka32x and DoomFX8
https://sites.google.com/site/computers ... book-cs-13
200 elo difference is only just my guess. There is no interest about its tournament performance
Last year (Feb 09, 2016) Mark said "Hiarcs experimental engine versions are hugely stronger than any previous Hiarcs so you can be assured it will be a very big jump in strength. The release however is still sometime away and we want it "
http://www.hiarcs.net/forums/search.php ... 0&start=15