Here the final result of Gull 2.
http://ls-ratinglist.beepworld.de/
Next test: Black Mamba 1.2b
Stefan
LS ratinglist: Gull 2
Moderator: Ras
-
- Posts: 2808
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
-
- Posts: 2285
- Joined: Fri Jul 14, 2006 7:56 am
- Location: London, England
- Full name: Jim Ablett
Re: LS ratinglist: Gull 2
Hi Stefan,
I have compiled some new JA builds of Gull II including SSE4.2 (popcount) enabled versions.
https://dl.dropbox.com/u/5047625/gull-II-ja.zip
Mirror:
http://cl.ly/LxXK/gull-II-ja.zip
Jim
I have compiled some new JA builds of Gull II including SSE4.2 (popcount) enabled versions.
https://dl.dropbox.com/u/5047625/gull-II-ja.zip
Mirror:
http://cl.ly/LxXK/gull-II-ja.zip
Jim
-
- Posts: 2285
- Joined: Fri Jul 14, 2006 7:56 am
- Location: London, England
- Full name: Jim Ablett
Re: LS ratinglist: Gull 2
Sorry, forgot to include the src code.
https://dl.dropbox.com/u/5047625/gull-II-ja.zip
Mirror:
http://cl.ly/LyCh/gull-II-ja.zip
Jim.
https://dl.dropbox.com/u/5047625/gull-II-ja.zip
Mirror:
http://cl.ly/LyCh/gull-II-ja.zip
Jim.
-
- Posts: 2808
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: LS ratinglist: Gull 2
Hi Jim,Jim Ablett wrote:Hi Stefan,
I have compiled some new JA builds of Gull II including SSE4.2 (popcount) enabled versions.
https://dl.dropbox.com/u/5047625/gull-II-ja.zip
Mirror:
http://cl.ly/LxXK/gull-II-ja.zip
Jim
can you say how much faster the popcount-compile is (compared with the 64bit-compile from Gull-homepage, which I tested in the LS ratinglist)?!?
Stefan
-
- Posts: 1833
- Joined: Thu Jun 22, 2006 12:07 am
Re: LS ratinglist: Gull 2
I found that the original is 1-3% faster.pohl4711 wrote: Hi Jim,
can you say how much faster the popcount-compile is (compared with the 64bit-compile from Gull-homepage, which I tested in the LS ratinglist)?!?
Stefan
-
- Posts: 2285
- Joined: Fri Jul 14, 2006 7:56 am
- Location: London, England
- Full name: Jim Ablett
Re: LS ratinglist: Gull 2
Martin Thoresen wrote:I found that the original is 1-3% faster.pohl4711 wrote: Hi Jim,
can you say how much faster the popcount-compile is (compared with the 64bit-compile from Gull-homepage, which I tested in the LS ratinglist)?!?
Stefan
I did the compiles for older machines primarily. On my Athlon64 the 64 bit GCC compile is way faster than the
official one. The popcount versions were untested. I still have no SSE4.2 Cpu. Here are the benchmarks > (from opening position)
Code: Select all
-------------------------------------------------------
JA : Gull II (32 bit) Intel = 331/knps
JA : Gull II (32 bit) MSVC = 368/knps (Best on my Athlon-XP/m)
JA : Gull II (32 bit) Mingw = 327/knps
Original : Gull II (32 bit) MSVC = 342/knps
JA : Gull II (64 bit) Intel = 634/knps
JA : Gull II (64 bit) MSVC = 610/knps
JA : Gull II (64 bit) Mingw = 690/knps (Best on my Athlon64)
Original : Gull II (64 bit) MSVC = 610/knps
-------------------------------------------------------
Jim.
-
- Posts: 2808
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: LS ratinglist: Gull 2
Thanx. At the moment my popcount-hardware does the PanChess 00.250-testrun. After finishing that, I will take a look on Gull 2 JA, but I doubt, if a new testrun is really necessary...Martin Thoresen wrote:I found that the original is 1-3% faster.pohl4711 wrote: Hi Jim,
can you say how much faster the popcount-compile is (compared with the 64bit-compile from Gull-homepage, which I tested in the LS ratinglist)?!?
Stefan
Best - Stefan