krazyken wrote:Hmm, so probably you have a Penryn processor. Is the Windows computer you are comparing to also a laptop?
Hello Kaj,
As far as I can see, you didn't answer the above question from Kenny. It's an important question: Unless the Windows computer you are comparing to also uses a laptop CPU, it is no surprise that chess programs runn slower on your MacBook Pro.
I wonder if there is a considerable difference in the quality of the compiles between the windows and mac version.
I tend to think you are right about the compiles. I've been doing some reading about that stuff, and what i see is that there can be significant speed differences depending on compiles. Maybe for the Mac there are some constraints that are difficult to master, or the worse alternative would be that programmers do not pay as much attention to product quality for the Mac and other smaller platforms than for Windows. As i cannot really believe that the latter option is the case, my guts says here is work to do on the technical side of things.
The Leopard version of Glaurung is compiled using the Intel compiler (unlike the Tiger version, which is compiled with gcc), and is very fast. What N/s count do you get with a single thread from the opening position? I get 1.14 MN/s on my MacBook Core 2 Duo 2.4 GHz.
Tord
Hi Tord,
Both are laptops using Core 2 duos as i told Kenny.
Glaurung 2.1 at 1 thread gives about 610 kn/s from the opening position on my MacBook Pro at 2x2,53 ghz. Definitely lower than on your Mac. Your kn/s are closer to what i would have expected.
Kaj Soderberg wrote:
I tend to think you are right about the compiles. I've been doing some reading about that stuff, and what i see is that there can be significant speed differences depending on compiles. Maybe for the Mac there are some constraints that are difficult to master, or the worse alternative would be that programmers do not pay as much attention to product quality for the Mac and other smaller platforms than for Windows. As i cannot really believe that the latter option is the case, my guts says here is work to do on the technical side of things.
Cheers,
Kaj
As far as different compiles go, compiles for Windows and compiles for Mac are frequently done by different people. Some are much better at optimizing than others it seems.
One thing I did see, A MacBook will run slower if you are running it without a battery. You aren't one of the people who likes to take the battery out are you?
Both are laptops using Core 2 duos as i told Kenny.
Glaurung 2.1 at 1 thread gives about 610 kn/s from the opening position on my MacBook Pro at 2x2,53 ghz.
In that case, something is definitely wrong, at least if you are using the Glaurung binary from my web site. If the same pattern holds for other chess programs, I suspect that your CPU is not running at full speed, possibly because of some sort of hardware problem.
Both are laptops using Core 2 duos as i told Kenny.
Glaurung 2.1 at 1 thread gives about 610 kn/s from the opening position on my MacBook Pro at 2x2,53 ghz.
In that case, something is definitely wrong, at least if you are using the Glaurung binary from my web site. If the same pattern holds for other chess programs, I suspect that your CPU is not running at full speed, possibly because of some sort of hardware problem.
Tord
Yep, as i see now at around 50% speed. Back to the Mac store this week.
Thanks Tord.
Open your System Preferences and click Energy Saver. In Energy Saver, click the Options tab. Now check the setting for Processor Performance. If it’s set to Reduced, change it to Highest.
Open your System Preferences and click Energy Saver. In Energy Saver, click the Options tab. Now check the setting for Processor Performance. If it’s set to Reduced, change it to Highest.
OK, thanks Kenny.
I'll have a look at this when the Mac comes back from the hardware doctor.
Just thought i'd let you know that all is well with the Mac after changing the hard drive and the battery, and i'm seeing some awesome Glaurung chess almost at daily basis. An aggressive bastard it is, and therefore great fun.
Looking forward to what the Smaug development will bring us.
Jim Ablett wrote:I compiled (by request a while ago) win32 versions of Glaurung & Stockfish with reduced uci parameters which fit in 16:9 screens. Download:
Glaurung 2.2 16:9 http://www.mediafire.com/?1oyxnkyyjyi
Unfortunately, Jim, this glaurung22_(16-9)_win32_ja.exe version (01/01/2009) does not behave the same as your original glaurung22_win32_ja.exe (20/12/2008)
Here are 2 examples (running on 1 cpu only, for reproducibility, and 512 MB hash):
1. Start_position
rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
Jim Ablett wrote:I compiled (by request a while ago) win32 versions of Glaurung & Stockfish with reduced uci parameters which fit in 16:9 screens. Download:
Glaurung 2.2 16:9 http://www.mediafire.com/?1oyxnkyyjyi
Unfortunately, Jim, this glaurung22_(16-9)_win32_ja.exe version (01/01/2009) does not behave the same as your original glaurung22_win32_ja.exe (20/12/2008)
Hi Ernest,
Looks like removing some of the uci options has altered the the search/eval defaults.
Odd. I would have expected the defaults to be initialized elsewhere.
Jim Ablett wrote:
Looks like removing some of the uci options has altered the the search/eval defaults.
Odd. I would have expected the defaults to be initialized elsewhere.
Thanks, I'll check it out.
Jim.
In Glaurung+derivatives defaults are initialized when ucioptions are loaded. The ones hardcoded in source are meaningless (and often misleading). If you remove uci options you need to manually fix the hardcoded defaults.
Tord said sometime ago that he has plans to fix this.
