On Tue, Oct 19, 2010 at 10:44 PM, David Fotland <fotl...@smart-games.com>wrote:
> Mobile phones have very limited memory. On the iPhone I only have about > 20 MB. This is a big limitation for UCT programs. > But you need much less memory when you are running 6-7 times slower. I browsed around on the web and found people complaining that the typical android device only leaves them with something like 150 MB free (which is 7x more than iphone) and that is probably not too bad given a 6x slower search (if that is what it is.) That is like having close to a gigabyte free on a PC if you assume memory use is linear with speed. Everything I've read about the iphone indicates that it is a popular toy but not very well suited to high performance computing (compared to other small devices.) For example I have read that it sucks memory and gets worse over time and that it's not a real OS (to me "real" mean you can run tasks in the background) etc. In the performance numbers for chess programs that Ingo quoted the top program was on an ipaq not an iphone. I also saw that Android seems to be outselling iphone and probably the majority of phones will be Android based soon. Don > > > David > > > > *From:* computer-go-boun...@dvandva.org [mailto: > computer-go-boun...@dvandva.org] *On Behalf Of *Don Dailey > *Sent:* Tuesday, October 19, 2010 9:48 AM > *To:* computer-go@dvandva.org > *Subject:* Re: [Computer-go] Candidates + Territory visualisation in > Valkyria > > > > > > On Tue, Oct 19, 2010 at 12:28 PM, "Ingo Althöfer" <3-hirn-ver...@gmx.de> > wrote: > > -------- Original-Nachricht -------- > > Datum: Tue, 19 Oct 2010 11:48:57 -0400 > > Von: Don Dailey <dailey....@gmail.com> > > ... > > > > If we assume 400 ELO difference that is approximately equivalent to > > 6 or 7 doublings - does that sound about right for the mobile phone > > hardware you are comparing? > > I am not a hardware expert, so I do not know. > I took the data from the SSDF list at > http://ssdf.bosjo.net/list.htm There we have > 1 Deep Rybka 3 x64 2GB Q6600 2,4 GHz 3213 +25 -23 ... > .. > 35 Pocket Fritz 3 Hiarcs Ipaq 214 624 MHz 2722 +53 -51... > > I'm not trying to refute you or claim you did anything wrong. I'm > just suggesting a refinement to get a more accurate estimate of how much we > can expect to lose going to the mobile phone platform. My refinement is > more appropriate for what to expect when porting a GO program to the mobile > platform. > > > > Trying to extrapolate what I know from your data and the "rybka > adjustment", you get this: As a general rule of thumb, if you port your go > program to a mobile platform you will get the same performance that you > would get if you played on the PC at a time control 6 or 7 times faster. > So for example if you set your program to play a game in 3 minutes on the > Q6600 platform and other programs are playing 20 minute games, you should > come relatively close to what you can expect on the mobile platform. > > > > Don > > > > > > > > I think, this Hiarcs version was especially tuned for "limited" hardware. > > > > They probably took the program that was easiest and most sensible to port. > I believe Hiarcs is 32 bit program which will port to mobile phone > hardware with the least amount of ELO loss. A 64 bit program would take an > additional hit in performance being ported to 32 bit hardware. > > > > > > > Ingo > > PS: The reason to mention go programs for mobiles was that Valkyria's > graphic seems ideally suited for small screens. > > > -- > GRATIS! Movie-FLAT mit über 300 Videos. > Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome > > _______________________________________________ > Computer-go mailing list > Computer-go@dvandva.org > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go > > > > _______________________________________________ > Computer-go mailing list > Computer-go@dvandva.org > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go >
_______________________________________________ Computer-go mailing list Computer-go@dvandva.org http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
