Einzelnen Beitrag anzeigen
  #3899  
Alt 12.05.2020, 15:06
Tibono Tibono ist offline
Resurrection
 
Registriert seit: 22.05.2018
Ort: Frankreich
Alter: 63
Land:
Beiträge: 585
Abgegebene Danke: 3.394
Erhielt 1.389 Danke für 466 Beiträge
Aktivitäten Langlebigkeit
6/20 7/20
Heute Beiträge
0/3 ssssss585
AW: Re: Mess Emulator für diverse Schachcomputer ist fertig!

 Zitat von hap Beitrag anzeigen
In reality, V9 will be slower than V10.
68040 is a faster CPU than 68030. MAME's timing emulation of it is not that accurate (neither for 68030, and especially not for 68060)

BCLK(eg. access to RAM/ROM is slowed) is not emulated in MAME.
Hi hap,
I think Eastnor is right - a quick test using Franz's CB-Emu and fixed depth search levels reveals current V10 emulation runs slower than the V9 one. I checked using "show FPS" option my own CPU achieved the expected emulation performance (100%).
I understand the 68030 (v9) emulation is not that accurate, better than nothing (and I am more than happy to use it); but the v10 68040 emulation speed seems badly wrong!

Just using Wiki Elo-List, at activ chess
v9 : CElo = 2150
v10: CElo = 2204 (+54)

We can assume the program to be the same and guess a speed ratio: v10/v9 = 10^(54/[Elo gain]*log(2))
-minimum value assuming 80 Elo points gain per speed doubling: 1.60
-medium value assuming 70 Elo points gain per speed doubling: 1.71
-highest value assuming 60 Elo points gain per speed doubling: 1.87

So yes, we can expect the v10 to run significantly faster than the v9, which the emulation does not seem to achieve.

Kind regards,
Eric
Mit Zitat antworten
Folgender Benutzer sagt Danke zu Tibono für den nützlichen Beitrag:
Eastnor (14.05.2020)