which ELO for a mephisto milano nigel short 16 Mhz

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
Post Reply
yoyo_chessboard
Member
Posts: 126
Joined: Mon Oct 17, 2016 5:44 pm
Location: South of France - cote d'azur

which ELO for a mephisto milano nigel short 16 Mhz

Post by yoyo_chessboard »

hi

i have modified my mephisto milano nigel short and it is now running at 16 Mhz instead of the 4.915 Mhz.
it is computing 3.2 times the computed nodes of the standard machine.

from a given position it takes 14,2 hours to computes 100 000 000 positions.
info from the board itself.

which ELO is it now supposed to have ?
against which machines could i test it ?

does it make it the strongest of the chessboards based on 6502 CPU ?
how to verify that point ?

against which 68000 based chessboard could it win now ?
User avatar
Scally
Full Member
Posts: 970
Joined: Sun Jun 05, 2016 5:44 pm
Location: Bermondsey, London
Contact:

Post by Scally »

Hi Yoyo,

This is an educated guess based on the Wiki ELO List.

Nigel Short 5 MHz is rated at 2028
Nigel Short 10 MHz is rated at 2083

This is a 55 ELO increase for 5 MHz, so 11 ELO per MHz

So this would put your 16 MHz version around 66 ELO stronger at around 2149 ELO

This would put it in the Mephisto TM Roma 6820, Mephisto Portorose 6820 & Mephisto Lyon 6800 range.

Of course this is total speculation and I’m open to any better suggestions ....


Cheers,

Al.
Larry
Senior Member
Posts: 2269
Joined: Wed Aug 01, 2007 8:42 am
Location: Gosford, NSW Australia

Post by Larry »

Scally wrote: Nigel Short 5 MHz is rated at 2028
Nigel Short 10 MHz is rated at 2083

This is a 55 ELO increase for 5 MHz, so 11 ELO per MHz

So this would put your 16 MHz version around 66 ELO stronger at around 2149 ELO
I'm wondering if you are asking for trouble clocking a 6502 that high.
For a given processor model, some individual processors will take a
higher clock speed than others. You might get away with it, but others
might find their 6502 will fail at that speed.
Regarding extra playing strength, I'm pretty sure you will find rating
gains are not linear as speed increases. There seems to be about a
60elo gain for the first doubling, but sharply diminishing returns on
later doublings. And even this depends on who's program we are
talking about. Some 'scale', ie, respond to speed increases, more
readily than others.
L
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

Scally wrote:
Nigel Short 10 MHz is rated at 2083
So this would put your 16 MHz version
well this has to be a first
estimating the rating of a modified computer ..by using a rating of another modified computer

Just Sayin Regards
Steve
User avatar
Scally
Full Member
Posts: 970
Joined: Sun Jun 05, 2016 5:44 pm
Location: Bermondsey, London
Contact:

Post by Scally »

Hi Guys,

I knew you can’t use the same formulae forever for doubling up nor do I condone modifications, I was just trying to be helpful. I did say It was an educated guess and invited others to correct me.

I guess I’m out of my league here .....


Cheers,

Al.
yoyo_chessboard
Member
Posts: 126
Joined: Mon Oct 17, 2016 5:44 pm
Location: South of France - cote d'azur

Post by yoyo_chessboard »

for Larry (and others)

my mephisto milano has new components.
in fact i modified my 3 mephisto milano in nigel short 16 Mhz.
one, on the operation desk, is even running at 18 Mhz but i limited myself to 16 Mhz

i didn't overclock the original IC (lazy approach) but bought new IC (CPU, RAM, ROM) meeting the new requirements.
also some 74HCxx have been changed.


so the chessboard run with stability and it is its first quality.
i make it run for 24hours periods to verify the stability.

the engine takes 3 times less to explore a given number of nodes.
so it has a deeper view in a given amount of time and commit less mistakes due to horizon effect
(in french : effect d 'horizon)

regards
User avatar
paulwise3
Senior Member
Posts: 1505
Joined: Tue Jan 06, 2015 10:56 am
Location: Eindhoven, Netherlands

Post by paulwise3 »

Hi Yoyo,

It is some rule of thumb that doubling the speed gains about 50 elo-points. That differs per program. So in your case the first doubling gains 55 points. But to think linear, you would need another doubling for another 55 points. It also depends how the programmed algorithms work. For instance pure brute force would need maybe a factor 4 to 6 speed increase to calculate one ply deeper in the same time...
My "educated" guess would be a 55+30=85 extra elo-points because you do not have a second doubling in speed.

So the Saitek Brute Force, and Fidelity's Mach IV and 2325 Master 68020 could also be good opponents. Of course it also depends on the match conditions: playing at tournament level (40 moves in 2 hours) or active level (30 sec/move average or 60 moves in 30 minutes).

Have fun with matchplaying regards,
Paul
2024 Special thread: viewtopic.php?f=3&t=12741
2024 Special results and standings: https://schaakcomputers.nl/paul_w/Tourn ... 25_06.html
If I am mistaken, it must be caused by a horizon effect...
Post Reply