Christian,
Would you care to provide some more detail on your implementation for the
playouts? Your results are very impressive. At 19x19 Go using bit-boards,
your implementation is roughly 7x as fast as the bitboard implementation I
presented just a few weeks back, and also outperforms libEgo by a
Very interesting stuff. One glimmer of hope is that the memory situations
should improve over time since memory grows but Go boards stay the same size.
Christian Nentwich wrote:
Mark,
let me try to add some more context to answer your questions. When I say
in my conclusion that "it's not wo
Thank you Christian, for taking the time to write an extensive reply.
I still don't understand how you come to conclude that the CPU, at 94K
playouts with two cores, is twice as fast as the GPU doing 170K
playouts per second. Sounds like the reverse to me. Or you meant to
say "more than half the s
Just the info I needed -- thanks!
Politeness to human opponents aside, it sounds like a safe policy is
to never pass if there are dead enemy stones on the board.
Peter Drake
http://www.lclark.edu/~drake/
On Sep 9, 2009, at 4:44 AM, Jason House wrote:
If you're asking about stone cleanup i
Mark,
let me try to add some more context to answer your questions. When I say
in my conclusion that "it's not worth it", I mean it's not worth using
the GPU to run playout algorithms of the sort that are in use today.
There may be many other algorithms that form part of Go engines where
the
I'm trying to understand your conclusion. The GPU is more than 3 times
faster than the CPU, yet you don't think it's worth it. You also say
the card has only 9% occupancy.
I know next to nothing about GPU programming, so take my questions in
that stride.
>>Optimal speed was at 80 threads per bloc
Steve,
assuming that you can find something meaningful for the GPU to do, the
overhead of this should be relatively negligible.
Option one:
- Copy the board grid to the GPU. Since this is at most 21 x 21 integers
if the CPU maintains it in uncompressed form, it should take unnoticable
amount
thanks for taking the time to do these experiments and to report your
results. it will (has) saved a nightmarish-sounding investment of
time to learn the order of the speedup for this particular problem.
how much penalty do you estimate there is to pass a board from, say, a
program running on the
Interesting stuff. Thanks for reporting your results.
- Dave Hillis
-Original Message-
From: Christian Nentwich
To: computer-go
Sent: Wed, Sep 9, 2009 11:54 am
Subject: [computer-go] CUDA and GPU Performance
I did quite a bit of testing earlier this year on running playout algori
I did quite a bit of testing earlier this year on running playout
algorithms on GPUs. Unfortunately, I am too busy to write up a tech
report on it, but I finally brought myself to take the time to write
this e-mail at least. See bottom for conclusions.
For performance testing, I used my CPU
Once your bot has an account you can send email to the admins and ask them
to make your bot rated.
> -Original Message-
> From: computer-go-boun...@computer-go.org [mailto:computer-go-
> boun...@computer-go.org] On Behalf Of Isaac Deutsch
> Sent: Wednesday, September 09, 2009 3:56 AM
> To:
If you're asking about stone cleanup in the KGS tournaments, then yes,
there is a special phase. See Nick Wedd's site for full details.
Here's the basics:
1. Play normally until two passes
2. Respond to final_status_list dead
[done if opponent supports final_status_list dead and there is
agr
Ah, that explains it; should get my bot rated then... Thx!
Erik
On Wed, Sep 9, 2009 at 12:55 PM, Isaac Deutsch wrote:
> That's only possible in free games, but not possible in rated games.
>
> Am 09.09.2009 um 11:56 schrieb Erik van der Werf:
>
>> Last time I tried my program on kgs human player
That's only possible in free games, but not possible in rated games.
Am 09.09.2009 um 11:56 schrieb Erik van der Werf:
Last time I tried my program on kgs human players could simply declare
all bot stones dead and win regardless of the position. Did this
change?
Erik
On Wed, Sep 9, 2009 at 8
Last time I tried my program on kgs human players could simply declare
all bot stones dead and win regardless of the position. Did this
change?
Erik
On Wed, Sep 9, 2009 at 8:28 AM, David Fotland wrote:
> Dead stones are removed by agreement. If there is no agreement, the human
> can continue pl
15 matches
Mail list logo
