TalkChess.com

I just tried several consecutive 120s searches of the opening position with SF10 at 32 threads on my Ryzen 3950x at 4.0 Ghz. I got 41895 Knps and had maximum package power of 139 watts and temperature of 70C. My Vcore droops from 1.20 to 1.16 volts when running all cores. My memory is set to 3200 Mh...

That's a very nice overclock! The Ryzen's do something called "clock-stretching" where they may slow the clock down to prevent crashing, or something like that. I found that with Vcore below about 1.175V my CPU seemed stable at 4.1Ghz, but my benchmark speeds were slower, or at least no faster. So I...

For my Ryzen 3950x, in the BIOS I set Vcore=1.20 and Multiplier to 40 and don't enable PBO. This gives an all-core frequency of 4.0 Ghz and when running Wasp with 32 threads the total CPU package power is about 132 watts. With the Noctua UH-12S air cooler the cpu temp stays under 70C. I really like ...

I probably have tried doing a reduced-depth search on fail-low nodes. But, my current version just stops searching the first move once it's failed low at a score of about 1.5 pawns below the target. It then sets a "panic" time of 20-25% (I forget) of the time remaining on the clock and starts search...

I have never tried reducing the depth in the ID loop after a fail-high at the root. I guess I should try this. For fail-low at the root I've tried lots of stuff. What I'm doing now is when the first move fails low, increase the window a few times to see if it's within about 1-2 pawns of the score fr...

You can pass parameters to cutechess by putting option.param=value after the -engine conf command. Like this:

-engine conf="Wasp 3.85" option.OwnBook=false option.Contempt=0 option.MoveOverhead=25 option.Log=true

John

Here's what I use in Wasp for both Windows and Linux. Reset != 0 starts the timer and reset == 0 returns the elapsed time in seconds (floating point). I think its accurate to well under a millisecond. John void GetElapsedTime(int reset) { struct timespec ts; static double start = 0.0; if (reset) { c...

Hi Bob, As I mentioned in my previous post, I don't think storing a 64 bit hash signature is enough even now on hardware like at TCEC. Wasp used "only" 16G of hash memory = 1G positions, so 30 bits is used up in the address. Since the table is always full of entries, that leaves 34 bits of useful ve...

Wasp had 3 crashes in the recent TCEC tournament where it used 128 threads and a 16 Gbyte hash table. I was unable to duplicate the crashes using 32 threads and an 8 Gbyte hash table even after many tries searching the position prior to the crash to fill the hash table and then the position Wasp cra...

I've never been able to get any gain from singular extension so I don't use it in Wasp. I must not be implementing it correctly or it doesn't play nicely with my extensions and reductions.

John