New AMD Ryzen™ Threadripper™ PRO 3995WX (Windows and Multithreading Problem)

[Zerbinati]

When you reduce the hash size to get higher NPS, is the time to depth smaller or larger on average?
I think it is possible that NPS is a red herring.
Depending on how they are counted, if a hash hit is not counted as a node, then it might really be a big slowdown.
If the hash table is pinned in memory and large, it makes no sense to me that reducing its size means more efficient computation.

@Dann Corbit sorry I missed your message ..
I agree with you ..
I think the problem is the RAM channels
this system supports 8 channels
currently i am only using 2 channels (2 slots occupied) 64gb x2

now i have ordered 16gb x 8 in order to use all 8 channels
as suggested by Tomasz Sobczyk
(Sopel97)
https://github.com/official-stockfish/S ... -828459351
I will update the thread as soon as I can verify by mounting them if I have solved it

[Zerbinati]

Filled all slots as suggested by Tomasz (Sopel97) I solved!
Thanks everyone for your help.

Regards

Marco

[MikeB]

Filled all slots as suggested by Tomasz (Sopel97) I solved!
Thanks everyone for your help.

Regards

Marco

+1 great news Marco!
Please post some new bencmarks when you have a chance.

[Zerbinati]

Filled all slots as suggested by Tomasz (Sopel97) I solved!
Thanks everyone for your help.

Regards

Marco

+1 great news Marco!
Please post some new bencmarks when you have a chance.

Hi Michael,
with stockfish or with asmfish and the bench of the ipman site?
if you are referring to Stockfish write me a reference bencmark for possible comparisons.