Hayperthreading experiment

[Hugo]

Greetings to all computerchess enthusiasts

Since 2 weeks I am running a hyperthreading experiment.
Webpage with updates and informations is:

http://www.clemens-keck.de/htEng.html

I will update every week.

Kind regards, Clemens Keck

[BrandonSi]

Thanks Celemens! I'm very interested in your results, and have been following the website.

I have long believed that HT does not inherently effect engine performance negatively as is widely believed. In my opinion, it is the implementation of SMP/MP/threading in the engine itself that determine in HT effects performance positively or negatively.

So far, your Fritz results are consistent with my opinion, but obviously more data is needed before reaching a conclusion.

I especially look forward to the Shredder and Stockfish results.

Brandon

[Dann Corbit]

Greetings to all computerchess enthusiasts

Since 2 weeks I am running a hyperthreading experiment.
Webpage with updates and informations is:

http://www.clemens-keck.de/htEng.html

I will update every week.

Kind regards, Clemens Keck

Prediction:
Hyperthreading will add less than 5 Elo, and so you won't be able to measure any difference.

[bnemias]

Prediction:
Hyperthreading will add less than 5 Elo, and so you won't be able to measure any difference.

I thought hyperthreading increased nps, but also increased the tree size enough that the small increase in nps wasn't enough to even break even.

It should reduce ELO, I thought.