24 threads
Deep Rybka 4 SSE42 x64
843 kns
Deep Shredder 12
8558 kns
Zappa Mexico
13165 kns
Hyperthreading Disabled
12 threads
Deep Rybka 4 SSE42 x64
1142 kns
Deep Shredder 12
9230
Zappa Mexico
11236 kns
Compare Hyperthreading
Moderators: hgm, Rebel, chrisw
-
kgburcham
- Posts: 2016
- Joined: Sun Feb 17, 2008 4:19 pm
Compare Hyperthreading
Hyperthreading Enabled
24 threads
Deep Rybka 4 SSE42 x64
843 kns
Deep Shredder 12
8558 kns
Zappa Mexico
13165 kns
Hyperthreading Disabled
12 threads
Deep Rybka 4 SSE42 x64
1142 kns
Deep Shredder 12
9230
Zappa Mexico
11236 kns
24 threads
Deep Rybka 4 SSE42 x64
843 kns
Deep Shredder 12
8558 kns
Zappa Mexico
13165 kns
Hyperthreading Disabled
12 threads
Deep Rybka 4 SSE42 x64
1142 kns
Deep Shredder 12
9230
Zappa Mexico
11236 kns
-
alpha123
- Posts: 660
- Joined: Sat Dec 05, 2009 5:13 am
- Location: Colorado, USA
Re: Compare Hyperthreading
That's odd. I thought engines were supposed to be faster with hyperthreading on, but it made the search tree was distorted so it ended up being weaker.
Peter
Peter
-
bob
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: Compare Hyperthreading
What is the point of quoting Rybka NPS/node-counts since it has been proven to be a completely bogus number???kgburcham wrote:Hyperthreading Enabled
24 threads
Deep Rybka 4 SSE42 x64
843 kns
Deep Shredder 12
8558 kns
Zappa Mexico
13165 kns
Hyperthreading Disabled
12 threads
Deep Rybka 4 SSE42 x64
1142 kns
Deep Shredder 12
9230
Zappa Mexico
11236 kns
NPS is the wrong measure to see of hyper-threading works. You want to search several different positions to fixed depth. Measure time with HT on and OFF. Overall it will always be faster without, although individual positions might vary from that.