CPU: 13th Gen Intel(R) Core(TM) i9-13900K
So far I am seeing 200K PPD @5.4MHz out of the box. No tweaking yet.
Motherboard: MSI Z790 Carbon WiFi
RAM: 16G DDR5 Kingston Fury Beast
GPU: RTX4090
Cooler: MSI Coreliquid 280R
PSU: MSI MPG A1000G
Raptor Lake folding - i9-13900K
Moderator: Site Moderators
Forum rules
Please read the forum rules before posting.
Please read the forum rules before posting.
-
- Site Moderator
- Posts: 6349
- Joined: Sun Dec 02, 2007 10:38 am
- Location: Bordeaux, France
- Contact:
Re: Raptor Lake folding - i9-13900K
If it is a single folding slot using all available threads, it's definitely not good ... limit the slot thread number to the number of performance threads.
-
- Posts: 37
- Joined: Sun Apr 22, 2012 1:35 pm
- Hardware configuration: MSI Z790 ACE | 12700K | Cooler Master Cosmos | TeamGroup 32GB | RTX 4080/4090 | Win10 Pro
- Location: Indy
- Contact:
Re: Raptor Lake folding - i9-13900K
I spent a little time with it. When set to -1, the cpu:30. I experimented with setting the cpu to 8, 16 and 20 and didn't see any real difference in performance. Project 16969.
-
- Posts: 31
- Joined: Sat Mar 14, 2009 2:16 pm
- Hardware configuration: HP DL380g8 x12
- Location: Virginia
- Contact:
Re: Raptor Lake folding - i9-13900K
That sounds either problematic or disappointing.
My ~8-year-old e5-2690v0 @ 2.9GHz (turbo boost OFF, DL380g8) running CPU:30 gets ~270k PPD on p16969.
FWIW I have found that v0 (32nm) cpus far outperform v1 and v2 (22nm) cpus for F@H. I have only worked with the Xeons of this generation, not the consumer level chips.
Ben N1NP
My ~8-year-old e5-2690v0 @ 2.9GHz (turbo boost OFF, DL380g8) running CPU:30 gets ~270k PPD on p16969.
FWIW I have found that v0 (32nm) cpus far outperform v1 and v2 (22nm) cpus for F@H. I have only worked with the Xeons of this generation, not the consumer level chips.
Ben N1NP
Antonomasia Productions
New Release 2022: Re-Entrant
New Release 2022: Re-Entrant
Re: Raptor Lake folding - i9-13900K
You should see most performance when limiting the number of threads to fit your P-cores only. As FaH overall performance is capped by the performance of your slowest CPU core, you don't want to share work between P-cores and E-cores.
Things are complicated by Windows 11 dumb scheduler which doesn't understand complications like this, which might get into its head to put work on E-cores if you run FaH on more than 8 threads. Try setting 16 threads (assuming no GPU folding!) and disable E-cores altogether and see what happens... Because of the limited smarts of the scheduler, I'm not sure how effective it will be to create two FaH CPU cores, one with 16 threads for your P-cores and one with however many threads you want to dedicate on the E-cores. Who knows where each thread will end up, as Windows tends to throw threads around all over the CPU for no seemingly good reason.