• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

I've read somewhere that HT hinders DC

ribbon13

Diamond Member
Since running two units at once with HyperThreading each unit takes longer. Can anyone point me to any article or provide more details?
 
Link to forum thread at F@H

EDIT:
BTW: I take advantage of HT for the sake of points... even if the folks at F@H would like us to give them their WU's a little faster.

I don't think you can say that HT "hinders" DC though... It may not be the optimal way to go scientificly speaking, when the subject is Folding@Home, but I think that for the majority of DC-projects it is an advantage to use HT, as you get more work done per time unit, if you're running more than one instance of the DC-project of course.
 
I believe that my P4 3.4GHz puts out more points with HT on, although each instance is slower and rated lower than a single instance if HT was off. Like what MechEng said.
 
I guess it really depends on the project, but IIRC the gains are from aroun 15-20% with F@H or FaD up to 70% with SETI for example. Then there are projects that can't even run two instances like ClimatePrediction.Net classic. As one of the CPDN WU takes almost a month even on really fast machine, you also wouldn't want to run two on HT and thus take two months to finish these WUs.

In the end it boild down to these two points:

- Are you a statswhore (like me)? --> HT, two instances, more points, slower results
- Are you a scientis? --> no HT, one instance, less points but faster research

:beer:😀:beer:
 
If you're running windows and got HT, it doesn't make sence to still run CPDN "classic", since the Intel-compiler used for the BOINC-version reportedly gives upto 30% faster crunching on Intel-cpu's... The BOINC-version can run 2 or more instances, so should also give an extra 20% or something improvement due to HT.


Have never used HT, but AFAIK seti will also get around 20-30% increase due to HT, and not 70%... Even a "real" Intel dual-cpu doesn't manage 70% increase if only got 256 KB cache-memory... To get more increase, you needs 512 KB or more cache-memory, or use Amd...


Have also seen comments there CPDN is on some computers is so memory-limited that no benefit from enabling HT, while in Einstein@home you're very often cpu-limited so not always you get any benefit here either from HT...

Depending on which part of the cpu different projects is using, it's sometimes better to run 2 different projects on dual-cpu to maximize total production...
 
I find that hard to believe. In my experience HT helps A LOT! I've run SETI with and without HT. On a 3.06GHz PC I was getting ~10 wu's per day with HT off, with HT on that jumped to ~16 wu's per day. Also, from what I've seen with FaD so far, my HT machines are MUCH more productive than my non-HT ones running at about the same clock speed.
 
Back
Top