• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Hyperthreading

I don't think I ever turned it off, I just don't do two instances on one HT core. Was a recomendation from Stanford to get the work back earlier.
 
There isn't much use in running two clients on a hyperthreaded machine. In my experience, it doesn't garner any more points and might actually slow things down a bit. I keep it enabled on my machine at work but only run a single client on it.

There was a time when you could get more PPD out of the machine with two clients. That was when they had the timeless tinkers. With no deadlines you could set them for low CPU usage and since they didn't use SSE, it wouldn't fight with the other client running gromacs or other cores which use SSE. If you try to crunch two WUs and they both use the SSE instructions, the clients fight over the SSE part of the CPU and slows things down badly.

 
Was it only with the Tinkers that HT benefited running more than one client? I used to have a couple of P4s and always enabled HT with two clients on each machine. I'm certain I saw slightly better production running two clients as opposed to one irrespective of the WU. I saw threads where members were recommending that configuration, but this was around the 2004 time frame. I don't know how this setup will produce with modern WUs, though. It doesn't hurt to try both configurations and see the results.
 
My old computer was much faster with HT off... 3.06ghz p4...
 
APOLLO said:
Was it only with the Tinkers that HT benefited running more than one client? I used to have a couple of P4s and always enabled HT with two clients on each machine. I'm certain I saw slightly better production running two clients as opposed to one irrespective of the WU. I saw threads where members were recommending that configuration, but this was around the 2004 time frame. I don't know how this setup will produce with modern WUs, though. It doesn't hurt to try both configurations and see the results.
Click to expand...

If you looked at the total output of the machine running two standard clients you could see up to a 15% point increase from that machine. That was on the very old clients like the tinkers etc.

The issue came to a head when Stanford noticed too many results from the same team member and machine coming back more slowly. In other words slightly more work, far longer to do it and there were no real benefits to anyone.

It was at that point the powers at Stanford asked the teams please not to use that methodology anymore because what we turn in work wise gets processed almost immediately.

This issue was originally tried with the Pentium-IV and as it showed no benefit the idea was dropped. There is no real reason to turn HT off if you still have it as it still helps a bit in everyday work and doesn’t slow the current single client.


 
BillR touched on what really happens at the end of the post. You can leave HT on to benefit your day to day user experience on the computer and then run 1 instance of F@H. With how hyperthreading works F@H will still use 100% of the cpu even though taskmanager says it's only using 50%.

 
You must log in or register to reply here.
Back
Top