Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I too had a pretty big boost to my output when I updated the scripts from 4 to 5 on my G5 dual 2ghz. I think I hit a high of 448ppd with just that one machine...

I hit 37k points in seven months with just that machine folding 24/7. So I averaged around 202ppd. However I saw an increase of about 2000 points per month when I updated the scripts.
 
Hey, guy(s) wondering about the HT Processors working with F@H, would you object to disabling the hyperthreading? I've noticed on my P4, when HT was enabled, even when only one F@H process was running, it would never go about 50% of the CPU. Instead of worrying about two different clients, I just disabled the HT thing, and am now running 1 F@H console at 100%. If you're interested in how to do this, post back here or PM me and I'll explain it. (It's actually pretty easy)
 
Dreadnought said:
Well, my G5 gets special power out of the wall! ;)
Is that, like, wonder twin power?
wondertwin_sm.jpg
:)

I really don't know how. Maybe because I have set my processorspeed to high, maybe it's only folding and I don't do anything else with it, and maybe I got lucky with the WU's I get.
I was under the impression that setting your processor speed to high only helped with initial ramp up of the processors when a process was started (such as with xbench), but that once the process was running, the processor would be on high. As a result, I always leave them on automatic.
 
RugoseCone said:
I too had a pretty big boost to my output when I updated the scripts from 4 to 5 on my G5 dual 2ghz. I think I hit a high of 448ppd with just that one machine...

I hit 37k points in seven months with just that machine folding 24/7. So I averaged around 202ppd. However I saw an increase of about 2000 points per month when I updated the scripts.
Can you just stop the currently running WUs, upgrade the to version 5 and start them back up again? I always just let them run, so I don't really know how to get them to complete the current WU and stop until I'm ready to begin the next WU.
 
Maybe I answered my own question ^^. Can I use the fah4 command with the -pause option, even though there are WUs in progress? Will that cause them pause after they complete, until I kick off the next WU?

TIA
 
daveL said:
Maybe I answered my own question ^^. Can I use the fah4 command with the -pause option, even though there are WUs in progress? Will that cause them pause after they complete, until I kick off the next WU?

TIA

Sorry, but you'll lose any uncompleted WU's. You have to use the "rid" command to uninstall the previous version of mc68k's scripts and then install the new version. This will eliminate all traces of the current work in progress.

However, if you are using the client as provided by Stanford, you might be able to do what you propose...

You'll just have to watch for when your WU's are nearing completion. I tried to time my upgrade and it sort of worked. I was about 25% through a 63 point WU and had just begun a 300 pointer when I did mine.
 
Littleodie914 said:
Hey, guy(s) wondering about the HT Processors working with F@H, would you object to disabling the hyperthreading? I've noticed on my P4, when HT was enabled, even when only one F@H process was running, it would never go about 50% of the CPU. Instead of worrying about two different clients, I just disabled the HT thing, and am now running 1 F@H console at 100%. If you're interested in how to do this, post back here or PM me and I'll explain it. (It's actually pretty easy)

Or you can do what I did at work and run dual F@H clients to cover both threads. Disabling HT or running dual clients will only net about a 15% boost though due to really being only one processor with two different pipelines feeding it. The processor gage in Windows isn't completely accurate in that it may say you are only using 50% of the processor but it is not true (surprised?). I've discussed how to get 2 clients running earlier in this thread.
 
atszyman said:
Or you can do what I did at work and run dual F@H clients to cover both threads. Disabling HT or running dual clients will only net about a 15% boost though due to really being only one processor with two different pipelines feeding it. The processor gage in Windows isn't completely accurate in that it may say you are only using 50% of the processor but it is not true (surprised?). I've discussed how to get 2 clients running earlier in this thread.
Let's just say that the Windows Task Manager doesn't fully understand HyperThreading. Basically, HyperThreading allows a CPU to become 115% of its former self, so when it says it's using 50%, it's actually 57.5% (115/2) because the CPU has 115% resources available. Confused? You should be...

I still haven't decided whether or not to reinstate F@H on my iMac, and, if I do, whether or not to use a (slightly) different user name.
 
Littleodie914 said:
Hey, guy(s) wondering about the HT Processors working with F@H, would you object to disabling the hyperthreading? I've noticed on my P4, when HT was enabled, even when only one F@H process was running, it would never go about 50% of the CPU. Instead of worrying about two different clients, I just disabled the HT thing, and am now running 1 F@H console at 100%. If you're interested in how to do this, post back here or PM me and I'll explain it. (It's actually pretty easy)

Regarding this and the post replying to it...I'm guessing this only really applies to how Windows 'handles' HT? One of my folding machines is a 2.4GHz w/ HT running Linux and it seems to be running at 100% just fine.
 
As far as I can tell the Windows task manager seems to recognize pipe utilization rather than processor utilization. Thus a single pipe fully fed with nothing utilizing the other pipe will give you about 87% of the processor for that process but only show 50% in the task manager. By shutting off Hyperthreading or feeding the other pipe you get the extra cycles that were normally dedicated to the other pipe back and can get 100% utilization.
 
atszyman said:
As far as I can tell the Windows task manager seems to recognize pipe utilization rather than processor utilization. Thus a single pipe fully fed with nothing utilizing the other pipe will give you about 87% of the processor for that process but only show 50% in the task manager. By shutting off Hyperthreading or feeding the other pipe you get the extra cycles that were normally dedicated to the other pipe back and can get 100% utilization.
That's correct. Like I said, this behavior is due to the way Windows Task Manager deals with HyperThreading, which I think is strange and counterintuitive.
 
Strange that Windoze....

lssmit02 said:
I was under the impression that setting your processor speed to high only helped with initial ramp up of the processors when a process was started (such as with xbench), but that once the process was running, the processor would be on high. As a result, I always leave them on automatic.

I leave it always on at the highest. Sometimes when I look in the taskmanager I FAH getting up to 105%! I wonder how accurate the taskmanager of the Mac is?
 
RugoseCone said:
Sorry, but you'll lose any uncompleted WU's. You have to use the "rid" command to uninstall the previous version of mc68k's scripts and then install the new version. This will eliminate all traces of the current work in progress.

However, if you are using the client as provided by Stanford, you might be able to do what you propose...

You'll just have to watch for when your WU's are nearing completion. I tried to time my upgrade and it sort of worked. I was about 25% through a 63 point WU and had just begun a 300 pointer when I did mine.
Fortunately my 2 WUs were running very close and completed at almost the same time. As soon, as the next 2 WUs started, I killed them, wiped out the old version and did the upgrade to version 5. Although it was stated above that release 5 seems to complete the WUs faster, I don't really understand why that would be, since the same cores are running as in fah4, and they do the work. Am I missing something?
 
Dreadnought said:
maybe an optimised code? or we are getting better WU's! Less effort, more points!!
My suspicion is that the v5.0 core has a different (and better) method of deciding what WUs to give to which machines, thus improving productivity on the whole. The work cores themselves don't need to be updated to take advantage of such optimizations, thus explaining the performance boost.

EDIT: My iMac G4 is folding once more.
 
I know this is obnoxious, but I have to toot my horn, after sitting on #501 last night, I made top 500 this morning! (Casey Kasam) Entering the charts at number 486 with 3749 pts, is stubeef and the 3 lil processors that could.

That was a goal for my self, I am happy, and having to look over my shoulder now cause of altair and sparkleytone.

Off we go to top 400! My goal is top 100, after that it is all gravey.
 
stubeeef said:
I know this is obnoxious, but I have to toot my horn, after sitting on #501 last night, I made top 500 this morning! (Casey Kasam) Entering the charts at number 486 with 3749 pts, is stubeef and the 3 lil processors that could.

That was a goal for my self, I am happy, and having to look over my shoulder now cause of altair and sparkleytone.

Off we go to top 400! My goal is top 100, after that it is all gravey.

I think that having a goal is a good thing. You really have quite lofty goals. I just happy to be the the top 200. :)
 
atszyman said:
I've done this on my work machine. You will only see about a 15% increase in productivity due to the fact that Hyperthreading is still only one processor, just two instruction pipelines.

1. You cannot have the GUI running.
2. You will need 2 copies of the console app in different folders.
3. Run the console app from each folder with the -local and -config option
4. When asked if you want to change advanced options type y or yes
5. Set the processor ID of one of the console apps to 0 the other to 1 (or 1 and 2 I can't remember if it's zero indexed or not)
6. Once configuration is done when you need to restart you only need the -local flag.

Hope this helps.
Thanks. If that is the case with only a 15% increase, would it be better to use this, or to just turn hyperthreading off?

PS. #107 now and rising!
 
GeeYouEye said:
Thanks. If that is the case with only a 15% increase, would it be better to use this, or to just turn hyperthreading off?

PS. #107 now and rising!

Congratulations, it would seem that you are a hard core folder. You must have a large farm.
 
How far along are we in folding all the proteins required to find the cure?
 
I congratulate anyone fortunate enough to be able to fold on more than one processor at the same time. I can only afford the PB that I have, running 24/7 ---- and here I hang out around #280 for long time.

ps. What happened to World Wizard? May I use his machines?
 
Lacero said:
How far along are we in folding all the proteins required to find the cure?

If I recall, over a year ago, the first one was complete. The project has a long way to go.
 
wdlove said:
Congratulations, it would seem that you are a hard core folder. You must have a large farm.
Funny thing, not really. Just a PowerBook which is often used for other stuff and a P4 2.8C which is pretty much used for gaming a couple hours a day (and for quite a while, not even that) and folding 22/7. It's odd actually, my PowerBook has done 38 units since June, but my desktop has done 130 since about late september. I also have a couple of old machines at home folding, but it takes weeks to get a single unit out of them. (iMac DV+ and an iBook)
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.