| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home
Thread: Windows 7 / 64 bit / Lower Credit? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 60
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Ah, that you did not mention in any of your previous posts... the multiboot back into a 32 bit OS and wanting to retain result compatibility.
----------------------------------------Give me a shout if you wish to progress to a solution beyond the "you (wcg) are giving the wrong client out to w7 (or 64 bit) users".
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Highwire
Cruncher Joined: Aug 18, 2006 Post Count: 39 Status: Offline Project Badges: 
|
Well I didn't want to sound cheeky, sorry if it came across that way! But surely you see my point, and you guys should be told about these things?
I can't understand why you're asking me to switch to that client, other than 'who knows it might help', and I've reasons not to. And I can't understand why switching OS would appear to have a fairly large effect on an application that basically does a lot of maths over and over, which should have pretty much nothing to do with the OS. It's not like it depends on some wierd graphics driver or directx / open gl version or anything like that. Same goes for whatever app happens to kick the application off. From what I've seen (I certainly need to confirm this though) HCC doesn't seem to be suffering this problem. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Anyone else running Win7 and FAAH?
Have you noticed a substantial lengthening of runtime? Do you have all the boinc folders turned off from your AV scan? Have you added FAAH to your Trusted programs (or whatever your AV calls it)? Surely there is something odd as your HCC times seem reasonable but the FAAH do not (especially in comparison to the HCC). I would think that if it was systemic that there will be someone posting soon with a brief recap of their latest results. |
||
|
|
Highwire
Cruncher Joined: Aug 18, 2006 Post Count: 39 Status: Offline Project Badges:
|
AV : Not turned off. I can't remember turning any exlusions on for that computer on earlier installation. Is same AV. I'll check. [Edit it's a Free AV and doesn't have exclusions on it - I've turned it off completely for a while to see if things speed up.] I did have a look today to see if I could tone anything else down but couldn't see a setting.
----------------------------------------Doubt it would make THAT much difference? But I'll try that. What's interesting is the estimated to-end times seem ramping up, I'd kind of assumed they were just a simple benchmark * X thing, but maybe they take past record into account. Yes the HCCs seem fine. I've since properly compared HCC results to P4 computer that runs both tasks, it's much more reasonable in both time and credit, so I get feeling it might be just FAAH not a general BOINC thing. Searching web for similar problems seems pretty sparce so maybe IS just FAAH .. somehow. I enabled kernel times in Task Manager to see if it's getting thrashed and it's not. If it was just wrong credit that would be one thing but the implication is that it's crunching them a lot slower, which is far worse. [Edit 1 times, last edit by Highwire at Feb 1, 2010 9:05:22 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Just something to compare to, both from p4- 2.1-2.4 GHz machines running Win XP. These results are from 2 differant boxes in my cabbage patch (not big enough to call a farm..... YET)
faah10451_ ZINC19681765_ xMut_ md07200_ 02_ 0-- 607 Valid 1/27/10 20:21:52 2/1/10 05:26:20 10.29 129.1 / 120.6 faah10479_ ZINC19772274_ xMut_ md07200_ 00_ 0-- 607 Valid 1/29/10 04:26:38 2/1/10 04:54:10 12.12 154.5 / 149.4 |
||
|
|
Highwire
Cruncher Joined: Aug 18, 2006 Post Count: 39 Status: Offline Project Badges:
|
Fredski : Yes, as I mention above I've a P4 2.8 (non HT) myself that's sometimes on (and is now, as 'heating' :)) and it's e.g. 11.54 98.5.
I'd a theory that maybe they allocated 'big' tasks to more powerful machines but Ive since thrown that out. I didn't check before installation what old times were for FAAH but .. 8 hour rings a very faint bell? I could boot back into 32 bit but I upgraded for a reason on that machine (I'd rather not have had to!). I might do that though later in week for a bit though. |
||
|
|
Highwire
Cruncher Joined: Aug 18, 2006 Post Count: 39 Status: Offline Project Badges:
|
Well I disabled the AV and let it run all night, the FAAH tasks didn't speed up one little bit.
Given the HCC are fine (in fact UNDER claiming), I've been running the computer 24/24 last few days and stats are STILL going down, I can only assume that my credit claimed IS ball park correct - because the FAAH tasks are taking sooo long. Which means, if you factor in the new lower total credit has 'good' HCC credit mixed in, the FAAH are taking 1.5x - 2x as long and would be even worse if I was just running FAAH. Recent results: Hours Claimed/Given 15.49 263.2 / 148.3 12.49 210.4 / 110.0 21.31 346.6 / 165.4 12.88 207.1 / 138.7 13.62 222.7 / 145.9 23.39 376.2 / 244.7 This isn't just about credit / points measuring, it points to a lot of wasted CPU? I've a pIII 600 that can knock out FAAHs in 25.3-28.3 hours! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
your run times appear waay to high to me, I also have a Piii/600 and like you say, 25-28 hrs there is normal, my 2.4 ish boxes are doing 10-12 hars consistently. Not running any W7 boxes, but have you looked at task manager, under performance to see where the cpus are really running, or may get a clue as to what is eating cycles, your times vs mine it appears that you are only running at 75%
|
||
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: 
|
This isn't just about credit / points measuring, it points to a lot of wasted CPU? I've a pIII 600 that can knock out FAAHs in 25.3-28.3 hours! Check your power-saving-settings in windows, make sure you're using the correct settings, and especially check "advanced power savings", "Processor power management", and make sure both minimum and maximum processor state is set at 100%. Also check the BIOS, in case there is something here that interferes. Also make sure there's not some drivers or utilities installed by your mainboard that somehow screws-up things for you. It is likely possible some of these mainboard-utilities will throttle the cpu to save power, even win7 is configured to use 100% cpu... Hmm, since the system-temps seemed ok it shouldn't be throttling... but if nothing else pops-up double-checking this also would be an idea. Another idea is to run something besides WCG to measure the speed, since let's face it, WCG-wu's can be "all over the place" as far as run-times goes... One possibility is to use another BOINC-project to do this testing, probably a fairly good choise at the moment is SIMAP, since they currently has work, and each wu takes roughly 1 hour to run it shouldn't take too much time to run a couple tests. Or, you can use something like example GROMACS Stresscpu2 and run a test for example 1 hour under both win7 and your old OS, and see if there's a huge difference in how many floating-point-operations each has done in the same time... Just remember, Stresscpu as default runs at normal priority, so in case the problem is only happening when runs at idle priority, like BOINC does, you'll need a tool to manually change thread-priority on the threads...  "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
Highwire
Cruncher Joined: Aug 18, 2006 Post Count: 39 Status: Offline Project Badges:
|
Re advanced power settings, max was 100% min was 5%. As to what minimum means, quick search ofthe web and it appears .. nobody knows?! I can't think myself what that means. Max is obvious to anyone familiar with speedstep etc, but minimum??
Let's see if makes difference in remaining task but I've switched over to just HCC on this computer which appears well behaved. Want to see if credit goes shooting up. I doubt it's throttling? It's not a laptop, it's on power, and as I say (and in answer to other poster) it's on 100%, all 4 cores at max, and temp is sort of temp it was before - and I pay attention to it as a little overclocked. I'll lower to normal again before summerm, winter brings out the overclocker in me ;) What gets me though is I thought the credit was clever enough to work out if you weren't using max CPU, otherwise it'd be easier to 'diddle'. I.e. credit for a job taking 10 hours @ 50% would be same for one taking 5 hours @ 100%. And as I say, HCC credit is fine. For example it's whacking them through (at same time as other cores are doing FAAH) in half the time the P4 is taking, BUT, the P4 is faster at doing the FAAH tasks. THAT is a bit of a clincher! |
||
|
|
|