Discussion in 'World Community Grid (WCG)' started by Chicken Patty, Feb 20, 2009.
I'll use a 7950 for an FX8350 build, the other will fold
Reaching for 3.6GHz again... IBT doesn't crash or fail, my computer simply turns off. Temps too high? They were up in the low 80s... I dusted my case out a little what I could with a can of compressed air (I really need to get a shop vac)...
To the 32/64 question I'm blank but I see a minute decrease (1% or so) going from a fast CPU to a slow and less than that going from 16X to 8X . Going from a slow HD to a fast SSD makes not difference. Yes you need to reserve one core/thread for folding regardless of the GPU being red or green. The 7950s can make +110k PPD each so welcome to folding
mmm... I think I'll use my spare C60 board then.
I've seen some decreases on K8 systems folding on Tahitis, at least in the estimated PPD, about 20-30K loss per card. This includes a single core s754 3400+ 2.4GHz running one HD 7950 and a s939 X2 4400+ 2.2GHz with an HD 7950 + HD 7970, both running Windows 7 and 2GB DDR, even though the CPUs don't seem loaded much beyond the start of the WU.
Do you think using a lower GPU would make a better macth? I could use my HTPC's 7870 instead... or would the 7950 perform better even with the handicap?
I'm thinking about making a "folding box" using a slim mATX case I have lying around. The 7950s come with PCIe risers so I could install the mITX C60 board and the GPU lying on its back beside it.
Another question, does the folder GPU need to be the primary GPU? I would use the IGP to display.
It's worth a try, I'm honestly not sure why the K8 systems handicapped the cards when the CPU usage was generally low after the first few minutes. I wish I had a more scientific answer though, but when I moved the Radeons off the A8N32-SLI which was not crunching to a 3770K crunching at 87.5% with 4GB DDR3-1333 on an ASRock Z77 Extreme6, I instantly saw estimates of 250-270K PPD and have continued to do so, when the 7950 only estimated 90K on the 3400+/ASRock K8NF6P and the A8N32-SLI was hanging around 200K.
I usually use IGPs on whatever rigs I can, my dad's using a 3000HD on a 2600K while folding a 7770, mother's office computer is folding on two GTX 460s while using the Radeon HD 3300 (790GX IGP) and can't run AMD GPUs in there due to AMD's idiotic driver fork (though I could try modded drivers), etc.
I have a problem with getting new jobs on a fairly new rig (172 results returned). It doesn't matter what I put in as buffer value I keep getting the message - Don't need. Well at least I now know that the rig idles at 88 W with a 7970. Any thoughts?
Not sure what's going on there... I just dumped some work on the i3-2100 rig and had no issues ( 14 wu's reported, 14 wu's received). Did you try to remove the project then add it back in?
Thx but it didn't work. I tried four times, with different profiles (work, default ect) that include all the active projects. It is a shame since it was a good WCG chip but now it will rest until it get cold and then do some assistance to 2x7970 folding. A bit of overkill, I could have used my i7-920 for that.
Edit: Posted on the WCG forum and one guy suggested to copy the start-up log to same. When I booted up it downloaded new work. Go figure.
All new installs takes a few days to land on a steady point/hr figure. This one started out very good (48.5 point/hour/tread for the first 10 "days") but is now down to 32.5. My e8400 @ 3.8 GHz does more and less the same. Hmm. 4.59 Ghz 4790k let me down a bit.
Might hit 3.6GHz... IBT is passing, OCCT is passing... at lower vcore than I tried before too! I raised NB and CPU VTT two notches, as I've done for higher FSB speeds in the past. I didn't think I would need to for 400x9 as I didn't for 400x8 but perhaps I do... we'll see after more extensive testing.
I win! I've hit 3.6 stable... as stable as stable can be anyways. The test of time remains to be taken but I have passed 100 iterations of LinX (maxed out problem size) and I have passed 10 hours prime95 blend (I chose LinX and prime95 as my testing tools over IBT and OCCT at some point yesterday during the testing process)
It took NB and CPU VTT +2 notches (1.41v/1.3v respectively I believe) and 1.4125v vcore to accomplish this without any sort of error, bsod, crash or anything. Max temp was 73c, during LinX. prime95 pushes it to the high 50s/low 60s.
To anyone who may have been reading my bantering about my OC adventures here in this thread and may be interested... this Q6600 is a G0, though it is a clunkier 1.3125 VID model.
There has not been a crash. Price is still $600. Gpu's are obsolete for mining in all currencies.
Pictures will come soon of a new rig I have sitting near me. I am not sure I will have time to properly play with it till the weekend though. Can anyone guess what it is? I know @ThE_MaD_ShOt can.
ok, GPU cryptocurrency minning crash... still insane prices. Got a Powercolor 7970 Vortex for $124 Going by the rest of the guy's auctions, he's parting with an X58 system so it wasn't used for crypto (not that that's a bad thing but one could need replacement fans).
Gonna switch the 7870 on my HTPC for a 7950 and the Vortex will go into my 8350 build. It'll be a nice color match with the Corsair fans
Can you get the X58 board for a good price?
All his auctions ended about the same time so no.
Desktop's down, I just installed the 280X, but the Thermaltake Water 2.0 Performer that has been slowly running dry is now, well, dry after leaving my computer off for just 15 minutes to swap cards. The pump/block is the highest point in the loop, so the pump won't prime anymore. Guess I need to go find my TPC-812 or a new watercooler, don't think I can really fix this one.
EDIT: Took a shower and decided to tilt my Define XL onto its back, and possibly broke the bottom dust filter, but it managed to prime itself after a few minutes and I was relieved when the pump grew quieter and finally stopped rattling. It buys me some time to figure out what to do at least, and at least the Haswell survived extreme throttling at 125c (horrific, I'm well aware, and likely shaved some life off of the chip).
Grab some distilled water and slice it up! Why buy a new car just because the old one ran out of gas?
No reason to give up on the cooler just because it needs to be refilled! Find the service port, add coolant/distilled water, and put it back to work.
Plan B- RMA the thing and get a replacement.... there's a Tt rep on the forum that's helped out a whole bunch of members here
Do Asetek AIO coolers have service ports? Also, I am not the original owner and I'm using different fans, just threw out one of the original fans as it seized up. It would be neat to refill it, though I just have Prestone yellow car coolant available right now, and not sure how I can safely remove the existing tubing and put it back on.
Could do what I did with an old cooler. Slice the cable down the middle, stick in a plastic tee, plum it up to a reservoir.
Edit: That's a big ol' negative on using car coolant, from what I hear. Someone may say otherwise and if so, they're probably smarter than me, so trust them instead. It does have some things you want in your loop, like glycol (anti-bacterial?), but I think some of the other things are not as good.
Well, before I do anything with the cooler, I'll need to get these things figured out, like the tubing diameter and what liquid to use (is plain distilled water fine, or where would I get replacement water cooler grade Propylene Glycol?)
Talking about glycol made me a bit curious. I found this: http://www.engineeringtoolbox.com/ethylene-propylene-glycol-d_904.html
WalMart has a bottle of what appears to be Ethylene Glycol in distilled water for ~$25. I'm currently looking for a less expensive alternative. Distilled is typically just fine, but you might get some bacteria growth. PT Nuke is the generally accepted option followed by a silver kill coil. Water wetter is the other well liked addition, which where the glycol comes in. I don't think the glycol has any antibacterial properties.
Speaking of which, I need to clean my loop. I'm running straight distilled and I've ended up with more discoloration than I'd like to admit. Took about 6 months.
Edit: Google-fu found these, so I am eating my words
Commercial products have mainly glycol and say to mix with distilled water.
Ethylhexanoate is.. uh.. Wiki for hexane says "They are widely used as cheap, relatively safe, largely unreactive, and easily evaporated non-polar solvents." ..so, uh, okay.
Para-hydroxybenzoic acid "is isomeric with 2-hydroxybenzoic acid, known as salicylic acid, a precursor to aspirin." so you can, I don't know, drink it if you get a headache.
Diethylene glycol, "A dilute solution of diethylene glycol can also be used as a coolant; however, ethylene glycol is much more commonly used. Most ethylene glycol antifreeze contains a few percent diethylene glycol, present as an inadvertent byproduct of ethylene glycol production."
The main point being commercial solutions shouldn't be too terribly mysterious.
On my first watercooling loop, I did use regular antifreeze. It was an aluminum block and a copper rad, so I used it partially for corrosion prevention. IIRC, some Tt or Zalmn watercoolers do used mixed metals, so do a little research on yours.
I seem to be rock solid... I've been running Handbrake all day and that has failed to crash the system. I sort of use Handbrake as a test because it's crashed my system before. I may or may not have used IBT to test it at that time, but if I did I'm sure I didn't run it like I ran LinX this time around (maxed out ram).
I used Handbrake to convert a project I've already done, but the first time I had it set to very fast... this time I have it set to slower. There are 7 pending episodes left to convert. I'm gonna let it finish since I'm curious to see what differences I can spot between very fast and slower... not much quality wise I'm sure since I'm coming from already compressed AVI files anyway, but I'm curious about filesize mostly...
Since the system has passed my tests it'll be time to restart WCG soon... at 3.6GHz. Better than 3.0 or 3.2... I'll make up for lost time eventually with the boosted speed.
Separate names with a comma.