Printable View
Mabyboi ... I am running an i7-920 at 4.0 GHz which helps and sometimes it is days between when I actually plug in a video cable and look at that machine :-)
I am also running it with HT off anf most importantly, because this is a dedicated crunher (cpu doing WCG) I am running Windows XP which helps reduce runtimes by ~20 minutes per wu.
Hey Otis,
Just a little update.
My RAC climbed a little bit to 26650 PPD. I haven't changed anything on the clocks.
Looking at the graph in BOINC managaer it seems that the average is still growing with a small rate - max not reached. I guesstimate it will level a little bit above 27K.
just got my air back, gonna let the desktop sit and accumulate points now :D bumped the clock up to 717mhz too.
Otis11 ... :toast:
I may be old but if you beat me hard enough with a stick long enough I can learn new tricks :up:
You got me thinking about the interaction between the GPU components and I realized I was blinding myself with the *shaders shaders shaders* mantra.
While not for the same reasons you state in your ratio theory (harmonic distortion, frequency collisions ... dude... what??? not saying you are wrong, I'm just not smart enough for that), I decided to see if I had my shaders so high it was preventing my memory OC which in turn was starving the very shaders I think are the catz azz.
Damn , damn, damn ... I gotta listen better to you whippersnappers ... I turned my shaders down one notch (1620) and my memory up (1215). While I did get a couple of errors this morning I have since been returning a little FASTER than before (~10 min per WU). So yes, utlimately I think ratios are important and will continue working on finding the sweet spot for my cards.
Since upping my core clock from 707 to 717 i've gone from 6k PPU up to 6.8K PPU! a whole .8 from that little?! WHAT DID I DOOO :shock:
which PPU are you referring to?
PPU Pay Per Use
PPU Physics Processing Unit
PPU Palestine Polytechnic University
PPU Power Processing Unit (solar electric propulsion)
PPU Picture Processing Unit
PPU Public Protection Unit (UK)
PPU Price Per Unit
PPU Point Park University (Pittsburgh, Pennsylvania)
PPU Professional People in Urbe
PPU Portable Pilot Unit (Microsoft)
PPU Physical Planning Unit (Grenada)
PPU Procurement Policy Unit (UK)
PPU Peoples Potential Unlimited (record label that reissues vinyl records)
PPU Portable Production Unit
PPU Pending Pick Up
PPU Prime Power Unit
PPU Packet Pick-Up (racing events)
PPU Please Pick Up
PPU Peripheral Processor Unit
PPU Protocol Processing Unit
PPU Pre Purification Unit
PPU Post Processing Unit
PPU Pre Processor Utility
PPU Postpartum Unit
PPU Pixelpushers United (pixel art commune)
PPU Primary Place of Usage
PPU Preproduction Unit
PPU Production Prototype Unit
PPU Powerplay Unit (hockey)
PPU Programme de Prévention Urbaine (French)
My programme went up by .8! but then i realised its cause of the new boosts...Quote:
Originally Posted by Snow Crash
So you meant Points Per Unit.
Not only are they changing because of the new bonus scheme but they vary between WU types anyway.
That's why I try to always test from the same *type*, they are usually consitant and not just in runtime but also in how they use the GPU itself.
Riddle me this: If everyone says that GPUgrid is the toughest thing for your gpu, and if its stable here, its stable everywhere.. and my 717/1600/1233 is stable here.. but then when video gaming. it crashes... why is that? it does the driver reset thingy and does half of what its supposed to..
Mabyboi- maybe because the work the Grid does is consistent; my own experience game graphics are dif kind of intensive
GPUGrid and F@H use the cards differently than games do. I had a 260 that was fine in games and certain F@H WUs but not GPUGrid and most F@H WUs. I also think F@H is much more demanding than GPUGrid on most WUs. GPUGrid does not generate as much heat either.
GPUGrid also only stresses the Shaders... sure it uses the other stuff, but they don't feel the pressure. Games make the core work and can even max the memory - Two things Grid will never do.
I do maintain though... If you're GPUGrid stable you're pretty flippin stable.
and my games are coming out crappy, even at stock settings... :shrug: dead card?
Ya might try CUDA Memory Tester. Tests GPU memory. Let it run like 1000 or more loops.Quote:
Originally Posted by Mabyboi
Quote:
Originally Posted by PoppaGeek
When gaming, it makes parts of the screen look like they're artifacting, and tearing.. its not good... would this help?
Artifacts a good or I should say bad sign of memory problems. OCing can cause artifacts. The tearing may be syncing. Try setting game to sync with monitor. Nvidia drivers call it "Vertical sync". Set to yes and also in game. If you can limit it to a little less than htz of monitor. My LCDs are 60HZ so i set game to 56 and sync frame or horizontal.
How many results completed do you want to get an average ppd?
I'm seriously impressed with this so far ...when I first used my 295 on here I was getting around 36k/day...from 4 results today I average over 31k on a 260 55nm
PPD just started to settle in after the last new app but bonus points changed yesterday :up: so they are changing again.
On top of that they just put up a new beta (6.16 for windows only) that is suppose to be even faster still :shocked:
PPD are going to in flux for quite some time. All moving upwards so it's all good :party:
Yeah, unfortunately because of that, ppd is not a good gauge of how effective the OC is... :rolleyes:Quote:
Originally Posted by Snow Crash
So any ideas on how to do this strategically? Test the effects of OC?
I try not to use PPD to gauge how effctive my OC test settings are for two main reasons. 1. It simply takes way to freakin long for PPD to normalize again after changing settings and 2. I'm just not that patient. I typically look at runtimes of individual WUs of the same type within a given application version, then I try to see if I can get a couple in a row to get a better sense of an average. This way I can use WUs I already crunched as my baseline and realisticaslly only need between 1 and 4 WU of the same type to see if/ how much of a difference I am making. I will say there is one downfall to this approach and that is that you are onlyntesting against one WU typre and they are all a little different so you might not be as stable as you think you are until your run a couple of each.
So it should be updated... Please let me know if I missed something, cuz that is VERY likely... Tried me best though. :p:
Also, anyone figure out anything?
As far as the timing issue goes - I talked to a researcher about it... He said the OC is only going to help if that component is being stressed as newer cards have advanced interfaces between the different components. But it might have an impact on older cards... Not that we have many of those... (idk where that distinction is though...)
Hey Otis,
Since we get more points for the WUs I think we can scratch all PPD/RAC from the first message.
My latest OC success: 702/1656/1296; fan 55; temp 70 (GPUGrid stable :up:, but not FurMark :shakes:).
Points will need some time, because the machine will be on/and off (so no 24/7 right now). But from what I see it's maybe some 3% faster than just using the shaders. Only time will tell.
Considering your previous message, I really think to revert back to stock and just leave it running at that speed.
Cheers
If anyone is interested, my 295 is now up for sale in the for sale section: http://www.xtremesystems.org/forums/...d.php?t=248061
i wasn't sure if i should use EVGA precision or riva tuner, so i picked evga precision for the sake of simplicity, and it also works my zotac gtx260 :lol:. either i got lucky or this software is not exclusive to evga cards only.
so i got lazy and ramped up my clocks to Otis11's max OC on his gtx260. i'm now crunching at 700/1507/40, fan on auto at 40% 60C for 24/7 load win7 64bit. so far, 2 WU's has been complete at these clocks and they are both vaid. i'll report back if i get any errors. thanks Otis11!
Huh, good to know Precision works with zotac... i think you just got lucky, but hey, if it works. :up:
Ya it did not seem to work with my Gigabyte card- though funny thing I was able to OC the fan?! So I used RivaTuner, took me a bit to get used to it but it does cover everything, I must say Precision is very simple
I've been doing a little research and came to the conclusion that all 200 series should work at 685/1477/1160 and most should work at 702/1512/1188.
Anyone willing to test that statement and validate it for GPUGrid? I know it works on mine! (especially need validation on 250's and 295's as those are the extremes)
If we can prove this, that's a really good note for our team as everyone at stock willing could safely get a 19-22% boost in output.
Also, need updates on the RAC of OCs as the points have changed...
I'm gonna guess that Precision will work with any reference card cause they really are all the same, just different stickers and tighter bins.
dam, i may have spoken too soon. 700/1500/1180 might be stable just for GPUgrid, but it's not stable for 3dmark vantage. now i'm having doubts as to which application is more demanding.Quote:
Originally Posted by Otis11
i think a good compromise would be having high shader clocks for GPUgrid, while only having moderate core and memory clocks to satisfy everything else. Otis11, you pointed out earlier in the thread that it's probably best in terms of stability and clocking to keep the shader and core linked, so hopefully 685/1515/1400 would work for all applications (moderate core OC, high shaders, stock memory). i'm testing that out right now.
edit: it's also a pain in the arse every time 3dmark vantage gets errors, because then the GPU downclocks and i can't seem to bring it back up to even stock speeds unless i reboot the computer
1400 on the mem?!?!?! Are you kidding? Is that will double data rates or before you consider that? The highest I've gotten mine is 1188... haven't pushed it, but 1400 seems really high.Quote:
Originally Posted by WhiteFireDragon
And my clock is rock solid on any test i've done... pretty sure that included furmark.
If you have time, would you mind finding out what part causes it to be unstable? Would 685/1477/1160 work on it?
oops! sorry typo, memory was stock 1050, not 1400 lol.
yeah i'm trying to find the edge of stability for all applications right now. i just tested it with 685/1500/1050 and 3dmark vantage still errored at 1440x900 resolution, so it's either the high shader clocks or the moderate core clock since memory was stock. 700/1500/1180 passed vantage at standard resolution though, but with higher res, it'll fail. i have the zotac AMP edition, so i kinda expected it to clock a little higher than reference, but i guess not. if i have enough data test points, i can make a small chart. i will keep editing this post with my own personal results.
GTX260 (216)
700/1500/1180: GPUgrid *stable so far* .... 3dm vantage- stable 1280x1024, unstable 1440x900
685/1500/1050: GPUgrid *stable so far* .... 3dm vantage- stable 1280x1024, unstable 1440x900
680/1485/1050: GPUgrid *stable* ............. 3dm vantage- stable 1280x1024, stable 1440x900
680/1500/1050: GPUgrid *stable* ............. 3dm vantage- stable 1280x1024, unstable 1440x900
680/1490/1050: GPUgrid *stable* ............. 3dm vantage- stable 1280x1024, stable 1440x900
700/1490/1050: GPUgrid *stable* ............. 3dm vantage- stable 1280x1024, stable 1440x900
715/1490/1050: GPUgrid *stable* ............. 3dm vantage- stable 1280x1024, unstable 1440x900
715/1485/1050: GPUgrid *stable* ............. 3dm vantage- stable 1280x1024, stable 1440x900
the last line is my final stable clock where it's crunching at.
So what happens if you keep the core at 680 and push the shaders up to 1500? will that pass?
If not we know where the error is. Also might try 1485 shader and 702 core... if it passes both of those, it should work at 702/1500, if not, you know which one is the error.
:up:
What I found was that both evga and xfx 260's will fold/grid @ 1620 IF the ambient is below 22deg but because that temp is on the cusp of normal house temp it is best to drop 1 step to 1584 (this with 100% fan and core @ 590 (586 real) and mem @ 999) then again I don't game ...all settings/readings from precision
my results so far answer this, it's the shaders. 1500mhz shader will not run, but 1485mhz will. i will lower it by 10mhz only, now testing at 1490mhz shader, and i think 1490mhz - 1495mhz will be the max.Quote:
Originally Posted by Otis11
Ok, well your shaders only run at 1476(1468-1494) or 1512(1495-1532)... you can change the slider to whatever you want, but it really runs only at those values, so bumping by just a few notches won't make a difference.
OC, how'd you get them so high? bumped the volts up something serious?
i have a gtx295 co-op running at 740mhz core and whatever the corresponding linked shader is. the fan is on auto, which is around 50%, and temps are around 63C, and on stock volts. what is the max you guys are getting with these 295's? ironically, this actually worries me because it's hard to believe i can run these high clocks stable without touching the voltage. i have not tested it higher, but i think it can even OC higher. what is the default voltage? i'm just afraid as i increase the clocks, the card might be overvolting itself.
OK, so by my guess you're at 1584 for your shaders... which is kinda high on stock voltage, but not unheard of.
Mabyboi did manage to pull it up to 1600 on shaders before getting majorly unstable... What are you using for your tool? The first page has some good ones if you need them.
Edit: and I should really update this thread - anyone who would post the info I'm going to clear the point values and start again. Any and all info welcome!
well the big difference is that he watercooled his. this one is bare stock cooling with no tweaks at all, fan only runs at 50% on auto, only tool i use is evga precision to OC the card. usually i'm worried with my CPU or GPU can't OC for crap, but in this case, i'm a little skeptical it can OC so high. when i get the chance, i'll trying maxing it out to see how high it can really go.
My EVGA 295 (vanilla) is rock solid stable at 1518.
If I bump to 1138 with EVGA voltage tuner I am 1656 stable.
Some GPUGrid WUs are less stressful than others but if you let it run a few days you will probably hit each different type and claim stable :up:
I always double check the WUs that fail against the results from other machines to make sure it is not a bogus WU but those have been quite rare lately.
Sorry if this is OT but ive noticed that the max gpu usage i see thru boinc is ~73% it hovers consistently between 67-73 , is there a way to increase this to increase productivity. I am reporting gpu usage numbers from precision.
EDIT - scratch the above i just noticed that one gpu 2 is sitting at 50% ~ . Im guessing it varies from WU to WU.
different WU's will vary, but XP will use between 70-95% of the GPU, while win7 will only use 50-75%. that's why XP will consistently give you faster WU completion times.
giving 1500 shader a go.
Shaders on the 295 move in "straps"
Setting 1500 will really move it to 1512 (I typoed in my last post)
If you get errors there move down to 1476 and see if that is better for your card.
Good luck, let us know how it works out :up:
I have the setting still with core and shaders linked and Voltage already set at 1.138 , I know I am going to use that since I do want to see how much I can push out of this , temps are going to be a limiting factor ultimately.
I think with lower volts and unlinked core (1500 shader 600~core) I am getting errors so I thought wth and bumped volts and kept them linked for now.
Hopefully I have the patience to run like this for a couple of days before trying the next step.
Core doesn't help in GPU crunching so you can unlink and save the power and heat, did I post that using an extra fan on the backside of the card helps tons???
So you mean a fan on the outside to pull air out ? makes perfect sense it puts the gpu heatsink in a push pull config
I have at other times benched with various fans strapped to it :P , I am also waiting on a shipment of 2 case fans which will pull air out from above the gpu (side panel fans the one that will do the work is a 80mm silverstone that can do a 80 cfm if i need it to the other is a 120 mm thin scythe nothing bigger will fit atm)
I just checked out how fast your 295 completes tasks and if I am correct it shows half the card at a time , its outscoring the 285 (half a card) you have, guess XP has a huge impact on performance , they should really optimise it more for win7. Oddly the averages dont reflect the superiority of the 295 system > 285 , I am missing something obvious :D
I cant go back to xp now I use this rig all the time (might try a dual boot so before I sleep or go out for my classes I make it crunch as efficiently as possibly) im sure they are working on making it as efficient on 7.
I will take your advice and drop the core I just feel with my card if I drop the core it will error out , I have not tried dropping the core since increasing the volts .
Silly me it seems I HAVE dropped the core a tad bit :P it is at 670c/1512s/1100m (precision settings) was 700c earlier.
The next shader strap that is worth trying for is 1600 , not sure if it can do that with current volts.
Could you suggest how far I can drop core for 1512 and 1600 shader settings. I was thinking 600 core for 1500 and 670or700 for 1600 though pulling off 1600s on this card seems unlikely at this point in time.
Edit - gonna try out 700c/1620s for a few hours and see how it goes.
Actually I have my fan blowing against the side at ~60 degree angle from the pcie plugs pointed towards card's exhaust vents (typically the back of the case but I'm not running in a case which helps keep temps down).
I move my cards around to much to have any of the stats make much sense.
The current difference you are seeing between the latest runtimes on my 295 and 285 is that the 285 is downclocked during the day because it is upstairs in my study (I'm not spending extra money on more AC than I really need to) and the 295 is in the basement which is cool enough so no AC used there. I just moved the 295 so am running it at 1476 which is 1 strap below my stock volts max (1512).
I'm thinking I might move the 285 onto that same board, breakdown the 480 and re TIM it all with IDC to see if that helps. Anyone know if leaving the shroud off entirely helps?
The optimization actually needs to happen at the driver level between NVidia and MS because the driver mode changes that MS made at the OS level, GPUGrid can't do anything about it.
Because these cards are used for crunching and daily driving (I'm not a gamer) I use Precision to underclock the core as much as it will let me (just pull the slider as low as it will go) and I have never had a problem with stability because of it.
Upping the core freq will only create more heat which in turns brings the max OC of your shaders down ... dont do it :rofl:
I seem to crash if I dont have the core and shader somewhat in each others vicinity , this is probably because of the straps mentioned in the first page of this thread.
Damn you have some insane cards for someone who doesn't game though I should get used to that this being GPUGRID :P
I too have played with fans blowing on the side of the card it helps all right. But the fan i have now for that is too darn noisy , I sleep in the same room and when I used to run that fan my ears were buzzing so I had to kinda stop that. Im hoping to get my fans in a couple of weeks I dont think it will help a LOT because my ambient temp is always close to 30 , card runs at high 80's and low 90's depending on the day. 295 temps were pretty exciting this summer :P
:rofl: Otis11, were you too tired when posting this? wrong thread buddy, but thanks for the update
Zloyd, i think you can up your gtx295 a bit higher. my gtx295 has shaders at 1585mhz, and my gtx275 has shaders at 1656mhz right now. all my cards are stock cooling and voltages.
Damn ive noticed that i lose work when i stop boinc and re-enable sli to game. I lose 1 gpu thread it errors out and starts on a new work unit :(
I usually right click on boinc and hit exit before re-enabling sli and I disable it again before i start boinc.
But the last 2 times its been crapping out. I think I will try a system shutdown and then restart to enable sli . Anyone else notice this issue ?
Are you getting the exit confirm dialog, are you choosing "Stop running etc"?Quote:
Originally Posted by Zloyd
Sometimes it takes a few for the drivers to reset so maybe wait a minute or two between switching back and forth.
I was not getting the exit confirm dialogue have now enabled it.
Still getting errors , also dont seem to be getting a improvement in performance while crunching. Am I running a bad OC ? I am comparing to Snowcrash's and Whitefiredragon's 2xx cards but they are on xp64. Wonder if this is just win7 32 holding me back.
yeah your errors are most likely because of the overclocks because the error happens after a few minutes of the WU. if you get an error in under 15 seconds, the it's most likely because of the WU, but this doesn't seem like the case. lol it's a bad idea to clock your cards to try and match my numbers. my 295 is most likely a better clocker than the average 295. i'm running it at 1585mhz shaders with this co-op version, and XP also gives a good boost to the GPU utilization. if you switch to XP, you WU completion time will closer to mine. i have a regualr dual pcb 295 coming in soon, so i'll try to max out the clocks to see how it compares to the co-op version. maybe that one will be a better comparison to yours. what's your max stable OC?
btw, what's the difference between the gtx285 vs gtx275 in terms of crunching? both have same shader cores, with the 285 having barely faster shader clocks. any kind of shader overclocking will cancel out this advantage, and memory bandwidth and capacity does not matter, so is it safe to say a 275 = 285 when crunching?
Im thinking there is no way of really telling if I can come close to those numbers as long as I run 32 bit. Dont think im going to get close to those numbers with 32 , irrespective of the oc the numbers that are being churned out are more or less similar. Im not sure how much moving to win 7 64 will help this.
Where can I see projected gpu stats ?
Sorry White I didn't answer your question last time round but my stable gaming oc is 626c1100m corresponding shader for crunching I'm playing with the shader now but before I try something blindly I want to check my oc against a projected stat or something. I saw a page in the 480 thread or somewhere cant find it.
Have run into some driver issues. I cant get my 295 to sit at 3d clocks when I disable SLI , it used to happen before quite easily but now its a major pain in the arse and if I try to set the power management mode to max performance it goes to SLI on its own it seems.
I have tried driver sweeper but with no success , I have retried it last night and installed the latest whql drivers but the same symptoms remain. Any suggestions short of a complete reinstall ?
When the clocks drop does the WU error? - Might be clocked to high...
That's the only thing I can think of... Not familiar with 295s
Nope its happening at stock clocks. Its dropping even before i can start boinc. And once i start boinc only 1 gpu clocks up and accepts work the other is just idling.
hmm weird problem. try enabling SLI and run a few stress test (3dmark and furmark) while monitoring the clocks and GPU usage with evga precision. if this seems fine, then disable SLI and crunch on it to see what happens.
Hey guys, was just updating this and needed some information.
Since points seem to change quickly as they update the client, I axed that portion of the entries, but I'd really appreciate some more numbers for the cards we're running here at XS. (Especially the 240, 250, 270, 280 and all 4xx)
I also need a stock baseline of PPD for every card if possible and a PPD for the Strongest OC you've got (trying to get a range of what can be expected - but please don't exaggerate, this is all to get max ppd/$ for any given set up. People may use the info you submit to choice what to buy)
This is all the info I need:
Core/Shader/Mem - Your Name.......... PPD
.......... Brand: xxxx.......... Cooling: Stock fan @ xx% .......... Voltage: Stock .......... Temp: ???
Finally, does anyone know the steppings for OCing the 4 series?
Thanks for the help guys!
i'll try to posting results on all my cards, but i don't have a lot of results for baseling PPD, none of my CPU's and GPU (except one) are running stock :lol:. here's the first results on the minimum at maximum for my gt240's:
this is the worst case scenario for a gt240:
- Galaxy gt240 ddr3 512mb
- all stock clocks and voltages
- fan auto (40%) at 58C load
- 15.7k PPD (taken from RAC over a couple months of WU)
- notes: win7, WCG also loaded, machine used for daily tasks also
best case scenario for gt240:
- EVGA gt240 ddr5 512mb
- 405/1595/1701 clocks on stock voltages
- fan 85% at 67C load (temps a little unfair, it's sandwiched between two gtx295's with almost no room to breathe)
- 20.1K PPD (point projection using many different WU's types)
- notes: winXP, WCG not loaded, but it's sharing resources with 4 other GPU's in the system. you may get a little more if the GPU was by iteself
will look at stats for my other cards tonight
Updated. - also added OS for reference.
Thanks for the info. :up:
GTX 460 MSI 1gig DDR
800/1600/1800(mem stock)(volts stock)
Vista 64, Fan on auto - 59% with GPU @ 65% usage temp 48c room at 25.5c
Almost 30k a day but some days I play LOTRO for 1-3 hours.
Not using SWAN_SYNC, 6 core running WCG.
Thanks :up:
Thanks for all you do for the team. :clap:Quote:
Originally Posted by Otis11
ok, i put one of my gtx275's at stock clocks for a few days just to test for the lower end of PPD for a gtx275.
this is the worst case scenario for a gtx275:
- evga gtx275 ddr3 896mb
- all stock clocks and voltages (648/1458/1188mhz)
- fan auto (40%) at 71C load
- 34.7k PPD (taken from PPD projection averaged from several types of WU)
- notes: win7, WCG also loaded, machine used for daily tasks also
here's the upper end for my other gtx275, but not maxed out yet:
- EVGA gtx275 ddr3 896mb
- 468/1620/1134mhz, on stock voltages
- fan 80% at 69C load, in a small mATX case
- 42.0K PPD (point projection using many different WU's types)
- notes: winXP, WCG also loaded, sharing resources with another crunching gts250 in the system. PPD will be a little higher with HT off, no WCG, and single GPU
will look at stats for my gtx295's, gts250, and gtx260 later. btw Otis11, you have a typo on the first page. the gts240 you labeled is not retail parts you can buy, and it's based on the older g92 chip. i think you meant gt240. and brand for my second upper end gt240 is evga, not galaxy. there's also no such thing is a gtx270 :p:. i know these aren't important typos, but still might confuse some people trying to pick up a gts240 or gtx270 and can't find anywhere selling them :lol2:
I was wonderin' why everyone was doing such a poor job of gettin stats for those. :ROTF:Quote:
Originally Posted by WhiteFireDragon
Thanks. Anything I can do to make this better is always appreciated. :up:
I've clocked my 275 a bit. Just upped the shaders and downed the core clock. I haven't touched the memory - there is no point right?
How do I find out expected ppd?
485/1630/1152 - ??? PPD
OS : xp 32bit .......... Brand: BFG.......... Cooling: Stock fan @ 65% .......... Voltage: Stock .......... Temp: 70
Good work :up: ... and you're correct, there is not much gain (maybe 1% at best) for OCing memory and you're running the best OS for GPUGrid :yepp:Quote:
Originally Posted by Kryja
As for expected PPD ... that's kind of tricky, some WUs point better than others and some (although not recently) are double length so run a really long time and don't point particularly well, how stable is the card long term, it this a dedicated cruncher, etc. A ballpark firgure based on your last returned WU (first one with the new shader setting) gets to about 43k per day, however those WU types point the best so you might want to lower your expectations to about 40k per day.
Cool thanks Snow. I've upped the fan to 70% now, gpu core is sat on 67. Nice little drop - just depends if I can sleep with it.
You think upwards of 1650 would be achievable?
2XX shaders move in straps so your 1630 is really running at 1620.
(see the first post in this thread)
The next strap is at 1656 which, if you are not volting up will be pretty tight.
Then after that move on to 1674 and kill em :explode:
Here is how I test ....
Suspend Network Activity in BOINC.
Suspend Activity in BOINC.
Stop BOINC.
Make a copy of your BOINC data directory.
Mess with your OC and Restart BOINC.
If you get a Comp Error stop BOINC.
Delete BOINC data dir
Rename your backup to BOINC
MAKE ANOTHER COPY (can you guess I've forgotten this step more than once :rofl:)
Repeat steps 4 and 5 until you are stable.
Once you are good then turn Network Activity back on.
This prevents you from pulling too many WUs and GPUGrid shutting you off for the day :up:
Let us know how high you can go :sonic:
KASHIF_HIV are the most demanding and TONI_CAP is about the average.
Cheers Snow :up:
I'll get to work after dinner and report back http://omg.wthax.org/smilies/emot-salute.gif
1656 shader is safe. Ran a KASHIF_HIV for a couple of hours and no errors. Halted everything, copied my BOINC folder and upped to 1669 and it errored my wu after a while. Wiped my boinc, coppied my backup in its place, still had the errored wu :shrug:
I'm happy sticking at 1656 as I don't really want to start pumping volts into it. Not exactly going to get pie with just a 275 now am I :rolleyes: Just happy to be putting some work out. Temps were @ ~70 with 70% fan. Good enough me thinks.
A 275 at 1656 is real nice ... you'll get some good point for sure :up:
1656 - on stock volts?!? Dude that's pretty sweet!
And do we have a ballpark for overclocking the 4xx series?
Sorry I had not posted ...
480s get interesting as each card has been hand set for their own default v at the factory so we all have a different base to start from
My base is 1038 (ouch, I've seen people bragging about much lower but you get ewhat you get:shrug:)
Anyway ... at 1138 mv I can go up to 1736
My sweet spot is 1050 mv for 1595 which I think is :up:
It is nice clock for 1.05v. My 470 works at 1530 at same voltage.
I am testing 1560 and it looks good for now.
http://img695.imageshack.us/img695/2...ture016.th.jpg
Is it possible to get help on a good and stable OC for two GTX 465 and yes they are wc, They are both MSI N465GTX-M2D1G . So far they have been crunching since I got them xD I feel like I need more crunching powah !!!, I barerly get to play on them because of college ( 9 hours ) .
Thank you
You can try to unlock it to a GTX470 to get some more points, possibly some PIE.:DQuote:
Originally Posted by LC_Nab
: O... I tought that this models cant be unlocked but hey, Ill try it, also yes Pie please !!! :DQuote:
Originally Posted by lkiller123
Galaxy GTX 480
803/1606/1900 - Plan.B - last 7 day average is 50,152 ppd
Volts: Stock
Cooling: watercooling, gpu only
Temps: 107 - 109 F loaded
This is a 7 day 24/7 stable loaded oc. I've just started playing with this card, but it overclocks nicely so far. Watercooling the gpu, with the rest of the card naked, dropped the gpu and pcb temps by over 20C.
Swan_sync on?
59.8k yesterday. Not bad for 470 in win7.
LC Nab: A pair of 465s ... I bet you will be able to get some nice points with those. What tools and experience do you have overclocking?
Plan B. - Nice .. we like stable better than fast :up:. I'm with lkiller123 in thinking you are not using SWAN_SYNC=0. We realize this uses a full CPU thread in conjuntion with your GPU and some people don't want to do that. Either way ... nice card!
roki977: Good score ... let it run for a week straight out and your average PPD will climb higher.
I have no experience at all on GPU ocing ( I know I'm a nub :( ), only tools I have is the MSI afterburner thats about it xDQuote:
Originally Posted by Snow Crash
That is all you need. Unlock voltage (if your gpu use Voltera regulation u ll be able to use volt. reg. ) controll and try ocing. What card u use?
Snow Crash, I would love to let my 470 run for a week straight but i own only one pc for everything..
Not yet. The setup with WCG running was easy without digging into the settings, so I just let it rip for a week or so to test oc stability and temps. The next step is set swan-sync on and start bumping up the oc.Quote:
Originally Posted by lkiller123
I'm using two of these VGA MSI| N465GTX-M2D1G, I have read about the unlock to a 470, But im skeptical about it, luckily my wc blocks just arrived, so I shall find out if they have 10 mem chips or just 8 .Quote:
Originally Posted by roki977
Update:
My 275 errored on a KASHIF_HIVPR wu at some point monday. I backed my clocks down to 485/1652/1152 monday evening, just a 4mhz drop on the shadders, and it has successfully completed another KASHIF_HIVPR wu and been going strong since. I'm still in the 1656 strap so it's all good :cool:
I think you can unlock shaders on every 465 regardless of mempry chips and only shaders are important for gpugrid.Quote:
Originally Posted by LC_Nab
http://img441.imageshack.us/img441/520/gtx465ocs.th.png
Under the water you can use 1.087v with no problems and you shoud get at least 1500mhz or more if you are lucky
Fermi loves cold. My 470 under ln2 with 1.15v goes over 1900mhz easy(1730 air), with stock mem. memory freq even more. If you overclocking for gpugrid leave mem. on stock.
Galaxy GTX 480
809/1618/1954 - Plan.B
Volts: Stock
Cooling: watercooling, gpu only
Temps: 107 - 109 F loaded
Swan_sync=0
Win Vista 64
Daily average over the last week is down to 42.5k
???
GPU usage as per EVGA Precision is at 65-66%
WCG is running, 7 wcg threads and 1 gpug
disabled hyper threading to increase ppd a bit
even without disabling HT if you tell Boinc to use one less thread than you really have then GPUGrid will use it and you will get much better scores.
btw, cpu is i7 960 (ES) at 4.0 GHz. RAC is climbing, up to 48k now. I probably haven't let it run long enough on the new settings to get a good daily average.
Hello everyone,Quote:
Originally Posted by Snow Crash
A couple of months ago I picked up a second EVGA GTX275 SC (648/1458/1188 clocks), primarily because I wanted to ditch SoftTH in favor of nVidia surround for my triple monitor setup. We finally got some cooler weather here, so the other day I fired up GPUGRID to see what the cards would produce. I ran them both between 8pm - 2pm for a couple of days (got too hot in the man-cave during the afternoons), then we had more cool weather on 10/2 and I've been running them 24hrs ever since. Looks like they're each averaging ~33k/day (while running 8 WCG threads concurrently), which is pretty much in line with my expectations .
But I'd like to see if I can get a little more out of them. Snow Crash suggested bumping the shader clocks a little, and posting in this thread if I need any assitance. Now in the past, I've briefly experimented with a few MILD overclocks on my cpu, but I've never tried anything on a GPU. In other words, I'm a complete n00b when it comes to this sort of thing.
:am:
I've done a little reading up on GPU overclocking; here's my current understanding of the process at a high-level:
Run GPU stress/stability tool (I have FurMark)
Run OC software (I have Rivatuner)
Unlink core/shader/memory clocks and increase MHz values in small increments
If artifacts/system lockups occur, back down to last stable clock value/restart system if necessary
Watch temps (keep below 80c?)
Is this roughly accurate?
Most of the OC guides I've come across seem to be oriented toward increasing gaming performance - they talk about stepping the core clocks first, then shaders, then memory. But in reading through this thread, I guess it's only the shaders that really matter for GPU computing. So before I try my hand at this, I have a couple of basic questions:
1. Is Rivatuner ok, or should I use EVGA precision? (I'm on 258.96 driver)
2. I'm assuming I should step the shaders according to the "Scale for G200b" table in the first post...correct?
3. I guess I leave the core clock alone, but do I need to bump the memory clock each time I step up the shader clock - i.e. do these need to be kept in sync?
4. After increasing shader MHz, how long should FurMark run before considering the frequency stable and moving to the next step?
5. Since I'm working with dual cards, should I OC one then the other, or both in tandem?
6. What kind of temps should I be concerned about for these cards? It's currently ~72F ambient and the cards are running 67C with GPU utilization at 65-67% running GPUGRID (which seems OK). I ran FurMark when I bought the second card and GPU utilization was 94% or more (I don't recall what the temps were)...what temps should I expect when running FurMark?
I'm very much a crawl/walk before run kind of guy, so I'm looking for a "mild" to maybe "medium" result (at least to begin with). Also, please let me know of any pitfalls I may encounter. Any guidance would be greatly appreciated!
:help::help::help:
EDIT: System specs -
ASUS P6T6 WS Revo
Intel i7 920 (stock clocks)
6GB OCZ3X1600LV6GK @1600MHz
EVGA GTX275 SC x2
Corsair H50
Corsair TX750
Regards,
1. I prefer EVGA precision - It has just worked better for me in the past. (But to be fair I had more experience when I used it)
2. Yes... And you can probably just jump the shaders to 1512. (1476 is you want to walk first :p:)
3. I like to bump the memory also and keep the core close to what it should be if it were locked, but that's just me. Many here leave the mem at stock and drop the core below stock.
4. Furmark - yeah I mean if you really want to. I have yet to see a card that can't do 1476, so I would recommend jumping there and testing, then going to 1512. Also, GPUGrid itself is a much better test of shaders than Furmark. I would just try it (disable network so you don't burn through your Queue though)
5. When I had two, I did both in tandem until I hit a problem, dropped both one step and bumped up each. (Turned out neither could get above 1512 without extra volts though... and one card couldn't overvolt)
6. I don't worry about Furmark temps...If it's below 70C on GPUGrid I'm happy. I even run up to 80s in the summer - but not for long stints. Only gets that hot during mid day. But then again I'm not very cautious about temps. I say it's going to be old in 3-4 years so I'll get rid of it before I'll ever kill it. :shrug:
I believe that's everything, but I'll go through this more thoroughly later. Feel free to point out anything I overlooked.
(And get that 920 OCed! :p: Vcor 1.275, VQPI 1.300 mult 21 bus 180+ easy - maybe even 200 for 4.2 Ghz. With your cooling you're golden :up:)
WTF ... I wrote a big long post yesterday and was certain I put it up???
Odd how it looked and read so much like Otis so no real loss :rofl:
I'm not married to FurMark by any means. I just got it because the most recent 275 came from EVGA's "B-Stock", and I wanted to stress the card to make sure everything was kosher with it. If I can do without it, GREAT.Quote:
Originally Posted by Otis11
So as far as testing goes, do I just start GPUGRID, step up shader clocks 1 notch and watch for errors in BOINC Manager? How long should it run before considering the OC stable? Do I need to complete the WUs and check for errors at the GPUGRID site before moving to the next step?
Sorry for all the questions, I'm sort of a nervous ninny about this. Maybe I'm trying to make it too complicated? :shrug:
Hehe, I should've known better than to admit to having an i7 at stock around here. :shakes: I was going to delve into CPU OCing using the dedicated cruncher I picked up last May (AMD 1055t), but the wife loaned it out to one of her (recently divorced) friends because "...her kids need their own computer..." :upset: I told her that Santa had better bring those kids their own for Christmas...Quote:
Originally Posted by Otis11
Anyway, thanks for your help!
Regards,