GTX 645 question
20 Aug 2013 15:36:46 UTC
Topic 197151
(moderation:
hi folks, recently got an XPS 8700, it came with a GTX 645 ive ran furmark on it and it doesnt seem to have the "power" i thought it would, some where in the 880 range which looking at the comparison table for other GTX 6x and 5x units, appears to be rather low so im wondering if it would even be worth enabling boinc on it.
but my main question is this, during burn in i noticed temps 97c (206f) that seem a bit hot to me (really i nearly ran to get a bucket of ice :D ) so just wondering if that kind of temp is normal, and would running it at that temp 24/7 cause problems?
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
GTX 645 question
)
The specification for the GTX 645 says the max temperature is 97 degrees. So I guess the low performance is because the card is throttling down due to poor cooling.
Running the card at such high temperatures is not advised. It will lower the lifetime of the card and be much slower because of throttling.
Is the fan on the card working at all?
yup, evga precision shows it
)
yup, evga precision shows it is any way, its at 30% idle, when i run furmark it only goes up to about 35% however, even running it at 90% the temps still hit 97c, furmark doesnt show throttling unless i try to o/c (it was only +20mhz just to see if it would do it) at which point it shows a throttling notice.
possible i got a dud? or the fan came loose in shipping? though with out any of dells tests throwing an error convincing them of that would be a pain.
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
Hi Paul, I'd check the
)
Hi Paul,
I'd check the card to ensure that the heatsink is making contact with the gpu - 97C is way too hot. If it's not then you'll have to make some adjustments, re-paste, reassemble & test again. If you can't get the oem cooler to perform, consider using an Arctic Accelero Twin Turbo. I use this cooler on my 550Ti cards, and with 97% GPU load, temp is 43C and there is zero fan noise. This is a 15C drop from what I was able to get out of the oem cooler running 90% fan speed.
Gord
hey, sorry about the late
)
hey, sorry about the late reply. wanted to say thanks for the suggestions.
finally got around to opening the case, the cooler on this thing is dinky, maybe half inch thick and it cuts short from the back by at least an inch, so no external exhaust. i didnt see a way to remove the cooler to change out the paste (ive got some arctic silver ceramigue around here some where i think might work) i did try reseating the card, but im having other issues (like a checker board screen then crash, which to me points to the GPU) so im tryin to talk dell in to a replacement even with out an error code..
later on im going to upgrade to a GTX 650Ti, i would do the cooler but figure i could put the cash toward the new card and if i get the evga version have a rear exhaust cooler all ready on it (though, knowing my pention for wanting things to run as cold as possible, id probably still change it out lol)
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
Hi Paul, Check the back side
)
Hi Paul,
Check the back side of the card for mounting screws. If they aren't there, then they might be hidden under the front cover - looks like 6 screws to remove the front cover...
Your cpu is capable of driving more card than a GTX 650Ti. ASUS DC2 dual fan cards cool really well oob, so that might be something to consider. The machine I'm typing this on has an ASUS DC2 GTX 670, running 97% gpu load, fans are 68% and temp is 50C.
Gord
97°C with the fan "stuck" at
)
97°C with the fan "stuck" at 35% sounds like the fan is just not spinning up as it should. Could be the cards bios or the graphics driver. Besides: Furmark is not a proper test any more, since it generates far more heat than any real application. Especially Einstein is not very stressful for the GPU, so I'd try if the temperatures there are safe anyway (<70°C).
MrS
Scanning for our furry friends since Jan 2002
RE: Hi Paul, Check the back
)
yeah i went looking at GPU's while i was waiting on the replacement (caught my self drooling at the GTX titans, if only!). my bottle neck appears to be the PSU which is 450watts. upgrades are a ways off but ive all ready made up my mind to upgrade the PSU to go for the biggist GPU the CPU (and my budget at the time) can handle :)
and your right, the screws are on the back. oddly i couldnt find a brand any where, not even on the PCB, but going by the GPU's they offer im betting this ones a PNY
i came to that conclusion as well, no matter how hard i pushed it that fan just would not go over 35% using evga precision and manually adjusting the fan the fan sped up and temps dropped significantly. but i still wanted a replacement given how it was crashing (and lets face it, it was only about a month old so that really shouldnt have been happening under normal load) the new card maxed out at 70c under the same tests, the fans still "stuck" at 35% im thinking its a poorly designed fan curve in the video bios. maybe they wanted it to stay quiet no matter what even at the sacrifice of much higher temps? though i have to say even at 99% you really cant hear the GPU fan unless your trying to.
as for the load, i was actually trying to find a windows based cuda benchmarking program that basically does what furmark does but would be closer to boincs cuda use.. i guess the best way to do that is just load up my temp monitoring software and enable GPU processing when i have some time to watch it :)
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
You checked the fan curve in
)
You checked the fan curve in PrecisionX? Set "allow software fan control" then click the box for automatic, then apply?
Could you make the fan run faster manually?
My gtx 650 would run at about 70c and 25% fan happily. I set it to use the precision x fan curve now, but I'm not sure if that mattered since I wasn't hitting the thermal limit or getting errors.
I cool my cards the old
)
I cool my cards the old fashioned way of setting the Precision X @ 50% and then have the side panels off all of the box's with a 20in fan blowing across my row of 5 (and run my laptop in the coldest room with one of those small fans blowing on it 24/7)
I have 2 650Ti's and I don't really like them as much as the 2 550Ti 2Gb's or the 660Ti 2Gb OC'd
But yes the PSU is better if you are more like 750watt or better in fact I have one 850 that I use as the main for my 660Ti box and also have the 650Ti from another box plugged in that one since that box has a cheap PSU and it is in a box so small I had to remove the side panel and cut the top off just to plug in the video card.
I have two 650 Ti (s). One
)
I have two 650 Ti (s). One is an Asus 650 Ti 1G variant with two fans (long card) and the other is a new (yesterday) EVGA 650 Ti 2G with one fan (short card).
They are in separate machines with 750W power supplies. The Asus card runs at ~57C with 45% fan speed. The EGVA card runs at 44C with 21% fan speed.
They both crunch for multiple projects but the GPUs on both support E&H (one machine is 100% E&H while the other is a 50/50 split between E&H and another project) and are set with a "utilization factor of: 0.33. This means that both GPUs are chewing away at 3 jobs each plus some CPU work.
Each box runs Linux Ubuntu 12.04.
Don't know if this will provide some useful info or not but ....