Also don't forget I'm running this with the stock fan at 100% and purely for these benchmark runs - no way would you want to do with this 24/7 with the stock cooler unless you like the sound of a hovering harrier jump jet!
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
****OFFICIAL AMD 6000 series overclocking and benchmark results thread****
- Thread starter bigjimmyauk
- Start date
Also don't forget I'm running this with the stock fan at 100% and purely for these benchmark runs - no way would you want to do with this 24/7 with the stock cooler unless you like the sound of a hovering harrier jump jet!
No but the problem is the fan doesnt seem to react fast enough at those clock speeds and stays at about 27% which isn't anywhere near enough so you reach a point where it will just crap out. I'm itching to tinker with the fan speed mappings in the bios only I can't flash it as yet - I can edit the bios but have nothign capable of flashing the new image back yet.
Give me five minutes and I will furmark it with those speeds and the fan on auto to see what temps it hits.
Gpu hits 66 degrees doing a burn in test with the fan on auto and the fan stays at around 27% but it will still crap out under extreme burning mode. According to gpuz the vrm's are all reasonably cool (all well under 70) so I think its the memory thats over cooking itself at 1200 without cranking the fan up - doesnt appear to be a mem temp sensor showing on gpuz so hard to tell.
Yup its the memory, just lowered it to 1100 and it doesnt crap out.
They would be but I'm not sure I actually believe them. Something heat related is causing the card to crap out at those clock speeds without the fan manaully set to 100%
The gpu temp I may believe but something still smells a bit fishy. I need to do some more messing around with furmark and lower ram speeds to see if it is the gpu or ram. I will post back
Ok not sure what to make of this one..
If I leave the fan on auto and crank the gpu up to 1000 it will run the burn in test until it hits 76 degrees on the gpu then it will crap out and the catalyst drivers will recover. If I do the same again only I manually set the fan speed to anywhere over 50% its fine.
Whilst running furmark and keeping the gpu at stock and messing with the memory speeds I kept an eye on the min and max framerates, If I push the memory to 1200 the minimum dives occassionally which suggests to me that I'm trying to push too far and that the memory is trying to self correct.
Best guess I can come up with and it is a guess at best is that it needs more voltage to the ram at those clock speeds and possibly the gpu too (yes I know more juice = more heat but its the only explanation I have).
If I leave the fan on auto and crank the gpu up to 1000 it will run the burn in test until it hits 76 degrees on the gpu then it will crap out and the catalyst drivers will recover. If I do the same again only I manually set the fan speed to anywhere over 50% its fine.
Whilst running furmark and keeping the gpu at stock and messing with the memory speeds I kept an eye on the min and max framerates, If I push the memory to 1200 the minimum dives occassionally which suggests to me that I'm trying to push too far and that the memory is trying to self correct.
Best guess I can come up with and it is a guess at best is that it needs more voltage to the ram at those clock speeds and possibly the gpu too (yes I know more juice = more heat but its the only explanation I have).
Associate
- Joined
- 12 Aug 2010
- Posts
- 125
GPU: 980 / 1000
Memory 1200 / 1250
Result = Stable!
Just crashed at GPU 1000, memory 1220, so calling it at that, and pretty happy with it.
Memory 1200 / 1250
Result = Stable!
Just crashed at GPU 1000, memory 1220, so calling it at that, and pretty happy with it.
Out of interest what are you using to judge that it's stable?
Associate
- Joined
- 12 Aug 2010
- Posts
- 125
Using Metro 2033, loaded up a few of the more graphically intensive scenes at 1920x1080 at very high detail and ran fine without any errors, as opposed to running the same scenes with a higher overclock which resulted in the screen freezing. Not very technical, but good enough for me.
Try a furmark extreme burning test because I seriosuly doubt those speeds are doing what you think they are.
I found doing a windowed extreme burn test was a great help at showing when the memory speed was pushed too far, when the min fps starts to nose dive but the max stays good the memory is basically erroring and trying to recover - you may have a better max fps but your average will take a massive hit from the spikes. If you are just running benchmarks that wont load the gpu to 100% you may not notice it but sooner or later it will cause you a problem.
I found doing a windowed extreme burn test was a great help at showing when the memory speed was pushed too far, when the min fps starts to nose dive but the max stays good the memory is basically erroring and trying to recover - you may have a better max fps but your average will take a massive hit from the spikes. If you are just running benchmarks that wont load the gpu to 100% you may not notice it but sooner or later it will cause you a problem.
As an example I benched heaven plenty of times at 1000 gpu and 1200 mem but furmark was showing dire min fps on the extreme burning runs - turns out the sweet spot for the memory on my card is actually around 1160 (no fps dips).
And having said all that I still run it at stock speeds because other than benchmark runs its really not worth it - the card is plenty fast at 1080p
And having said all that I still run it at stock speeds because other than benchmark runs its really not worth it - the card is plenty fast at 1080p
Associate
- Joined
- 12 Aug 2010
- Posts
- 125
Knew it was to good and to simple to be true! lol. I'll give it a go and report back, however just doing some uni work at the moment, but will try and take a look over the next couple of days. 
Associate
- Joined
- 12 Aug 2010
- Posts
- 125
Curious Andy, what did you settle for in the end in terms of the GPU clock? Excluding that you said you normally play on stock settings
Caporegime
- Joined
- 18 Sep 2009
- Posts
- 30,505
- Location
- Dormanstown.
I can't wait for waterblocks to come out for these.
WC'ing my two 6870's.
WC'ing my two 6870's.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
Are any games not stable at a higher speed though, this I don't get, Furmark and other benchmarks(non game ones generally) put a massively higher load on than actual games. If I went by Furmark I wouldn't run my 5850 above 1000Mhz, but not a single game has ever crashed or shown dodgey fps at aroun1 1075/1250, furmark gets upset around 1025/1200 or so.
People should be less worried about whats stable for something they never intend to use again, and more worried about what they can use with actual software they'll use.
Sometimes it will be a big difference due to temps being borderline in furmark and not in games, with great cooling the difference might be lower.