• 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.

My 7970 OC doesn't seem that good!

Associate
Joined
16 Mar 2012
Posts
1,524
Location
Chelmsford,Essex,UK
Hi a lot of people seem to get good OC on 7970's, i bought a Gigabyte windforce which is 1000mhz stock i was told 1225 would be ok voltage wise but even then in Heaven benchmark i can only get a 1150mhz OC without artifacts and stock mem of 1375, where a lot of people seem to get 1200mhz on a lot lower voltage also my ASIC is 67.5% whats peoples opinions?
Thanks
 
send it back under dsr if ur unhappy
Hi that's what I was thinking but I thought that might be over the top,what do other people think? Would you dsr because you can't reach 1200mhz on one of these cards at 1.225 volts?
What in real terms would this make in fps would it make much difference?
My ASIC is 67.5%
Thanks
 
1150 is still a good overclock, you need to reamber the stock 7970 speed is 925mhz so that is still a 225mhz speed bump over a stock card which is ok.

My advice is try playing afew games see if you like the perfomance of the card first.
 
with that spec graphics card, you should easily get max fps in any game, even when it or you are lagging, so why overclock it :)? Its already overkill.
 
I can't get anything over 1050/1450 stock clocks on my VTX3D. Should be able to get a stable 1100+ core but need lots of volts. Still happy though, tis a good card.
 
What's your max temp while running heaven? You should still have some voltage headroom. Get that memory up to 1500mhz for starters too.
 
What are your temps like? In my experience, these cards are sensitive to temperature whilst overclocking high.

For example, I had a VTX X Edition 7970, would run 1150 on stock voltage no problem, but any bump in core voltage would increase the temps above 62. As soon as the temperature hit 62 i'd start seeing artefacts, increasing as the temperatures rose.
 
Mine will do 1125/1536 at stock easy but cant get stable 1200 no matter what voltage i use, its just the luck of the draw, i only oc for benching as stock speeds are fine for all my games
 
Hi a lot of people seem to get good OC on 7970's, i bought a Gigabyte windforce which is 1000mhz stock i was told 1225 would be ok voltage wise but even then in Heaven benchmark i can only get a 1150mhz OC without artifacts and stock mem of 1375, where a lot of people seem to get 1200mhz on a lot lower voltage also my ASIC is 67.5% whats peoples opinions?
Thanks
Your ASIC percentage is VERY low hence the poor overclocking performance.
 
Yes I also think that asic quality has no bearing on maximum overclock. Seems like a pointless number really.

It does seem to have a bearing on the stock voltage by the look of things, but yes, maximum overclocks it doesn't seem to make a difference.
 
Yes I also think that asic quality has no bearing on maximum overclock. Seems like a pointless number really.

I think the idea of the quality rating is fine, it's just that GPU-Z doesn't seem to be able to read/calculate this particularly well.

We talked about this in another thread as some people had over 100% which is a bit ridiculous and others who went from 70% to 80% with the latest release of GPU-Z so I'd take anything to do with the asic with a pinch of salt, at least until this can be found to be a bit more reliable.
 
also, not sure how valid this is but amd have started binning for ghz edition cards, and presumably this means any non-ghz branded reference pcb is going to be guaranteed to overclock less than previously released cards, rather than just a maybe.

gigabyte is also preparing a SOC which is binned, but that's non-reference pcb so not sure how that affects windforce versions...

so if i'm right about the ghz edition being binned then even if you return you might not get legendary overclocks lol. i think your current one is worth keeping anyway
 
I think the idea of the quality rating is fine, it's just that GPU-Z doesn't seem to be able to read/calculate this particularly well.

We talked about this in another thread as some people had over 100% which is a bit ridiculous and others who went from 70% to 80% with the latest release of GPU-Z so I'd take anything to do with the asic with a pinch of salt, at least until this can be found to be a bit more reliable.

the over 100% problem was to do with GPU-Z not being updated for new GPU's but being used on those GPU's... basically all they are doing is reading a value from a register on the card and then converting it to a number - if they change the way the conversion is done based on new information (I wouldn't imagine they are getting much in the way of support from AMD or Nvidia on this) then obviously it will change

GPU-Z is not "testing" anything to come up with the number, purely reading a value and displaying it

having said that - different people have come up with different explanations on what the number ends up meaning for people, but as soon as you say "low means this and high means the other", someone pops up and says "but mine is low and I get X"

there might be a trend (which the makers of GPU-Z have mentioned), but there are always outliers that don't conform to this

the trend seems to be that low asic figures mean higher OC on water, and high ASIC means good OC on air... but it depends on where the faults are on the chip as to wether this is true...

a quality test might come up with a low figure because the chip is really bad in one spot but generally good - this chip will have difficulty OCing at all because it has a weak spot that will give up before the rest of the chip wants to

equally the chip might have a high ASIC % because most of the chip is really good, but has this bad spot - on average the chip is "good" but the weak spot still means it fails early

any test that results in single number to try to cover the entire chip will be averaged and as such an average can be hiding something or skewed in a particular direction by an odd result somewhere in the actual data
 
Back
Top Bottom