Whats your GPU-Z ASIC score ?

how long does this test run for? So far 15 mins and nothing.
 
Got it :) 72.4%

overclocked to 1125/1575 (havnt tried higher) and stock voltage on asus 7970
 
MSI GTX 560 ti TFIII PE scored 70%.

Stock voltage 1v. I just got the card and I just did a Kombuster benchmark with the core at 900mhz. I didn't add voltage at all, just slid the core speed to 900mhz. Seems like a good GPU.

Completed a benchmark no problem.

http://www.ozone3d.net/msi_kombustor/score_200.php?id=36198


ASIC score doesn't seem to mean much or it is buggy.
 
Last edited:
One of my GTX580s scores 50% and the other 62%. Ouch. Both are overclocked to 875/2200 though.
 
my sapphire 7970 is 81.1% and I can hit 1250/1700 with stock voltages, i tested only 3dmark11, vantage and leo demo.. didn't test any further... but i tested alot in 1200/1600 and no problems at all...besides the damn coil whine.
 
Test with a game such as BF3 cause 3dmark11 isn't taxing enough, at least not for my setup.
At my OC the card would have no problem with 3dmark11, but with BF3 with SSAA enabled, the game would lockup and crash.
 
no 7XXX series card here (running 6970+6950 2GB Xfire)

but reading through all I could think about the entire thread is the Olestra "scandle" a decade ago on the bags of chips that said "May cause anal leakage"


heh leakage.
 
Here's where you find the option. It's not very clear...



I'm beginning to suspect AMD is hoarding the 85-100% ASIC quality (higher binned) chips for HD 7980 to compete with the GTX 680. Just a thought. It's very possible they are / aren't but that's just my thought. Considering how much leeway they have in the clocks they can easily add a 7890 with 1100 Core+.


GTX 560 TI hawk here.
88.6% ASIC.
Overclocks to 1050 in heaven, but BF3 works with this oc.
1000 MHz core @1.087v, 1165 MHz memory.
 
Just wanted to chime in since there still seems to be confusion on the matter, but in terms of leakage current on ASICs this is the amount of current that is flowing through transistors that aren't switching states (i.e. think IDLE or holding the same value for several clock cycles). In general higher leakage is undesirable since is sets the lower bound on your device power use, since even if the device is 100% idle it will pull X amount of power. How this number would directly relate to OCing bounds is unclear, especially given the fact that one man's 100% stable may not mean the same to another (i.e. 30 min loop in FurMark vs several hours playing the current favorite AAA title) and this also makes the assumption that the HEX value read by GPU-Z means what they think it means in terms of leakage targets.

As another mentioned before, any given TSMC wafer run has more than a little bit of "black magic" involved as they are constantly tweaking with process parameters in order to maximum the yields. In general though ASICs with higher leakage tend to pull more power from the wall and as such also generate more heat than an ASIC with less, but given these are not targeted at mobile devices and the efficiency of the default cooling system on the 7970 odds are most differences would simply be hidden by a slightly faster fan RPM (i.e since AUTO seems to stay below 30% most of the time anyway).
 
80%

Reference ASUS HD7970, 1125 and 1575 stable so far (10 Metro runs, 10 Unigine, 10 Crysis and shortly Furmark for temps).

Stock voltages 1.168V, GPU-Z reports 1.1 under load which makes sense due to vdrop. Idle it's 1.143V. Power usage set to 0% and no throttling so far.

Removed limits and clocked it to 1200MHZ with the ASUS tool, ran Unigine a few times, still stable at those volts. Didn't want to push beyond limits though so I'm running the maximum "default" clocks. And I didn't see an FPS jump worth mentioning (like 1-2FPS better), but then again I should have pulled memory along too I suppose.

Does overriding those limits void your warranty, by the way?
 
Kaynar said:
It is a fact that the ASIC score is not related to the quality of the card. My card has 77.7% and it overclocks greatly. Others with card scores near to 100% had much worst overclocking "walls" whilst other with high scores had great OC potential, and vice versa with low scores. Look on older threads on this forum about ASIC and see for yourself.
Click to expand...

yup. Mine's a 77% and max I can get is 1150 core on 1.3v.
 
Zotac GTX560ti , this is a "budget" card basically. Short PCB, 2 heat pipe single fan heatsink. Was $210 (bought shortly after launch 560ti launch) with no rebate and 3 bundled games. So yea budget.
ASIC 115.1%
idle 0.95v, load 0.975v
stock voltage OC test -
925mhz kombustor okay, BF3 fail
900mhz both okay.

This the only time I've ever OCed this card, just ballpark test figures.

edit: 920mhz BF3 seems to be okay, maybe it crashed at 925mhz for another reason (was alt tabbing a lot with other background apps). But back to stock I go lol...
 
Last edited:
Eluder79 said:
Test with a game such as BF3 cause 3dmark11 isn't taxing enough, at least not for my setup.
At my OC the card would have no problem with 3dmark11, but with BF3 with SSAA enabled, the game would lockup and crash.
Click to expand...

my sapphire 7970 is 81.1% and I can hit 1250/1700 with stock voltages, i tested only 3dmark11, vantage and leo demo.. didn't test any further... but i tested alot in 1200/1600 and no problems at all...with all tests above plus BF3 1920x1080 all on ultra plus AA, also tested 5760x1080 in a 64 player server, Skyrim, APB Reloaded and some more
 
I don't know if this carries over to GPU design but Intel has talked about how they tune their chips to run cooler or at a certain clock speeds. The same die can use different designs of the same basic transistor. He said they can use 'leaky' faster switching transistors or 'less leaky' slower switching transistors in different areas of the die depending on what is needed.

Seems to me the leaky one should get higher stable clock speeds if you have proper cooling.

Then there are other variables to throw in there like having a decent power supply with clean / stable power. PCB circuit design.

I am going with the Asus 7970 DirectCU II in a month or so. Hopefully by then, they will have some in stock.
 
This was a post by Unwinder (Afterburner author) in regards to Asic, @ Guru 3d

Originally Posted by Unwinder
Not exactly. Bits 4-13 of reg 174 are fuses reflecting ASIC leakage/quality, 100% quality is 3ff. So my ASIC (23e0) is 23e/3ff = 574/1023 = 56%.


AMD uses the following table to select VID depending on ASIC quality:

up to 2F90 (up to 75% quality) - 1.1750V
up to 34D0 (up to 80% quality) - 1.1125V
up to 3820 (up to 85% quality) - 1.0500V
up to 3A90 (up to 90% quality) - 1.0250V

But in the [H] Gigabyte review, their sample defaulted to 1.174mv, so it would be one of the ones that is less than 75%. I wouldn't read too much into it......Mines 71.6% and I can max the sliders in ccc at default voltages
 
The new GPU-Z 0.5.9 explains ASIC a little better now. It's basically the same theory that goes with CPUs and others hardware when you look at OCing/temp
 
My 7970 has 77.7% ASIC with 1.112v shown in MSI AB. I can max out to 1125/1575 with this voltage and pass 3dmark11, unigine heaven and furmark and I have even run several hours of gaming on this setting. With a 1575Mhz on the memory clock it artifacted in The Witcher 2 but was fine in everything else.

Question:

At stock settings (1000/1425 of my BEDD or even at 925/1300) will I be able to down-volt the GPU to, lets say, 1.075v and play still be stable? I wouldn't like to test it without anyone's opinion because I actually dont know what could happen when undervolting below the GPU's specification (I can't take it for granted that its the same as overvolting).

Why would I want to undervolt? Well many times I spend alot of hours in WoW or other online games that dont require so much "horsepower" so undervolting the card would reduce temps and therefore reduce the GPU fans spinning speed and may bring the card to a totally silent level even when playing these games for long hours.

Any suggestions?
 
It will be no different than overclocking, either you have enough voltage for a given speed, or you don't. The symptoms will be the same
 
Mine is 71.86%, runs 1200core/1600mem benchmark stable for multiple passes, no noticed artifacts on stock voltage 1.175.

I typically run the card 1000mhz, 98% load, 24x7 at 1.043 volts though.
 
My score was somewhere in the 55-59% range with 1.175 voltage ;p. I am able to overclock to 1100 core, 1700 memory at 1.181V, while keeping the fan below 50%.
 
Dont really think this crap means anything.. My lowest card is around 73% though i'm still able to overclock my Quad 7970s to 1200/1800+ memory.. prob gona end up with around 1900Mhz on the ram.
 
Why couldn't they just have added a simple button? Instead they make you dig through forum posts until someone gracefully tells you..

My GTX 480's are 83.5% each. So is this like capacitor aging or something similar? Or does this mean your GPU production was able to yield this average on each one. I can imagine people will now start asking for this number before they purchase a used card just to see if it means it can tolerate more abuse through say overclocking..
 
eclypse said:
Dont really think this crap means anything.. My lowest card is around 73% though i'm still able to overclock my Quad 7970s to 1200/1800+ memory.. prob gona end up with around 1900Mhz on the ram.
Click to expand...

to me it means 1125mhz on 1.13v! :p and lower power consumption over all.
 
Godmachine said:
Why couldn't they just have added a simple button? Instead they make you dig through forum posts until someone gracefully tells you..

My GTX 480's are 83.5% each. So is this like capacitor aging or something similar? Or does this mean your GPU production was able to yield this average on each one. I can imagine people will now start asking for this number before they purchase a used card just to see if it means it can tolerate more abuse through say overclocking..
Click to expand...

The value is just read off the card and is something set at the factory. It won't change nor is it actually testing anything to get this value. Of course how the program interprets the value can change, for the instance the new version of GPU-Z reads mine as a more reasonable 94.1% as opposed to 115.1%.
 
on My XFX R7970 BEDD with ASIC 77.7% and stock voltage 1.112v I have managed to be stable at clocks 1000/1425 with voltage 1.056v which is near to the supposed 100% ASIC. In fact I didnt even have to experiment with it, I straightly went down from default voltage to 1.056v and it worked (using MSI AB). I'm getting only 1-2c less temps though.

I'm just trying to make one more point about this weird ASIC and its supposed relation to voltage and overclockability.
 
Kaynar said:
I'm just trying to make one more point about this weird ASIC and its supposed relation to voltage and overclockability.
Click to expand...

Honestly, I think it's just one more ePeen number for nerds to fawn over when in reality it is barely understood and so far means nothing.
 
latest gpu-z gives you more details on what ASIC quality implies
Low Quality
Higher Default Voltage
Higher Power Consumption
Lower Overclocking on Air
Higher Overclocking on Water

High Quality
Lower Default Voltage
Lower Power Consumption
Higher Overclocking on Air
Lower Overclocking on water
 
MavericK96 said:
Honestly, I think it's just one more ePeen number for nerds to fawn over when in reality it is barely understood and so far means nothing.
Click to expand...

That can't be true because the ASIC rating is a specification made by AMD and the people who make GPU-Z simply claim that they found a way to decode this.

The thing is, that this ASIC numbers most certainly have large safety margins to avoid errors in the whole proccess of rating the card and therefore 2 identical ASIC rated cards will propably perform differently. In addition to that, I am certain (but with no proof) that the ASIC rating and the assossiated voltages are not the only factors that affect overclocking. There are so many things on these cards that there must be dozens of factors that will affect the potential. We are, therefore, giving too much attention to a number that describes only 1 factor and that is why people keep wondering about out-of-the-ordinary results compared to the "standard" description of the ASIC rating.
 
You must log in or register to reply here.
Back
Top