PDA

View Full Version : 7970 Matrix Platinum Overclocking - can't break 1180mhz



B0UJI
04-13-2013, 01:56 PM
Hey all,
Just wanted to inquire with the wisdom of the forum about what luck any one has had overclocking the Matrix 7970 platinum.

If I lock the GPU Clock + Voltage together and push to 1200 I get major artifacting in the Heaven 4.0 Benchmark and a raft of errors when using the GPU test of OCCT (http://www.ocbase.com/).

I've been fiddling with it over the last few days untying the clock + voltage to be independent and have ended up with the settings:

GPU Clock: 1180Mhz
GPU Voltage: 1325mV
VDDCI Voltage: 875mV (the default left it unchanged)
Memory Clock: 7000Mhz
Memory Voltage: 1620mV
Power Target: 120%
Load Line: 75%
VRM Clock: 705Khz

With these setting I'm stable in Heaven 4.0 (ran for an hour in loop no identifiable artifacting) and will pass and error check GPU test in OCCT fine for an hour too. Temps are also fine, set up a custom fan profile to kick it up higher earlier and sits around 62* c on load.

What does anyone think? Any settings look wrong? Anything I can try adjust to get 1200Mhz GPU Clock stable?

I'm happy with 1180Mhz if that's all I'll get but just thought I'd throw it out see if there's something obvious I've missed.

Thanks!

HiVizMan
04-13-2013, 03:36 PM
Nope I never had much luck with any of my Matrix cards when it came to overclocking them either. Zka17 has three as well and he did not get spectacular results either.

Myk SilentShadow
04-13-2013, 03:43 PM
My best guess, would be depending on when you purchased your Matrix, if you got one in the 1st deployment you'd have been golden...but then, AMD decided it would be a good idea to put a voltage lock on the 7xxx series after a certain time...which was, I think what affected the 2nd deployment of the Matrix cards, if memory serves correctly it was that time the voltage lock was enforced.

B0UJI
04-13-2013, 04:04 PM
Fair enough, thanks for the quick responses. It's not voltage locked for sure as I can push the voltage up and down independently. But second I go to 1190 it's starts getting errors in OCCT fast no matter how higher I push the voltage.

Does the power target need to be 120% too? Or can I lower you reckon? Or does it not matter as it only draws more if it needs it.

HiVizMan
04-15-2013, 11:33 AM
The fact of the matter is that any overclocking head room on a graphic card is a bonus. To put that in perspective the first few of cards I had were bought the week the HD7970 were released. They were stock reference cards. 4 out of the 5 cards did 1300MHz Unigine Heaven. None of my so called special cards that I bought later, Lightnings or any others have come close to clocking as well on air as those first cards.

On LN2 it was a whole different ball game. Because the big difference was the power delivery system. So getting close to 1750MHz on the core was possible using LN2 and adding some voltage.

But air clocking nope the first batch were great for me and nothing has come close since then.

twisted1
04-15-2013, 12:04 PM
HiVizMan that's very true, they Matrix ie. is really made for sub zero, low ASIC gpu's and the 20 phase powerdelivery is so overkill it can easily power two 7970's @1.2-1,3Ghz. The Ares 2 has the exact same powerdelivery, and that's got two gpu's :P

My Matrix is really lovely tho, i'm getting it above 1300Mhz benchmark stable. And thats w/o hotwire and OCP mod's. So I'm really happy with mine. Only thing I dont like is that I want another one but can't get a hold of one.

Got a dirt-cheap Dcu2 top instead and lucky me, seems to be a good sample. Not quite keeping up with the matrix but rock stable up too 1250Mhz on gpu. Need to hotwire that so I can get Vmem and PLL adjusted.

OP: Lover your mem clocks, mine can sometimes artifact above 7000mhz on mem. The hynix mem. is actually only specced for 1600Mhz if i remember correctly and it's already OC'd with stock settings on the matrix. So leave the mem alone and just tweak gpu, when you find the limit for gpu you can tweak mem.

Powertarget always maxed out when OC.

B0UJI
04-15-2013, 01:29 PM
Allrighty I'll keep it at 120% power target.

Thanks guys it's been very informative. Quite happy now with my 1180 overclock.

Twisted1 - I haven't seen any artifacting and the OCCT GPU test reported no memory errors after an hour, also ran heaven for an hour numerous times + just tried the Metro and Bioshock Infinite benchmarks and all seemed fine (no artifacting).

Nice boost in performance too! Did notice that possibly my "minimum" framerates may have dropped a bit but maximums are WAAAAY up. Any one seen this behavior with OC's?

Zka17
04-15-2013, 02:16 PM
Zka17 has three as well and he did not get spectacular results either.

Thank you, HiVizman... - and I was thinking that I'm doing good... :D But yes, you're right! I was expecting more off these cards...

However, all what I did until now is that I mounted waterblocks on them, no hotwire or other mods yet (collecting more infos about those)... - I may be lucky, but all three of my cards can do 1250MHz on the GPU on air and 1300MHz easily (except Vantage) on water...

BOUJI, 62C is pretty high for those cards... I assume, you meant 62C on the GPU... - then the VRMs are way too high (you can not measure that unfortunately) and the card is downclocking...

twisted1
04-15-2013, 04:20 PM
Once you get the hot-wire set up you'll get alot more voltage headroom, and you can disable OCP. OCP is just good for messing upp OC's ;)

B0UJI
04-16-2013, 02:49 AM
BOUJI, 62C is pretty high for those cards... I assume, you meant 62C on the GPU... - then the VRMs are way too high (you can not measure that unfortunately) and the card is downclocking...

It doesn't seem to be down clocking as everything is showing the CoreClock stable and Memory at 7000 - not dropping down at any point. If i see any artifacting or any halts I'll drop the Memory Clock back to 6600.

Or should I try drop the VRM clock and see if it's still stable?

twisted1
04-20-2013, 12:01 AM
Your voltage is abit on the high side for 1180Mhz. Try reducing the voltage a bit.

And about OCCT, that is one killer test, it puts furmark-like loads on the card. You will never even get that kind of load on your card when using it for anything else.

A clock that will fail after a few min in OCCT, can be perfectly stable for gaming and benchmarks, but not torture tests like OCCT/furmark. And there's no real point in having it OCCT stable if you don't run OCCT all day long which i doubt :)

The artifacting in Heaven, do you get it right away or does it take a few loops before it appears?

B0UJI
04-20-2013, 01:13 AM
Usually see it about half-way in. See some of the clouds distorting and flashing on some of the textures (most at a distance).

When I had the voltage linked and pushed it to 1250MHZ I saw colour blurring immediately on the screen (like a massive sky ray) and after a bit it skipped then dropped the video driver.

It doesn't crash OCCT or dump the video driver but I get errors star racking up in the "error" check mode.

MykaAurora
04-20-2013, 07:07 AM
Sorry topic owner for hitch hike your topic. I also wanted to know, any of you've tried to overclock matrix with EK waterblock fitted? How is the difference?

twisted1
04-20-2013, 07:57 AM
MykaAurora: Zka17 is running fullcover blocks on his matrix cards. Sure he will answer your question soon :)

BOUJI: Half way in as in halfway in one run och halfway off 1 hour looping?

Anyways the fact that it does not occur pretty much right away could very well point towards heat problems.

Zka17
04-20-2013, 09:19 AM
As I said before, the most important temp at these card is the VRM's temp... - which we can not measure directly, but a lot depends on it...

Check my posts #567 and #568 here: http://rog.asus.com/forum/showthread.php?20047-The-Chronicle-of-My-TestBench/page57

MykaAurora
04-20-2013, 12:25 PM
As I said before, the most important temp at these card is the VRM's temp... - which we can not measure directly, but a lot depends on it...

Check my posts #567 and #568 here: http://rog.asus.com/forum/showthread.php?20047-The-Chronicle-of-My-TestBench/page57

The VDDC + VDDCI aren't the same as VRM temp?

Zka17
04-20-2013, 12:27 PM
Nope, at least I don't think so...

MykaAurora
04-20-2013, 12:31 PM
VDDC always showing 80c on load ( @ 32c ambient ). I've figured it was the VRM. Guess not.

Btw, I don't think ASUS can track whether the stock cooler has been removed or not. :D

Edited : VDDC 'is' VRM

http://rog.asus.com/forum/showthread.php?28574-What-is-VDDC

Zka17
04-20-2013, 12:51 PM
"VDDC always showing 80c on load ( @ 32c ambient )." - wow, you do have a very high ambient temp! That can easily explain why you can not OC higher...

"I don't think ASUS can track whether the stock cooler has been removed or not." - don't be so sure about that... don't forget that Asus has very highly qualified engineers who know their products better then us... (on the other hand, not sure how many of the RMA'd cards are really getting back to Asus - the RMA centers are not the same as Asus...)

MykaAurora
04-20-2013, 01:01 PM
"VDDC always showing 80c on load ( @ 32c ambient )." - wow, you do have a very high ambient temp! That can easily explain why you can not OC higher...

"I don't think ASUS can track whether the stock cooler has been removed or not." - don't be so sure about that... don't forget that Asus has very highly qualified engineers who know their products better then us... (on the other hand, not sure how many of the RMA'd cards are really getting back to Asus - the RMA centers are not the same as Asus...)

Agreed to both. And have to try and outsmart answer 2, to solve answer 1 problem.

Going to fit waterblock for my matrix. :)

Thanks @Zka17 , hopefully this give some info for topic owner too, so my topic hitch hike do contribute to something. :D

Zka17
04-20-2013, 01:11 PM
No problem, mate! :)

Also, be aware (regarding your ambient temp) that even water-cooling, with regular radiators, can not bring your temps below ambient...

MykaAurora
04-20-2013, 02:05 PM
No problem, mate! :)

Also, be aware (regarding your ambient temp) that even water-cooling, with regular radiators, can not bring your temps below ambient...

I know, studying engineering. One of most basic law of nature. Cooling temperature can't be below ambient temperature without assisted cooling ( eg. R134a refrigerant cycle ). :D

twisted1
04-20-2013, 08:02 PM
I know, studying engineering. One of most basic law of nature. Cooling temperature can't be below ambient temperature without assisted cooling ( eg. R134a refrigerant cycle ). :D


If someone can do that, we'll have to rethink everything. Imagine free energy for everyone, from a cpu cooler!! :D

MykaAurora
04-20-2013, 11:09 PM
If someone can do that, we'll have to rethink everything. Imagine free energy for everyone, from a cpu cooler!! :D

Haha, yes! Imagine what if....

100% efficiency system? Haha! Are we living in vacuum space and there's no gravity? < even that still not 100% efficient. :D

B0UJI
04-21-2013, 04:12 AM
MykaAurora: Zka17 is running fullcover blocks on his matrix cards. Sure he will answer your question soon :)

BOUJI: Half way in as in halfway in one run och halfway off 1 hour looping?

Anyways the fact that it does not occur pretty much right away could very well point towards heat problems.

About halfway into the first loop.

Yes this all has been informative, I've got GPU Tweak monitor running as a widget in my second screen now to help watch things. So VDDCI and Memory Temp are different things? What's key and what should the be at under load?

I'll try drop my GPU Voltage a bit but what should Memory Voltage, VDDCI and VRM Clock be at? Does anyone tweak those?

Zka17
04-21-2013, 04:18 AM
I pushed the memory voltages on the Matrix cards to 1650mV and it seems that 1900MHz (7600 effective) is a no problem - but, remember, I'm on watercooling...

I also have played a little bit with the VDDCI and VRM clocks... - but could not find yet a real benefit... maybe, with the VDDCI at very high GPU and VRAM settings, but no luck with the VRM clock yet... I'm starting to think that those may be for the LN2 cooling...

B0UJI
04-21-2013, 01:43 PM
Yeah I've reset back to defaults. Had a grey screen of death, with lines all over the screen today. Windows lists a Video Hardware Issue - but it was while watching YouTube. The great googled internets tells me it may have been Flash video acceleration so I've turned that off too for good measure.

SO! This hasn't perturbed me though. I'll try tweaking up the GPU clock again, but leave the Memory Clocks alone see if now with Memory Down i might be able to get it stable at a lower voltage. Also realized a mistake I made in OCCT testing, had a frame cap of 60fps. At that, it was fine no errors. When I realized this and took it off I got errors during the testing pretty quick. BUT! Again I'll try with tweaking but leaving the Memory at default.

Also when I ran the OCCT test with no framecap I noticed the clocks reported as 1050MHZ (the non-boost frequency). Is this due to the nature of the test it realizes and clocks it down? Does this then mean it's not an accurate test of the Overclock?

Definitely want to thank everyone for your inputs this is all fantastic info!

twisted1
04-23-2013, 02:25 PM
I was testing with diffrent VDDCI settings(VDDCI is the same as PLL if I'm not mistaken) and got it through 3dmark @1330 which is higher than before at least. Had VDDCI set at 905.

B0UJI
05-01-2013, 01:09 AM
To update, latest testing with OCCT framecap removed!

1100/1257 - Stock Settings
1130/1275 - OK - 10MINS OK!
1140/1275 - Error - LL% - 75%
1140/1288 - Error - LL% - 75%
1140/1288 - Error - LL% - 100%
1140/1300 - Error - LL% - 75%
1140/1313 - Error - LL% - 75%
1140/1313 - Error - LL% - 100%
1140/1325 - Error - LL% - 100%
1140/1338 - Error - LL% - 100%

So far it looks like voltage increasing isn't helping. 1130Mhz is the highest stable with no errors, checked for 10mins.

Next I'll try that voltage for 1140 but try increase the VDDCI/VRM settings. See if they help!

Also, load line %. Should that be at 100% or 75%? Does it help keep the clocks stable?

*After my spectacular crashing with the new drivers in another thread I was wondering if anyone knows how to enable voltages in Afterburner for the ASUS 7970's?

twisted1
05-10-2013, 08:37 AM
As far as I know only GPU Tweak works. AB only works with ref. or ref. compatible cards. And MSI cards of course.

You should use a practical test like gaming or whatever you use your card for. OCCT test is only good for testing if its stable in OCCT.

And keep a close eye on temps when using OCCT, you can easily overheat the VRMs with OCCT. The only reason for using it would be if you run OCCT all day long for some strange reason :)

Your voltages are really high, for 1200Mhz I set it at about 1270-1280mV, and powertarget always maxed out when OC.

Just tried OCCT on my DCU2, massive errors at 1100Mhz, app crash. I can play BF3 for hours @1200Mhz no problem. And 3dmark11 @1260 no problem.

B0UJI
05-11-2013, 09:15 AM
Ahhhh okay. Well makes me a bit more confident knowing others get errors in that test O/C'd too. Okidoki I'll try your settings and game for a while see how it goes!

(on a side, you're not water-cooled are you?)

twisted1
05-11-2013, 10:11 AM
Ahhhh okay. Well makes me a bit more confident knowing others get errors in that test O/C'd too. Okidoki I'll try your settings and game for a while see how it goes!

(on a side, you're not water-cooled are you?)

No stock cooler :)

B0UJI
05-11-2013, 10:24 AM
Wicked. I will give that a shot, just not until next GPU Tweak revision or ATI Driver Update. The current ones cause my machine to grey screen out when GPUTweak is installed.

twisted1
05-11-2013, 10:28 AM
What? Strange...

Win8?

B0UJI
05-11-2013, 10:36 AM
Nope Win7 x64 Pro, yep just started happening when I reinstalled the latest driver. Tested all sorts of combinations, tested turning off overclock and in the end uninstalled it and had no problems. Installed it again after that, first reboot grey screened again. WIndows logs showed a GPUTWEAK error not a Video Hardware error too.

It is quite weird!

twisted1
05-11-2013, 10:40 AM
Strange.

I've never had a problem with gpu tweak, I see alot of people having probs tho.

I'm on 13.5 beta2 and latest gpu tweak

B0UJI
05-11-2013, 11:27 AM
Must just hate me.

Or it's a combination of things.. and the latest drivers kicked it in the pants.

B0UJI
06-09-2013, 04:57 AM
Okdidoki so the new drivers and GPU Tweak play nice now so i've given my OC's another crack. I Ran Heaven for 4hours on stock and took the temperatures. I then used your OC twisted1 and it worked great for about 2hrs on Heaven then it started Artifacting like balls.

Scary though was I set it back to stock straight after exiting and tried Heaven again on stock a few minutes later and it was still artifacting. I shut down and waited an hour or so then started up again and ran Heaven stock and it behaved fine then. Is this likely a heat issue anyone reckon?

I'm going to try again though now with Fans forced to 100% and see if it lasts 4hrs ok on Heaven like stock and if it was a heat issue. It did get a good bit hotter (VDDCI was hitting 73*c/75*c and core / mem 70* range)

*Is VDDCI temperature the VRM temp? I've always thought it was. If the artifacts don't show up until a few hours in likely these are getting hot? And taking a while to cool down?

twisted1
06-23-2013, 01:47 PM
VDDCI is a voltage, VDDCI is the same thing as PLL. It should not be useful for normal OC, more for extreme OC.

I have VDDCI power temp, however the unit the value is reported in is %. So that doesnt say alot aboult the temp.

And you with your OCCT again :) I promise you my friend, if you can run OCCT for no less than four hours straight, you can play games with that same clock guaranteed 100% stable for four weeks.

I would take it easy with these tests if i was you, with fan on auto the card gets pretty hot, if you run at 100% fanspeed, if should be ok. Remember hearing protection tho hehe. If i run a single platinum for games I run gpu @1200mhz(1295mV, the auto setting sets way to high vcore)

In CF i downclock to about 1000Mhz, and gpu usage is below 50% on both cards, all running bf3 @1440p

VDDC power temp, should be on of the VRM temps(vddc beeing the vrms that provide gpu Vcore) But the unit for this is %, and w/o knowing the temp spec's that dont say a whole lot.

Mem power temp is always at 53 or 54% for me, bug? Thats the vrm for Vram. Thats the two VRM's situated a few cm from the bios switch.

Mine are sensitive to mem OC, fullcover blocks should fix that. When i bench in low ambient temp, i can crank up RAM muuuch higher w/o probs.

B0UJI
06-24-2013, 06:59 AM
Yeah i'm not sure about those readings. It really sucks that GPU Tweak is the only think that can read it as GPU-z sensors would be great to monitor it.

I read the % as possibly just as a *C just badly formatted but if not then yes it's useless. Also if not, we have no way of knowing real VRM temps and that's not good.....

On a side, my Matrix has an odd issue that pops up sometimes (even @stock) after like a whole day of gaming where all of a sudden i'll get crazy flickering and artifacting. If i shutdown and reboot it's gone. Not sure if it's heat as Core temp is like 34*/40* and i've seen it happen. Might just be a *quirk* of the card.

*In before OCCT ****ed it it has done this before I discovered the program

twisted1
06-27-2013, 12:21 AM
I've now had the same problem, crash when gpu tweak starts.

Just don't set gpu tweak to remember settings over reboots. That's what causes the problem, at least on my sytem.

It has worked before but not now, I tested and replicated the problem.

MykaAurora
07-23-2013, 07:56 AM
I've sunk my matrix. Blegh, not much gained. 1210MHz @ 1.363vCore ( in ROG Tweak ) , and 7000MHz Effective @ 1.65v < Is this can be optimized more?

Straata
09-18-2013, 07:41 PM
I just got a new M7970 Plat, RMA'd the first due to massive heat issues (couldnt run above stock without hitting like 70C on aggressive fans)

New card is better, but still not as impressive as the reviews that I used to choose this card. I am also getting 1210 GPU and 7000 mem max. Been using Heaven 4.0 for tests, Ultra and Extreme Tessellation. Was hoping for at least 1275 on the GPU, but no amount of voltage will allow me anything higher.

I guess the AMD voltage lock I think I read about is still in effect?

EDIT:

Using GPU Tweak 2.4.3.1 and AMD Driver 13.8 Beta

EDIT 2:

After more game testing I got a MASSIVE display crash @ 1210, and was forced to clock down to 1175 again :( At least the temps are stable and acceptible, but considering I got the card for its OC potential and I cant do much OCing with it I am considering switching to a Lightning BE so I can use Afterburner to its max.

MykaAurora
09-19-2013, 10:59 PM
I just got a new M7970 Plat, RMA'd the first due to massive heat issues (couldnt run above stock without hitting like 70C on aggressive fans)

New card is better, but still not as impressive as the reviews that I used to choose this card. I am also getting 1210 GPU and 7000 mem max. Been using Heaven 4.0 for tests, Ultra and Extreme Tessellation. Was hoping for at least 1275 on the GPU, but no amount of voltage will allow me anything higher.

I guess the AMD voltage lock I think I read about is still in effect?

EDIT:

Using GPU Tweak 2.4.3.1 and AMD Driver 13.8 Beta

EDIT 2:

After more game testing I got a MASSIVE display crash @ 1210, and was forced to clock down to 1175 again :( At least the temps are stable and acceptible, but considering I got the card for its OC potential and I cant do much OCing with it I am considering switching to a Lightning BE so I can use Afterburner to its max.

I've given up OCing Matrix for now. We're the latecomers in buying this Matrix card. Apparently good OC yield for early batch cards. Not going to repeat same mistake again, and going reference+water block from now on. :D

Zka17
09-20-2013, 01:43 AM
OK, trying to post again...

So, with the Matrix Plats temps have a huge impact on OC'ing... not so much the GPU temp (that is important too), but the VRM temps... - if those heat up, the card will start throttling, then crashing... - so, all I can say is to try improving the cooling...

B0UJI
09-26-2013, 06:17 AM
Yeah and mem temp. I think that's related to a lot of people's issues with it. I'm testing with having my memory downclocked to 6000Mhz as I think the stock 6600 is making it overheat after a while.

twisted1
10-12-2013, 07:23 AM
I've sunk my matrix. Blegh, not much gained. 1210MHz @ 1.363vCore ( in ROG Tweak ) , and 7000MHz Effective @ 1.65v < Is this can be optimized more?

It's your mem OC that's too high. Leave mem @stock and OC the gpu only. When you reach the OC limit you can tweak the mem OC.

And you voltage is way high, to high voltage can be just as bad as too low voltage. I use 1288mV for 1200Mhz, rock stable. For 1300Mhz I set it around 1330mV.

MykaAurora
10-20-2013, 03:36 PM
It's your mem OC that's too high. Leave mem @stock and OC the gpu only. When you reach the OC limit you can tweak the mem OC.

And you voltage is way high, to high voltage can be just as bad as too low voltage. I use 1288mV for 1200Mhz, rock stable. For 1300Mhz I set it around 1330mV.

Solved the problem. :D Can't overclock well with 32-35c ambient. :D ( Tropical climate )

1200MHz stock vCore . 6.6k MHz memory @ 24c ambient. Rock solid stable. ( Air conditioned room )

After the air cond switched off, back to unstable mayhem. :D

Tom8989
11-02-2013, 04:58 PM
HI all this driver working for me OK :) windows 7 (64) windows 8 (64) test and NO ARTEFACT only Win 8.1 (64)
stop working (driver) try:

http://sendfile.pl/85020/AMD_VGADriver_Win832_64_VER9_02.rar

tic42
12-31-2013, 03:36 AM
Hi,

I know this thread is kind of old. I hit the wall with default voltage (1.25V) on the matrix 280x at about 1110 mhz. At 1150, it wanted 1.30V to keep it stable. The interesting part is, I can turn the voltage down at 1110 to min allowed 1.188 and it still works perfectly fine at 1110, temps go lower too. Perhaps I will install a full cover EK but people don't seem to gain much.

Question, what is VDDCI in the Asus GPU tweak, the core is under 'GPU Voltage'. So what does VDDCI set?

Panaani
01-17-2014, 02:34 PM
Old topic, but just tested my 7970 Matrix Platinium. Stuck at 1190Mhz, GPU tweak monitorin says voltage is unstable varies 1.25v - 1.275v.
Lack of power maybe. Using only 600w psu, 12v2 rail gives 19A.

twisted1
01-18-2014, 05:14 AM
Which program are you using for stability testing? How does the instability show it self, arfifacts? shutdown?

I have 2 7970 plats, they both benched @1300Mhz on air, and for gaming I ran them at 1200Mhz in a regular case with good airflow.

I also have a 7970 dcu2T and that clocks about the same, not as high on mem though if you don't use VGA hotwire so you can up the mem voltage. They're all on water now and gaming for hours @1300Mhz is no problemo. And benching @1350-1400Mhz depending on wich bench I'm running.

Panaani
01-18-2014, 09:28 PM
Using Far Cry 3 and Crysis 3. If I go past 1190Mhz after some time artifacts appear (red, green, blue spots). Case side open, fans 100% and 2x80mm fans blowing back side of card. On Cpu ( i7-860) is watercooling. So temperature is not problem. Gpu usage only 79% on 1080p MSAA x4, fps limited to 60. Gaming temps on Gpu is 77 C. Maybe too much?

If i look card from side, looks like its bend little. That can cause overheating...
No dust on cooler.

Do you have EK-blocks?

Panaani
01-19-2014, 01:30 AM
I was inspecting my card for some kind of reason why it fails. I noticed that every bolt or screw was loose. Cooler didn't properly fit on Gpu.
I changed the cooling paste to Arctic Silver 5 and put together. Tightened every screw.

My gaming temp dropped to 55 C! 22C lower than before. Nice!
But still only 1190Mhz after that artifacts. It's not the heat causing this.

Downclocked my CPU from 4,2Ghz to 3,2Ghz, not helping.
Maybe this card is just bad version.

Voltage monitor says 1189mW - 1218mW on Gpu when its set to be 1257mW.
If i raise the voltage with GPU Tweak or board + button, voltage raises ok, but not helping.
My PSU 12V line is only 11,70V. Maybe that is the problem?

Edit. measured on the card 11,6V