Sign in with
Sign up | Sign in
Your question

7870 OC Problems, Unstable even @ 1175 - 1.3v

Last response: in Graphics & Displays
Share
a b U Graphics card
April 16, 2012 4:17:25 PM

Evening folks.

Ive been running my Sapphire HD 7870 @ 1250/1450, 1200/1450, 1200/1400, 1175/1400 all @ 1.3v

Same crap happens on every clock - Crash crash crash crash crash.

Could be 3 days, a week or 5 minutes between crashes but it does it.

I'm slightly miffed if its the card itself, Its supposed to have excellent overclockability and this one is just chucking that notion right back into my face.

On another note, I had issues with my old GTX460s when they were overclocked even the slightest bit, I got crashes regularly - this was after a power surge in the house.

Taking that into account, could it be my 12v rail causing the issue? or my PCI-E slot?

I'm trying it as we speak @ stock volts 1100/1300 but I dont really want to spend all the time I get away from work sitting watching furmark run for 8 hours.

April 16, 2012 4:25:38 PM

Overclocking is never guaranteed, no matter how good on average a card overclocks. I also don't think it's a good idea to just jump your card up to 1.3V... particularly if you have a reference card (or reference PCB/VRMs even with a custom cooler).

What is your power supply?

Also, personally I don't find furmark the best test for GPU stability. I know it's the "standard" stability test, but I prefer to loop things like Unigine or 3dmark and/or play games... furmark/OCCT don't place realistic loads on the GPU and can cause weird issues that you either may or may not get during gameplay.
a b U Graphics card
April 16, 2012 5:29:00 PM

please don't run furmark for hours on end, it's really not healthy. Also it doesn't find all instabilities and finds some that don't exist. I usually run furmark for a minute or two making sure it doesn't crash or show any artifacting. Then i just play games and adjust the clock whenever it crashes.

Also gpus are not cpus. Increasing voltage doesn't always increase stability. Most cards have a "useful max" for voltage where the card stops benefiting from voltage increases.

Try lowering the voltage slightly (like to 1.25ish) and see if you can reach the same maximum clock as before (as in see what clock rate will run a minute of furmark at 1.3v, then 1.25v). If at 1.3v you can run at a higher clock, then you know your voltage increase is actually helping, otherwise you are just annoying your card and it could even be responsible for the instability. (i learned this when i started wc and temps ceased to be a limit.)

I'll offer other suggestions if you tell me the following:
What temps are you getting? what psu? what kind of crashes are happening?
Related resources
a b U Graphics card
April 16, 2012 6:36:22 PM

Temps top out at 75C during Furmark @ 1.3V 1250/1450 with fan speed on 40% + Auto adjust.

PSU is an antec CP-850 (plenty of power, but as mentioned, problems with old cards started after power surge)

Crashes are the general screen going black and PC staying on, with the complimentary buzzing noise from the speakers.
April 16, 2012 6:39:28 PM

Stop with the crazy application of 1.3V and see what happens. It sounds to me like you kinda just tossed that voltage at the card and hoped for the best... that's not a good procedure to get a stable overclock.
a b U Graphics card
April 16, 2012 6:45:04 PM

i doubt it's the psu but if you have another on hand, you could always try swapping. A decent 450W will run that setup.

Your crashes sound like my 6950s when i had too much vcore, randomly turning the screen black after anywhere from an hour to many hours of gaming. Does it also have trouble starting up right after crashing? It could be another problem but try lowering the vcore and see if you get the same crashes.

also, download hwmonitor and watch the temps of the other internal gpu components (it shows voltage reg temps and such). See if any of them are unreasonable.
April 16, 2012 6:46:56 PM

slicedtoad said:
i doubt it's the psu but if you have another on hand, you could always try swapping. A decent 450W will run that setup.

Your crashes sound like my 6950s when i had too much vcore, randomly turning the screen black after anywhere from an hour to many hours of gaming. Does it also have trouble starting up right after crashing? It could be another problem but try lowering the vcore and see if you get the same crashes.


+1

lower vcore... reference cards are not designed to have that much voltage thrown through them for extended periods of time
a b U Graphics card
April 16, 2012 8:28:31 PM

you happen to know why too much vcore can cause problems regardless of heat? I'm basing this on my observations and the observations of others but no one's ever explained the mechanics of it.

On a CPU too much vcore can lead to two things: overheating and degradation over time (could be instant if there's way too much voltage) due to electron, uhh, i want to say erosion but that's wrong.
It's different on a gpu though since it can become unstable without taking any damage simply due to too much vcore. Clearly something else is factoring in here, no idea what though.
April 16, 2012 8:33:25 PM

well, for one thing - reference graphics cards almost never have strong enough VRMs to handle such a large voltage increase

It's kind of like trying to do a large CPU overclock on a cheap motherboard... it just isn't a good idea.
a b U Graphics card
April 17, 2012 7:27:24 AM

These cards start at 1.21v, I'm applying a 900mv increase as stated by all the reviewers on all the reference cards.

It may be too much voltage your right, but I was following guides and reviews of all the reference designs.
April 17, 2012 1:52:22 PM

You are not following guides to get a stable overclock. Reviewers do that just to see what they can get to run for 30 minutes through their benchmarks. And obviously your card does that. Reviewers do not generally go through the process necessary to find a 100% stable overclock (which can be very time consuming).

Also, 90mv is a large increase in voltage when you are dealing with a reference PCB/VRM design that isn't designed to handle extra voltage at all. Just because it sounds like it's a small increase doesn't mean it actually IS a small increase.

Again, think about overclocking a CPU on a cheap motherboard. Yes, you can do it. No, it's not a good idea, and you will eventually fry something because it's not REALLY designed to do it even though it CAN.

Lower your vcore, follow good overclocking practices, and then see if you still have problems.
a b U Graphics card
April 17, 2012 2:32:12 PM

yeah, i have to agree. I'm all for pushing a card to it's limits and basically ignoring the risks but other than voltage and clock speeds, there really isn't anything else you can tweak on your gpu, it just isn't stable at that voltage. Try a lower vcore and see what clocks you can get.

If you get the same crashes with stock settings then it might be your cpu/ram misbehaving. Usually a black screen is associated with the gpu but you never know.
April 28, 2012 8:46:41 AM

Hello guys!New here!
Yesterday i bought an Gigabyte 7870 OC 2GB and i have some issues.
I stayed late last night searching about my problem.Well heres whats happening:

After all the unistalling (nvidia before) and installing the new drivers (ATI) i tried my newest game Battlefield 3..I Tried at once at Ultra graphics and i could tell i was soooo excited (yes,i had an Zotac 8800GT 512mb before running at low..!!!!) about the quality..I had some bugs so i lowered a bit the graphics to abjust them at the best -eye quality and performance(generally at High and some ultra)..

Eitherway after some minutes the game crashed..I tried again lowering the graphs and again..I tried Skyrim..Ultra graphics and everything at full..Perfect..i played around 20 minutes without problems and also perfect eye fps..

So after a lot of google searching and after a lot of nvidia cleaning (driver sweeper) lots of Ati Driver unistalling and reinstallings and all the time with the help of GPUZ i started thinking that my problem should be my PSU..How?I played some BF3 and saw the VDDC Current at arround 23A in average and some times even 13A!Before the game crashed (before some secs) the VDDC Current went 15.8 and the 12 Voltage rating was at 11.53 in average..

So i think that these are some low ratings for this card to work properly.But in contrary why the skyrim played at ultra without problems?Propably i should let it working for some more.

So my PC parts are a bit old:
-Intel 775 LGA 2.66 E6750 core duo,
-2x2GB OCZ DDR2 PC2-8500 / 1066 MHz / Reaper HPC Edition / Dual Channel
-Gigabyte 7870 OC (GV-R787OC-2GD)
-Asus P5Q-WS motherboard
-And OCZ 520W modstream [ +3.3V(28A), +5V(52A), +12V(28A) ] and 620 Peak for 60 sec
with one PCIe 6 pin.The other 6 pin connector is the adapter that included in the Graphics card(2x4pin to PCIe 6pin).

And about the antec CP-850 deadjon dont you think that 25A of this PSU are few for this card?Thats urs right? http://www.jonnyguru.com/modules.php?name=NDReviews&op=...
April 28, 2012 10:52:27 PM

Also i just noticed that at the GPU-Z the bus interface sais: PCI-E 3.0x16@x16 1.1 and when i point the mouse on it it sais that the gpu is 3.0 pcie and currently works at 1.1!!Is this normal??I just played Crysis on 1920 at hardcore sets and this card rocks!!!Not even a glitch..!Why at BF3?Compatibility issues?
!