Gigabyte GTX 770-OC-2GD Overclocking...

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690
Hello to all my friends here.

So, I've read pretty much all the posts here on tom's regarding the general PSU requirements for this GPU but did not find anything specifically concerned with overclocking it. Well first things first, below is my rig:

i5 3570
8GB Corsair Vengeance 1600 DDR3
Asus B75M-A mobo
Gigabyte GTX 770 WF3
Xonar DX
1TB HDD
80GB GDD
Asus DVDRW
CM Silent Pro Gold 550W

Questions:
1. As one of the posts here suggested that this particular PSU does meet the requirements to run this GPU in normal circumstances (having 42A on a single +12V rail and all), what about OC'ing it using the same PSU? Will I be safe? I really can't upgrade my PSU at this time.

2. I also learned that EVGA's Precision X is much preferred over gigabyte's OC Guru II. Could someone please reconfirm it ?

3. Using any of the OC software (in case the current PSU allows me), what are the power / voltage / temp. / clock / vram (etc. etc.) targets that I should keep in mind to hit, without frying up the GPU, or my whole rig for that matter?

As I'm totally new to this overclocking world and have no idea what I'm talking about, any help would be greatly appreciated.

Thanks very much in advance.
 
Solution
kombutstor, furmark, occt... their all the same thing. its a furry algorithm meant to stress the video card much more than most applications ever will. but i wouldn't run them constantly as they are not very reflective of daily use. some manufactures recommend not using them. other benchmarks like unigine or 3dmark give you scores so you can see your relative increases in performance from overclocking. sometimes certain overclock settings can actually decrease performance.

afterburner has the temp target, your just not seeing it. there is a little arrow in the right side of the power target that drops down the temp target bar. apply overclocking at system startup is safe and means exactly what it says, it overclocks to whatever...

That_Guy88

Distinguished
Dec 31, 2011
76
0
18,660
Clearly OC'ing a card while you don't have a recommended PSU is going to be risky. I think you are probably OK to play normally with that PSU, but OC'ing is going to be risky. The 770 can pull as much as 240 watts, so OC'ing will increase that.

http://www.anandtech.com/show/6994/nvidia-geforce-gtx-770-review/17
This page shows an OC'd system using the GTX 770 pulling about 400 watts. That is getting pretty close to the limit of your PSU (right at 500).

Could it work? Maybe. Could it burn out you PSU and other components? Yes.
You have to ask yourself, is the extra 5-10% gain worth possibly burning out your PC.
If it were me I wouldn't do it.
 

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690

First of all thanks man.

So how would you quantify that 5-10% gain? I mean how many frames per sec. tops I could gain?

Also, what choices would you have recommended (brands/capacity) if I could afford to upgrade the PSU?

Thanks again.
 
1. As one of the posts here suggested that this particular PSU does meet the requirements to run this GPU in normal circumstances (having 42A on a single +12V rail and all), what about OC'ing it using the same PSU? Will I be safe? I really can't upgrade my PSU at this time.
if it can handle it stock, it will handle it overclocked. I'm not sure of the quality of CM supplies though.

2. I also learned that EVGA's Precision X is much preferred over gigabyte's OC Guru II. Could someone please reconfirm it ?
It's the only program I use on my nvidia cards.

3. Using any of the OC software (in case the current PSU allows me), what are the power / voltage / temp. / clock / vram (etc. etc.) targets that I should keep in mind to hit, without frying up the GPU, or my whole rig for that matter?
Every single card is different. But Power; 106%, Voltage; Stock, Temp; <85c. The clocks depend on the card, I would go for 1150 core and leave memory stock though. My 770 hits 1400/9Ghz, but it's golden. Others do not move past stock. If your boost is already at 1150 or higher try for a 20-50Mhz bump.
 

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690

Ok.

And what about the cooling? Is it a must for OC'ing?
 

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690


It's a 770 WF3 2GB model.

Ok then. So i'll try it today and will get back with the data regarding how far mine could be stretched.

Thanks once again my friend.
 

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690

Well as per your suggestions, using Precision X, I set the temperature at 85C (the power target automatically set itself at 104%) and increased the clock gradually to 50mhz (my boost is 1189 by default). At 70 I started seeing red dots (in the rain drops scene) so I reverted it back to 50. And I was about to drop it to 30 and finalize it.

These I suppose are really modest settings but only 15 minutes into the process, I felt something was burning (I was testing with Unigine Valley and the temperature never rose above ~68C); immediately stopped and closed everything and restarted the PC right away :D ...

tested Far Cry 2 after that and thank God everything is fine (@ ultra settings, 1080p, 8xAA)...never doing it again I love my 770 LOL...

What do think must have happened? I'm sure this burning smell at around 70C is not normal.
 
your power supply is fine, it gets decent reviews. it has a single 42amp 12v rail. you can overclock the cpu and gpu and have no problems. msi afterburner is the best overclocking tool as it can be modded to allow for certain settings on certain model cards that other overclocking software cannot. none of the overclocking software, afterburner/precision, will allow you "fry" your card. if you find a custom bios and mod afterburner, then you could run into trouble if you dont know what your doing. just use the unigine valley benchmark and let afterburner/precision log your temps and make sure you dont go over 80c on the core or above 85c vrm's(if you can even read vrm temps) because thats your thermal limit.
 

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690

1. Whether with Afterburner or Precision, did I have to I take the power and temperature targets to the maximum first, like this guy here (http://www.youtube.com/watch?v=4uYEJuLA8os) is doing? Is it safe?

2. By logging in Afterburner/Precision, do you mean the "Log history to file" option in monitoring tabs in both of them? The monitor-able item list does have GPU temp. but not memory/vrm temp. What other options should I log?

As you can imagine this OC'ing stuff is tempting. I never wanted to do it again but could you please suggest what gain am I looking for here? I know those gains will be different in every game. But still I wouldn't want to tinker with it if the maximum I'm gonna get out of it is 2-5 fps.
 
its likely you dont need to change the temperature target as your not going to reach it anyways. if you do the card will start throttling the core clocks which is annoying. the power target increases the tdp headroom, but you might not need to increase it.

i prefer msi afterburner for overclocking. when i say logging, msi afterburner has the side attached hardware monitor that gives you a host of measurements... mostly you will be paying attention to the core clock and gpu temp. you should download gpuz if you already haven't to check that your core clocks are registering when you change and apply them in afterburner.

go download unigine valley and run the extreme hd preset and benchmark it to check how your overclocking is affecting your score and min/max fps. when the benchmark is done... look at the hardware monitor and see what the max temp you got to was and look at the core speed graph and see if it was holding steady at the speed you overclocked it too. if its going up and down during valley, then your card is throttling and you need to increase the power target and or you were hitting 79c/80c, whatever the temp priority is.

if you want more information i would go here....
http://www.overclock.net/t/1396319/official-gtx-770-owners-club
 

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690
You mean to say power and temp. targets (in Precision) cause throttling of the core clocks once reaching their respective limits? So in that case if I move those sliders in Afterburner to the maximum, then the limit before throttling occurs should also be increased. right?

There is nothing to increase the temp. target in Afterburner and it too doesn't show anything anywhere regarding vrm temps. Also, is it safe to select the option "Apply overclocking at system startup"?

One thing I noticed last night, was that Kombustor kept increasing the temperature and capped it to 80C (although I believe before running Kombustor, I had pushed power and temp. sliders in Precision to maximum, that is 111% and 95C respectively). Whereas Valley never crossed ~68C. Could you please suggest a difference between Kombustor, Furmark, Valley and Heaven or whether they're used for different purposes? Which one is the most recommended?
 
kombutstor, furmark, occt... their all the same thing. its a furry algorithm meant to stress the video card much more than most applications ever will. but i wouldn't run them constantly as they are not very reflective of daily use. some manufactures recommend not using them. other benchmarks like unigine or 3dmark give you scores so you can see your relative increases in performance from overclocking. sometimes certain overclock settings can actually decrease performance.

afterburner has the temp target, your just not seeing it. there is a little arrow in the right side of the power target that drops down the temp target bar. apply overclocking at system startup is safe and means exactly what it says, it overclocks to whatever settings you have applied last automatically. for quick testing and benchmarking you can go over 80c but i wouldn't for very long. 95c is getting way too hot. for daily 24/7 overclock your going to want to be under 80c.
 
Solution

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690
Some final questions bro.

1. I doubt the card can consume more power than the maximum, that PCIe slot and the two PCIe power connectors can provide (which if I'm not wrong is ~275W). Then if the power target is set at, say 106%, does it mean the slot and power connectors could possibly provide the card 6% more power if need be, on top of ~275W? And in that case, will my 550W PSU be able to accommodate?

2. Will tinkering with voltage harm my GPU?

3. Is it true that some people recommend using the OC application from the card's OEM? Like is it better for me to use OC Guru for Gigabyte card, as compared to Afterburner?
 
1. I doubt the card can consume more power than the maximum, that PCIe slot and the two PCIe power connectors can provide (which if I'm not wrong is ~275W). Then if the power target is set at, say 106%, does it mean the slot and power connectors could possibly provide the card 6% more power if need be, on top of ~275W? And in that case, will my 550W PSU be able to accommodate?
550w PSUs can handle any single GPU setup. Even an unlocked TITAN only pulls ~450w with voltage at 1.4v. PCI-e power standards "limits" with 6+6+slot is 225w, but these standards are loose and many cards overdraw power. Even overclocked and overvolted my 770 can only get to 83% TDP.

2. Will tinkering with voltage harm my GPU?
It can. Nvidia has strict limits on these cards anyway, so it shouldn't be an issue outside of cards like the lightning that do have extended voltage control.
3. Is it true that some people recommend using the OC application from the card's OEM? Like is it better for me to use OC Guru for Gigabyte card, as compared to Afterburner?
People do say that, but I've never seen a problem with it. I use HIS iTurbo on my AMD cards, PrecisionX on my nvidia cards, and Afterburner as a fallback since it supports near everything.
 

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690
I'm really sorry guys. Although I've got all my major queries answered but I'm seeing this strange thing here.

In Precision logs/monitoring, @ 60mhz core offset, the TDP never crosses 76% (target = 106) which somehow I believe is preventing the clocks to properly boost. The temps. too never go above 76C (target = 80). Valley just keeps freezing for a couple of seconds and then continues; Precision logs show the GPU downclocking to 1215 after that freezing moment, while the power and temp. being way below the limits. How am I going to overclock properly if the power consumption/temp. limits aren't properly put to good use?

Am I missing something here? Do i need to modify the voltage?


It's not there man :). Exactly what you're hinting to here, appears to be present in Precision, not Afterburner. I've got AB 2.3.1 and it does not show anything for temp target. But I'm gonna prefer Precision X. Thanks anyway.
 
It is throttling temps. My 770 throttles at 90c (It was a test, usually in low 70s.) even though temp target is 95c. Try setting power limit to 106% and temp target to 95c. Make sure to watch temps though, as I don't know how the new windforce does with it's fancurve.

 

rlakhani11

Honorable
Aug 8, 2013
234
0
10,690

As you suggested, I set the power target to 106% and temp target to 95C (with +80 core offset this time; +60 went fine). This is what happens (OC Guru II logs):
GPU(MHz) Voltage(V) Power(%) Temp('C)
135 0.875 6 33
784 0.912 20 36
640 0.875 10 34
941 1.000 33 39
967 1.012 35 40
967 1.012 35 41
967 1.012 35 41
1293 1.200 54 47
1293 1.200 61 49
1293 1.200 58 49
1293 1.200 29 44
1137 1.125 23 41
1137 1.125 34 43
1215 1.200 65 51
1215 1.200 58 51
1215 1.200 50 49
1215 1.200 9 50
1215 1.200 50 50
1137 1.125 23 44

I doubt the formatting of the above table is how you'd desire. Apologies for that. But the point is that the temp. doesn't go beyond 51C and it starts downclocking for some reason! As you can see, there's never a need to throttle the temps by the look of things. What do you suggest is happening?

EDIT: Below are Precision logs which look quite drastically different:
Power GPU-clock Temp voltage
68 1293.387 52 1.200
68 1293.387 51 1.200
77 1293.387 54 1.200
82 1293.387 55 1.200
48 1136.612 45 1.125
48 1136.612 45 1.125

The TDP reaches 82% here, but the temp. is nowhere near the throttling point, where it actually starts throttling something/downclocking!