Is it okay to overclock my GPU without water cooling?

ProperlyPsychotic

Commendable
Nov 12, 2016
74
0
1,640
Hi, I currently have 2 fans in my system, one is a CoolerMaster SickleFlow 120, and the other one is the stock fan that came with my case (Cougar Spike) I would like to overclock my GPU which is the EVGA GTX 1060 6GB ACX 3.0 http://www.newegg.com/Product/Product.aspx?item=N82E16814487280 and I would like to know if this is safe. I am not overclocking by a lot. And, if it is safe, what are some good programs to overclock with?
 
Solution


No it doesn't matter. They both do the same thing. You could use precision x with an msi card and afterburner with an evga card. There should be no difference in your overclocks. I like afterburners monitoring...
Yes of course, the ACX 3.0 cooler is more than enough for a decent overclock. You likely wont get much out of it however. My msi gaming 1060 will boost to 1974mhz on its own without manual overclocking, I can push it to about 2050mhz before I start artifacting, too much past that and crash.

I prefer msi afterburner for overclocking software, not just because I have an msi card, I feel it is the best utility. However EVGA precision x will work as well.
 

Would the EVGA Precision be better due to having an EVGA card?
 


No it doesn't matter. They both do the same thing. You could use precision x with an msi card and afterburner with an evga card. There should be no difference in your overclocks. I like afterburners monitoring tools better and I feel the interface is less cluttered and more intuitive, at least in my opinion. Try them both see what you like.
 
Solution
The 1060 is 120 watts ... I'm gonna guess that you have a 90 or so watt PSU and maybe hitting 120 w/ your moderate OC. Are those 2000 rpm fans ?

For low noise systems, I recommend one 120mm fan (1200 rpm) for each 50-75 watts
For moderate noise systems, I recommend one 120mm fan (1650 rpm) for each 75-100 watts

Beyond that, it can get a bit noisy.

is this your case ?

https://www.techpowerup.com/reviews/Cougar/Spike/3.html


It would appear that the case can take 4 fans. Assuming all the same fans ... the inlet filter on the front will result in up to 1 33% cfm hit... so let's call it 1 fan in back and one in 0.67 fans in front. That means you have more air going out than in and therefore you have negative case pressure and you are sucking in dust from side and rear grilles. Add some hot PSU exhaust and GFX card exhaust to that dust on the rear grille.

Flow wise, a 2nd fan on the side panel would make it 1.33 in and 1.0 out. I would block the 2nd inlet on the side (Hey, black take on the inside works if ya have nuttin' bettah) so air blown in thru the 1st wouldn't just push right out the empty fan mount next to it. If ya can spring for a 2nd additional fan, 3 in w/ inlet filters would be equivalent to 2 in and 1 out and that's great. You can run them all a bit slower for less noise.

Now to an area of concern for EVGA cards which has existed since the 5xx series. EVGA has consistently paid little attention too cooling memory and VRMs. The 570 VRMs had a tendency to fry ... the 970's cooler missed the GPU and they have a fix out that requires an our of disassembly / reassembly to add / replace thermal pads on the 1070 / 1080 SC and FTW w/ ACX coolers

Note that the EVGA 1070 / 1080 ACX coolers have a deficient design where the lack of necessary thermal pads have caused overheating

https://www.youtube.com/watch?v=WAbl0fLY06U
https://www.youtube.com/watch?v=qZ0eAylzl0o

But the 1060 ACX has not been shown to be affected, [well until recently, see EDITS] and at just 120 watts you wouldn't expect it to be a problem ... whether OCing makes it a concern, I just don't know but a) I doubt it and b) you could add it in about an hour's effort if you are affected.

https://www.youtube.com/watch?v=Tdn5r___1Iw

So, if it were me, and assuming you have the case as linked, I'd add the two fans, set up your fan curve and then try your OCs seeing what temps you get . The MSI gets to 73C on the GPU.... and I wouldn't expect the EVGA SC / FTW ACX to get significantly hotter but the thermals on the VRM is something you should keep an eye on.

EDIT:

Found this, I didn't read thru it but you may want to ... apparently problems exist on some EVGA 1060s as well.

http://www.tomshardware.de/nvidia-geforce-gtx-1060-pascal-gp106-grafikkarten-roundup,testberichte-242152-12.html

The card is surprisingly quiet and is also in the trend in the trend, but suffers somewhat from the lack of cooling in the area of ​​the voltage transformers. With a few handies, however, the user could even help himself, especially since EVGA explicitly allows the removal of the radiators without loss of warranty.

More here ... again, not sure what models may be affected, you might be able to get more from those links.

https://www.reddit.com/r/nvidia/comments/59i24w/its_not_just_the_evga_10701080ftws_having_vrm/

http://forums.evga.com/EVGA-GTX-1060-6GB-ACX-20-blew-up-and-caught-fire-m2568927.aspx

That one shows a fried EVGA 1060 ACX same problem as EVGA 1070 and 1080.
 
Anybody regardless of knowledge, talent or experience can make a video. Putting it mildly, sites like Gamersnexus and JayzTwoCents are sites I'd only visit for "entertainment purposes". After all, how much trust can ya put in a guy who drills 4 holes thru an expensive X99 MoBo and thinks going in "what could happen ?" I find it "interesting" that the same guy who is saying "This VRM thing isn't a problem is the the same guy who posted a video of the repair process.

In this video, he describes as the thermal pad as the "recommended or required approach" to resolving the VRM thermal problems. Also says they've done a lot of them ... so if it's not a problem, ya have to wonder, why are they doing a lot of them and calling it the "required or recommended" solution ?

He also says "Still hitting 114C is really hitting a fine line for conditions that could trip a runaway thermal scenario especially since the backplate is effectively acting as a hotbox before applying thermal pads." If one tries to kill a card and can't, it's not a unknown fact that the quality, performance of all electronic components are variable. If they weren't we'd all have 5.0 Ghz overclocks on our CPUs. The 2% that can do that because they can reach that clock at lower voltages than most others and therefore they operate at lower temps which fall in the "acceptable" range.

Let's also remember that:

a) GN acknowledge 114C VRM temps which is a place no one should want to go. When they claim that because the VRMs are rated at 115C, 114C is OK.... does that mean running our CPUs at 99C is OK too ?

b) High VRM temps also limits the stability of card OCs

c) EVGA has a long history of "cheaping out" on the SC series where VRM and memory cooling is concerned. Historically, EVGA has used just a plain stock VRM / PCB with no additional cooling ... only the changed cooler allows it to be considered as an AIB card. On rare occasions they have made slight changes. I recall once that they swapped brands but used a competing VRM or comparable cost / quality and same number of phases. For the 970, they added 1 extra phase where their competitors all had more. Their SC line memory and VRM cooling has always been lacking compared to comparable cards from Gigabyte, MSI and Asus and it shows up in what the card can deliver overclocking wise. If you go back and look at forum posts from the 5xx era, you will see many, many posts from fried VRMs on the 570 SC as well as other cards with stock VRMs. The stock VRM was simply undersized and undercooled like most reference cards even today. And on the 970 SC, 1/3 of the heat sink actually missed the GPU. The FTW line was essentially the SC with better VRM and better memory / VRM cooling ... but with 10xx, the SC and FTW solutions are the same in this respect.

And if you watch that GN thermal imaging video, he puts up the picture on the cards for which EVGA is supplying thermal pad solutions....the 1060 is now included. We see this on the forums, we see this in youtube reviews ... when these things occur, someone always responds "Well I have that card and it's still working so the problem is imaginary." Not everyone has a chip from the same wafer, the same ambient, the same case / cooling solution, the same OC, the same loads. What percentage of cards must fail before it is accepted as a problem ?

The last FTW we had could not run at the factory OC. That FTW demanded a nice price premium and GPU temps were fine ... the most logical culprit therefore would be the VRM. After 18 months, 20 support calls and 5 RMAs we finally had something that worked at advertised speeds... but it was a next generation card that they substituted.

Anyway, getting back to the OPs problem .... these are the 1060 cards affected having the following part numbers.

03G-P4-6365
03G-P4-6167
03G-P4-6165

http://www.evga.com/thermalmod/

If you have this card, once this mod is done, should take about an hour, the card will be fine and perform comparable to other cards in its price range, If you find the process daunting, you can send back to EVGA and they will do it for you. If willing to brave it this guide's you thru ... was posted above, but for your convenience

https://www.youtube.com/watch?v=Tdn5r___1Iw





 
Well then... I was not trying to argue with you I just posted a video claiming this VRM debacle is not as bad of a problem as its made out to be, if you disagree fine, honestly I dont know enough about VRMs and gpu power delivery to say one way or the other.

Nor am I defending EVGA or claiming GN is 100% accurate I only posted that video in case you hadn't seen it. I absolutely agree you should do the thermal pad mod or at least get their new bios, I did not mention it before because I was not aware the 1060 was affected as well.
 
Yes, was news to me too.... With it's power draw, I am amazed that you could actually create a problem no matter what is ignored. But one thing great about this debacle is this ... it perfectly illustrates the uselessness of hybrid cooling solutions. You could swap out the ACX cooler for a hybrid cooling solution and unless that solution breaks the mold and actually addresses VRM / memory cooling, it will accomplish nothing.

With typical AIB air cooling ... you temps are in low 70s when OC'd... VRMs are up in the low 80s. Dropping it from 73C to 50C on the GPU won't help as 73C was no where near the throttling point, your card will not be affected by throttling. But your typical hybrid solution mostly leaves VRM cooling alone but in some cases makes it worse as that big fat heat sink and fans are gone.
 
Have to say guys, I would imagine OP is lost, and wondering what he did to deserve this lol, but I have had an education. All v interesting and anyone who wants to know more about GPU's should read this thread.

But you know what world class fighters do when that final bell has gone? They hug and shake hands. Thanks for the lesson! Feel like I learnt something today.