Upgraded PC, but now I have to lower my GPU overclock?

H

HockeyJon

2[H]4U
Joined
Dec 14, 2014
Messages
2,081
Hey guys,

I have a weird one for you, just wondering if anyone had any ideas. I decided to upgrade to a Ryzen 3700X from my old i5 4670k just recently, and went for an entirely new build, keeping my still-functioning older PC together and building on an entirely new platform. Thus, I built it in a new case and decided to use a new PSU.

Weird thing is I now have to lower the offset for overclocking on my GTX 1070Ti, and I can't figure out why. Previously I had it at +150 in Afterburner. I'm getting instability even at +120 now. My knee-jerk reactions would be either thermals or power delivery.

On the thermal side, I'm getting crashes quickly, before the GPU can even warm up. For example, using Firestrike as a test bed, it'll crash after 5-15 seconds, but the GPU only warms up to 55 degrees. Doesn't make a whole lot of sense. My old case had a side intake fan which my new case lacks, but I would only think that might matter if the GPU was actually getting hot.

For comparison on power delivery, my old PSU was an Antec HCG 750W, and I built this new PC with a Corsair RM650X 650W PSU, with plenty of spare wattage according to my calculations for peak system draw. I don't seem to have any PSU stats shown on HWinfo for whatever reason (not sure why), but the 12V, 5V, and 3.3V numbers in the motherboard section are all showing normal values. No distinct undervoltage present.

I tried boosting the voltage offset to 100% on Afterburner. This didn't help stability at all, but I did see that the GPU is now drawing the max voltage of 1.094V under load, as one would expect to see.

VRAM overclock offset remains stable at +375. I tried playing with that one as well, wondering if the lower GPU frequency might free up additional room there. No dice, it's at its limit.

Anyone have any ideas what might be going on? Is there something I'm not seeing?

Thanks!!!
 
My friend had this same issue when he upgraded his video card, even though the old PS was more than sufficient, it did not cope with the new component(s)

Don't rule out the power supply being the problem, just because its new doesn't mean its not faulty. Put your old Antec in and see if you still get crashes. Or return the Corsair and get another one. Personally I would go with a 750-850w at all costs. Better to have more than you need than not enough.
 
Can you put the card back in your old computer and see if you have the same issues?
 
Krenum said:
My friend had this same issue when he upgraded his video card, even though the old PS was more than sufficient, it did not cope with the new component(s)

Don't rule out the power supply being the problem, just because its new doesn't mean its not faulty. Put your old Antec in and see if you still get crashes. Or return the Corsair and get another one. Personally I would go with a 750-850w at all costs. Better to have more than you need than not enough.
Click to expand...

Yeah, I'm thinking I might reach out to Corsair and at least see if they have any suggestions. 650W is definitely oversized for my application, I calculated peak draw a hair under 400W, but I like to try to keep things in the peak efficiency range, so I went with the 650W.

Do you know of any software or anything I could use to check to see if there is a power delivery problem? As I mentioned, my PSU is not showing up on HWinfo, and I'm not entirely sure why, but the motherboard section of that program is showing steady voltages are the proper levels.
 
OK so update for you guys:

First off, one correction to my OP. Original offset was +130, not +150.

Anyway, I took you guys up on your suggestion, slapped my 1070Ti in my old PC, booted it up, and tried to run Firestrike. It crashed. I couldn't believe it. I had run that card at +130 since November, rock solid stable, never had an issue running any game (still don't), and had benched with Firestrike quite a few times according to my records as I was fiddling around tweaking my system.

On one hand, this is good, because I suppose my power supply is not the problem. On the other hand....am I insane? Unless the card itself is starting to have delivery issues VRM style, I can't see why this would crop up now. I presume Firestrike crashing to the desktop is an automatic sign of instability, although I had gamed on that for months without issue. Weird.

Anyways, thanks for the suggestions and help, guys!
 
It may well be a newer driver pushing things harder.

I find this happens with every gen. What you could overclock to from day 1 will always be higher by the time the drivers have matured. Usually you just start seeing artefacts a few months down the line, then comes the random crashing and it starts to feel like your card is dying.

You'd have to roll back to an older drier to test, would be interesting if it's reproducible in this case...o_O
 
spine said:
It may well be a newer driver pushing things harder.

I find this happens with every gen. What you could overclock to from day 1 will always be higher by the time the drivers have matured. Usually you just start seeing artefacts a few months down the line, then comes the random crashing and it starts to feel like your card is dying.

You'd have to roll back to an older drier to test, would be interesting if it's reproducible in this case...o_O
Click to expand...

Interesting.

Also, mirin' your overclocks in your sig, damn!!
 
HockeyJon said:
OK so update for you guys:

First off, one correction to my OP. Original offset was +130, not +150.

Anyway, I took you guys up on your suggestion, slapped my 1070Ti in my old PC, booted it up, and tried to run Firestrike. It crashed. I couldn't believe it. I had run that card at +130 since November, rock solid stable, never had an issue running any game (still don't), and had benched with Firestrike quite a few times according to my records as I was fiddling around tweaking my system.

On one hand, this is good, because I suppose my power supply is not the problem. On the other hand....am I insane? Unless the card itself is starting to have delivery issues VRM style, I can't see why this would crop up now. I presume Firestrike crashing to the desktop is an automatic sign of instability, although I had gamed on that for months without issue. Weird.

Anyways, thanks for the suggestions and help, guys!
Click to expand...
Could be that you just got a lemon, and the card's core is degrading faster than most do.

I do not think the issue is the card's VRM. They don't generally degrade that like this. They just keep working right up until something burns out.

I'm inclined to think that the issue is software, though. The driver has likely changed, and may be using the GPU a little different than it used to. Regardless, I think I'd just live with it and not worry. You're not going to notice a difference of 20MHz unless you're a competitive overclocker.
 
RazorWind said:
Could be that you just got a lemon, and the card's core is degrading faster than most do.

I do not think the issue is the card's VRM. They don't generally degrade that like this. They just keep working right up until something burns out.

I'm inclined to think that the issue is software, though. The driver has likely changed, and may be using the GPU a little different than it used to. Regardless, I think I'd just live with it and not worry. You're not going to notice a difference of 20MHz unless you're a competitive overclocker.
Click to expand...

Totally agree, that's what I basically concluded yesterday. I couldn't figure out how this would break down to a power delivery issue for the reason you stated, and I've just come to accept that I lost badly on the silicon lottery. I guess that'll just provide me a more compelling upgrade path for the future.

Good point about the software, I hadn't thought too much about that being a potential cause, but I'm inclined to agree that it's also within the realm of possibility at this point, after I clearly ruled out anything new in my new build holding it back yesterday.
 
You must log in or register to reply here.
Back
Top