Can overclocking damage GPU?

[Shing-Fhue]

I have a gtx 1050 ti and some stupid guy said that I could overclock it to 175 MHz to core and 700 MHz to clock, and it ran fine but programs stopped running and csgo was super glitchy, then I saw on youtube that another person did 150 MHz to core and 500 MHz to clock and showed that anything higher would cause exreme glitching. So even without overclock some games are really glitchy now, but I still get extremely high frame rates. Have I broken my GPU or are the games not optimised with the 1050 ti?

[[LTT] LinusFTW]

No it's totally fine however no card can get a +700 offset.

[Shing-Fhue]

I meant +700 to the memory clock my bad.

[Shing-Fhue]

My old integrated graphics and 745 OEM before that ran CS:GO better lol this card is really glitching it. So you think its just that it isn't optimised for the 1050 ti or the new Pascal architecture?

[SundownKid]

Yes, OCing can damage a GPU if you make the settings too extreme.

[[LTT] LinusFTW]

Originally posted by Nerfed2Much:

I meant +700 to the memory clock my bad.

+500 - +600 is the max for GDDR5 anything above it can cause the PC to crash or act weird.

[Shing-Fhue]

The +700 memory was no where near the max I can do. It was about half-way slid over. Is there a way I can post a screenshot?

[[LTT] LinusFTW]

Originally posted by SundownKid:

Yes, OCing can damage a GPU if you make the settings too extreme.

No

https://www.youtube.com/watch?v=-qn6sf-U2rQ

[Shing-Fhue]

Ok thank you for showing me this video I will watch it and see if it helps. :) Thank you for the help and happy holidays.

[®NapalmSunrise®]

Originally posted by Nerfed2Much:

The +700 memory was no where near the max I can do. It was about half-way slid over. Is there a way I can post a screenshot?

Just because you went say 50% across that slider doesnt mean your GPU can/will handle it all they way over to 100% of what it will allow you..

[Fluffy]

i dont recommend 700mhz on the memory 4-500 is the usual max depending on whether its hynix or samsung or other hynix has been the best for me

ive had gpus oced and used until too old with no issues of longevity

what are you using to oc and whats the current max boost clock reached?? monitored by gpu-z or another program

[Shing-Fhue]

MSI Afterburner is what I use. Max boost clock reached was 700 I never went furtyher than that. The specific GPU is the EVGA GTX 1050 ti SSC, it has 2 fans.

[Revelene]

Well, what exactly do you mean by "csgo was super glitchy" and "some games are really glitchy"?

You'd know if you damaged a GPU. Not only would games have issues, but also the desktop. Anything that overclocking would do would not only affect video games.

You could overheat the card, but that is about it. And even that would prove difficult to actually do any damage to a card, as it'll shut down long before damaging temps. Now, excess heat over long periods of time could potentially lower the lifespan of the components.

You don't have as much freedom as you think you do. Mainstream overclocking programs simply access what is already set in the firmware as maximum offsets allowed.

I'd say the issue is something else.

[Sticky Honeybuns]

Originally posted by LTT LinusFTW:

Originally posted by SundownKid:

Yes, OCing can damage a GPU if you make the settings too extreme.

No

https://www.youtube.com/watch?v=-qn6sf-U2rQ

Yes, you can damage a GPU if you up the voltage too much.

I like LTT but I'm far more educated than he is and I work with 22nm gate structures and smaller on a daily basis. We use focused ion beams to isolate gate structures then electrically overstress them to failure and then characterize these failure modes. We have been working on finfets(pascal) for the last two years with the same failures.

Point being, electrical overstress can cause catastrophic failures.

[Revelene]

Originally posted by Sticky Honeybuns:

Originally posted by LTT LinusFTW:

No

https://www.youtube.com/watch?v=-qn6sf-U2rQ

Yes, you can damage a GPU if you up the voltage too much.

I like LTT but I'm far more educated than he is and I work with 22nm gate structures and smaller on a daily basis. We use focused ion beams to isolate gate structures then electrically overstress them to failure and then characterize these failure modes. We have been working on finfets(pascal) for the last two years with the same failures.

Point being, electrical overstress can cause catastrophic failures.

Yeah, but without actual modification of the firmware, you'll never actually be able to have voltage be anywhere near harmful.

[Zireth]

temps are the damaging part. get over a point an that goes on the lifespan of your hardware.