Overclocking doesn't bring any benefits.
Overclocking doesn't bring any benefits.
Hello
I've been using my ASUS Strix 2080 Ti for a few years, but I decided to upgrade recently. Now I still see the same 3DMark score with VRAM at 15,800 MHz and 16,100 MHz. Why is that? It’s not crashing or anything. I updated the ASUS Matrix BIOS, thinking it has 300W TDP compared to Strix’s 250W, but how can I confirm its actual TDP and whether it’s limited?
In Time Spy, I’m getting around 10,000 CPU score and 16,432 GPU score, totaling about 15,000.
Here are the specs:
Asus x570 TUF
Gskill 4x8GB die, 3600MHz to 3800MHz at 16-16-16-36
Ryzen 3700X stock, 1900MHz clock
Seasonic Prime Gold 1200W
A few Samsung and WD SSDs, total 4TB.
You hot-rod a car, add a bigger fuel pump, cleanup and tune the exhaust, fine tune the timings, cold air induction, all the little extra HP tricks that add more speed to the top end. And that adds up to exactly nothing because the fastest you can go is @ 80mph on the highway.
Gpu is no different. You can trick up the vram, clock speeds and still you'll run into limitations. Whether that's power limits, transmission speed limits, software induced bandwidth limits, cpu fps limits, gpu silicon limits, vrm limits or any other limit doesn't matter, you'll still reach a limitation that prevents the card from going any faster.
There's only so much a card can do, regardless of internal, external or user implied boosts. Otherwise it'd be...
Occasionally checking the VRAM can lead to a lower score than a faster VRAM speed. I've noticed that users with 2080 Ti who run on watercooling can achieve high scores like 17000 in Timespy, while those using air cooling don't. I'm not interested in physically modifying my card or using water cooling.
Hi, Hitman2022,
Watch power limits and temperatures closely with and without the OCs.
The Gpu Boost algorithm is a GPU OC roadblock: hitting power limits more often and raising core temperatures can undermine your efforts.
Unlike some OC guides, there isn’t much room for improvement—this has been the case since the GTX 10 series.
Plus, aftermarket GPUs already come with factory OCs on top of what Gpu Boost provides.
For a 2080Ti, you might need a thermal probe or another device to check VRAM temperature, and keep in mind that memory OC stability is questionable at those settings.
You can use the Gpu-Z sensors tab or hwinfo to track the GPU’s parameters.
The variation in VRAM speeds is usually under 2%, which falls within acceptable limits. Moreover, boosting the speed doesn't always lead to proportional performance gains. Regarding lower scores paired with higher VRAM rates, running the VRAM at higher speeds can introduce intermittent issues that are fixable, though they may result in the GPU slowing down.
You hot-rod a car, add a bigger fuel pump, cleanup and tune the exhaust, fine tune the timings, cold air induction, all the little extra HP tricks that add more speed to the top end. And that adds up to exactly nothing because the fastest you can go is @ 80mph on the highway.
Gpu is no different. You can trick up the vram, clock speeds and still you'll run into limitations. Whether that's power limits, transmission speed limits, software induced bandwidth limits, cpu fps limits, gpu silicon limits, vrm limits or any other limit doesn't matter, you'll still reach a limitation that prevents the card from going any faster.
There's only so much a card can do, regardless of internal, external or user implied boosts. Otherwise it'd be possible to drive a 2060 to equal a 3090's output by extreme OC and some outrageous cooling.