How far can you push your GTX 1660 Ti?

[SomeTechScrub]

Well, I have been trying my best to push my GTX 1660 Ti for the past few days. Nothing crazy of course, just regular air cooling. I've also tried replacing the stock thermal paste. Still not able to get enough FPS to stop lagging some minor lag. So far I've produced the best results with +100 on core and +1000 on Memory. Just wanted to know what everyone else got so maybe I can try out what yall got. Game of focus for me would be Rainbow Six Siege.

[Jurrunio]

What is the actual reported frequency?

[SomeTechScrub]

2 minutes ago, Jurrunio said:

What is the actual reported frequency?

I jumped between 2010 - 2050 on core, 7000 on the memory

[Jurrunio]

2 minutes ago, SomeTechScrub said:

I jumped between 2010 - 2050 on core, 7000 on the memory

That's not very high for a Turing GPU, but I've seen worse. What exact model is this? How hot does it run?

[SomeTechScrub]

GALAX 1660 Ti. 65 degrees at max. But if I push the card further, I seem to get less FPS : /

 

[LukeSavenije]

@Schnoz your time has come

[SomeTechScrub]

2 minutes ago, Schnoz said:

2140 MHz core, 15 GHz effective memory. 55 C under load, liquid-metalled.

English please xd. I can't understand what your trying to say

[Zando_]

7 minutes ago, SomeTechScrub said:

English please xd. I can't understand what your trying to say

I assume they're saying they put liquid metal on the GPU and OCed it to that, not sure how that helps your situation. 65C is nowhere close to too hot, and the 1660 Ti doesn't make much more heat as you ramp up the power target. 

What do you mean by the fps gets worse after a manual OC? And what CPU/RAM are you running with it on what monitor? 

[SomeTechScrub]

3 minutes ago, Zando Bob said:

I assume they're saying they put liquid metal on the GPU and OCed it to that, not sure how that helps your situation. 65C is nowhere close to too hot, and the 1660 Ti doesn't make much more heat as you ramp up the power target. 

What do you mean by the fps gets worse after a manual OC? And what CPU/RAM are you running with it on what monitor? 

If I push above +100 on core and +1000 on memory, I get less frames. CPU would be a 2600X @ 4 Ghz Ram would be Corsair Vengeance @ 3133 or smth like that. I use a 1920x1080 165 hz monitor

[Zando_]

3 minutes ago, SomeTechScrub said:

If I push above +100 on core and +1000 on memory, I get less frames. CPU would be a 2600X @ 4 Ghz Ram would be Corsair Vengeance @ 3133 or smth like that. I use a 1920x1080 165 hz monitor

Are you sure your OC above +100 is stable? I normally run Unigine Heaven on a loop in the background while changing settings, if anything seems iffy or you see even slight artifacting, your OC isn't stable. If you wanna find the absolute max for your card, I'd put the voltage, power target, temp, and fan sliders up all the way (Nvidia won't let the card pull more voltage than is safe, you'd need hardware mods to do that IIRC), then have a go at it. I've found the voltage slider usually doesn't do much though, but it's worth a shot. Then once you find a stable max, you dial back the fan curve to keep it at a reasonable temp but not unbearably loud. My EVGA 1660 Ti XC Ultra will stay around 50-52C with the fans at 100%, but it's very loud so I dial them down to keep it in the mid 60s. 

[SomeTechScrub]

8 minutes ago, Zando Bob said:

Are you sure your OC above +100 is stable? I normally run Unigine Heaven on a loop in the background while changing settings, if anything seems iffy or you see even slight artifacting, your OC isn't stable. If you wanna find the absolute max for your card, I'd put the voltage, power target, temp, and fan sliders up all the way (Nvidia won't let the card pull more voltage than is safe, you'd need hardware mods to do that IIRC), then have a go at it. I've found the voltage slider usually doesn't do much though, but it's worth a shot. Then once you find a stable max, you dial back the fan curve to keep it at a reasonable temp but not unbearably loud. My EVGA 1660 Ti XC Ultra will stay around 50-52C with the fans at 100%, but it's very loud so I dial them down to keep it in the mid 60s. 

I seem stable, no artifacting from what I can tell

[Zando_]

1 minute ago, SomeTechScrub said:

I seem stable, no artifacting from what I can tell

Huh. I know Radeon VIIs report frequencies wrong (it can say it's 2100Mhz or whatever but get a lower score than stock), but to my knowledge and experience Turing doesn't do that. You definitely shouldn't drop performance when increasing clocks. What are you using to bench it for fps numbers? 

[SomeTechScrub]

Just now, Zando Bob said:

Huh. I know Radeon VIIs report frequencies wrong (it can say it's 2100Mhz or whatever but get a lower score than stock), but to my knowledge and experience Turing doesn't do that. You definitely shouldn't drop performance when increasing clocks. What are you using to bench it for fps numbers? 

I use the in game benchmark in R6:S and Heaven Benchmark and Furmark

[Zando_]

Just now, SomeTechScrub said:

I use the in game benchmark in R6:S and Heaven Benchmark and Furmark

Furmark is more of a VRM torture test, it creates a massive power draw and pushes the GPU to the highest power consumption (and thus, hottest temps) possible, it's not really a benchmark. The in-game benchmark for R6:S should be fine, I forget if R6 is hard on the CPU though. Heaven is pretty old, even Valley is but they're still purpose built GPU benchmarks. You basically want something that's gonna push the GPU really hard but not the CPU (like benching GPU overclocks in CS:GO or ARMA III would be a terrible idea). 

 

Do you have stock vs overclocked scores recorded somewhere? And how many runs did you do? You can drop fps on the first or second run sometimes, usually you do 2-3 and either pull the best score or average all of them. 

[SomeTechScrub]

2 minutes ago, Zando Bob said:

Furmark is more of a VRM torture test, it creates a massive power draw and pushes the GPU to the highest power consumption (and thus, hottest temps) possible, it's not really a benchmark. The in-game benchmark for R6:S should be fine, I forget if R6 is hard on the CPU though. Heaven is pretty old, even Valley is but they're still purpose built GPU benchmarks. You basically want something that's gonna push the GPU really hard but not the CPU (like benching GPU overclocks in CS:GO or ARMA III would be a terrible idea). 

 

Do you have stock vs overclocked scores recorded somewhere? And how many runs did you do? You can drop fps on the first or second run sometimes, usually you do 2-3 and either pull the best score or average all of them. 

Hmm.... Alright. It's quite late for me, I'll find time to slowly benchmark

[Zando_]

4 minutes ago, SomeTechScrub said:

Hmm.... Alright. It's quite late for me, I'll find time to slowly benchmark

If I had a CPU to pair with mine I'd boot it up again and do a proper bunch of benchmarks at stock vs OCed to see how it behaves, but I just sold my i5 8400 and only have old Xeons that'll probs bottleneck it a bit...  (main rig is on a custom loop now so I can't just swap out GPUs anymore). Let me know how it goes, and I may put together a rig with one of my Xeon rigs this weekend and see how it does regardless. 

[SomeTechScrub]

12 hours ago, Zando Bob said:

If I had a CPU to pair with mine I'd boot it up again and do a proper bunch of benchmarks at stock vs OCed to see how it behaves, but I just sold my i5 8400 and only have old Xeons that'll probs bottleneck it a bit...  (main rig is on a custom loop now so I can't just swap out GPUs anymore). Let me know how it goes, and I may put together a rig with one of my Xeon rigs this weekend and see how it does regardless. 

Alright, I spend a solid hour with +120 and +1000. Averaged around 2515. I let the PC cool down after each run (Did 3 runs). Its quite time consuming so I've abandoned my tests. I'll make another thread about how far I've pushed my GTX 1660 Ti and what I did during each run.