Question RTX 3060 Ti running twice as slow ?

[patrykupisz]

Recently I upgraded from GTX 1070 to an Gigabyte RTX 3060 Ti Gaming Oc. I started noticing that in most games my GPU usage with 1070 was around 100%, while with 3060 Ti it hovers around 60%. It results in a very similiar performance between these two cards. In the games I tested - Armored Core 6 and Remnant From the Ashes, the FPS were practically the same. In Cyberpunk 2077 and Deadlink, there is some improvement noticeable, however it is not as big as shown online. Additionaly, while performing benchmarks, I noticed that my results were sometimes as bad as 50% of the scores posted online. I tried DDU and clearing the BIOS, however nothing helped. What's more, the first time is switched the GPUs, my PC got stuck after the BIOS screen. I had to reset the BIOS to get it running.

In superposition 1080p extreme, most users got around 7000, while my result is 3921.
In Furmark 1080p benchamark most 3060 Tis score around 10k - 11k, while mine scored 6k.

My PC:
i5 8600k @4.5GHz
16GB ram 3200MHz
700W Silverstone PSU
2 x HDD 1 TB
2 x NVME 500GB

 

[rgd1101]

it is a pcie 4.0 x16, your system only use 3.0

 

[Avro Arrow]

Most of the time, when you upgrade your video card but get no increase in performance, it's because your CPU is holding it back. That CPU is 5 (almost 6) generations old and could be the problem. After all, your FPS output will only the output of the CPU or the GPU, whichever is slower.

 

[patrykupisz]

it is a pcie 4.0 x16, your system only use 3.0

Still, It should not affect the performance this drastically. From what I've seen it's 2-3 fps difference.

 

[patrykupisz]

Most of the time, when you upgrade your video card but get no increase in performance, it's because your CPU is holding it back. That CPU is 5 (almost 6) generations old and could be the problem. After all, your FPS output will only the output of the CPU or the GPU, whichever is slower.

I expected some bottleneck, but my performance being cut in half is a bit too much. Even in games, I don't have 100% CPU usage, so I don't think that's the problem. Most of the examples is could find showed around 10% performance differnce from modern CPUs.

 

[Avro Arrow]

I expected some bottleneck, but my performance being cut in half is a bit too much. Even in games, I don't have 100% CPU usage, so I don't think that's the problem. Most of the examples is could find showed around 10% performance differnce from modern CPUs.

Yeah, I have no idea then. It seemed plausible (but not really).

Where did you get this card? It sounds defective.

 

[SorryBella]

Additionaly, while performing benchmarks, I noticed that my results were sometimes as bad as 50%

What synthetic benchmark we are looking at here?

Also your gaming resolution might be a bit helpful to know.

 

[thestryker]

Without being able to see what the clockspeeds and power usage looks like it's hard to nail down a specific cause. I'd suggest running 3dmark because it'll show graphs of clocks/performance at the end of the run and you'll be able to compare the results to similar hardware. This might highlight where the problem lies.

 

[Peter Ferrari]

it is a pcie 4.0 x16, your system only use 3.0

that has no impact on gpu usage
he's bottlenecked by 6c/6t system. needs a 8700k at least, but best get a 9900 (k or non-k, small difference).

9700k with oc will do a lot better as well

https://allegro.pl/oferta/procesor-i7-9700kf-3-6-ghz-8-rdzeni-14-nm-lga1151-13956201122

 

[boju]

6 cores will hinder performance in some games but shouldn't compromise graphics card this much. Running 1080ti on my old 2600k (4.5 oc) system ran the card fine, the card was fully utilised in 1080p or 1440p, tested both resolutions in Ghost Recon WL and Doom 2016 at the time.

Check pcie lanes in gpuz, is the card running at 16x? Try reseat the card anyway.

Was 3060ti bought new or used?

 

[rgd1101]

that has no impact on gpu usage
he's bottlenecked by 6c/6t system. needs a 8700k at least, but best get a 9900 (k or non-k, small difference).

9700k with oc will do a lot better as well

https://allegro.pl/oferta/procesor-i7-9700kf-3-6-ghz-8-rdzeni-14-nm-lga1151-13956201122

op also said gpu benchmark is much lower.

 

[kerberos_20]

In superposition 1080p extreme, most users got around 7000, while my result is 3921.

open gpu-z
then run superposition 1080p extreme
once finished, post picture from gpu-z main page and sensors page

 

[patrykupisz]

Thanks everyone for help. I don't know what happened, but the issue resolved itself. The card performs now as it should. In the other games I mentioned, there is some bottleneck noticeable, however it feels much better now. However, shouldn't a bottleneck look like 99% cpu usage, 50% gpu usage? Because in Armored Core and Remnant I have both cpu and gpu usage at around 60%.

 

[Unolocogringo]

Only a few games can use all of the processor cores to 100%.
Most games use one main core to 100% and a few smaller use cores.
So unless you play one of those few games you will never get 100%CPU usage.
Just 100% on one or 2 cores. With the rest 20-30-50 or none.
Some games only use one core. which would be 16.666% CPU usage max for the game. so one core at 100% and the rest doing nothing for the game.
The only way to change this is with a faster processor.
FPS is mostly determined by the CPU. then the GPU.

The only time a GPU becomes the FPS limiter is if you set graphics details higher than the card can handle.
Like trying to run Rust @Max settings on an entry level card. Gt1030 or Rx 6400 type scenario.

 

[Peter Ferrari]

3060ti should run new-ish games at 90%+ usage at least. If it's not, you're leaving more on the table than you should.