Bottlenecking

[Solar_Beast]

So I built a system last year that featured a ryzen 5 2600x, GeForce 1660 ti, b450 tomahawk, and 32 gb of 3000mz ram. A situation arose and I was able to easily upgrade my gpu to a 2070 super. I fear that the CPU may be bottlenecking my gpu though by a significant amount. I don't get the same frames on games such as assassin's Creed Odyssey as other people's posts online. I get an average of 55 to 60 frames on Odyssey at ultra settings at 1920x1080. Something just doesn't seem quite right.

[Ralf]

So what's your cpu(on each core) usage and gpu usage in games such as assassin's Creed Odyssey?

[Solar_Beast]

39 minutes ago, Ralf said:

So what's your cpu(on each core) usage and gpu usage in games such as assassin's Creed Odyssey?

CPU is running on 75% to 80% at 4.07 GHz and gpu is at 80% to 90% utilization.

[Kasper_MC]

Sounds like there's another limiter, no? 

Maybe it could also be your RAM, even though you have a lot of it. Ryzen likes speed, though maybe 3000 and 3200 is almost equal, there could be something there - timings too maybe. Have you enabled XMP on RAM? 

The other people posting only, are those reviewers on YT? be aware they most often use an I9-k variant for max performance when testing GPUS, and vice versa a 2080ti when testing CPUs to eliminate the other being a bottleneck. In that sense, all CPU's are potential bottlenecks for a 2080ti, and most would be for a 2070s... But idk if you've only checked for comparable GPU? 

[Solar_Beast]

5 minutes ago, Kasper_MC said:

Sounds like there's another limiter, no? 

Maybe it could also be your RAM, even though you have a lot of it. Ryzen likes speed, though maybe 3000 and 3200 is almost equal, there could be something there - timings too maybe. Have you enabled XMP on RAM? 

The other people posting only, are those reviewers on YT? be aware they most often use an I9-k variant for max performance when testing GPUS, and vice versa a 2080ti when testing CPUs to eliminate the other being a bottleneck. In that sense, all CPU's are potential bottlenecks for a 2080ti, and most would be for a 2070s... But idk if you've only checked for comparable GPU? 

How does one fix the timings on ram? I do have my xmp set but on my monitoring software it does not show 3000mhz. It shows like 1260mhz or something like that.

[Kasper_MC]

Hm, seems like there's something funny with that XMP then, or somethings up with your software - which programme you using?
There's no suprise that you'd see a lower speed in CPU-Z than what's advertised: 3000 refers the "megatransfer pr. sec", and because it's "double data rate" that means that the RAM is truly only running at 1500, when it's set at 3000. So what's really 1500mhz is advertised as 3000, because it can transfer twice pr. cycle, so to speak. 
The weird thing is, that it's running 1260, which seems to suggest that there might be several XMPs on your board. Does it say 3000 in the BIOS when you set it up? 

If you wanna mess with RAM timings, I suggest you watch a video by Buildzoid or at least Gamers Nexus on the topic before venturing into it. What I meant was that the timings on your RAM might be loser than what other people have. Combine that with 3000 instead of 3200, might explain some of your trouble. 

What FPS are other people getting compared to yours? Is it something like LTT? 

Also: If you compare a benchmark run to actual gameplay, there might be great variation. For instance, I have a 1080ti and a 3700X, and when I run Far Cry 5 benchmark run, the FPS in generally higher with 5-7% than what I've ever seen in game. It's a select path it runs. 
So if others online is reporting maybe a difference of 5-10% of yours, or whatever, it might be because they're running the benchmark, compared to your actual gameplay? 

[Solar_Beast]

10 minutes ago, Kasper_MC said:

Hm, seems like there's something funny with that XMP then, or somethings up with your software - which programme you using?
There's no suprise that you'd see a lower speed in CPU-Z than what's advertised: 3000 refers the "megatransfer pr. sec", and because it's "double data rate" that means that the RAM is truly only running at 1500, when it's set at 3000. So what's really 1500mhz is advertised as 3000, because it can transfer twice pr. cycle, so to speak. 
The weird thing is, that it's running 1260, which seems to suggest that there might be several XMPs on your board. Does it say 3000 in the BIOS when you set it up? 

If you wanna mess with RAM timings, I suggest you watch a video by Buildzoid or at least Gamers Nexus on the topic before venturing into it. What I meant was that the timings on your RAM might be loser than what other people have. Combine that with 3000 instead of 3200, might explain some of your trouble. 

What FPS are other people getting compared to yours? Is it something like LTT? 

Also: If you compare a benchmark run to actual gameplay, there might be great variation. For instance, I have a 1080ti and a 3700X, and when I run Far Cry 5 benchmark run, the FPS in generally higher with 5-7% than what I've ever seen in game. It's a select path it runs. 
So if others online is reporting maybe a difference of 5-10% of yours, or whatever, it might be because they're running the benchmark, compared to your actual gameplay? 

Frequency is showing correctly now in cpu z.

[Kasper_MC]

Alright. Cool. So 1500? 

But, have you checked the sources that have higher FPS, that they're running the same GPU and CPU as you? I'm still thinking it might be, that they've used a 2080ti for the CPU benchmark, or that you've use a 9900k/10900k for the GPU benchmark. 
If so: How much higher are they getting? It might be that your hardware is running hot and therefore not boosting as higg as theres? That and RAM that's on a little bit of the slow side for Ryzen, might explain the difference? 

[Solar_Beast]

1 minute ago, Kasper_MC said:

Alright. Cool. So 1500? 

But, have you checked the sources that have higher FPS, that they're running the same GPU and CPU as you? I'm still thinking it might be, that they've used a 2080ti for the CPU benchmark, or that you've use a 9900k/10900k for the GPU benchmark. 
If so: How much higher are they getting? It might be that your hardware is running hot and therefore not boosting as higg as theres? That and RAM that's on a little bit of the slow side for Ryzen, might explain the difference? 

Yeah, when I upgrade, I plan to get a b550 motherboard, ryzen 7 3700x, and 16 gb of 3600mhz ram.

[Kasper_MC]

From what I can see, this guy/girl is getting a little hight than you, but on mixed settings, so you're not that far of: 

While this site is quoting ultra satting a fair bit lower than what you're seeing: 
https://www.gpucheck.com/game-gpu/assassins-creed-odyssey/nvidia-geforce-gtx-1660-ti/amd-ryzen-5-2600/

[Kasper_MC]

Well, as far as I can tell, there's no point in getting the B550 mobo - the B450 is plenty for even the 3800X. Also, depending on what you're doing, and save the money for the motherboard and get the 3600(X), if you're mostly just gaming. It's every bit as competent at gaming. The money you save you can put towards a faster GPU, or save. 
But also, if you have the 2600 and is not planning to do any kind of editing or other production on the machine, I would strongly suggest you wait for the 4000-SKUs of Ryzen, whould mean a good deal more for your money, from what I can tell by listening to reviewers. 

[Solar_Beast]

6 minutes ago, Kasper_MC said:

Well, as far as I can tell, there's no point in getting the B550 mobo - the B450 is plenty for even the 3800X. Also, depending on what you're doing, and save the money for the motherboard and get the 3600(X), if you're mostly just gaming. It's every bit as competent at gaming. The money you save you can put towards a faster GPU, or save. 
But also, if you have the 2600 and is not planning to do any kind of editing or other production on the machine, I would strongly suggest you wait for the 4000-SKUs of Ryzen, whould mean a good deal more for your money, from what I can tell by listening to reviewers. 

Yeah, that's very true.

[Solar_Beast]

11 minutes ago, Kasper_MC said:

From what I can see, this guy/girl is getting a little hight than you, but on mixed settings, so you're not that far of: 

While this site is quoting ultra satting a fair bit lower than what you're seeing: 
https://www.gpucheck.com/game-gpu/assassins-creed-odyssey/nvidia-geforce-gtx-1660-ti/amd-ryzen-5-2600/

Although, this guy is using a 1660ti not a 2070 super.

[Kasper_MC]

True, well:

And: 
gpucheck.com/game-gpu/assassins-creed-odyssey/nvidia-geforce-rtx-2070-super/amd-ryzen-5-2600x/ultra 



Also: 
- What are your temps? 

- and what sources have you seen with higher FPS? 

[Solar_Beast]

https://www.eurogamer.net/articles/digitalfoundry-2019-09-25-geforce-rtx-2070-super-benchmarks-7001 is a source showing more than 60 fps. My gpu is the msi twin frozr gaming x version and its at 57 degrees celcius.

[Kasper_MC]

Ok. I can't see their methodology, but my best guess is that they'd be using a 9900k, the best CPU for gaming at the time - would be most likely, to show the difference most consistently. So I'd guess that's why. It's not like your frames are tanked, as far as I can see. But ofcause, if you upgrade to the best of the best, it's fps is going to be higher. 

[Solar_Beast]

4 minutes ago, Kasper_MC said:

Ok. I can't see their methodology, but my best guess is that they'd be using a 9900k, the best CPU for gaming at the time - would be most likely, to show the difference most consistently. So I'd guess that's why. It's not like your frames are tanked, as far as I can see. But ofcause, if you upgrade to the best of the best, it's fps is going to be higher. 

Right, I looked at the CPU variations on the website you shared and it seemed to match my 2600x.

[Kasper_MC]

Good - sorry it's not faster! But! You could lover the quality one notch and see if it helps? 

[Solar_Beast]

1 hour ago, Kasper_MC said:

Good - sorry it's not faster! But! You could lover the quality one notch and see if it helps? 

I did and it's great! 100 fps! Thanks!

[Kasper_MC]

Great! Have fun