[SOLVED] From a 2070 to a 6800- weird performance ?

[bpanugget]

I just upgraded from a msi armor 2070 (vanilla), to a rx 6800 xfx merc319 Speedster. To make a long story short, the performance gains are strange. Playing at 1440p, my fps in doom eternal has gone from a little over 100 to over 200. But then, Fenyx Immortals Rising has stayed roughly the same, at right around 60. CoD Modern Warfare, seems to have gotten about a 20% performance increase, which is half of what I was expecting.

Why am I getting no boost in Fenyx? My CPU usage is never above 60% while playing, so I don't feel like it's throttling the 6800.

Other specs-
Ryzen 5 2600, oc'd on all cores to 4.1ghz
32 gigs ddr4 at 3066mhz
SATA SSD Samsung Evo 860
EVGA G3 Gold 650 watt psu
MSI X470 Gaming Plus

My buddy suggested it could be my power supply, as xfx recommends a 750 watt minimum. But, if that were the case wouldn't the performance be more consistently bad? I'm not seeing any power spikes using AMD's Adrenaline software. Regardless, I have a 850watt unit on the way.

Assuming it's not the PSU, what else could it be?

 

[cdrkf]

I just upgraded from a msi armor 2070 (vanilla), to a rx 6800 xfx merc319 Speedster. To make a long story short, the performance gains are strange. Playing at 1440p, my fps in doom eternal has gone from a little over 100 to over 200. But then, Fenyx Immortals Rising has stayed roughly the same, at right around 60. CoD Modern Warfare, seems to have gotten about a 20% performance increase, which is half of what I was expecting.

Why am I getting no boost in Fenyx? My CPU usage is never above 60% while playing, so I don't feel like it's throttling the 6800.

Other specs-
Ryzen 5 2600, oc'd on all cores to 4.1ghz
32 gigs ddr4 at 3066mhz
SATA SSD Samsung Evo 860
EVGA G3 Gold 650 watt psu
MSI X470 Gaming Plus

My buddy suggested it could be my power supply, as xfx recommends a 750 watt minimum. But, if that were the case wouldn't the performance be more consistently bad? I'm not seeing any power spikes using AMD's Adrenaline software. Regardless, I have a 850watt unit on the way.

Assuming it's not the PSU, what else could it be?

With regard to Fenyx.. looks to be a limitation of the game:

View: https://www.youtube.com/watch?v=NYaF59650mU


That is with an RTX 3080 and also only hits ~ 60fps at 1440p (in fact performance doesn't drop much at 4k so game is obviously limited by other factors).

I think the cpu is likely partly to blame as well - looking at the % cpu usage doesn't tell the whole story. Many games are not coded to fully utilise all cores, and any older titles that use DX9, 10 or 11 are coded to send data to the gpu using a primary thread (DX 11 can use up another couple of threads to support the primary render thread). This means that graphics performance is typically limited by single thread performance and leaves most of the rest of the cpu idle. Doom Eternal on the other hand uses Vulkan which (along with DX12) has the ability to properly distribute draw calls between all cpu cores - assuming the developers have chosen to do so.

The R5 2600 is probably a bit on the slow side to pair with an RX 6800 - it will vary by game, however I think you would benefit from moving to a cpu with higher single core performance - the Ryzen 3000 series are ~ 15% faster per core compared to the 2000 series, and the latest 5000 series cpu's are another 20% faster than the 3000 series. The good thing for you though is that MSI X470 board should work fine with any of the newer ryzen cpu's with just a bios update.

 

[bpanugget]

With regard to Fenyx.. looks to be a limitation of the game:

View: https://www.youtube.com/watch?v=NYaF59650mU


That is with an RTX 3080 and also only hits ~ 60fps at 1440p (in fact performance doesn't drop much at 4k so game is obviously limited by other factors).

I think the cpu is likely partly to blame as well - looking at the % cpu usage doesn't tell the whole story. Many games are not coded to fully utilise all cores, and any older titles that use DX9, 10 or 11 are coded to send data to the gpu using a primary thread (DX 11 can use up another couple of threads to support the primary render thread). This means that graphics performance is typically limited by single thread performance and leaves most of the rest of the cpu idle. Doom Eternal on the other hand uses Vulkan which (along with DX12) has the ability to properly distribute draw calls between all cpu cores - assuming the developers have chosen to do so.

The R5 2600 is probably a bit on the slow side to pair with an RX 6800 - it will vary by game, however I think you would benefit from moving to a cpu with higher single core performance - the Ryzen 3000 series are ~ 15% faster per core compared to the 2000 series, and the latest 5000 series cpu's are another 20% faster than the 3000 series. The good thing for you though is that MSI X470 board should work fine with any of the newer ryzen cpu's with just a bios update.

Thank you for the reply. I am waiting for MSI to release the BIOS update for my mobo to upgrade my CPU. Hopefully that'll fix it.

 

[Karadjgne]

You didn't get an fps increase in anything. When you switched to a far stronger gpu, you stopped being gpu limited and became cpu limited.

Basically you allowed the gpu to play catch-up and actually render all the frames the cpu was spitting out. In doom eternal, the cpu was kicking out over 200fps, but you were gpu limited at 1440p to just over 100fps. Now you aren't.

So in some games it'll look like you gained, but you really didn't, as shown in Fenyx, you gained nothing because in that game the 2070 was strong enough to already exceed cpu limits.

Only way to Increase fps is to increase the output of the cpu. The more frames a cpu can deliver, the more a gpu can render, but a gpu cannot render more than it's given.

Moving to a 5600x/5700x you'll see an increase in fps across the board, as long as the games are optimized to allow such.

 

[bpanugget]

You didn't get an fps increase in anything. When you switched to a far stronger gpu, you stopped being gpu limited and became cpu limited.

Basically you allowed the gpu to play catch-up and actually render all the frames the cpu was spitting out. In doom eternal, the cpu was kicking out over 200fps, but you were gpu limited at 1440p to just over 100fps. Now you aren't.

So in some games it'll look like you gained, but you really didn't, as shown in Fenyx, you gained nothing because in that game the 2070 was strong enough to already exceed cpu limits.

Only way to Increase fps is to increase the output of the cpu. The more frames a cpu can deliver, the more a gpu can render, but a gpu cannot render more than it's given.

Moving to a 5600x/5700x you'll see an increase in fps across the board, as long as the games are optimized to allow such.

That makes sense. I thought the oc I had would buy me more time with the 2600. Guess not.

 

[bpanugget]

You didn't get an fps increase in anything. When you switched to a far stronger gpu, you stopped being gpu limited and became cpu limited.

Basically you allowed the gpu to play catch-up and actually render all the frames the cpu was spitting out. In doom eternal, the cpu was kicking out over 200fps, but you were gpu limited at 1440p to just over 100fps. Now you aren't.

So in some games it'll look like you gained, but you really didn't, as shown in Fenyx, you gained nothing because in that game the 2070 was strong enough to already exceed cpu limits.

Only way to Increase fps is to increase the output of the cpu. The more frames a cpu can deliver, the more a gpu can render, but a gpu cannot render more than it's given.

Moving to a 5600x/5700x you'll see an increase in fps across the board, as long as the games are optimized to allow such.

I tested Fenyx by lowering the resolution scale by 50%. My FPS stayed the same. Definitely my CPU, as you had suggested. Thanks for the advice. I'll try to grab a 5800x when the x470 bios update for zen 3 comes out.