Performance questions.

exhumedye

exhumedye

exhumedye

Associate
Joined
26 Nov 2008
Posts
193
Hello, I recently upgraded my PC from a 5800X to a 9800X3D(haven't changed my sig yet). Prior to upgrading I ran some game benchmarks on my 5800X to compare with the newly installed 9800X3D, with the games being:

Cyberpunk 2077
Total War: Pharaoh
Total War: Warhammer III
Black Myth Wukong
Horizon Zero Dawn: Remastered
F1 2025
Doom: The Dark Ages

I tested them without using any upscaling or frame generation, I was never expecting massive increases in performance on most of these games, I was however, expecting **some** increase in performance.

The first game I ran was Cyberpunk 2077 and the result was more or less identical. I wasn't expecting much of a difference in frame rate to be honest as it's a GPU heavy game, but not completely identical.
image.png
image.png


The next game I tested was Black Myth Wukong, and again it was an identical result.

I will say that again I wasn't expecting a big increase in this game, but I was expecting to see something.
image.png
image.png


Next up was F1 2025, this was again almost identical with only the maximum fps showing a huge boost. But I assume this is an anomaly as it didn't change the average frame rate.
image.png
image.png


At this point I'm frustrated and I wonder why I spent £1000 on 3 components to 'upgrade'. So I picked a game I knew I'd see a performance increase in. Total War: Pharaoh and Warhammer III, which thankfully delivered some very nice results.
image.png
image.png
image.png
image.png


Horizon Zero Dawn: Remastered also saw some nice gains.
Point to note: I'm pretty sure the 'upscale quality' options is irrelevant in these tests, as I had the upscaling method off entirely.
image.png
image.png



I've not included the Doom screenshots as I forgot to save the 9800X3D benchmark, but it was again very similar to the 5800X result. So I'm curious, is this normal for results to be this identical in some cases? As I said earlier, I understand that they are GPU heavy games, but I expected some improvement across the board, even if it was just an average of 5 fps.

Perhaps there are some bios settings I should have on which I don't?
Perhaps the performance increases are only seen in upscaling and frame gen?
 
Looks like you might have v-sync on or something else capping performance in some of those screenshots - Horizon Zero Dawn says V-Sync on in the settings and some of the others seem to be clipping the v-sync multiples.

This is why I don't like a lot of gaming CPU benchmarks, though I get stick for it, 1080p low settings with a 4090 or 5090 does not represent what most people will see especially if you are at 1440p or above with high or above settings paired up with a GPU below the highest tier! though you should be seeing a bit more of an uplift than that especially to minimum FPS.
 
Last edited:
What kind of gpu utilisation were you seeing in the games with little difference before the upgrade? If it's near 100% then upgrading the cpu wouldn't really help as the gpu is the limiting factor and that hasn't changed.
 
Rroff said:
Looks like you might have v-sync on or something else capping performance in some of those screenshots - Horizon Zero Dawn says V-Sync on in the settings and some of the others seem to be clipping the v-sync multiples.

This is why I don't like a lot of gaming CPU benchmarks, though I get stick for it, 1080p low settings with a 4090 or 5090 does not represent what most people will see especially if you are at 1440p or above with high or above settings paired up with a GPU below the highest tier! though you should be seeing a bit more of an uplift than that especially to minimum FPS.
Click to expand...

I will double check the settings, but would vsync cap my frame rate below the refresh rate? I don't know much about vsync.
michty_me said:
Did you do a fresh install of windows rather than just installing new drivers etc?
Click to expand...
Yes, a fresh install of windows. I suffered through immense issues just to get a fresh install of windows. I had no idea that vertically mounting my GPU before installing windows would give me such stress lol.


GeorgeStorm said:
What kind of gpu utilisation were you seeing in the games with little difference before the upgrade? If it's near 100% then upgrading the cpu wouldn't really help as the gpu is the limiting factor and that hasn't changed.
Click to expand...
Unfortunately that's not something I took note of, I guess I'll keep that noted for next time. But I will say that I expected minimal gains in GPU intensive games, like maybe an average 3-7 fps gain, I just didn't expect identical results lol.
 
Looking at benchmarks between 5800X and 9800X3D, tends to show CP running near double the frames of the 5800X with the same GPU.

Have you tried some 3Dmark tests etc?
 
Last edited:
exhumedye said:
I will double check the settings, but would vsync cap my frame rate below the refresh rate? I don't know much about vsync.
Click to expand...

Depends on settings but by default V-Sync will cap the frame rate to the monitor refresh rate and when it can't hit that it will reduce frame rate to multiples of the refresh rate i.e. 120Hz V-Sync would either output 60, 90 (with triple-buffering) or 120Hz - reducing performance in benchmarks. If you have some form of adaptive sync though it will render at intermediate frame rates, with adaptive sync or fast sync you aren't necessarily capped to the monitors max refresh depending on settings.
 
michty_me said:
Looking at benchmarks between 5800X and 9800X3D, tends to show CP running near double the frames of the 5800X with the same GPU.
Click to expand...

That is usually using a 5090 with 1080p low to medium graphic settings - a 5070ti at 1440p and high settings you are probably looking at around 20% uplift in Cyberpunk 2077 - though that is off the top of my head and I might be wrong.

This is why I find many articles/benchmarks and people bleating on about "buts it's a gaming CPU so it makes sense to benchmark at 1080p and not bother with 4K", etc. a bit tedious.
 
Last edited:
Rroff said:
Depends on settings but by default V-Sync will cap the frame rate to the monitor refresh rate and when it can't hit that it will reduce frame rate to multiples of the refresh rate i.e. 120Hz V-Sync would either output 60, 90 (with triple-buffering) or 120Hz - reducing performance in benchmarks. If you have some form of adaptive sync though it will render at intermediate frame rates, with adaptive sync or fast sync you aren't necessarily capped to the monitors max refresh depending on settings.
Click to expand...
Good explanation, I learnt a bit through this too.

Sorry, The tests were in 4k with a 3090TI so not really an apples to apples comparison.
 
michty_me said:
Good explanation, I learnt a bit through this too.

Sorry, The tests were in 4k with a 3090TI so not really an apples to apples comparison.
Click to expand...

The only test I can find with a 5800X3D vs 9800X3D with CP2077 and a 5070ti shows a 32% performance uplift at 1440p but I can't see what settings they were running. I would have expected somewhere in the 20-30% range though depending on how much you are loading the GPU up settings wise.

EDIT: Also this video here with a 4070 Ti Super, not sure validity of it, but they are using medium settings with not RT so getting more like 30-40% uplift:

 
Last edited:
michty_me said:
Looking at benchmarks between 5800X and 9800X3D, tends to show CP running near double the frames of the 5800X with the same GPU.

Have you tried some 3Dmark tests etc?
Click to expand...
No I've not done any 3Dmark tests or anything like that, just video game benchmarks.
Rroff said:
Depends on settings but by default V-Sync will cap the frame rate to the monitor refresh rate and when it can't hit that it will reduce frame rate to multiples of the refresh rate i.e. 120Hz V-Sync would either output 60, 90 (with triple-buffering) or 120Hz - reducing performance in benchmarks. If you have some form of adaptive sync though it will render at intermediate frame rates, with adaptive sync or fast sync you aren't necessarily capped to the monitors max refresh depending on settings.
Click to expand...

Interesting, I didn't know this.
However, I just checked Cyberpunk and I had Vsync off for it, and in my Nvidia global settings.
 
What motherboard and RAM are you using?
Updated the BIOS to latest, Ensure X3d boost modes are OFF in bios etc. Maybe do a load optimised defaults and apply your EXPO profile?
 
If you were capping your GPU at at MAX before and after, then you're really only going to see a benefit in your minimum frame rate.
However, games like MMOs, RTS and city builders where a lot of CPU is used would also get an improvement.
 
Last edited:
WiGgLr said:
If you were capping your GPU at at MAX before and after, then you're really only going to see a benefit in your minimum frame rate.
However, games like MMOs, RTS and city builders where a lot of CPU is used would also get an improvement.
Click to expand...
Yeah this is true. It just seems very low going from 5800X to 9800X3d.

Edit: Just noticed the Psycho RT enabled in CP too.
 
Last edited:
michty_me said:
What motherboard and RAM are you using?
Updated the BIOS to latest, Ensure X3d boost modes are OFF in bios etc. Maybe do a load optimised defaults and apply your EXPO profile?
Click to expand...
MSI X870E Gaming Plus Wifi.
TeamGroup Vulcan EXPO 32GB 6000Mhz

I'm using the latest bios(updated it via the MSI Center). I'm using the default EXPO profile 1. RAM running at 6000 MT/s according to MSI Center.
I will go into bios and disable X3D boosts, then do some more tests.

iamgigglz said:
oof that'll do it. Looks like Black Myth Wukong was also cranked.
Click to expand...

The idea was to max it out and just see what it could do on the in-game benchmarks. If I play on those settings I will use frame gen and upscaling, assuming it doesn't look awful and have input lag. Otherwise I will lower the settings to a more playable state.

Just to reiterate, I never expected big gains in games like cyberpunk or black myth wukong when upgrading, as I know they are GPU heavy. I'm just surprised by the identical results lol.
 
You must log in or register to reply here.
Back
Top Bottom