Starfield CPU performance reviews

[LtMatt]

Rachet and Clank the difference between the 7900XTX and 4090 is actually only about 10%

Yeah, with RT off there's not a huge difference in performance in my testing between the two GPUs.

However, with RT on and DLSS Q/FSR Q enabled for both the difference is 58%.

Ratch and Clank 2160P Very High Settings + RT - Imgsli

 

[humbug]

Yeah, with RT off there's not a huge difference in performance in my testing between the two GPUs.

However, with RT on and DLSS Q/FSR Q enabled for both the difference is 58%.

Ratch and Clank 2160P Very High Settings + RT - Imgsli

Have you noticed something about these two sides?

The right side has more fine detail, is sharper while the left side is also washedout.

Look at the carpet, the brass on the floor, the scaffolding, the grain in the concrete. Are you running DLSS on the 4090? It look decidedly a lower render resolution. For sure....

 

[LtMatt]

Have you noticed something about these two sides?

The right side has more fine detail, is sharper while the left side is also washedout.

Look at the carpet, the brass on the floor, the scaffolding, the grain in the concrete. Are you running DLSS on the 4090? It look decidedly a lower render resolution. For sure....

Yeah, I don't think there is a sharpening pass for DLSS but there is for FSR, or that was the case when I tested it at launch.

That's also a screenshot of two videos from my YT channel, so some quality lost due to YT compression.

 

[3t3P]

so is my CPU toast for this game ? 6 cores all 5ghz?

 

[VinylRichTea]

My mate and I have the same Ryzen 7 5700X,and their RX6700XT is generally quicker than my RTX3060TI in the game using the same settings in the same locations

But but, mr painal face said otherwise

https://twitter.com/x/status/1699655687809470971

I joined his discord and left, he's been pushed to the point unfortunately with too many idiots joining his discord, where he snaps at anything and everything and throws out bans left, right and centre, which in his defence is not his fault, but it tends to be an echo chamber in there also with a lot of arse licking. He basically expects everyone to read his 80+ rules, pure madness...

 

[LtMatt]

As a rule of thumb, never trust anything he says about AMD, and AMD CPU/GPU performance. His audience and schtick is all centred around bad mouthing anything AMD and creating shock YT videos for clicks. This user has Nvidia GPUs being faster than AMD GPUs in COD, that's how bad his results are.

 

[Grim5]

As a rule of thumb, never trust anything he says about AMD, and AMD CPU/GPU performance. His audience and schtick is all centred around bad mouthing anything AMD and creating shock YT videos for clicks. This user has Nvidia GPUs being faster than AMD GPUs in COD, that's how bad his results are.

It's cause he uses a 13900k overclocked to the eyeballs, probably pulling 350watts and has ddr5 ram tuned to over 8000mhz, so in low resolution his 4090 gets performance well above what others do in cpu limited situations such as cod warzone

 

[LtMatt]

It's cause he uses a 13900k overclocked to the eyeballs, probably pulling 350watts and has ddr5 ram tuned to over 8000mhz, so in low resolution his 4090 gets performance well above what others do in cpu limited situations.

That has no bearing on what I said though in anyway shape or form, trust me. His COD numbers have always been very low compared to the numbers my 7950X3D puts out, like hundreds difference in average CPU FPS for example. And how he had Nvidia GPUs being faster in COD is criminal when I test the performance myself with both tuned.

Anyway, getting off track of the topic. Just don't trust anything AMD related from his chops is all I'm saying.

 

[Grim5]

I don't believe his benchmarks are fake, but he is ******** and his performance numbers are outside of reach for an average gamer. Actually so are yours too, you tune your PC hard and put up numbers here the average person can't get, so you're basically the AMD version of him

 

[KompuKare]

I don't believe his benchmarks are fake, but he is ******** and his performance numbers are outside of reach for an average gamer. Actually so are yours too, you tune your PC hard and put up numbers here the average person can't get, so you're basically the AMD version of him

I for one greatly appreciate the effort that @LtMatt and @Robert896r1 put into tuning and telling us the results.
Even if I am the kind of person who would take the simplicity the Ryzen 3D's offer at stock. Well, mostly at stock as LtMatt does get more by tuning too.

 

[LtMatt]

I don't believe his benchmarks are fake, but he is ******** and his performance numbers are outside of reach for an average gamer. Actually so are yours too, you tune your PC hard and put up numbers here the average person can't get, so you're basically the AMD version of him

At least I make sure both are tuned to perfection. Rather than somehow making Nvidia faster in a game where everyman and his dog knows that AMD GPUs are faster. Gotta get those YT clicks for money somehow I guess.

Meanwhile, in reality. (1440P Lowest settings + spot cache Ultra)

Spoiler

Spoiler

Feel free to see if you can beat my 4090 result, @Grim5. I'll be surprised if you get within 20 FPS of it. If you do, I'll give it another run as I can probably add a few more FPS onto the result. Regardless, It's not going to be faster than the AMD GPU - unless you are framechasers that is.

I for one greatly appreciate the effort that @LtMatt and @Robert896r1 put into tuning and telling us the results.
Even if I am the kind of person who would take the simplicity the Ryzen 3D's offer at stock. Well, mostly at stock as LtMatt does get more by tuning too.

Welcome. Robert's results I actually trust. regardless of what hardware he is using.

 

[Robert896r1]

I for one greatly appreciate the effort that @LtMatt and @Robert896r1 put into tuning and telling us the results.
Even if I am the kind of person who would take the simplicity the Ryzen 3D's offer at stock. Well, mostly at stock as LtMatt does get more by tuning too.

No worries. When i get my RPL-R and D5 build, I'll redo most of the test we normally use around here. Prob over xmas holidays.

 

[LumpyVR]

So I recently paired A new 7800XT with my 3600X and performance has been mostly pretty good for me (high preset @1440p).

Its normally sat around 70+ fps (much higher in space etc.) although I do see drops in crowded areas, Akila city was a noticeable drop to high 50’s I think.

Reading this thread and others I’m wondering whether it’s worth dropping a 5800X or 5800X3D in there. I was planning a full AM5 upgrade next year so wasn’t going to bother but if it would be a noticeable difference I may be tempted if one comes up on offer

 

[g67575]

If you're building an AM5 system, games like this seem to benefit from higher speed RAM (particularly 1% and 0.2% lows), at least running in gear 1 mode:

You could buy high speed RAM now, and just run it at 6400 MT/s + gear 1 mode.

Or wait for Ryzen 8000 series CPUs, which I'm sure will fully support even higher RAM frequencies.

I think higher speed RAM will serve you better than Vcache on a 5800X3D.

 

[opethdisciple]

So I recently paired A new 7800XT with my 3600X and performance has been mostly pretty good for me (high preset @1440p).

Its normally sat around 70+ fps (much higher in space etc.) although I do see drops in crowded areas, Akila city was a noticeable drop to high 50’s I think.

Reading this thread and others I’m wondering whether it’s worth dropping a 5800X or 5800X3D in there. I was planning a full AM5 upgrade next year so wasn’t going to bother but if it would be a noticeable difference I may be tempted if one comes up on offer

In general a 5800x3d will give you a nice boost over a 3600x. A 5800x3d is generally in the top 5ish gaming cpus.

I posted in the 5800x3d thread a chart showing its performance compared to other CPU's.

 

[cyber69]

In general a 5800x3d will give you a nice boost over a 3600x. A 5800x3d is generally in the top 5ish gaming cpus.

I posted in the 5800x3d thread a chart showing its performance compared to other CPU's.

I've ordered a 5800X3D to replace my 3700X, now not sure if I will see any benefit in Starfield as my current FPS is 45 average, CPU useage around 50% and GPU is 97% so looks like i'm GPU limited on the 3070TI.

 

[eeii]

I've ordered a 5800X3D to replace my 3700X, now not sure if I will see any benefit in Starfield as my current FPS is 45 average, CPU useage around 50% and GPU is 97% so looks like i'm GPU limited on the 3070TI.

You will see an improvement in the 1% lows.

 

[g67575]

I don't think a 5800X3D is worth it now (if the goal is high FPS), it looks like it struggles to get a steady 60 FPS in Starfield:

Starfield: 44 CPU Benchmark, Intel vs. AMD, Ultra, High, Medium & Memory Scaling

Thermal Grizzly: https://www.thermal-grizzly.com/produkte/522-cpu-contact-frame-for-12th-gen-intel-by-der8auerSupport us on Patreon: https://www.patreon.com/...

youtu.be

And it can't be combined with high end RAM. Only a bit cheaper than a 7700X also.

 

[Grim5]

I don't think a 5800X3D is worth it now (if the goal is high FPS), it looks like it struggles to get a steady 60 FPS in Starfield:

Starfield: 44 CPU Benchmark, Intel vs. AMD, Ultra, High, Medium & Memory Scaling

Thermal Grizzly: https://www.thermal-grizzly.com/produkte/522-cpu-contact-frame-for-12th-gen-intel-by-der8auerSupport us on Patreon: https://www.patreon.com/...

youtu.be

And it can't be combined with high end RAM.

Starfield likes strong single core performance so it's no surprise that Intel dominates and it's also no surprise that a AMd cpu focused on multithread performance with high cache struggles

 

[g67575]

If looking for a stop gap CPU on AM5, I'd consider the 7600X (as these are ~£215 atm), plus some decent DDR5 RAM. That ought to be good for gaming until the Ryzen 8000 series is released late next year.