News Cyberpunk 2077 CPU Scaling, What Processors Work Best?

[Admin]

We test a wide range of both Intel and AMD processors to determine which processors perform best in a variety of resolutions.

Cyberpunk 2077 CPU Scaling, What Processors Work Best? : Read more

 

[cewhidx]

I dont know yet what my average fps is but I can say that the game is running smooth as silk so far with my configuration:
Ryzen 5 3600
EVGA RTX 2080 XC Gaming
Corsair Vengeance DDR4 3200 32 GB
1440p GSync 144hz Monitor
Pretty much maxed out settings with version 1.04, All RTX options enabled with DLSS on balanced.
I'll update after trying 1.05, but so far after 3 hours of playing, Ive not had any bugs / slowdowns or anything.

 

[willburstyle06]

We test a wide range of both Intel and AMD processors to determine which processors perform best in a variety of resolutions.

Cyberpunk 2077 CPU Scaling, What Processors Work Best? : Read more

Was the 9900K tested at stock 4.7GHz all cores? In general were the CPU’s all tested at stock? I have mine at 5.1GHz and I’m trying to see how it compares to this test

 

[Evill-Santa]

What really poor selection of CPU's. You will see a high % of gamers have older CPU's as they get swapped a lot less than GPU's, especially intel's with their more expensive upgrade path. As per the Nov 2020 a bit under 75% of gamers have Intel CPU's yet you have chosen only 2 generations of intel CPU's. (though you have done better with AMD) So it's not really a "wide selection" of nearly 75% of gamers is it?

Looks like you have taken a wide selection of what is easy for you to test, rather than what would be useful for a large volume of your user base. You have provided good feedback for AMD CPUS's (~25% of market) but very shallow feedback on the intel generations.

I get it, AMD is all the rage and it will probably be my next CPU, but this would be a much more useful article if you catered for what would be the more likely scenario of the gamers CPU.

Using CPU performance websites is not really a great way for someone to work out where their CPU sits.

 

[Geef]

At a bare minimum, many enthusiasts enable the XMP profile in the BIOS

The only enthusiast who doesn't enable it are the <<Removed by moderator>> ones, or the ones who possibly forgot to last time they reset their settings and they smack their head when they realize what they've done.

 

[Conahl]

What really poor selection of CPU's. You will see a high % of gamers have older CPU's as they get swapped a lot less than GPU's, especially intel's with their more expensive upgrade path. As per the Nov 2020 a bit under 75% of gamers have Intel CPU's yet you have chosen only 2 generations of intel CPU's. (though you have done better with AMD) So it's not really a "wide selection" of nearly 75% of gamers is it?

Looks like you have taken a wide selection of what is easy for you to test, rather than what would be useful for a large volume of your user base. You have provided good feedback for AMD CPUS's (~25% of market) but very shallow feedback on the intel generations.

I get it, AMD is all the rage and it will probably be my next CPU, but this would be a much more useful article if you catered for what would be the more likely scenario of the gamers CPU.

Using CPU performance websites is not really a great way for someone to work out where their CPU sits.

or, what they tested, is what they had on hand to test, ever consider that ? what is on hand/easy to test vs what they have available to test, could be 2 different things.

 

[milaaaaaaan_63]

9700K is the best overall for a 199$ !!!

 

[Minh Nguyen]

the 1.05 patch pretty much does the hex fix, so the default will be the "modified" results

 

[PaulAlcorn]

Was the 9900K tested at stock 4.7GHz all cores? In general were the CPU’s all tested at stock? I have mine at 5.1GHz and I’m trying to see how it compares to this test

yes, the processors are all tested in stock configs. Head to our 9900K review to see what that chip looks like overclocked - you can kinda reverse engineer the numbers there to get a decent sense of how it does in gaming with an OC.

 

[PaulAlcorn]

What really poor selection of CPU's. You will see a high % of gamers have older CPU's as they get swapped a lot less than GPU's, especially intel's with their more expensive upgrade path. As per the Nov 2020 a bit under 75% of gamers have Intel CPU's yet you have chosen only 2 generations of intel CPU's. (though you have done better with AMD) So it's not really a "wide selection" of nearly 75% of gamers is it?

Looks like you have taken a wide selection of what is easy for you to test, rather than what would be useful for a large volume of your user base. You have provided good feedback for AMD CPUS's (~25% of market) but very shallow feedback on the intel generations.

I get it, AMD is all the rage and it will probably be my next CPU, but this would be a much more useful article if you catered for what would be the more likely scenario of the gamers CPU.

Using CPU performance websites is not really a great way for someone to work out where their CPU sits.

Time is a factor, plus the patch was looming, so I went with what I had. The patch doesn't seem to make any significant changes, so as mentioned in the article, we'll soldier on with 7th and 8th gen Intel next.

We don't suggest that you visit CPU benchmark sites for relative performance comparisons. We suggest that you go to our CPU benchmark hierarchy, which has testing performed with solid and repeatable test methodologies in a static test environment.

 

[TimmyP777]

The only enthusiast who doesn't enable it are the <<Removed by moderator>> ones, or the ones who possibly forgot to last time they reset their settings and they smack their head when they realize what they've done.

XMP has caused a plethora of problems over the last decade. Especially back when it was more an Intel feature being brought to AMD boards.

 

[itzmec]

I guess I'll have to do my own review. still running a 5930k. Can't believe it's been 5 years already.

 

[mitch074]

I guess I'll have to do my own review. still running a 5930k. Can't believe it's been 5 years already.

Me, I can't believe Intel hasn't updated its microarchitecture in 5 years. Haswell wasn't much slower than Skylake either so yeah, you're still good to go.

 

[madbiker]

Sadly, I've been dying to spend money on 'upgrades' that I dont need, thankfully, I cant even get. I wanted a 5900x so badly, but, I'm looking at this as my 9900K paired to TridentZ 3600 14-15-15 (Asrock z370 Phantom Gaming K6) seem to be as good as any other build for gaming.

 

[hannibal]

What really poor selection of CPU's. You will see a high % of gamers have older CPU's as they get swapped a lot less than GPU's, especially intel's with their more expensive upgrade path. As per the Nov 2020 a bit under 75% of gamers have Intel CPU's yet you have chosen only 2 generations of intel CPU's. (though you have done better with AMD) So it's not really a "wide selection" of nearly 75% of gamers is it?

Most likely because Intel cpus have been about the same Many years. So you can always pick Intel cpu that is similar to your own and you get very near estimation what you get.
In amd side these has been more development. Zen1 was meh, zen1+ Little improvements. Zen2 was big jump, Zen3 was another big jump.
I think that it was the Main reason for this test setup. I think that you overestimate amd. Most likely more than 85% of gamers still have Intel... I would not be surpriced if it is near 90%. Bus as I said, the last 6 years in Intel has been about the same. Small clock bump and 5% more ipc... nothing more so most Intel cpus Are Also almost the same considering the core count and clockspeeds!

 

[ingtar33]

I dont know yet what my average fps is but I can say that the game is running smooth as silk so far with my configuration:
Ryzen 5 3600
EVGA RTX 2080 XC Gaming
Corsair Vengeance DDR4 3200 32 GB
1440p GSync 144hz Monitor
Pretty much maxed out settings with version 1.04, All RTX options enabled with DLSS on balanced.
I'll update after trying 1.05, but so far after 3 hours of playing, Ive not had any bugs / slowdowns or anything.

same, I'm hitting a constant 60fps in 1080p with my r5 3600, rtx2700s and 32gb of ram on ultra, i think there is something wrong with their test bench.

 

[mitch074]

Most likely because Intel cpus have been about the same Many years. So you can always pick Intel cpu that is similar to your own and you get very near estimation what you get.
In amd side these has been more development. Zen1 was meh, zen1+ Little improvements. Zen2 was big jump, Zen3 was another big jump.
I think that it was the Main reason for this test setup. I think that you overestimate amd. Most likely more than 85% of gamers still have Intel... I would not be surpriced if it is near 90%. Bus as I said, the last 6 years in Intel has been about the same. Small clock bump and 5% more ipc... nothing more so most Intel cpus Are Also almost the same considering the core count and clockspeeds!

I wouldn't call Zen/Zen+ "meh" - it was equivalent to Haswell, but with up to double the cores. For productivity, it was great (past the first 6 months where BIOS updates righted most of the early problems) and you could still play games on it without breaking the bank.
As for the platform itself, to think that a cheapo B350 motherboard from 2017 can officially run a 12-core Zen2 CPU from 2020 with a BIOS update shows how stable the hardware specs were.

 

[BadMrSnowman]

Since these tests were done with an RTX 3090 GPU, can these charts be interpreted as the maximum FPS one can expect for the given processor?

 

[JarredWaltonGPU]

same, I'm hitting a constant 60fps in 1080p with my r5 3600, rtx2700s and 32gb of ram on ultra, i think there is something wrong with their test bench.

I always 'love' statements like this. Have you run OCAT to capture frametimes and then checked the results in a spreadsheet? Because even if Riva Tuner Statistics overlay, or whatever tool you're using to show fps, indicates an average of 60+ fps, unless you're WAY above 60 you'll almost always see some dips below that mark. I average 63 fps during our test sequence on an RTX 2070 -- but that's with XMP enabled. The 99th percentile (meaning 1 percent of frames are below this) was 50.4 fps.

 

[gggplaya]

same, I'm hitting a constant 60fps in 1080p with my r5 3600, rtx2700s and 32gb of ram on ultra, i think there is something wrong with their test bench.

They are getting 59fps average with the 3300x and 63fps with the 3600x, so you're getting what's right in line with their testing?

 

[JarredWaltonGPU]

Since these tests were done with an RTX 3090 GPU, can these charts be interpreted as the maximum FPS one can expect for the given processor?

Sort of. If you have faster RAM, with better timings, you can exceed these results by up to 10-15 percent in some cases. But at fully official stock clocks? Yes, this is the relative maximum performance you can expect (for the test sequence).

 

[Olle P]

Time is a factor, plus the patch was looming, so I went with what I had. ...

Instead of waiting for the patch before starting?

... The patch doesn't seem to make any significant changes, ...

Really? It's supposed to help Ryzen 3 and Ryzen 5 perform better.

 

[JarredWaltonGPU]

Instead of waiting for the patch before starting?

Really? It's supposed to help Ryzen 3 and Ryzen 5 perform better.

The 1.05 patch does exactly what the 'modified' EXE does (for AMD CPUs), as far as we can tell. So yes, it helps if you're not willing to hack the EXE file to change one byte, and you're using a 4-core to 8-core AMD chip (12-core and 16-core don't seem to benefit). It may help minimum fps a bit as well (99.9th percentile), but overall performance looks very close to 1.03 and 1.04. Plus, when you've already tested a dozen CPUs or more, scrapping all of that to restart isn't great. And it's supposed to fix some bugs, which aren't really performance related either.

 

[AgentLozen]

This article was posted several days ago so I realize I'm late to the party. Can anyone explain why Comet Lake cpus are performing so much better than Coffee Lake? The 1080p benchmarks indicate that Comet Lake smashes Coffee Lake on a per-core basis across the board. I'm surprised because I thought that their architectures were indistinguishable. I accounted for differences in cores and clock speed and there's not enough of a difference to explain the 30% performance gap. Any ideas?

 

[JarredWaltonGPU]

This article was posted several days ago so I realize I'm late to the party. Can anyone explain why Comet Lake cpus are performing so much better than Coffee Lake? The 1080p benchmarks indicate that Comet Lake smashes Coffee Lake on a per-core basis across the board. I'm surprised because I thought that their architectures were indistinguishable. I accounted for differences in cores and clock speed and there's not enough of a difference to explain the 30% performance gap. Any ideas?

Comet Lake supports DDR4-2933, Coffee Lake tops out at DDR4-2666. Also different motherboards and firmware patches -- could be something with the security vulnerability patches affects CFL more than CML. I'm not sure, but I definitely measured much higher performance than Paul when using DDR4-3600 XMP settings and timings for my RAM.