Cyberpunk 2077: PC graphics perf benchmark review

Fox2232

2020-12-10 17:01

Seeing 40% CPU utilization of 16C/32T CPU. I would like to see 1080p resolution and framerate scaling for: 4C/4T ; 4C/8T ; 6C/6T ; 6C/12T ; 8C/8T ; 8C/16T @cucaulay malkin : https://forums.guru3d.com/threads/new-games-added-to-the-dlss-list-incl-cyberpunk-2077.435605/page-3#post-5864704 (As example.) But it is as always pointless discussion. Some see differences, some can swear on grave of their beloved that there is none.

#5864766

Embra

2020-12-10 17:02

Seems the game does not benefit much higher core count cpus. Even the top recommended is 4790/3600. A bit surprised.

#5864768

DannyD

2020-12-10 17:06

pharma:

Take a read over here. People with lessor cards are getting higher fps, and quality.

Nice one, if i run into trouble further in i'll take a look at the Cascaded Shadows Resolution. 🙂

#5864770

cucaulay malkin

2020-12-10 17:07

cpu benches interesting 5800x can't keep a consistent 60 5600x and 3800xt do very poorly https://www.pcgameshardware.de/Cyberpunk-2077-Spiel-20697/Specials/Cyberpunk-2077-Benchmarks-GPU-CPU-Raytracing-1363331/

#5864772

DannyD

2020-12-10 17:09

cucaulay malkin:

cpu benches interesting 5800x can't keep a consistent 60 5600x and 3800xt do very poorly https://www.pcgameshardware.de/Cyberpunk-2077-Spiel-20697/Specials/Cyberpunk-2077-Benchmarks-GPU-CPU-Raytracing-1363331/

Cheers for this, you see folks it's a cpu hog.

#5864773

cucaulay malkin

2020-12-10 17:10

Fox2232:

Seeing 40% CPU utilization of 16C/32T CPU. I would like to see 1080p resolution and framerate scaling for: 4C/4T ; 4C/8T ; 6C/6T ; 6C/12T ; 8C/8T ; 8C/16T @cucaulay malkin : https://forums.guru3d.com/threads/new-games-added-to-the-dlss-list-incl-cyberpunk-2077.435605/page-3#post-5864704 (As example.) But it is as always pointless discussion. Some see differences, some can swear on grave of their beloved that there is none.

it's possible it gets patters confused but I thought you were gonna point out some bigger difference for 2x performance uplift form pic1 to pic2 I'll take that and won't complain

DannyD:

Cheers for this, you see folks it's a cpu hog.

9900k falls behind too with 2666 ram I suspect this could be what's keeping 6c/8c ryzens behind,memory latency huge gaming cache will run out at some point too 3600 vs 10400 is a massacre,30% faster on intel

#5864774

Fox2232

2020-12-10 17:10

DannyD:

Sry to hear that JB, i've been loving it and the 'slow pace' at the start suits me as there is so much to take in and i didn't even consider it slow i think it's perfect for me.

Slow != Bad. Like with Raised by Wolves. Quite a few people here wrote that it is bad. And I had low expectations as result. Even postponed watching till there were no good shows around. It is slow paced, but I enjoyed it from 1st moment.

cucaulay malkin:

it's possible it confuses patters but for 2x performance uplift form pic1 to pic2 I'll take that and won't complain

Time to finish DLSS process is affected by card. 3090 is able to finish in fraction of time it takes 3070 to finish DLSS processing due to number of available tensor cores.

#5864776

Denial

2020-12-10 17:12

Fox2232:

Slow != Bad. Like with Raised by Wolves. Quite a few people here wrote that it is bad. And I had low expectations as result. Even postponed watching till there were no good shows around. It is slow paced, but I enjoyed it from 1st moment.

That ending tho.. lol

#5864778

bobnewels

2020-12-10 17:17

cucaulay malkin:

cpu benches interesting 5800x can't keep a consistent 60 5600x and 3800xt do very poorly https://www.pcgameshardware.de/Cyberpunk-2077-Spiel-20697/Specials/Cyberpunk-2077-Benchmarks-GPU-CPU-Raytracing-1363331/

Thanks for the the benchmarks Who in the world is playing @ 1280x720 ,never mind only a Intel fan would post this crap. Come on no one is playing at 1280x720 with Intel/AMD high end CPU that is just ridiculous.

#5864780

cucaulay malkin

2020-12-10 17:18

Fox2232:

Time to finish DLSS process is affected by card. 3090 is able to finish in fraction of time it takes 3070 to finish DLSS processing due to number of available tensor cores.

what ? this is absoulutely not true and I don't know where you'd get an idea like that cause it's never been the case in any of dlss games I've seen benched. https://www.computerbase.de/2020-12/nvidia-geforce-rtx-3060-ti-asus-msi-test/2/#diagramm-control-raytracing-in-2560-1440 scroll down to dlss results 2070 S has a bigger perf uplift from dlss than 2080ti all ampere cards have equal number of tensor per sm

#5864781

Fox2232

2020-12-10 17:19

Denial:

That ending tho.. lol

Yep, they could have thought something better. Especially since that thing is actually copy.

bobnewels:

Thanks for the the benchmarks Who in the world is playing @ 1280x720 ,never mind only a Intel fan would post this crap. Come on no one is playing at 1280x720 with Intel/AMD high end CPU that is just ridiculous.

Resolution does not really matter much. It for example says that even on 720p, 6C/6T CPU like 9600K averages at 60fps and dips down to 35fps. Higher resolutions can only get worse from there. But 6C/12T 10400F (even while clocked bit lower) delivers average 74fps and dips to 46fps. So 6C/6T has around 1/2 of frametimes worse than 16.67ms (for 60fps target) and dips are quite bad. While 6C/12T has most of frametimes better than 16.67ms and occasional dips to 46fps may be quite acceptable for many. But 1st really good experience starts with 5800X/10700K. Those can deliver about solid 60fps and above. @cucaulay malkin DLSS been discussed to death. I have no reason to repeat same discussions again. You either know or don't.

#5864782

cucaulay malkin

2020-12-10 17:19

bobnewels:

Thanks for the the benchmarks Who in the world is playing @ 1280x720 ,never mind only a Intel fan would post this crap. Come on no one is playing at 1280x720 with Intel/AMD high end CPU that is just ridiculous.

okay boomer I don't do reviews blame pcgh for testing cpus in cpu limited scenarios

#5864783

Denial

2020-12-10 17:28

Look at the performance difference between 3200 and 3800 on that pcgh test. Drop 3800 on the 5800x and it will do 60 consistently easily.

#5864785

cucaulay malkin

2020-12-10 17:30

Denial:

Look at the performance difference between 3200 and 3800 on that pcgh test. Drop 3800 on the 5800x and it will do 60 consistently easily.

thanks,that's what I suspected too memory latency on ryzen is having a bigger impact than usual meanwhile 10700K on 2933 is fine,9900k on 2666 takes a hit very cpu intensive game if memory has such a big impact my 4133 c16 sticks may do a fine job here 🙂

#5864787

DannyD

2020-12-10 17:33

Memory will be my next upgrade seeing as i get by on current card and have no issues with DLSS balanced, i'll eye some nice 4600mhz sticks in March. Currently 3600 c15.

#5864788

cucaulay malkin

2020-12-10 17:34

DannyD:

Memory will be my next upgrade seeing as i get by on current card and have no issues with DLSS balanced, i'll eye some nice 4600mhz sticks in March.

3600 c15 is more than fine for intel's ring bus cpus especially if you oc the ring bus too

#5864789

DannyD

2020-12-10 17:35

cucaulay malkin:

3600 c15 is more than fine for intel's ring bus cpus especially if you oc the ring bus too

It's the weak point in my rig and they'll be super future proof. I originally had 4400Mhz in this rig.

#5864790

The Phoenix

2020-12-10 17:37

turning those cascaded shadows range and reso to low,makes a HUGE difference in performance as some have said

#5864792

cucaulay malkin

2020-12-10 17:39

The Phoenix:

turning those cascaded shadows range and reso to low,makes a HUGE difference in performance as some have said

what are cascaded shadows ?

#5864794

Undying

2020-12-10 17:42

DannyD:

Cheers for this, you see folks it's a cpu hog.

You want those 8c/16t cpus for this game.