Cyberpunk 2077 Hotfix 1.05 would improve performance of certain Ryzen CPUs

HybOj

2020-12-20 10:44

yeah, 30 patches to go... than IF they add the cut out content, so the game is not a hollow corpse, I will consider getting the game. I will not play it in current content state, which is a joke compared to 5 years old GTA

#5869318

Neo Cyrus

2020-12-20 10:45

I have no idea WTF CDPR are trying to pull here, and I'm not going to re-write everything I've written before but CDPR are 100% full of BS stating that it doesn't matter for higher thread count chips and they're insane leaving SMT disabled for those. After further tests, yes SMT matters, a lot, even on my 3900X. 5-15% performance increases to average FPS, even in areas that seem 100% GPU bottlenecked with an RTX 3080 that hits 2.1GHz. More importantly than the average, it's preventing, or at least dulling, the severe lows. Just the giant increase in CPU usage should be a hint that something is going on even if your average FPS is the same. CPU usage doesn't skyrocket if nothing is being processed. Same goes for the memory mod that totally does nothing because the files literally did nothing in the final build... except every test I've done, repeatedly back to back many times, show that it's not just reloading the game that changes performance, the file modifications, which totally do nothing to the game, make a difference. Honestly CDPR just lost all respect I had left for them. I don't believe a fucking word they say anymore.

#5869325

Evronius

2020-12-20 11:36

Wtf? 8 cores and more behaving as intended. That's totaly BS. Toms Hardware even tested this and confirms that even 8 cores get improvments in performance! Cyberpunk 2077 CPU Scaling, What Processors Work Best? | Tom's Hardware

#5869328

Valken

2020-12-20 11:45

Fellow Gurus who have this game, can you please take some simple benchmarks at your own maxed or playable settings BEFORE and AFTER patch, like for like? Just want to see if it made any performance increase as well as fixes. I am more interested in improve minimum and average FPS than peak FPS. Thanks!

#5869340

Rich_Guy

2020-12-20 12:47

Patch is out now, just updated. 🙂

#5869342

Dragam1337

2020-12-20 12:52

Valken:

Fellow Gurus who have this game, can you please take some simple benchmarks at your own maxed or playable settings BEFORE and AFTER patch, like for like? Just want to see if it made any performance increase as well as fixes. I am more interested in improve minimum and average FPS than peak FPS. Thanks!

Made no noteworthy performance difference for me, then again, it didn't mention any performance improvements in the patch notes.

#5869350

Fox2232

2020-12-20 13:30

Some jokes are just that: "Removed the use of AVX instruction set thus fixing crashes occurring at the end of the Prologue on processors not supporting AVX." Like what? Freaking hell, it is 2020. If CPU has AVX, use it, if it does not, use alternative code path. Which in most cases can be optimized by compiler.

#5869351

JAMVA

2020-12-20 13:31

Removed the use of AVX instruction set thus fixing crashes occurring at the end of the Prologue on processors not supporting AVX My sandy Bridge from 2011 supported AVX , Instead of actually fixing it , they just ripped the code out for everybody 🙄

#5869354

Neo Cyrus

2020-12-20 13:48

JAMVA:

Removed the use of AVX instruction set thus fixing crashes occurring at the end of the Prologue on processors not supporting AVX My sandy Bridge from 2011 supported AVX , Instead of actually fixing it , they just ripped the code out for everybody 🙄

That's a seriously bizarre and lazy thing to do. I hope they mean that it simply won't try to run AVX instructions on CPUs that don't support it, but they are wording it as if they just outright removed it for everyone.

#5869357

Stormyandcold

2020-12-20 14:16

I'm enjoying this game tbh, but, the performance needs improving. It's as if occlusion culling isn't working properly and that's why performance isn't the best.

#5869360

DannyD

2020-12-20 14:22

Neo Cyrus:

.. Same goes for the memory mod that totally does nothing because the files literally did nothing in the final build... except every test I've done, repeatedly back to back many times, show that it's not just reloading the game that changes performance, the file modifications, which totally do nothing to the game, make a difference. Honestly CDPR just lost all respect I had left for them. I don't believe a procreating word they say anymore.

Yup, my game is back to being more jittery after they removed the memory pool file. I don't believe a word either.

#5869363

rflair

Moderator

2020-12-20 14:34

The game never had any AVX code, it was an oversight. Maybe its a compiler thing. There have been a few games where AVX was 'patched' out and the game ran exactly the same, with the benefit of some older CPUs with enough grunt able to play the games afterward.

#5869376

asturur

2020-12-20 15:15

Neo Cyrus:

I have no idea WTF CDPR are trying to pull here, and I'm not going to re-write everything I've written before but CDPR are 100% full of BS stating that it doesn't matter for higher thread count chips and they're insane leaving SMT disabled for those. After further tests, yes SMT matters, a lot, even on my 3900X. 5-15% performance increases to average FPS, even in areas that seem 100% GPU bottlenecked with an RTX 3080 that hits 2.1GHz. More importantly than the average, it's preventing, or at least dulling, the severe lows. Just the giant increase in CPU usage should be a hint that something is going on even if your average FPS is the same. CPU usage doesn't skyrocket if nothing is being processed. Same goes for the memory mod that totally does nothing because the files literally did nothing in the final build... except every test I've done, repeatedly back to back many times, show that it's not just reloading the game that changes performance, the file modifications, which totally do nothing to the game, make a difference. Honestly CDPR just lost all respect I had left for them. I don't believe a procreating word they say anymore.

I think they know theyr game more than you or some other website do.

#5869383

theoneofgod

2020-12-20 16:07

asturur:

I think they know theyr game more than you or some other website do.

Well..... I don't know.

#5869384

Calmmo

2020-12-20 16:07

Memory file made 0 changes and had 0 impact. SMT also has no noticable impact on average framerate and sites that tested it show reduced lows with it on. 5900x and the 3900x (had the 3900x installed the first couple of days of CP) showed no perf. difference. Some people just like to yell into the social media ether and cry wolf.

#5869391

Alessio1989

2020-12-20 16:40

Evronius:

Wtf? 8 cores and more behaving as intended. That's totaly BS. Toms Hardware even tested this and confirms that even 8 cores get improvments in performance! Cyberpunk 2077 CPU Scaling, What Processors Work Best? | Tom's Hardware

"confirms" so, now THW is science? xD

HybOj:

yeah, 30 patches to go... than IF they add the cut out content, so the game is not a hollow corpse, I will consider getting the game. I will not play it in current content state, which is a joke compared to 5 years old GTA

still less bugged than "critical acclaimed" bethesda TES/FALLOUT games

#5869404

Evronius

2020-12-20 17:21

Alessio1989:

"confirms" so, now THW is science? xD still less bugged than "critical acclaimed" bethesda TES/FALLOUT games

I linked to the test they did and it shows that CDPR has wrong. But I guess that you didnt bother to take a look and see for your self.

#5869406

kapu

2020-12-20 17:22

I think they lied a bit on that 8 core part. Tomshardware claims fix is working . But check that 3300X and 5600X performance 😀 AMD 6 core matching intel 8 core. https://scontent.fktw5-1.fna.fbcdn.net/v/t1.15752-9/131657567_216670309989267_3195463003867249411_n.png?_nc_cat=107&ccb=2&_nc_sid=ae9488&_nc_ohc=sTfn5ikqF5kAX-DNC5S&_nc_ht=scontent.fktw5-1.fna&oh=42cf742b1bb1feea53204de8792c65ea&oe=6005D9D7

#5869436

kapu

2020-12-20 19:22

I wonder if anyone can explain poor 9900k performance .... It completely sux vs almost any AMD offering.

#5869446

Netherwind

2020-12-20 19:52

Valken:

Fellow Gurus who have this game, can you please take some simple benchmarks at your own maxed or playable settings BEFORE and AFTER patch, like for like? Just want to see if it made any performance increase as well as fixes. I am more interested in improve minimum and average FPS than peak FPS. Thanks!

Slight difference with less lows perhaps but no major difference with a 5900X.

Alex13:

Memory file made 0 changes and had 0 impact. SMT also has no noticable impact on average framerate and sites that tested it show reduced lows with it on. 5900x and the 3900x (had the 3900x installed the first couple of days of CP) showed no perf. difference. Some people just like to yell into the social media ether and cry wolf.

Do you agree that the game is more demanding on your CPU than GPU? I'm running my 3080 undervolted but cannot really say that GPU is stressed to the absolute maximum, like in Control for example, rather sitting at some 96-97%. Feels like the game is crazy heavy on the CPU which is understandable with all the things going on at the same time.