It seems that you're using an outdated browser. Some things may not work as they should (or don't work at all).
We suggest you upgrade newer and better browser like: Chrome, Firefox, Internet Explorer or Opera

×
This is the patch note on AMD SMT:

[AMD SMT] Optimized default core/thread utilization for 4-core and 6-core AMD Ryzen(tm) processors. 8-core, 12-core and 16-core processors remain unchanged and behaving as intended. This change was implemented in cooperation with AMD and based on tests on both sides indicating that performance improvement occurs only on CPUs with 6 cores and less.


I was confused and disappointed to see that this change only affects 6 core cpus or less. I have an 8 core cpu(16t) I have monitored my cpu usage and it seems to be under-utilized compared to other games. I have also heard the hex edit had helped with cpus that were more than 6 cores only it had less of an impact.

So I guess my question is

1) if SMT is working as intended why does this game seem to utilize my cores/threads less than other games?
2)has anyone with a 8 core AMD cpu or more used the hex edit trick and if so what were your results? I'd like to verify what GoG is saying here if possible because it doesnt seem to add up.
avatar
jayseearr: 1) if SMT is working as intended why does this game seem to utilize my cores/threads less than other games?
2)has anyone with a 8 core AMD cpu or more used the hex edit trick and if so what were your results? I'd like to verify what GoG is saying here if possible because it doesnt seem to add up.
1) it could simply mean that the CPU is not the bottleneck

2) I have 12 (Ryzen 9 3900X) and honestly I didn't notice any change with or without the hack.
avatar
jayseearr: 1) if SMT is working as intended why does this game seem to utilize my cores/threads less than other games?
2)has anyone with a 8 core AMD cpu or more used the hex edit trick and if so what were your results? I'd like to verify what GoG is saying here if possible because it doesnt seem to add up.
avatar
Gersen: 1) it could simply mean that the CPU is not the bottleneck

2) I have 12 (Ryzen 9 3900X) and honestly I didn't notice any change with or without the hack.
ok so you actually did the hex edit and didn't notice anything with a 3900x? interesting, ok thanks for the input

more input welcome from anyone/everyone who has tried the hex edit with a 8 core or more AMD cpu
Post edited December 19, 2020 by jayseearr
avatar
jayseearr: ok so you actually did the hex edit and didn't notice anything with a 3900x? interesting, ok thanks for the input
I tried both the hex, rollback the changes, and the DLL injection one. No difference, at least on my system.
3700x here - after a play through free from major bugs - I edited the file and tried it for a while.

Whether by accident or the edits, there was no performance difference but I started to notice bugs. Played for a few hours and rolled it back to original. Performance remained same and bugs decreased.

Not saying the edit caused the bugs necessarily, but it definitely didn't make a change to performance.
avatar
jayseearr: 2)has anyone with a 8 core AMD cpu or more used the hex edit trick and if so what were your results? I'd like to verify what GoG is saying here if possible because it doesnt seem to add up.
I have a Ryzen 5 2700x and I did see a huge bump in utilization. Before the hack it was in total of 19-24% and after it was 24-40%.

According to Fraps the performance however seem to be largely the same.

Due to updates I've moved the game to an HDD so there are some loading issues, but yeah, since I started using this and the memory-pool hack I've been getting crashes (3 so far). Didn't have any crashes before the hack, not even before 1.04 came.

So no, doesn't seem to be working on a double CCX CPU or with more cores...
Post edited December 19, 2020 by sanscript
avatar
jayseearr: 2)has anyone with a 8 core AMD cpu or more used the hex edit trick and if so what were your results? I'd like to verify what GoG is saying here if possible because it doesnt seem to add up.
avatar
sanscript: I have a Ryzen 5 2700x and I did see a huge bump in utilization. Before the hack it was in total of 19-24% and after it was 24-40%.

According to Fraps the performance however seem to be largely the same.

Due to updates I've moved the game to an HDD so there are some loading issues, but yeah, since I started using this and the memory-pool hack I've been getting crashes (3 so far). Didn't have any crashes before the hack, not even before 1.04 came.

So no, doesn't seem to be working on a double CCX CPU or with more cores...
CDPR said the memory pool file was a leftover from a prevoius version and had no effect. I think it's in the 1.05 post notes
.
avatar
smuggly: CDPR said the memory pool file was a leftover from a prevoius version and had no effect. I think it's in the 1.05 post notes
Ah, haven't gotten that memo yet. Thanks for clearing that up.
Thank you for the input so far guys. I was web browsing and i found an interesting article that tested various different amd cpus with the hex edit and the results clearly showed that some 8+ core cpus were seeing performance increases. The same article also said the hex edit doesn't seem to help the ryzen 9 series and in some cases actually hurt ryzen 9 performance which would explain why the guy with the 3900x didn't see any improvements, but it would also indicate that there is still some performance being left on the table for other amd cpus.


*edit*
I can't post links on here but if you google cyberpunk ryzen hex edit results i think the second result is a arcticle/experiment by overclock3d if you want to see for yourselves and it seems to show that there is (potential) performance to be gained from all core counts not just 6 or less.
Post edited December 19, 2020 by jayseearr
https://www.gog.com/forum/cyberpunk_2077/further_investigate_amd_smt_fix_because_it_helps_6c_cpus_too_proofs
I tried it with a 3900X. It did indeed utilize all 24 threads. Not sure if performance was affected, I've since rolled back the edit. I'm heavily GPU bound anyway (2070 Super with RT on)

It makes sense that utilizing all threads might be detrimental on high core count CPUs, as Windows normaly schedules the best performing cores for heavy loads.
Post edited December 19, 2020 by AZaph_gog