Rpcs3: God Of War 3 physics bug

[Wizzy2018]

Rpcs 3 0.0.6-7947 version

Just started playing God of War 3 and when i reach the first Boss suddenly Kratos dissapears! And a screen goes black and it says You Died! After that screen stuck and emulator don't respond to anything i need to close it to be able to use it again.

CPU-Ryzen 7 1700
RAM-DDR4-2400Mhz
GPU-RX 470 4Gb

Here is a screenshoot
A resolution is default 1280x720

Here is a part when Kratos dissapears

[RainbowCookie32]

Attach a log when creating an issue.

Reproduce the issue, close RPCS3, grab RPCS3.log from the emulator's folder and add it to your original post

[Asinin3]

Log won't help with this. But either way it's a random bug that happens from time to time. You just need to try a few more times and eventually you will be attached to the ground on this section. It's one of many reasons the game is not listed as playable. (Although normally you just fall to your death and it doesn't crash).

[Wizzy2018]

Here it is a log file and the screenshoot

RPCS3.log.gz

[QuaiGoner]

I have that issue too, i think this happens because of the low framerate and Kratos spawns somewhere behind the wall that you should climb or something and starts to fall where on real HW you need to climb

[MSuih]

It's a known fact that GOW 3 physics break if perfomance is not ideal. The game can be played from start to finish on overclocked 8700k but anything slower than that will probably run into issues. Improving SPU performance should eventually fix this.

[RandomP8]

The game can be played from start to finish on overclocked 8700k...

this is crazy!

[Wizzy2018]

I have now another problem.
I getting this error now on latest build of RPCS 3 0.0.6-8002
·F 0:06:26.511480 {RSX [0x016e5bc]} RSX: class std::runtime_error thrown: Unknown stencil op 0xe400
(in file c:\projects\rpcs3\rpcs3\emu\rsx\gcm_enums.cpp:706)
Here is a log file
RPCS3.log.gz

[kd-11]

Retest. All problems here are caused by sub-optimal performance; maybe the recent improvements may have alleviated most of them.

[Wizzy2018]

I test it soon and report back to you

On Thu, Jul 4, 2019, 13:30 kd-11 [e-mail] wrote:

Retest. All problems here are caused by sub-optimal performance; maybe the
recent improvements may have alleviated most of them.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
https://github.com/RPCS3/rpcs3/issues/5806?email_source=notifications&email_token=AKRWCNRD6MH33RUBURSIEXDP5XNMXA5CNFSM4HER27F2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZHFCPY#issuecomment-508449087,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AKRWCNVUSXE7UPTJRE2QKL3P5XNMXANCNFSM4HER27FQ
.

[QuaiGoner]

Ryzen 5 2400G
GTX 1050
Latest Drivers

Same issue, perfomance is still too low. Maybe something chaged for more powerful processors that could not get through that sequence

[Wizzy2018]

I retest it.On my Ryzen 7 1700 3.2@Ghz same issue nothing changed.One thing i noticed when i disable in cpu option a Thread scheduler i gain 5-6 more fps and my fps is more stable.When i enable thread scheduler it going very slow and fps jumping from 1-15fps every sec.

[Wizzy2018]

I just think this emulator is not optimized for Ryzen cuz it was mostly favored Intel.I don't believe 8 core 16 thread Cpu can't run it well with 3.2 Ghz and God of War usses only 50 % of Ryzen cpu so there is enough room left for a boost

[kd-11]

Thread scaling does not work like that. You cannot split a PS3 hardware thread into multiple simulated threads so 50% is about where a PS3 system would map to a Ryzen 7 (2 PPE + 6 SPE threads). 3.2 Ghz is also very slow, consider overclocking to 3.6-3.9. The reason newer intel performs so much better is because of TSX technology, for which AMD has no alternative in the real world. TSX makes a huge difference because it closely matches PS3 behavior and does not require complicated workarounds. I have a Ryzen processor as well so this claim about rpcs3 being anti-AMD is just not true. While AMD has proposed processor extensions for transactional memory, none have been actually implemented in hardware to date. See https://en.wikipedia.org/wiki/Advanced_Synchronization_Facility

[Wizzy2018]

Thx for explain KD-11.
Do you think is it possible somewhere in future we see Ryzen 7 better
performance??

On Thu, Jul 4, 2019, 16:15 kd-11 [e-mail] wrote:

Thread scaling does not work like that. You cannot split a PS3 hardware
thread into multiple simulated threads so 50% is about where a PS3 system
would map to a Ryzen 7 (2 PPE + 6 SPE threads). 3.2 Ghz is also very slow,
consider overclocking to 3.6-3.9. The reason newer intel performs so much
better is because of TSX technology, for which AMD has no alternative in
the real world. TSX makes a huge difference because it closely matches PS3
behavior and does not require complicated workarounds. I have a Ryzen
processor as well so this claim about rpcs3 being anti-AMD is just not
true. While AMD has proposed processor extensions for transactional memory,
none have been actually implemented in hardware to date. See
https://en.wikipedia.org/wiki/Advanced_Synchronization_Facility

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
https://github.com/RPCS3/rpcs3/issues/5806?email_source=notifications&email_token=AKRWCNVOQNIPQJBOT4K7E4TP5YAXJA5CNFSM4HER27F2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZHQ4PA#issuecomment-508497468,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AKRWCNQMYJFI5AS2BMIKIHTP5YAXJANCNFSM4HER27FQ
.

[kd-11]

Well, yes of course, its just a lot harder than on intel where we have hardware assistance to emulate difficult features. Performance on all processors does improve just slower on some processors than others. Compared to where performance was at ryzen launch, performance has gone up an order of magnitude in most games. Its just that this specific game is one of the last ones that are still slow.

[Wizzy2018]

I want to say Thx KD-11 for a hard work you doing with this emulator 😘
Love you bro <3

On Thu, Jul 4, 2019, 16:31 kd-11 [e-mail] wrote:

Well, yes of course, its just a lot harder than on intel where we have
hardware assistance to emulate difficult features. Performance on all
processors does improve just slower on some processors than others.
Compared to where performance was at ryzen launch, performance has gone up
an order of magnitude in most games. Its just that this specific game is
one of the last ones that are still slow.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
https://github.com/RPCS3/rpcs3/issues/5806?email_source=notifications&email_token=AKRWCNQJD5MOCMZWUO4QTKDP5YCVPA5CNFSM4HER27F2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZHSEXA#issuecomment-508502620,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AKRWCNXAC5DGZZ3DD7S6KODP5YCVPANCNFSM4HER27FQ
.

[Wizzy2018]

So i tested again with Overclocking my R7 1700 to 3.7Ghz and i past a screen where i was stuck but after a fight with a water horse is done i got this screen and i stuck there.
God of War® III [BCES00510] 2019-07-12 09-28-49.zip

[jeremybris]

Try this game again. Someone reported the bug has been fixed recently with "SPU Decoder - Interpreter (Fast)"

Edit: Try with this https://ci.appveyor.com/api/buildjobs/pe9poxs8twn8nig3/artifacts/rpcs3-v0.0.8-6fb3307a_win64.7z (elad's fix preview)

[AniLeo]

Accuracy differences between Interpreters and Recompilers are known, may not be fixed

[RainbowCookie32]

Just played through this area and couldn't reproduce the issue with the same CPU.

Issue can probably be closed, unless OP can still reproduce it on latest master.

[Xcedf]

Oh come on, the game just needs to be fast enough
here's the example, issue still here when i use SPU Interpreter and it runs very slow
even ASMjit is enough for me to pass it
https://youtu.be/qBWbjji2U1g

[Asinin3]

It happens on my R7 1700 @3.9ghz consistently even when my fps is 23-26 there.

And yet I've seen another person with an 8700k having the issue too, though his 8700k was at stock he was using the right settings. I still think it's performance related, but it seems to require quite a lot of FPS for it not to happen.