-
-
Notifications
You must be signed in to change notification settings - Fork 1.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
rsx: Fixes #4611
rsx: Fixes #4611
Conversation
|
Fixes flickering during Tekken Tag Tournament 2 Intro on pre-rendered videos but also the "press start" screen. Didn't find any regressions with the following games (excluding the lockups): Skate 3, Persona 5, Ratchet & Clank 1 (one fps improvement), Ninja Gaiden Sigma, Teslagrad, Zone of the Enders 1, Godzilla, Wangan Midnight, Double Dragon Neon, Dreamworks Superstar Kartz, Dynasty Warriors 6, Asura's Wrath, Knights Contract, Driver San Francisco and Uncharted 1 |
|
Asinin3, is Ninja Gaiden Sigma movies fixed too? |
|
Fixes flickering in Uncharted series and Sly Cooper Thieves in Time |
|
Damn squashed commit made tests more difficult, yet the issue with Metro caused by |
|
Uncharted and other games movie flickering is fixed by using strict rendering mode.. That is how master also behaves |
|
Conan [BLES00076] freezes upon initial rendering of gameplay now (4/4 times) and kills the emulator as well (need to task manager > kill process). That means I don't get a log after to upload here. Copied the last few lines from the console below. Runs fine on Master and marked as Playable on the compat. list (only missing a few gfx/textures). Let me know if you need anything else. E {PPU[0x100001d] Thread (FMOD thread for FMOD_NONBLOCKING) [0x00d12090]} 'sys_fs_open' failed with 0x80010006 : CELL_ENOENT, “/dev_hdd0/game/BLES00076/sounds/conan_ambient_cold_rain.fsb” [1] [Update] Fixed on new commits/rebase. |
|
@legend800 i may be mistaken, but the issue you've discribed look's like #4615, which is nothing to do with this PR, and you had the issue here cause this PR is not yet rebased on latest master, as you said yourself it works on master |
|
Possible I guess. I'll retest after a rebase. |
3bba005 to
e3c1263
Compare
|
Strict rendering is need to fix lighting |
|
@MarioSonic2987 That build is wrong. It does not have the last commit added to master which fixes that issue. Rpcs3 appveyor did not compile this but my repo did - https://ci.appveyor.com/project/kd-11/rpcs3/build/0.0.0.6-6822 |
|
p5 mem access error Edit Wrong report i was testing spu llvm i forgot to use asmjit |
|
@spyropt you're using spu llvm, your crash is not related to these changes |
|
@13xforever Sorry i forgot that i was testing spu llvm |
|
So this happens very reliably if you go to the Diner in Shibuya on a rainy day and order a Hot Coffee |
|
Deadlocks have been fixed |
1. rsx: Rework section synchronization using the new memory mirrors
2. rsx: Tweaks
- Simplify peeking into the current rsx::thread instance.
Use a simple rsx::get_current_renderer instead of asking fxm for the same
- Fix global rsx super memory shm block management
3. rsx: Improve memory validation. test_framebuffer() and
tag_framebuffer() are simplified due to mirror support
4. rsx: Only write back confirmed memory range to avoid overapproximation errors in blit engine
5. rsx: Explicitly mark clobbered flushable sections as dirty to have them
removed
6. rsx: Cumulative fixes
- Reimplement rsx::buffered_section management routines
- blit engine subsections are not hit-tested against confirmed/committed memory range
Not all applications are 'honest' about region bounds, making the real cpu range useless for blit ops
- Allows grabbing an unsynchronized memory block if overwriting contents anyway - Allows flushing only specified range of memory
…sx::thread - Defers on_notity_memory_unmapped to only run from within rsx context - Avoids passive_lock + writer_lock deadlock
|
Using 66ca3e2 6600k(tsx) win10-x64 Getting a huge number (1000s) of these: Edit: this occurred prior to this commit, but not to the extreme level it does now. The game accepts no input after the fifo spam begins. |
|
Regressions will have to be tracked separately before this grows unmaintainable |
|
@jon-reid This PR addresses a specific type of lockup, my guess is that the same error happens before this PR. |
|
With this PR MSi Afterburner overlay, finnaly started to work. |
|
@Asinin3 - Ok, i reported it 4 months ago, regardless Geforce Experience overlay FPS counter is still broken - it shows 30 fps regardless of real FPS with Vulkan. |











This PR will cause deadlocks on non-TSX cpus due to a vm lock contention issue. That will hopefully be fixed soon.(Deadlocks fixed)