Commit graph

1003 commits

Author SHA1 Message Date
kd-11 bff6e0d277 rsx: Don't assert when surface and texture cache overlap lockable memory
- It's not a serious enough problem and the root cause is understood.
2023-07-11 04:09:30 +03:00
kd-11 c325017675 rsx: Preserve the texcoord transform around destructive modifications 2023-07-11 04:09:30 +03:00
kd-11 e7ac84fca3 rsx: Fine-tune the subimage view bezel dimensions 2023-07-06 14:06:30 +03:00
kd-11 b9ea6be842 rsx: Fix AA factor calculation 2023-07-06 14:06:30 +03:00
kd-11 d46dbc0fa8 rsx: Fill section dimensions when converting atlas op to copy op 2023-07-06 14:06:30 +03:00
kd-11 79e000f63d rsx: Fix virtual coordinate clamping 2023-07-04 09:31:51 +03:00
kd-11 d13cf0e29a rsx: Fix shader interpreter
- It broke ages ago
2023-07-04 09:31:51 +03:00
kd-11 69eb1401a9 vk/gl: Fix debug overlay stats 2023-07-04 09:31:51 +03:00
kd-11 aa06ac119f rsx: Fix sampled descriptor address corruption 2023-07-04 09:31:51 +03:00
kd-11 1afee59c95 rsx: Insert missing texture barrier 2023-07-04 09:31:51 +03:00
kd-11 4decad6f6b rsx: Force compatibility checks when applying fast-path optimizations 2023-07-04 09:31:51 +03:00
kd-11 a37f786a37 rsx: Properly fill in the cyclic ref field for framebuffer references 2023-07-04 09:31:51 +03:00
kd-11 8f98e572e1 rsx: Use designated initializer for the dangerous copy_region_descriptor type
- We need to move more dangerous braced initializations to c++20
- Also adds a base address variable which will come in handy
2023-07-04 09:31:51 +03:00
kd-11 1d004f2788 rsx: Lower single-image-blit to a scaled coordinate fetch 2023-07-04 09:31:51 +03:00
kd-11 cef4ea786a rsx: Trivially detect single-image-blit transfer operations 2023-07-04 09:31:51 +03:00
kd-11 69b02f7e46 rsx: Implement atlas trivialization pass
- Decompose sequences of copies into the minimally required set.
  This does 2 things:
  a. Reduces GPU workload by doing less transfers
  b. Allows the clipping optimization to kick in and skip a ton of GPU work
2023-07-04 09:31:51 +03:00
kd-11 b788e05396 rsx: Implement GPU copy ellision through coordinate transform
- TODO: Some corner cases still exist where format may not be a match after a cache merge.
2023-07-04 09:31:51 +03:00
kd-11 d77a78cdf1 rsx: Rework texture coordinate handling to support clamping and a more sane scale-bias setup 2023-07-04 09:31:51 +03:00
kd-11 ca472b6750 Silence log spam about copy ellision 2023-06-27 16:41:30 +03:00
kd-11 2df7e46cb1 rsx: Import robin-hood unordered map which is much faster than std 2023-06-10 01:36:10 +03:00
kd-11 bbbc5f6e6c rsx: Write a warning to the log if we decide to collapse surfaces to reduce VRAM usage. 2023-06-01 19:25:50 +03:00
kd-11 12f213ffad rsx: Trim the number of in-flight invalidated resources (temp cache)
- This drastically improves memory allocation behavior.
  Holding too many invalidated resources can lead to a cascading overallocation error as old resources hold refs to even older resources and nothing gets deleted.
2023-06-01 19:25:50 +03:00
kd-11 c99ef4b09f vk: Refactor descriptor handling 2023-05-30 16:50:08 +03:00
kd-11 7e131f8fb6 rsx/util: Add some c++ guards around simple_array class 2023-05-30 16:50:08 +03:00
kd-11 793a21a825 rsx: Fix build after variable rename 2023-05-18 18:10:46 +03:00
kd-11 edb2b60f2f rsx: Perform region clipping in a normalized coordinate space
- Depending on what the caller wants to do, the scaling will be different for src vs dst calculations.
  We cannot reliably handle this in the callee
2023-05-18 18:10:46 +03:00
kd-11 b00b8ae71b rsx: Allow swizzle flag to propagate as an override to surface properties 2023-05-03 01:26:14 +03:00
kd-11 45f416c4f1 rsx: Apply swizzle flag on data source as well
- The fact that the attribute is tacked on to the dest is a semantic error.
  It's the context that is swz\ln, not the source or dest. They both are one or the other.
2023-05-03 01:26:14 +03:00
kd-11 c58149f5b6 vk: Propagate swizzle flag in hw blit engine 2023-05-03 01:26:14 +03:00
Megamouse 6555de6ef5 fix warnings 2023-04-21 16:41:06 +02:00
Ivan Chikish 1f19804cab BufferUtils: fix SSE2 path for byteswaps 2023-04-18 16:46:18 +03:00
kd-11 a3ededdcf1 rsx: Implement deferred buffer allocations for GPU write 2023-03-02 12:48:58 +03:00
kd-11 0bb8127372 rsx: Use emplace when constructing in-place.
- Unexpectedly high gains for msvc builds. GCC doesn't care, clang shits itself.
2023-03-01 01:10:06 +03:00
kd-11 6adcabda29 rsx: Fix graphics state foot-gun 2023-01-26 11:42:39 +03:00
kd-11 29c1b20b41 Fix compilation 2023-01-11 16:48:53 +03:00
kd-11 439bdde849 rsx: Fix printing of expected values 2023-01-11 16:48:53 +03:00
kd-11 a272f3e3b9 rsx: Improve performance by using an integral type to indicate error 2023-01-11 16:48:53 +03:00
kd-11 eae1ac6558 refactor: Fix build 2023-01-11 16:48:53 +03:00
kd-11 0b019401bd Refactor gcm enums 2023-01-11 16:48:53 +03:00
kd-11 04fb86556a rsx: Fix surface metadata life-cycle
- Beware of clone operations. Blindly inheriting the parent's metadata is wrong.
- It is possible, especially when reusing a pre-existing slice, that the parent and child info has diverged
2022-12-17 20:16:58 +03:00
kd-11 90cf47cdce rsx: Handle some corner cases in surface locking 2022-12-17 20:16:58 +03:00
kd-11 66dc1cc15d rsx: Conditionally skip flush if no new data was introduced 2022-12-17 20:16:58 +03:00
kd-11 a05e3f02b8 rsx: Avoid expensive protection scan by sharing some data between surface and texture cache 2022-12-17 20:16:58 +03:00
kd-11 a0ef1a672c rsx: Implement interpolation using barycentrics 2022-12-11 15:21:58 +03:00
Nekotekina 7c15001042 Implement read_from_ptr<>() util
Doing std::bit_cast on a "span".
Should be usable in constexpr.
2022-11-26 09:30:11 +03:00
Nekotekina ae809ad320 Unexpected bugfixes
Mostly unaligned memory access.
Also includes workarounds for ubsan execution.
2022-10-31 14:20:02 +03:00
kd-11 73784b9e12 Fix GCC build 2022-10-03 12:57:16 +03:00
kd-11 533f960854 rsx: Handle some more corner cases 2022-10-03 12:57:16 +03:00
kd-11 765208a181 rsx: Avoid clobbering CELL memory when splitting fbos 2022-10-03 12:57:16 +03:00
kd-11 4417701ea7 rsx: Track orphaned surfaces' parent addresses 2022-10-03 12:57:16 +03:00