@@ -16,7 +16,7 @@ Copyright (c) 2015-2025 [The Brenwill Workshop Ltd.](http://www.brenwill.com)
16
16
MoltenVK 1.4.0
17
17
---------------
18
18
19
- Released 2025-07-29
19
+ Released 2025-08-20
20
20
21
21
- Add support for _ Vulkan 1.4_ .
22
22
- Add support for extensions:
@@ -43,6 +43,8 @@ Released 2025-07-29
43
43
- Fix high latency on present wait.
44
44
- Fix shader cache misses within pipeline cache.
45
45
- Fix misalignment between pipeline pixel formats and dynamic rendering attachments.
46
+ - Fix attempting to add memoryless images to residency sets.
47
+ - Fix color-depth image copies and image copies using buffer with multiple layers.
46
48
- Enable ` shaderRoundingModeRTEFloat16/32 ` .
47
49
- Move ` vkGetPhysicalDeviceMetalFeaturesMVK() ` from ` mvk_private_api.h ` to ` mvk_deprecated_api.h ` .
48
50
- Re-expose proc addr for deprecated ` VK_MVK_moltenvk ` extension functions.
@@ -56,11 +58,18 @@ Released 2025-07-29
56
58
- MSL: Implement clustered subgroup operations for sizes other than 4.
57
59
- MSL: Fix generation of the ` WorkgroupSize ` built-in.
58
60
- MSL: Implement accurate ` OpSMod ` .
61
+ - MSL: Always emit gl_Layer for multiview.
59
62
- MSL: Don't emit const for BDA pointers.
60
63
- MSL: Fix loading BDA from ` std140 ` UBO.
64
+ - MSL: Fix issues with ` fp16 ` trancendentals.
65
+ - MSL: Fix edge case where a reference is taken of packed vector element.
66
+ - MSL: Fix crash when using force-native-arrays in some cases.
67
+ - MSL: Don't use fast::normalize for half at all.
61
68
- MSL: Implement ` atomic_thread_fence ` for MSL 3.2+.
62
69
- MSL: Always emit access qualifier when emitting coherency.
63
70
- MSL: Support SPIR-V Fast Math execution modes and decorations.
71
+ - MSL: Deal with spec constant arrays which depend on composite extract.
72
+ - MSL: Attempt to deal with float controls and trancendentals.
64
73
- MSL: Rethink ` add_pragma_line() ` and ` add_typedef_line() ` .
65
74
- GLSL/MSL: Add support for ` SPV_KHR_quad_control ` .
66
75
0 commit comments