Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 23, 2020

• Replaced the static reference to vkGetBufferDeviceAddress with a dynamically loaded function to fix NVIDIA#100

  apanteleev

  committed
  0a16c73

  Copy full SHA for 0a16c73

Commits on Dec 22, 2020

• Increased the size of the AS build scratch buffer as per request in NVIDIA#99

  apanteleev

  committed
  514b136

  Copy full SHA for 514b136

Commits on Dec 19, 2020

• Added support for pooled memory allocations with greater than 4 KB alignment.

  apanteleev

  committed
  56ebf92

  Copy full SHA for 56ebf92

• Revert commit 5be2b3e "Store the world tangents in oct-encoded format."

  apanteleev

  committed
  12e7878

  Copy full SHA for 12e7878

• Fixed the previous version of encode_normal on the CPU.

  apanteleev

  committed
  3071bd9

  Copy full SHA for 3071bd9

• Revert commit fe82ddb that replaced the encode/decode_normal functions, because that caused a crash on AMD.

  apanteleev

  committed
  ae06b4d

  Copy full SHA for ae06b4d

Commits on Dec 18, 2020

• Fixed the bloom output jittering when DRS is used.

  apanteleev

  committed
  4ead437

  Copy full SHA for 4ead437

• Fixed the non-TAAU upscaling when DRS is enabled and its maximum scale is set to lower than 100%: NVIDIA#96

  apanteleev

  committed
  7330ad7

  Copy full SHA for 7330ad7

• Store the world tangents in oct-encoded format.

  apanteleev

  committed
  5be2b3e

  Copy full SHA for 5be2b3e

• Replaced the encode_normal and decode_normal with more straightforward and slightly faster versions, and fixed the CPU implementation of encode_normal which was broken because abs instead of fabsf.

  apanteleev

  committed
  fe82ddb

  Copy full SHA for fe82ddb

• Store the normals and tangents of instanced models in oct-encoded format, which saves about 80 MB of GPU memory with the current limits.

  apanteleev

  committed
  1f08b3f

  Copy full SHA for 1f08b3f

• Moved the model tangent computation into the instange_geometry shader, thereby avoiding the need to store them in model VBOs.

  apanteleev

  committed
  78ca94e

  Copy full SHA for 78ca94e

Commits on Dec 17, 2020

• Improved the texture memory allocator to allow better packing of textures into allocation blocks, therefore using less memory for the same textures.

  Show description for 36ea9d9
  apanteleev

  committed
  36ea9d9

  Copy full SHA for 36ea9d9

• Implemented denser packing of the visibility buffer data to reduce memory consumption.

  apanteleev

  committed
  170ad19

  Copy full SHA for 170ad19

• Made the game exit on GPU memory allocation failure to prevent render corruption, and improved the device allocator host memory handling.

  apanteleev

  committed
  ebcb065

  Copy full SHA for ebcb065

Commits on Dec 16, 2020

• Fixed the SBT size for hit and miss shaders.

  apanteleev

  committed
  9f1844c

  Copy full SHA for 9f1844c

• Added parentheses around the DESTROY_LATENCY macro, as noted in NVIDIA#84

  apanteleev

  committed
  7940f07

  Copy full SHA for 7940f07

Commits on Dec 10, 2020

• Fixed the Lanczos final blit filter that was broken since commit 5406a1b.

  apanteleev

  committed
  243ccbb

  Copy full SHA for 243ccbb

• Use cached memory for image readbacks - this makes taking screenshots much faster.

  apanteleev

  committed
  bd91eaf

  Copy full SHA for bd91eaf

Commits on Dec 9, 2020

• Added opaque flags to the geometry that should be considered opaque for the purposes of ray tracing. Improves performance a little.

  apanteleev

  committed
  8d4aa98

  Copy full SHA for 8d4aa98

• Added a second driver version check for KHR_ray_tracing_pipeline.

  apanteleev

  committed
  a4d5017

  Copy full SHA for a4d5017

• Improved motion vectors for multiple refraction, in particular when thick glass is enabled.

  apanteleev

  committed
  d337bff

  Copy full SHA for d337bff

Commits on Dec 4, 2020

• Fixed a few profiler issues.

  apanteleev

  committed
  35ca528

  Copy full SHA for 35ca528

• Fixed the tone mapping curve shader - it was missing some barriers between shared memory reads and writes. Also, kept the subgroup use switch for potential later experiments.

  apanteleev

  committed
  9b13106

  Copy full SHA for 9b13106

• Added a warning about the validation layer being enabled.

  apanteleev

  committed
  6ca0b2c

  Copy full SHA for 6ca0b2c

• Replaced the global RT pipeline with multiple pipelines, with a single RGEN in each one.

  apanteleev

  committed
  6e198fb

  Copy full SHA for 6e198fb

• Increased the sizes of descriptor pools to actually fit the descriptor sets.

  apanteleev

  committed
  4177261

  Copy full SHA for 4177261

• Reduced the set of requested device features.

  apanteleev

  committed
  5a64117

  Copy full SHA for 5a64117

Commits on Dec 3, 2020

• Added KHR_deferred_host_operations extension, which is required for KHR_acceleration_structure.

  apanteleev

  committed
  37892f9

  Copy full SHA for 37892f9

• Fixed a buffer overflow in extension init.

  apanteleev

  committed
  6679d9e

  Copy full SHA for 6679d9e

• Fixed the default value for flt_taa.

  apanteleev

  committed
  043bc13

  Copy full SHA for 043bc13

• Validation layer can now be enabled through the vk_validation cvar.

  apanteleev

  committed
  525bc3b

  Copy full SHA for 525bc3b

• Added support for the final official KHR ray tracing extensions (v11). Squashed from a feature branch.

  Show description for 5406a1b
  apanteleev

  committed
  5406a1b

  Copy full SHA for 5406a1b

Commits on Nov 20, 2020

• Implemented dynamic switching between TAA and TAAU depending on the resolution scale: TAAU is used when scale is <= 100%, TAA otherwise. Both are processed by the same shader now, so the old TAA sh…

  Show description for f98111e
  apanteleev

  committed
  f98111e

  Copy full SHA for f98111e

• Optimized the TAAU shader some more by also preloading the motion vectors into shared memory.

  apanteleev

  committed
  e12a072

  Copy full SHA for e12a072

Pagination

Previous

Next