summaryrefslogtreecommitdiffstats
path: root/builtin-programs/gpu (follow)
Commit message (Collapse)AuthorAgeFilesLines
* gpu/draw: Refresh rate hackOmar Rizwan2026-04-021-1/+2
|
* gpu: Fix image limit so Pi GPU worksOmar Rizwan2026-04-021-0/+3
|
* gpu: Replace texture descriptor set lock with serial handlingOmar Rizwan2026-04-013-80/+147
| | | | should fix weird texture flickers.
* Tracy fixes. Makefile: Reduce sampling Hz.Omar Rizwan2026-03-241-6/+0
| | | | | | | (this is so we don't overwhelm with sampling data and overhead.) gpu/draw: Drop zones that would sometimes fail and that we weren't really using.
* gpu/textures: Go back to ~old way; new way wasn't correctOmar Rizwan2026-03-241-19/+12
|
* gpu/draw: Don't kill Folk on GPU failure.Omar Rizwan2026-03-241-2/+2
|
* Speed up texture loading into GPU by removing fence waitOmar Rizwan2026-03-232-15/+25
|
* c: Use -O2 and -march=native and fix how that breaks drawOmar Rizwan2026-03-221-5/+8
| | | | Should make lower-end systems a lot faster? But need to check.
* gpu/draw: Fix crash on glfw failureOmar Rizwan2026-02-231-0/+1
|
* gpu/canvases: Fix display canvas blink (titles, points-at)Omar Rizwan2025-12-241-31/+37
| | | | | | Handles one remaining case where we would miss the Collect statement for a canvas and treat it as "blink out" instead of reuse previous Collect.
* gpu/vma: Remove extra putsOmar Rizwan2025-12-191-1/+0
|
* Fix VMA on Linux. This is WAY faster with camera slicesOmar Rizwan2025-12-122-4/+24
|
* WIP: Finish moving to VMA? Add -noload to C.Omar Rizwan2025-12-123-34/+28
|
* WIP: Continue switching to VMAOmar Rizwan2025-12-123-73/+84
|
* WIP: Start switching to VMAOmar Rizwan2025-12-122-33/+128
|
* gpu/enumerate: Unbreak macOSOmar Rizwan2025-12-121-3/+6
|
* gpu: Lock out gpu until enumeration doneOmar Rizwan2025-12-122-1/+4
| | | | Some weird race on Pi means it never sees displays.
* gpu/draw: Retry if display not foundOmar Rizwan2025-12-081-11/+24
|
* virtual-programs -> builtin-programsOmar Rizwan2025-12-046-0/+2847