mirror of
https://github.com/tangerineturtle1/NX2navi.git
synced 2026-11-19 19:29:57 -04:00
Keep CPU helper threads on Automatic and Start next frame early on by default in 0.10.1, as in 0.10.0, and give 0.10.1's notes the video memory fix.
The release commit for 0.10.1 had turned both off by default (replay_workers = 0, tail_early = 0). They stay as 0.10.0 has them: replay_workers = auto, half the processor's physical cores less one (3 on 8 cores), and tail_early = 1. A player's session with all four Testing settings at 0 filled the video memory as before (bug capture 7: 15.5 GiB in use, 20.7 frames a second); what filled it were the images of moved render targets, which the two commits before this one fix. runtime/cfg.c, runtime/ui/menu.cpp, runtime/vk_exec.c and nx2navi.ini.example are back as they were at 0.10.0. The notes of 0.10.1 lose the entry about the defaults and take the fix's entry from the draft for 0.10.2, which goes. Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
This commit is contained in:
@@ -12,13 +12,19 @@
|
||||
game and 1.5 GB for the launcher; a GPU with Vulkan 1.3 and `VK_EXT_external_memory_host` (current NVIDIA and AMD
|
||||
cards).
|
||||
|
||||
## Changed
|
||||
## Fixed
|
||||
|
||||
- **Every setting under the Game tab's Testing heading is off by default again:**
|
||||
- *CPU helper threads* (`replay_workers = 0`; it was Automatic in 0.10.0);
|
||||
- *Start next frame early* (`tail_early = 0`; it was on in 0.10.0);
|
||||
- *Helpers work ahead* and *Helpers take small jobs* (`replay_pipe` and `replay_small`, off as before).
|
||||
- **Video memory no longer fills up while you walk or turn the camera** (by epicmeme546). In 0.10.0 the video memory
|
||||
in use climbed by several gigabytes within seconds of moving, in any area, until the card was full: the frame rate
|
||||
fell to 2 to 4 frames a second and the game could end. The game lays its render targets out anew in every frame, one
|
||||
behind the other. One of them, a shadow map, changes its size with the view, and every target behind it then lies at
|
||||
a new address in each frame. A moved target should take over the image of the one it replaces, but that image counted
|
||||
as in use until the game recycled the old target's texture object, so every frame got about a dozen new images and
|
||||
each old one stayed for 120 frames or more: 6.7 GiB of them at 3416x1440 and 12.2 GiB at 3840x2160, on top of the
|
||||
10 GiB an area needs at those sizes. Such an image now goes to the next target of its size once nothing has drawn
|
||||
with it or sampled it for 16 frames, and sooner once the card is past 80 % of its memory. Walking through Kokiri
|
||||
Forest at 3416x1440 on a 20 GiB card: 15.9 GiB in use before, 9.7 to 10.7 GiB now. The same walk at 3840x2160 with
|
||||
the memory budget of a 16 GiB card: 10.9 to 11.4 GiB, where a player's 16 GiB card had 22.2 GiB asked of it.
|
||||
|
||||
They are experimental and were measured on few cards. A value you chose yourself, in the settings menu or in
|
||||
nx2navi.ini, stays as you set it: to keep 0.10.0's behaviour, set *CPU helper threads* to Automatic and *Start next
|
||||
frame early* to On. Everything else is as in 0.10.0; see its notes.
|
||||
Everything else is as in 0.10.0, the settings under the Game tab's Testing heading included: *CPU helper threads* is
|
||||
Automatic (half the processor's cores less one: 3 on 8 cores) and *Start next frame early* is on. See 0.10.0's notes.
|
||||
|
||||
Reference in New Issue
Block a user