Rewrite engine renderer to lag less.
asiekierka opened this issue · 10 comments
Right now the entire Engine is a TESR.
It needs to be split into an ISBRH and a TESR, with severe display list (or even VBO) based caching for the moving part due to the fact engines are used in large arrays.
If anything is being a render issue its the quarry. A single one of them is chewing up about 10% of my render time. A single engine is negligable.
I've seen 20-30 engines take up quite a lot of rendering time on BC 6.4 on
chilm's machine, IIRC.
Also, I think adding display list support to RenderBoxProvider might fix
this one to an extent. That, or maybe it's the mechanical arm?
2016-03-12 15:04 GMT+01:00 AlexIIL [email protected]:
If anything is being a render issue its the quarry. A single one of them
is chewing up about 5% of my render time. A single engine is negligable.—
Reply to this email directly or view it on GitHub
#3042 (comment)
.
Right now the Engine is split into an ISBRH and a TESR! This is better, but not quite there yet.
Actually we need to remove the block model part so it's only a TESR. With fry's new fastTesr class it should speed things up but it has a tonne of bugs with using both.
On 12 Mar 2016, at 11:39, Adrian Siekierka [email protected] wrote:
Right now the Engine is split into an ISBRH and a TESR! This is better, but not quite there yet.
—
Reply to this email directly or view it on GitHub.
That's a bad idea IMHO. FastTESRs still have a massive load compared to ISBRHs, it's just smaller than regular TESRs - and if FastTESRs need to send 2-3x more polys, well...
I'd rather wait until FastTESRs are fixed up. :)
This isn't opinion, it's comparing to when I split them up. We need to use a TESR at all times anyway (changing the block model a lot is a very bad idea).
On 12 Mar 2016, at 12:16, Adrian Siekierka [email protected] wrote:
That's a bad idea IMHO. FastTESRs still have a massive load compared to ISBRHs, it's just smaller than regular TESRs.
I'd rather wait until FastTESRs are fixed up. :)
—
Reply to this email directly or view it on GitHub.
Yes, but then we send more data a frame to the graphics card. This might cause issues.
I'm not sure, someone'd have to benchmark it.
