!MFogdGJfnZLrDmgkBN:matrix.org

WebGPU

119 Members
The future of graphics and compute on the Web - unofficial channel. Related rooms: https://matrix.to/#/+webgraphics:matrix.org35 Servers

Load older messages


SenderMessageTime
25 Sep 2020
@kvark:matrix.orgkvarkIf you have the budget left, will it be spent on those limits? And on which exactly?23:32:34
@kainino:matrix.orgkaininoif you have the budget left for all of them, it will be spent on all of them, and if not, then you're out of budget and device creation fails23:34:00
@kainino:matrix.orgkaininoor the UA can try to be smarter and figure out exactly how much budget you spent learning that device creation failed23:34:53
@kvark:matrix.orgkvarkThat sounds tricky, non-obvious from the user perspective.23:39:41
@kainino:matrix.orgkaininoI think privacy budgeting is always tricky23:40:12
@kainino:matrix.orgkaininothe important thing is that it may be non-obvious from the user perspective, but from the application code's perspective nothing particularly weird has happened23:40:34
@kainino:matrix.orgkaininoit's just as if the adapter didn't support what you requested23:40:52
@kainino:matrix.orgkaininoso even if the dev doesn't think about it, the app will behave in a reasonable way23:41:16
@tojiro:matrix.orgtojiroI'm a little concerned about the behavior of frameworks with this pattern. For example: Three.js right now goes out and requests a bunch of extensions right away because they're used for various features in the library, even if you aren't using one of those features yet. (Because who knows! You might?) 23:42:19
@tojiro:matrix.orgtojiroSo they're going to be incentivised to spin through all possible extensions up front and enable as many of them as the library can possibly take advantage of.23:43:10
@kainino:matrix.orgkaininoin practice I think there are a lot of situations where you can enable quite a few extensions without spending much budget23:43:47
@tojiro:matrix.orgtojiroThey could shift the responsibility to the devs, and say "You have to request what you want to use" but that's not always going to be an obvious GPU feature -> Library feature mapping23:43:52
@kainino:matrix.orgkaininoa lot of extensions will be available on almost every desktop device23:43:59
@tojiro:matrix.orgtojiroSo the budget is a dynamic thing that depends on rarity?23:44:25
@kainino:matrix.orgkaininoin an ideal world, I assume so23:44:34
@tojiro:matrix.orgtojiroEG: The cost of checking for Anisotropic filtering (to pick a random GL extension) is ~0 because it's literally everywhere.23:45:07
@kainino:matrix.orgkaininothe budget is supposed to run out when k (as in k-anonymity) goes down to a certain value23:45:10
@tojiro:matrix.orgtojiroBut the cost of checking for raytracing is presumably higher?23:45:21
@kainino:matrix.orgkaininoin practice I don't think any of these questions have been figured out yet23:45:35
@kainino:matrix.orgkaininoat least, conclusively23:45:53
@kvark:matrix.orgkvark
In reply to @kainino:matrix.org
so even if the dev doesn't think about it, the app will behave in a reasonable way
No. Suppose the app uses 5 bind groups but doesn't check for it. Suddenly, it may work here but not there, depending on the privacy budget size, and what have been queried.
23:53:48
@kainino:matrix.orgkaininothat's reasonable, because if they did that then it wouldn't work on system that can support 4 but it would on a system that can support 823:54:29
@kainino:matrix.orgkaininothey still have to requestDevice with 5 bind groups23:54:47
@kvark:matrix.orgkvarkTrue23:55:18
@kvark:matrix.orgkvarkOk, I finally understand it:)23:56:07
@kainino:matrix.orgkainino😀23:56:24
27 Sep 2020
@michail:ru-matrix.orgmichail 17:48:56
28 Sep 2020
@cwfitzgerald:matrix.orgcwfitzgerald
In reply to @kangz:matrix.org
That seems like a lot no? Creating descriptors should be cheap-ish.
so I eliminated a bunch of issues and took a look in a proper sampling profiler, and now 256 textures shows 0.12ms, which is a lot better
04:57:40
@benkant:matrix.orgbenkant joined the room.10:07:46
29 Sep 2020
@kunigunde:kde.orgkunigunde 19:49:50

There are no newer messages yet.


Back to Room List