1928829 - intel m3-8100y with Intel HD Graphics 615 not fully supported

Status: UNCONFIRMED

(Core :: Graphics: WebRender, defect)

Description

[Elle Solomina (she/her)]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:132.0) Gecko/20100101 Firefox/132.0

Steps to reproduce:

GPU #1
Active Yes
Description Intel(R) UHD Graphics 615
Vendor ID 0x8086
Device ID 0x591c
Driver Version 31.0.101.2130
Driver Date 8-13-2024
Drivers igdumdim64 igd10iumd64 igd10iumd64 igd12umd64 igdumdim32 igd10iumd32 igd10iumd32 igd12umd32
Subsys ID 591c8086

Actual results:

WEBRENDER_SCISSORED_CACHE_CLEARS
default available
env blocklisted Blocklisted by gfxInfo Blocklisted due to known issues: bug 1603515

Comment 1

[Teodor Tanasoaia [:teoxoy]]

WEBRENDER_SCISSORED_CACHE_CLEARS is a specific feature that doesn't work properly on intel GPUs as found in Bug 1603515.

As far as I see things are working as expected. Am I missing something?

Comment 2

[Elle Solomina (she/her)]

Yes, there is a different device and a new driver.

Comment 3

[Teodor Tanasoaia [:teoxoy]]

Bug 1638672 switches Windows/Intel to use quad rendering for scissored clears. This is based on the knowledge that Intel can't do fast-clears if scissor is set. We need to consider doing this on other platforms for Intel as well.

I'm not familiar with the details of this issue but there is this comment in Bug 1643838; do newer drivers support fast-clears if scissor is set?

Comment 4

[Teodor Tanasoaia [:teoxoy]]

Glenn, you might know more about this - just putting it on your radar. Feel free to dismiss the NI.

Comment 5

[Glenn Watson [:gw]]

We'd need to get that hardware + driver combo locally and test, or provide a test build with that enabled for the reporter, and see if any of the driver bugs have been resolved recently. If so, we could unblock that feature. Having said that, it's questionable whether it's worth worrying about - that feature is mostly only useful on older low-end iGPUs that have very low memory bandwidth and we saw some performance benefits from using it.

Comment 6

[Elle Solomina (she/her)]

(In reply to Glenn Watson [:gw] from comment #5)

We'd need to get that hardware + driver combo locally and test, or provide a test build with that enabled for the reporter, and see if any of the driver bugs have been resolved recently. If so, we could unblock that feature. Having said that, it's questionable whether it's worth worrying about - that feature is mostly only useful on older low-end iGPUs that have very low memory bandwidth and we saw some performance benefits from using it.

I can give you access, but I don't know how to do this.