karolherbst has quit [Remote host closed the connection]
karolherbst has joined #etnaviv
embden has joined #etnaviv
_whitelogger has joined #etnaviv
_whitelogger has joined #etnaviv
pcercuei has joined #etnaviv
<mntmn>
am i missing some out-of-tree patch to mesa that's important for GC7000L? i unfortunately lost my mesa patchstack due to a failing SSD and rebuilding it now.
_whitelogger has joined #etnaviv
<mntmn>
problem goes away when XWAYLAND_NO_GLAMOR env var is set
<mntmn>
ok i found the problem. it's the glClear() in glamor_pixmap_clear_fbo() causing etnaviv to go into a corrupted state, still
<cphealy>
With these Vivante GPUs and their associated HW counters, are there any that could be accessed independent of using Mesa? On the NXP i.MX8M, I have access to read the DRAM interface counters using the "perf" tool. I'm wondering if support could be added to read the GPU host interface counters through the "perf" tool while still retaining the existing counter access through Mesa.
<cphealy>
My long term vision is to be able to have a single tool that on these i.MX based platforms we can show system wide utilization in real-time. CPU load, DDR bus load, video encoder/decoder load, and GPU load. For GPU, it would be good to have an overall busy % but also read/write and busy time for the GPU sub-components.
<cphealy>
austriancoder: do you have example output from this?
<austriancoder>
You can load some in the web UI.. I am currently not at home
<daniels>
cphealy: yep that's what we've been working on! :)
<cphealy>
daniels: Do you mean your working on perfetto.dev implementation that fits with what austriancoder is doing?
Chewi has joined #etnaviv
<daniels>
cphealy: right, including Weston timeline which we could previously only get through wesgr
<daniels>
cphealy: I can send you a draft blog post we have about it tomorrow or Monday, or I'm pretty sure it's in your email somewhere if you search for 'perfetto'