Skip to content

arm64: dts: exynos: google: zumapro-tegu: Enable the GPU - #85

Closed
W4HO-ham wants to merge 1 commit into
derekross:omamobilefrom
W4HO-ham:tegu-dt-gpu
Closed

W4HO-ham wants to merge 1 commit into
derekross:omamobilefrom
W4HO-ham:tegu-dt-gpu

Conversation

@W4HO-ham

@W4HO-ham W4HO-ham commented Oct 27, 2026 •

Copy link
Copy Markdown
Contributor

Stacked on #81 and #80 (the board). It also needs #84 (clk-acpm's zumapro variant) at run time for the gpu clock. Only the last commit is new.

This turns on CMU_G3D and the Mali-G715 for the Pixel 9a:

The 9a's table goes up to 940 MHz. I've left 723-940 out until a GPU thermal zone is in the DT, because nothing upstream throttles the GPU yet. The palmtop runs those levels today with an out-of-tree thermal zone behind them.

On the palmtop: these are the GPU node's clocks and the first nine OPPs of the table it runs now. With #84's clk-acpm and zumapro_gpu_max_khz set, panthor's devfreq moves through them under load. With the parameter at its default (read-only), the GPU stays at the bootloader's 150 MHz, as on gs201.

CHECK_DTBS=y W=1 for zumapro-tegu.dtb is clean, and checkpatch --strict is clean.

@derekross

Copy link
Copy Markdown
Owner

Thanks. The nine levels are the shader column of #84's pair table, which matches Google's gpu_dvfs_table_v2, and holding 723 MHz and above back until #87's thermal zone exists is the right order. Without #84 the GPU simply waits for its clock provider, and with #84's default cap the clock stays read-only, so nothing changes until the cap is set.

One change: on #84 I've asked for a google,zumapro-acpm.h id header, as gs101 and gs201 have. Once it's in, please use its name for the paired clock here (ZUMAPRO_CLK_ACPM_GPU or whatever you call it) instead of 17.

This lands after #81 and #84. The tegu device package will also need an options clk_acpm zumapro_gpu_max_khz=… line, like cheetah's, before the GPU actually scales. That's an omarchy-mobile change.

@W4HO-ham
W4HO-ham force-pushed the tegu-dt-gpu branch 2 times, most recently from 8499d30 to 8bbbc11 Compare October 27, 2026 20:19
@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Done: the GPU's core clock is <&acpm_ipc ZUMAPRO_CLK_ACPM_GPU> from #84's new google,zumapro-acpm.h, and the stack is rebased onto #84's header commit and today's omamobile. The options clk_acpm zumapro_gpu_max_khz=… line will come in the tegu device package.

@derekross

Copy link
Copy Markdown
Owner

Thanks, ZUMAPRO_CLK_ACPM_GPU with the include is just what I meant. This will land with the rest of the stack through #87, after the boot on #80.

@derekross

Copy link
Copy Markdown
Owner

Thanks. Your GPU commit is unchanged since I last checked: same patch, same tree at this commit (850e4cdb089b), still <&acpm_ipc ZUMAPRO_CLK_ACPM_GPU>. The only change underneath is the board commit's "Tested on" paragraph.

As on #80, the branch needs the Makefile rebase before CI can run. With that resolved, the stack builds at every commit without warnings and the 9a's DTB validates with no messages.

The boot of #96's head will be this node's first run in the form that lands, so in that smoke report I'd like the GPU row to show a render node (/dev/dri/renderD128) and panthor bound. The GPU staying at the bootloader's 150 MHz is expected until the tegu device package sets options clk_acpm zumapro_gpu_max_khz=…. That line isn't in the tegu package under review on omarchy-mobile yet, so it's still a to-do there, not here.

This lands with #96.

Turn on CMU_G3D and the Mali-G715 for the Pixel 9a, with its core clock
on ACPM's paired "gpu" clock (index 17), which sets the shader (G3D) and
L2 (G3DL2) domains together, and coregroup and stacks on CMU_G3D's gates.

The operating points are the shader rates of the 9a's GPU DVFS table up
to 649 MHz. The levels above it, to 940 MHz, wait for a GPU thermal
zone, since nothing throttles the GPU yet.

Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Rebased onto today's omamobile: the touchscreen commit is already in your branch, so it's dropped and only the GPU commit remains. The board device tree compiles and checkpatch --strict is clean.

@derekross

Copy link
Copy Markdown
Owner

Thanks for the rebase. Same patches and trees as before, and the gate passes on today's omamobile. This still lands through #96 once its display and GPU boot is in.

@derekross

Copy link
Copy Markdown
Owner

Its patches landed unchanged through #96, so I'm closing this one. Thanks!

@derekross derekross closed this Nov 1, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants