glxgears fullscreen test shows unstable and low initial FPS

### Enhancement Proposal

When running the glxgears benchmark in fullscreen mode recent generation CPU systems, the initial frames per second (FPS) is unstable and significantly lower than expected.

This behavior is directly linked to the CPU's power-saving state. If the CPU is simultaneously put under load using another tool (like stress-ng), the glxgears FPS is high and stable from the very beginning.

Reproduce steps:
1. Ensure there are no other significant CPU loads running.
2. Run the glxgears command in fullscreen mode $glxgears -fullscreen

Results:
Fail Rate: 1/3
Kernel Version: 6.14.0-1011-oem

**CID: 202508-37850**
**CPU:Intel(R) Core(TM) Ultra 5 225T (10x)
GPU: 00:02.0 VGA compatible controller [0300]: Intel Corporation Device [8086:7d67] (rev 06)**

prime offload env: {'DRI_PRIME': 'pci-0000_00_02_0'}
No prime-select, it should be ok to run prime offload
running command:[['glxgears', '-fullscreen']]
Checking success:
Offload process:[['glxgears', '-fullscreen']]
Card ID:[0000:00:02.0]
Device Name:[Arrow Lake-U [Intel Graphics] [8086:7D67]]
Running synchronized to the vertical refresh. The framerate should be
approximately the same as the monitor refresh rate.
301 frames in 5.0 seconds = 60.176 FPS
229 frames in 5.0 seconds = 45.651 FPS
250 frames in 5.0 seconds = 49.839 FPS
250 frames in 5.0 seconds = 49.839 FPS
291 frames in 5.0 seconds = 58.014 FPS
Test finished

**CID: 202508-37860
CPU: Intel(R) Core(TM) Ultra 7 265 (20x)
GPU: 00:02.0 VGA compatible controller [0300]: Intel Corporation Device [8086:7d67] (rev 06)
03:00.0 Display controller [0380]: Advanced Micro Devices, Inc. [AMD/ATI] Device [1002:743f] (rev ce)
amd-vbios: b'BRT148347.001'["743F"]**

prime offload env: {'DRI_PRIME': 'pci-0000_03_00_0'}
No prime-select, it should be ok to run prime offload
running command:[['glxgears', '-fullscreen']]
Checking success:
Offload process:[['glxgears', '-fullscreen']]
Card ID:[0000:03:00.0]
Device Name:[Navi 24 [Radeon RX 6400/6500 XT/6500M] [1002:743F]]
Running synchronized to the vertical refresh. The framerate should be
approximately the same as the monitor refresh rate.
301 frames in 5.0 seconds = 60.180 FPS
265 frames in 5.0 seconds = 52.818 FPS
186 frames in 5.0 seconds = 37.077 FPS
249 frames in 5.0 seconds = 49.637 FPS
243 frames in 5.0 seconds = 48.436 FPS
Test finished

The issue names no repository file or test entry point. Start by reproducing `glxgears -fullscreen` under the reported Intel and AMD configurations, then trace the Checkbox test that launches this command; done means determining and addressing the initial FPS instability while preserving the expected synchronized refresh behavior.

Written by the indexing model from the issue text.