GPU freeze for over 3 seconds

Greetings,
Enjoying a lot using vello, very nice package.

I have a strange situation that I am struggling to get to the bottom of.
I have a wasm client that is rendering some graphics using vello, and there is a significant delay (over 3 seconds) between the initial page load and when the graphics appears. I am using gRPC to retrieve the data, and initially thought there was some oddness with futures not being run (and so I move most of that into a separate web worker instance). The delay also seems unrelated to the amount of graphics being rendered e.g I have something that is drawing 60-70 rectangles, and another that is drawing 1000's of elements, and the time spent actually rendering by the CPU is quite small (10's of milliseconds). So I have been doing some Chrome performance analysis, and I see that nearly all the time (consistently about 3.2 seconds) is taken up by the 'GPU' element in the performance debug tool.
I have enclosed a screen shot of the performance display, and a chrome://gpu dump that does some errors, but I am not sure these are relevant.

I am not sure where to start even going about debugging this. Potentially I can put together some minimal reproducing set of code, but initially wanted to get an opinion on this data I have captured to see if it is a known issue, or maybe some advice about how to debug this.

about-gpu-2025-04-24T00-44-00-546Z.txt

Start with the attached Chrome performance capture and chrome://gpu dump, then reproduce the wasm client's roughly 3.2-second delay with the reported rendering cases. A minimal reproducing example would help isolate whether the GPU timeline or reported Chrome GPU errors are relevant; done means identifying the cause or narrowing it to a reproducible GPU issue.