User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101 Firefox/78.0

Steps to reproduce:

Using FF with same profile as ever - now Webrender in this version is enabled by default - in former versions i had enabled by about:config

Actual results:

Now, enabled by default, GPU process is growing up to more than 3 GB of RAM (NOT GPU Memory!) and when Browser gets idle this GPU process is eating up 25% of my Qud Core CPU (AMD Ryzen 4700U).
First when i start using FF, just only selecting one TAB and doing nothing else inside, CPU is going down again, RAM for the GPU stays as it is.

Expected results:

Clearly no RAM consumtion if more than 3G, in former Versions RAM consumption of my FF with that profile was never higher than 1.4G

Bugbug thinks this bug should belong to this component, but please revert this change in case of error.

Just discovered a potential thing:
This site, a larger german on-line shop, was the selected(foreground) tab.
In this case RAM of GPU process increases approx. 0.5 MB per second.

https://www.tchibo.de/meine-glanzleistung-t400116124.html

If i tick any other tab the GPU RAM usage stays more or less constant. I do nothing inside this tab/website, just keep it in foreground.
Even during this time tab is selected CPU of this process is at around 5% on my system.
When selecting another tab CPU goes down to ~0%

Closing this tab frees quite some memory on the GPU process and from then FF behaves normaly.

The memory reports shows it is almost entirely heap unclassified. The STR suggests this may be another case of bug 1640564 in disguise.

Hi Frank, can you please check this issue again and see if the issue still occurs now that the Bug 1640564 is fixed ?

Unfortunately exactly this webpage is no longer available, so in the end i cannot test this.
And it was exactly on this specific page of that web shop, other pages did not lead to the described behavior.
Means there was some resource on that pages which leads to the memory problem

Since this issue may have been fixed by Bug 1640564 and it does not occur on any other page I will close this issue as Works for me but in case a different page might cause the same issue or a new Test cases is provided we can definitely reopen it.

Thank you for all the info on this issue.