Reporting Metal rendering bug in Fusion?

[pcx2020]

What's the best way to report a rendering bug for 3D accelerated graphics in Fusion?  I put the VM on a Windows host and confirmed that the rendering bug isn't present there, so I believe it has to do with Fusion's Metal translation layer.

Given its free status, I don't know whether anyone at VMware/Broadcom is interested in fixing obscure bugs in Fusion?

[Technogeezer]

Given that this forum is the only support avenue that Broadcom provides for the free desktop hypervisor products, I would recommend posting whatever you find here in this forum. Include as much detail as you possibly can (including the logs found in the .vmwarevm bundle).

Given Broadcom's level of investment in the desktop hypervisor products, it's unknown how or if they will respond. Broadcom employees have been known to be found here in the community from time to time but not to the level that we would all like. You might catch their interest however. 

[pcx2020]

There's nothing informative in the VMware logs.

However, using Jose Fonseca's apitrace, I isolated several sample traces of guest OpenGL activity that demonstrate the issue.

I don't see any way to attach those traces here (and they're 150MB in total), but I can at least provide video captures of the problem.

First, here's one of the samples, with hardware acceleration enabled, showing how it's getting rendered in Fusion:

That exact same trace renders correctly with hardware acceleration disabled in Fusion (and, for good measure, VMware Workstation even with hardware acceleration):

I confirmed that the rendering bug happens on both x64 and Apple Silicon, so it's something in the Metal translation layer.

I can provide the traces to VMware if they provide me some way to do so.

[pcx2020]

Update: I've been contacted and have provided my traces.

[Neha Bhende]

Thank you so much for providing apitraces. It has been really helpful. We have fixed this issue internally. It will fixed in one of future releases of VMTools

[pcx2020]

Thank you, Neha.

For anyone who's affected by this, the fix doesn't appear to be in 26H1 (25388279), as it was announced only 6 days later.