If it's of any help, I have managed to narrow down the first output appearance of these interlacing lines between events glDispatchCompute(80, 45, 1) and second call of glDrawElementsBaseVertex(48).

Refreshing the upload due to the link expiring: https://www38.zippyshare.com/v/d0oWhmbR/file.html

Just a remark, the capture was taken with RenderDoc 1.1 and the bug still occurs on Mesa 18.3.1.

Situation still occurs on Mesa 19.1.3, refreshed RenderDoc capture at https://www47.zippyshare.com/v/fL6gXv7u/file.html

I might be conflating things, but there's also a shader scheduling error:

##post_scheduler: unscheduled pending instructions :MOV     R27.x.5@R1.w,    R17.x.1@R1.x
MOV     R27.x.3@R1.y,    R17.w.1@R0.z
MOV     R17.x.1@R1.x,    A9.y[R16.x.1@R0.w]_73F@R1.y
                                            rels: A9.y[R16.x.1@R0.w]_73F@R1.y :  <= R9.y.1F, R10.y.1F, R11.y.1F, R12.y.1F, R13.y.1F, R14.y.1F
MOV     R17.w.1@R0.z,    A9.x[R16.x.1@R0.w]_62F@R1.x
                                            rels: A9.x[R16.x.1@R0.w]_62F@R1.x :  <= R9.x.1F, R10.x.1F, R11.x.1F, R12.x.1F, R13.x.1F, R14.x.1F
MOV     R14.y.1F,    4.2039e-45|00000003
MOV     R13.y.1F,    4.2039e-45|00000003
MOV     R12.y.1F,    7.00649e-45|00000005
MOV     R11.y.1F,    5.60519e-45|00000004
MOV     R10.y.1F,    4.2039e-45|00000003
MOV     R9.y.1F,    4.2039e-45|00000003
FLT_TO_UINT     R16.x.1@R0.w,    C4102.x
MOV     R14.x.1F,    1.4013e-45|00000001
MOV     R13.x.1F,    0|00000000
MOV     R12.x.1F,    0|00000000
MOV     R11.x.1F,    0|00000000
MOV     R10.x.1F,    5.60519e-45|00000004
MOV     R9.x.1F,    7.00649e-45|00000005

I don't read mnemonics, but I'm guessing that could be "Shader 9120" from the capture, which also causes the interlacing lines. I'm attaching it shortly.

Created attachment 144946 [details]
Shader suspected of causing defect

Can confirm that the program with the attached compute shader is producing these lines, compared to nvidia.

-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/648.