Hi Christian, Ah thanks. In that case nothing trivial pops out. Our init/start functions are not verbatim copies of one another. Tom ________________________________ From: Koenig, Christian Sent: Thursday, June 16, 2016 08:56 To: StDenis, Tom; amd-gfx at lists.freedesktop.org Subject: Re: [amd-gfx] golden registers for CIK devices Keep in mind that radeon uses byte addresses for register offset while amdgpu uses dword addresses. So the different values are correct and intentional. Christian. Am 16.06.2016 um 14:21 schrieb StDenis, Tom: Another thing I noticed is that in the ring test on amdgpu side we write a different value into the buffer, e.g. - amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START)); + amdgpu_ring_write(ring, ((scratch - PACKET3_SET_UCONFIG_REG_START) >> 2)); So far I've tried keep/change golden registers to match radeon, change this ring value, and for kicks disabled CG. The ring test still fails. I was really hoping this would be a simple copy/paste error from the radeon side :-) Tom ________________________________ From: amd-gfx <amd-gfx-bounces@xxxxxxxxxxxxxxxxxxxxx><mailto:amd-gfx-bounces at lists.freedesktop.org> on behalf of StDenis, Tom <Tom.StDenis at amd.com><mailto:Tom.StDenis at amd.com> Sent: Thursday, June 16, 2016 7:17:45 AM To: amd-gfx at lists.freedesktop.org<mailto:amd-gfx at lists.freedesktop.org> Subject: [amd-gfx] golden registers for CIK devices Hi team, Is there a reason the golden registers have different register addresses on amdgpu's cik.c versus radeon's cik.c? Thanks, Tom _______________________________________________ amd-gfx mailing list amd-gfx at lists.freedesktop.org<mailto:amd-gfx at lists.freedesktop.org> https://lists.freedesktop.org/mailman/listinfo/amd-gfx -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://lists.freedesktop.org/archives/amd-gfx/attachments/20160616/561113ce/attachment.html>