; Cc: Kuehling, Felix
> Subject: Re: [PATCH] drm/amdgpu: improve debug VRAM access
> performance using sdma
>
> Ah, yes! GART doesn't need to be read to make a GTT allocation.
>
> When GART becomes ready it will be filled with all the buffers which were
> allocated befo
; amd-gfx@lists.freedesktop.org
Cc: Kuehling, Felix ; Koenig, Christian
Subject: Re: [PATCH] drm/amdgpu: improve debug VRAM access performance
using sdma
Caution: This message originated from an External Source. Use proper
caution when opening attachments, clicking links, or responding.
I don't
3 1:02 PM
> To: Quan, Evan ; Kim, Jonathan
> ; amd-gfx@lists.freedesktop.org
> Cc: Kuehling, Felix ; Koenig, Christian
>
> Subject: Re: [PATCH] drm/amdgpu: improve debug VRAM access performance
> using sdma
>
> Caution: This message originated from an External Source. Use pro
I don't think so. Have we recently re-ordered something here?
Christian.
Am 20.03.23 um 08:05 schrieb Quan, Evan:
[AMD Official Use Only - General]
I happened to find the sdma_access_bo allocation from GTT seems performing
before gart is ready.
That makes the "amdgpu_gart_map" is skipped sinc
[AMD Official Use Only - General]
I happened to find the sdma_access_bo allocation from GTT seems performing
before gart is ready.
That makes the "amdgpu_gart_map" is skipped since adev->gart.ptr is still NULL.
Is that done intentionally ?
Evan
> -Original Message-
> From: amd-gfx On Be
our suggested cleanup for review out soon.
Jon
-Original Message-
From: Koenig, Christian
Sent: January 12, 2022 2:33 AM
To: Kim, Jonathan ; amd-
g...@lists.freedesktop.org
Cc: Kuehling, Felix
Subject: Re: [PATCH] drm/amdgpu: improve debug VRAM access
performance using sdma
Am 04.01.22
freedesktop.org
> Cc: Kuehling, Felix
> Subject: Re: [PATCH] drm/amdgpu: improve debug VRAM access
> performance using sdma
>
> Am 04.01.22 um 20:12 schrieb Jonathan Kim:
> > For better performance during VRAM access for debugged processes, do
> > read/write copies over SDM
Am 04.01.22 um 20:12 schrieb Jonathan Kim:
For better performance during VRAM access for debugged processes, do
read/write copies over SDMA.
In order to fulfill post mortem debugging on a broken device, fallback to
stable MMIO access when gpu recovery is disabled or when job submission
time outs
Am 2022-01-04 um 2:12 p.m. schrieb Jonathan Kim:
> For better performance during VRAM access for debugged processes, do
> read/write copies over SDMA.
>
> In order to fulfill post mortem debugging on a broken device, fallback to
> stable MMIO access when gpu recovery is disabled or when job submiss