Background:
Gfx Duty Cycle Scaling(DCS) is applied on the small power limit skus.
When the current/power/temperature exceeds the limit with the heavy workload,
the gfx core can be shut off and powered on back and forth.
The ON time and OFF time is determined by the firmware according to
the accumul
[AMD Public Use]
Reviewed-by: Tao Zhou
> -Original Message-
> From: Kenneth Feng
> Sent: Wednesday, January 27, 2021 4:17 PM
> To: amd-gfx@lists.freedesktop.org
> Cc: Zhou1, Tao ; Feng, Kenneth
>
> Subject: [PATCH v2] drm/amd/pm: Enable gfx DCS feature
>
> Background:
> Gfx Duty Cycle
[AMD Public Use]
Have some comments, please find inline <>
-Original Message-
From: amd-gfx On Behalf Of Kenneth Feng
Sent: Wednesday, January 27, 2021 1:47 PM
To: amd-gfx@lists.freedesktop.org
Cc: Zhou1, Tao ; Feng, Kenneth
Subject: [PATCH v2] drm/amd/pm: Enable gfx DCS feature
Backgr
Move all the dummy functions in amdgpu_amdkfd.c to
amdgpu_amdkfd.h as inline functions.
Signed-off-by: Lang Yu
Suggested-by: Felix Kuehling
---
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 87
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h | 151 ++---
2 files changed,
Am 26.01.21 um 17:15 schrieb Harry Wentland:
On 2021-01-26 9:51 a.m., Ernst Sjöstrand wrote:
No problems with amdgpu, but when my headphone port is glitching I
often end up with some HDMI audio output selected instead.
Wouldn't this be better fixed in the audio management service, like
Puls
On Tue, Jan 26, 2021 at 02:23:17PM -0600, Bjorn Helgaas wrote:
> From: Bjorn Helgaas
>
> The double negative makes it hard to read "if (!ACPI_FAILURE(status))".
> Replace it with "if (ACPI_SUCCESS(status))".
Acked-by: Mark Brown
signature.asc
Description: PGP signature
___
On Wed, Jan 27, 2021 at 06:14:03PM +0800, Yu, Lang wrote:
> Move all the dummy functions in amdgpu_amdkfd.c to
> amdgpu_amdkfd.h as inline functions.
>
> Signed-off-by: Lang Yu
> Suggested-by: Felix Kuehling
Reviewed-by: Huang Rui
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 87
Just some additional thoughts...
The amdgpu driver already supports this as I mentioned, however only for
the non-DC codepath.
"Set HDMI/DPAudio. Only affects non-DC display handling. The default is -1
(Enabled), set 0 to disabled it."
Linux doesn't have a good infrastructure to disable sound car
Am 27.01.21 um 13:23 schrieb Ernst Sjöstrand:
Just some additional thoughts...
The amdgpu driver already supports this as I mentioned, however only
for the non-DC codepath.
"Set HDMI/DPAudio. Only affects non-DC display handling. The default
is -1 (Enabled), set 0 to disabled it."
I've added
Hey Daniel, just a ping.
Andrey
On 1/25/21 10:28 AM, Andrey Grodzovsky wrote:
On 1/19/21 8:56 AM, Daniel Vetter wrote:
On Mon, Jan 18, 2021 at 04:01:10PM -0500, Andrey Grodzovsky wrote:
On device removal reroute all CPU mappings to dummy page.
v3:
Remove loop to find DRM file and instead ac
Wave limiting can be use to load balance high priority
compute jobs along with gfx jobs. When enabled, this will reserve
70% of waves for compute jobs
Signed-off-by: Nirmoy Das
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 1 +
drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c| 16
dr
Enable gfx wave limiting for gfx jobs before pushing high priority
compute jobs so that high priority compute jobs more resources
to finish early.
Signed-off-by: Nirmoy Das
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c | 9 +
1 file changed, 9 insertions(+)
diff --git a/drivers/gpu/drm/amd
Am 27.01.21 um 15:56 schrieb Nirmoy Das:
Wave limiting can be use to load balance high priority
compute jobs along with gfx jobs. When enabled, this will reserve
70% of waves for compute jobs
Can this be changed on the fly? I was always assuming this must be
something static.
Christian.
S
On Tue, Jan 26, 2021 at 02:23:17PM -0600, Bjorn Helgaas wrote:
> From: Bjorn Helgaas
>
> The double negative makes it hard to read "if (!ACPI_FAILURE(status))".
> Replace it with "if (ACPI_SUCCESS(status))".
>
> Signed-off-by: Bjorn Helgaas
> ---
>
> This isn't really an ACPI patch, but I'm se
[AMD Official Use Only - Internal Distribution Only]
The series are:
Reviewed-by: Leo Liu
-Original Message-
From: amd-gfx On Behalf Of Alex Deucher
Sent: January 27, 2021 12:33 AM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander
Subject: [PATCH 4/4] drm/amdgpu: bump driver ver
On Wed, Jan 27, 2021 at 4:16 PM Bjorn Helgaas wrote:
>
> On Tue, Jan 26, 2021 at 02:23:17PM -0600, Bjorn Helgaas wrote:
> > From: Bjorn Helgaas
> >
> > The double negative makes it hard to read "if (!ACPI_FAILURE(status))".
> > Replace it with "if (ACPI_SUCCESS(status))".
> >
> > Signed-off-by: B
On Wed, Jan 27, 2021 at 04:44:02PM +0100, Rafael J. Wysocki wrote:
> On Wed, Jan 27, 2021 at 4:16 PM Bjorn Helgaas wrote:
> >
> > On Tue, Jan 26, 2021 at 02:23:17PM -0600, Bjorn Helgaas wrote:
> > > From: Bjorn Helgaas
> > >
> > > The double negative makes it hard to read "if (!ACPI_FAILURE(statu
On Wed, Jan 27, 2021 at 5:06 PM Bjorn Helgaas wrote:
>
> On Wed, Jan 27, 2021 at 04:44:02PM +0100, Rafael J. Wysocki wrote:
> > On Wed, Jan 27, 2021 at 4:16 PM Bjorn Helgaas wrote:
> > >
> > > On Tue, Jan 26, 2021 at 02:23:17PM -0600, Bjorn Helgaas wrote:
> > > > From: Bjorn Helgaas
> > > >
> >
On Wed, Jan 27, 2021 at 12:57 AM Feng, Kenneth wrote:
>
> [AMD Official Use Only - Internal Distribution Only]
>
> Hi Alex,
> Actually the FA DCS is dependent on the workload type.
> FA DCS is applied only when there's 3D fullscreen workload or VR workload.
> So we need to disable it if there's a
Am 2021-01-27 um 7:33 a.m. schrieb Qu Huang:
> Amdgpu driver uses 4-byte data type as DQM fence memory,
> and transmits GPU address of fence memory to microcode
> through query status PM4 message. However, query status
> PM4 message definition and microcode processing are all
> processed according
On some boards the rpm interface apparently does not work at all
leading to the fan not spinning or spinning at strange speeds.
Both interfaces work properly on the boards I have.
Let's try and use the percent interface instead.
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1408
Signed-off-
Am 2021-01-27 um 5:14 a.m. schrieb Lang Yu:
> Move all the dummy functions in amdgpu_amdkfd.c to
> amdgpu_amdkfd.h as inline functions.
>
> Signed-off-by: Lang Yu
> Suggested-by: Felix Kuehling
Just a some nit-picking inline, other than that the patch is
Reviewed-by: Felix Kuehling
> ---
>
[AMD Official Use Only - Internal Distribution Only]
Hi Alex,
Yes, SMU acts differently with regard to DCS when you set different workloads
via the workload interface. So, Kenneth seems right.
But I think the dcs feature disablement needs to be placed before
SMU_MSG_SetWorkloadMask(some code as
[AMD Public Use]
Thanks for Felix's review, I will update soon.
Regards,
Yu Lang
-Original Message-
From: Kuehling, Felix
Sent: Thursday, January 28, 2021 8:22 AM
To: Yu, Lang ; amd-gfx@lists.freedesktop.org
Cc: Huang, Ray ; Deucher, Alexander
Subject: Re: [PATCH] drm/amd/amdkfd: adj
[AMD Official Use Only - Internal Distribution Only]
Hi Alex,
Yes, SMU acts differently with regard to DCS when we set different workloads
via the workload interface.
The mapping is as below.
Default, power saving, video, custom -> DCS is disabled in SMU firmware
internally though it's enabled f
Amdgpu driver uses 4-byte data type as DQM fence memory,
and transmits GPU address of fence memory to microcode
through query status PM4 message. However, query status
PM4 message definition and microcode processing are all
processed according to 8 bytes. Fence memory only allocates
4 bytes of memo
the kernel ftrace can better help analyze the kernel running status.
add some trace events to support TTM.
add trace events list:
ttm:ttm_bo_add_mem_to_lru
ttm:ttm_bo_del_from_lru
ttm:ttm_bo_move_mem
ttm:ttm_bo_wait
ttm:ttm_bo_evict
ttm:ttm_bo_swapout
ttm:ttm_bo_device_init
ttm:ttm_bo_device_rele
add ttm bo related trace event support
trace events:
ttm:ttm_bo_add_mem_to_lru
ttm:ttm_bo_del_from_lru
ttm:ttm_bo_move_mem
ttm:ttm_bo_wait
ttm:ttm_bo_evict
ttm:ttm_bo_swapout
ttm:ttm_bo_device_init
ttm:ttm_bo_device_release
ttm:ttm_bo_init_reserved
ttm:ttm_bo_validate
ttm:ttm_bo_release
Signed-of
add ttm bo VM related trace event support
trace events:
ttm:ttm_bo_mmap
ttm:ttm_bo_vm_fault
ttm:ttm_bo_vm_access
Signed-off-by: Kevin Wang
---
drivers/gpu/drm/ttm/ttm_bo_vm.c | 12 -
drivers/gpu/drm/ttm/ttm_trace.h | 78 +
2 files changed, 88 insertions(+), 2
add ttm memory related trace event support
trace events:
ttm:ttm_shrink
ttm:ttm_mem_global_reserve
ttm:ttm_mem_global_free
Signed-off-by: Kevin Wang
---
drivers/gpu/drm/ttm/ttm_memory.c | 7
drivers/gpu/drm/ttm/ttm_trace.h | 70
2 files changed, 77 insert
[AMD Official Use Only - Internal Distribution Only]
Reviewed-by: Evan Quan
-Original Message-
From: amd-gfx On Behalf Of Alex Deucher
Sent: Thursday, January 28, 2021 6:41 AM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander
Subject: [PATCH] drm/amdgpu/swsmu: use percent rather
Not a bad start, but that needs quite some more work.
First of all please rebase on top of current drm-misc-next, a whole
bunch of the stuff you want to trace here was already removed or is
about to be removed.
Then concentrate on the necessary trace points, for example
ttm:ttm_bo_device_ini
32 matches
Mail list logo
