On Wed, 16 Apr 2025 16:16:05 +0100
Steven Price wrote:
> On 15/04/2025 10:47, Boris Brezillon wrote:
> > On Mon, 14 Apr 2025 16:34:35 +0100
> > Steven Price wrote:
> >
> >> On 14/04/2025 13:47, Boris Brezillon wrote:
> >>> On Fri, 11 Apr 2025 16:39:02 +0200
> >>> Boris Brezillon wrote:
> >
On 15/04/2025 10:47, Boris Brezillon wrote:
> On Mon, 14 Apr 2025 16:34:35 +0100
> Steven Price wrote:
>
>> On 14/04/2025 13:47, Boris Brezillon wrote:
>>> On Fri, 11 Apr 2025 16:39:02 +0200
>>> Boris Brezillon wrote:
>>>
On Fri, 11 Apr 2025 15:13:26 +0200
Christian König wrote:
>>
On Mon, 14 Apr 2025 09:03:55 -0400
Alyssa Rosenzweig wrote:
> > Actually, CSF stands in the way of re-allocating memory to other
> > contexts, because once we've allocated memory to a tiler heap, the FW
> > manages this pool of chunks, and recycles them. Mesa can intercept
> > the "returned chunk
On Mon, Apr 14, 2025 at 01:22:06PM +0200, Boris Brezillon wrote:
> Hi Sima,
>
> On Fri, 11 Apr 2025 14:01:16 +0200
> Simona Vetter wrote:
>
> > On Thu, Apr 10, 2025 at 08:41:55PM +0200, Boris Brezillon wrote:
> > > On Thu, 10 Apr 2025 14:01:03 -0400
> > > Alyssa Rosenzweig wrote:
> > >
> > >
Hi,
On Mon, 14 Apr 2025 at 16:34, Steven Price wrote:
> On 14/04/2025 13:47, Boris Brezillon wrote:
> > Hm, I might have been too prompt at claiming this was doable. In
> > practice, doing that might regress Lima and Panfrost in situations
> > where trying harder than GFP_NOWAIT would free up som
On Mon, 14 Apr 2025 16:34:35 +0100
Steven Price wrote:
> On 14/04/2025 13:47, Boris Brezillon wrote:
> > On Fri, 11 Apr 2025 16:39:02 +0200
> > Boris Brezillon wrote:
> >
> >> On Fri, 11 Apr 2025 15:13:26 +0200
> >> Christian König wrote:
> >>
>
> > Background is that you do
On 14/04/2025 13:47, Boris Brezillon wrote:
> On Fri, 11 Apr 2025 16:39:02 +0200
> Boris Brezillon wrote:
>
>> On Fri, 11 Apr 2025 15:13:26 +0200
>> Christian König wrote:
>>
> Background is that you don't get a crash, nor error message, nor
> anything indicating what is happeni
On Mon, 14 Apr 2025 16:34:47 +0200
Simona Vetter wrote:
> On Mon, Apr 14, 2025 at 02:08:25PM +0100, Liviu Dudau wrote:
> > On Mon, Apr 14, 2025 at 01:22:06PM +0200, Boris Brezillon wrote:
> > > Hi Sima,
> > >
> > > On Fri, 11 Apr 2025 14:01:16 +0200
> > > Simona Vetter wrote:
> > >
> > > >
On Mon, Apr 14, 2025 at 01:22:06PM +0200, Boris Brezillon wrote:
> Hi Sima,
>
> On Fri, 11 Apr 2025 14:01:16 +0200
> Simona Vetter wrote:
>
> > On Thu, Apr 10, 2025 at 08:41:55PM +0200, Boris Brezillon wrote:
> > > On Thu, 10 Apr 2025 14:01:03 -0400
> > > Alyssa Rosenzweig wrote:
> > >
> > >
On Mon, Apr 14, 2025 at 02:08:25PM +0100, Liviu Dudau wrote:
> On Mon, Apr 14, 2025 at 01:22:06PM +0200, Boris Brezillon wrote:
> > Hi Sima,
> >
> > On Fri, 11 Apr 2025 14:01:16 +0200
> > Simona Vetter wrote:
> >
> > > On Thu, Apr 10, 2025 at 08:41:55PM +0200, Boris Brezillon wrote:
> > > > On T
> I'm still trying to see if we can emulate/have incremental-rendering on
> JM hardware
I guess since we don't advertise vertex shader side effects... Lol,
maybe that could work...
> Actually, CSF stands in the way of re-allocating memory to other
> contexts, because once we've allocated memory to a tiler heap, the FW
> manages this pool of chunks, and recycles them. Mesa can intercept
> the "returned chunks" and collect those chunks instead of re-assiging
> then to the tiler
On Fri, 11 Apr 2025 16:39:02 +0200
Boris Brezillon wrote:
> On Fri, 11 Apr 2025 15:13:26 +0200
> Christian König wrote:
>
> > >
> > >> Background is that you don't get a crash, nor error message, nor
> > >> anything indicating what is happening.
> > > The job times out at some point, bu
Hi Sima,
On Fri, 11 Apr 2025 14:01:16 +0200
Simona Vetter wrote:
> On Thu, Apr 10, 2025 at 08:41:55PM +0200, Boris Brezillon wrote:
> > On Thu, 10 Apr 2025 14:01:03 -0400
> > Alyssa Rosenzweig wrote:
> >
> > > > > > In Panfrost and Lima, we don't have this concept of "incremental
> > > > > >
On Fri, Apr 11, 2025 at 12:55:57PM +0200, Christian König wrote:
> Am 11.04.25 um 10:38 schrieb Boris Brezillon:
> > On Fri, 11 Apr 2025 10:04:07 +0200
> > Christian König wrote:
> >
> >> Am 10.04.25 um 20:41 schrieb Boris Brezillon:
> >>> On Thu, 10 Apr 2025 14:01:03 -0400
> >>> Alyssa Rosenzweig
On Fri, Apr 11, 2025 at 02:50:19PM +0200, Christian König wrote:
> Am 11.04.25 um 14:01 schrieb Simona Vetter:
> > On Thu, Apr 10, 2025 at 08:41:55PM +0200, Boris Brezillon wrote:
> >> On Thu, 10 Apr 2025 14:01:03 -0400
> >> Alyssa Rosenzweig wrote:
> >>
> >> In Panfrost and Lima, we don't hav
On Fri, Apr 11, 2025 at 09:52:30AM -0400, Alyssa Rosenzweig wrote:
> > 2. Device Lost
> > --
> >
> > At this point we're left with no other choice than to kill the context.
> > And userspace should be able to cope with VK_DEVICE_LOST (hopefully zink
> > does), but it will probably not
Am 11.04.25 um 15:00 schrieb Boris Brezillon:
> On Fri, 11 Apr 2025 14:45:49 +0200
> Christian König wrote:
>
>> Am 11.04.25 um 14:02 schrieb Boris Brezillon:
> I guess this leaves older GPUs that don't support incremental
> rendering in a bad place though.
Well what's the handlin
On Fri, 11 Apr 2025 15:13:26 +0200
Christian König wrote:
> >
> >> Background is that you don't get a crash, nor error message, nor
> >> anything indicating what is happening.
> > The job times out at some point, but we might get stuck in the fault
> > handler waiting for memory, which is pre
> 2. Device Lost
> --
>
> At this point we're left with no other choice than to kill the context.
> And userspace should be able to cope with VK_DEVICE_LOST (hopefully zink
> does), but it will probably not cope well with an entire strom of these
> just to get the first frame out.
>
>
On Fri, 11 Apr 2025 14:45:49 +0200
Christian König wrote:
> Am 11.04.25 um 14:02 schrieb Boris Brezillon:
> >>> I guess this leaves older GPUs that don't support incremental
> >>> rendering in a bad place though.
> >> Well what's the handling there currently? Just crash when you're
> >> OOM?
Am 11.04.25 um 14:01 schrieb Simona Vetter:
> On Thu, Apr 10, 2025 at 08:41:55PM +0200, Boris Brezillon wrote:
>> On Thu, 10 Apr 2025 14:01:03 -0400
>> Alyssa Rosenzweig wrote:
>>
>> In Panfrost and Lima, we don't have this concept of "incremental
>> rendering", so when we fail the allocat
Am 11.04.25 um 14:02 schrieb Boris Brezillon:
>>> I guess this leaves older GPUs that don't support incremental
>>> rendering in a bad place though.
>> Well what's the handling there currently? Just crash when you're OOM?
> It's "alloc(GFP_KERNEL) and crash if it fails or times out", yes.
Oh, pl
On Fri, 11 Apr 2025 12:55:57 +0200
Christian König wrote:
> Am 11.04.25 um 10:38 schrieb Boris Brezillon:
> > On Fri, 11 Apr 2025 10:04:07 +0200
> > Christian König wrote:
> >
> >> Am 10.04.25 um 20:41 schrieb Boris Brezillon:
> >>> On Thu, 10 Apr 2025 14:01:03 -0400
> >>> Alyssa Rosenzweig
On Thu, Apr 10, 2025 at 08:41:55PM +0200, Boris Brezillon wrote:
> On Thu, 10 Apr 2025 14:01:03 -0400
> Alyssa Rosenzweig wrote:
>
> > > > > In Panfrost and Lima, we don't have this concept of "incremental
> > > > > rendering", so when we fail the allocation, we just fail the GPU job
> > > > > wi
Am 11.04.25 um 10:38 schrieb Boris Brezillon:
> On Fri, 11 Apr 2025 10:04:07 +0200
> Christian König wrote:
>
>> Am 10.04.25 um 20:41 schrieb Boris Brezillon:
>>> On Thu, 10 Apr 2025 14:01:03 -0400
>>> Alyssa Rosenzweig wrote:
>>>
>>> In Panfrost and Lima, we don't have this concept of "inc
On Fri, 11 Apr 2025 10:04:07 +0200
Christian König wrote:
> Am 10.04.25 um 20:41 schrieb Boris Brezillon:
> > On Thu, 10 Apr 2025 14:01:03 -0400
> > Alyssa Rosenzweig wrote:
> >
> > In Panfrost and Lima, we don't have this concept of "incremental
> > rendering", so when we fail the all
Hi Christian,
On Thu, 10 Apr 2025 20:52:27 +0200
Christian König wrote:
> Am 10.04.25 um 19:20 schrieb Boris Brezillon:
> > [SNIP]
> Faulting is only legal when you have something like HMM, SVM or
> whatever you call it. And then you can just use a plain shmem
> object to provid
Am 10.04.25 um 20:41 schrieb Boris Brezillon:
> On Thu, 10 Apr 2025 14:01:03 -0400
> Alyssa Rosenzweig wrote:
>
> In Panfrost and Lima, we don't have this concept of "incremental
> rendering", so when we fail the allocation, we just fail the GPU job
> with an unhandled GPU fault.
>
Am 10.04.25 um 19:20 schrieb Boris Brezillon:
> [SNIP]
Faulting is only legal when you have something like HMM, SVM or
whatever you call it. And then you can just use a plain shmem
object to provide you with backing pages.
I mean we could in theory allow faulting on GEM obj
On Thu, 10 Apr 2025 14:01:03 -0400
Alyssa Rosenzweig wrote:
> > > > In Panfrost and Lima, we don't have this concept of "incremental
> > > > rendering", so when we fail the allocation, we just fail the GPU job
> > > > with an unhandled GPU fault.
> > >
> > > To be honest I think that this is
> > > In Panfrost and Lima, we don't have this concept of "incremental
> > > rendering", so when we fail the allocation, we just fail the GPU job
> > > with an unhandled GPU fault.
> >
> > To be honest I think that this is enough to mark those two drivers as
> > broken. It's documented that thi
Hi Christian,
On Thu, 10 Apr 2025 18:43:29 +0200
Christian König wrote:
> Hi Boris,
>
> Am 10.04.25 um 17:53 schrieb Boris Brezillon:
> > Hi Christian,
> >
> > On Thu, 10 Apr 2025 17:05:07 +0200
> > Christian König wrote:
> >
> >> Hi Boris,
> >>
> >> thanks for looping me in. Can you also se
Hi Boris,
Am 10.04.25 um 17:53 schrieb Boris Brezillon:
> Hi Christian,
>
> On Thu, 10 Apr 2025 17:05:07 +0200
> Christian König wrote:
>
>> Hi Boris,
>>
>> thanks for looping me in. Can you also send the full patch set to me since I
>> don't see that on the mailing list (yet maybe).
>>
>> Am 10
Hi Christian,
On Thu, 10 Apr 2025 17:05:07 +0200
Christian König wrote:
> Hi Boris,
>
> thanks for looping me in. Can you also send the full patch set to me since I
> don't see that on the mailing list (yet maybe).
>
> Am 10.04.25 um 16:48 schrieb Boris Brezillon:
> > +Christian, Alyssa and F
Hi Boris,
thanks for looping me in. Can you also send the full patch set to me since I
don't see that on the mailing list (yet maybe).
Am 10.04.25 um 16:48 schrieb Boris Brezillon:
> +Christian, Alyssa and Faith, as suggested by Sima. I'll make sure to
> Cc you on v4, but before that, I'd like t
+Christian, Alyssa and Faith, as suggested by Sima. I'll make sure to
Cc you on v4, but before that, I'd like to get your opinion on the
drm-gem/drm-gem-shmem changes to see if sending a v4 is actually
desirable or if I should go back to the drawing board.
On Fri, 4 Apr 2025 11:26:26 +0200
Boris
This patch series is a proposal for implementing sparse page allocations
for shmem objects. It was initially motivated by a kind of BO managed by
the Panfrost/Panthor and Lima drivers, the tiler heap, which grows on
demand every time the GPU faults on a virtual address within the heap BO
range.
Be
38 matches
Mail list logo