On 20/08/2019 06:23, Tomeu Vizoso wrote:
> On 8/16/19 11:31 AM, Steven Price wrote:
>> The hardware has a set of '_NEXT' registers that can hold a second job
>> while the first is executing. Make use of these registers to enqueue a
>> second job per slot.
>
> I like this in principle, but upon som
On 19/08/2019 18:02, Rob Herring wrote:
> On Mon, Aug 19, 2019 at 11:58 AM Rob Herring wrote:
>>
>> On Fri, Aug 16, 2019 at 4:31 AM Steven Price wrote:
>>>
>>> The hardware has a set of '_NEXT' registers that can hold a second job
>>> while the first is executing. Make use of these registers to e
On 8/16/19 11:31 AM, Steven Price wrote:
The hardware has a set of '_NEXT' registers that can hold a second job
while the first is executing. Make use of these registers to enqueue a
second job per slot.
I like this in principle, but upon some quick testing I found that Mesa
is around 10% slow
On Mon, Aug 19, 2019 at 11:58 AM Rob Herring wrote:
>
> On Fri, Aug 16, 2019 at 4:31 AM Steven Price wrote:
> >
> > The hardware has a set of '_NEXT' registers that can hold a second job
> > while the first is executing. Make use of these registers to enqueue a
> > second job per slot.
> >
> > Si
On Fri, Aug 16, 2019 at 4:31 AM Steven Price wrote:
>
> The hardware has a set of '_NEXT' registers that can hold a second job
> while the first is executing. Make use of these registers to enqueue a
> second job per slot.
>
> Signed-off-by: Steven Price
> ---
> Note that this is based on top of
The hardware has a set of '_NEXT' registers that can hold a second job
while the first is executing. Make use of these registers to enqueue a
second job per slot.
Signed-off-by: Steven Price
---
Note that this is based on top of Rob Herring's "per FD address space"
patch[1].
[1] https://marc.inf
