Am 27.05.19 um 20:23 schrieb Kuehling, Felix:
On 2019-05-27 7:51 a.m., Christian König wrote:
That idea sounds sane to me as well.
By the way, do we somewhere signal to userspace if atomics are
supported or not?
Yes. KFD topology (a flag in the iolink) provides that information to
user mode.
On 2019-05-27 7:51 a.m., Christian König wrote:
> That idea sounds sane to me as well.
>
> By the way, do we somewhere signal to userspace if atomics are
> supported or not?
Yes. KFD topology (a flag in the iolink) provides that information to
user mode.
Regards,
Felix
>
> I mean would be n
That idea sounds sane to me as well.
By the way, do we somewhere signal to userspace if atomics are supported
or not?
I mean would be nice to keep the state inside adev if this fails for
some reason.
Christian.
Am 27.05.19 um 13:16 schrieb Zhang, Hawking:
How about put pci_enable_atomic_o
How about put pci_enable_atomic_ops_to_root ahead of
amdgpu_device_ip_early_init, while move pci_atomic_requested from kfd device to
kgd device ? In such way, we can avoid duplicate atomic request from both
amdgpu and amdkfd.
Regards,
Hawking
-Original Message-
From: amd-gfx On Behalf
