[PATCH i-g-t] lib/intel_compute.c : Increase WMTP loop count

Dandamudi, Priyanka priyanka.dandamudi at intel.com
Tue Mar 25 13:54:18 UTC 2025



> -----Original Message-----
> From: Dugast, Francois <francois.dugast at intel.com>
> Sent: 25 March 2025 02:22 PM
> To: Dandamudi, Priyanka <priyanka.dandamudi at intel.com>
> Cc: igt-dev at lists.freedesktop.org
> Subject: Re: [PATCH i-g-t] lib/intel_compute.c : Increase WMTP loop count
> 
> Hi Priyanka,
> 
> On Tue, Mar 25, 2025 at 10:59:01AM +0530, priyanka.dandamudi at intel.com
> wrote:
> > From: Priyanka Dandamudi <priyanka.dandamudi at intel.com>
> >
> > Preemption is not happening and getting failed at the check for long
> > kernel with current WMTP loop count. So, increasing the loop count.
> >
> > Signed-off-by: Priyanka Dandamudi <priyanka.dandamudi at intel.com>
> > ---
> >  lib/intel_compute.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/lib/intel_compute.c b/lib/intel_compute.c index
> > 28149db53..76d435794 100644
> > --- a/lib/intel_compute.c
> > +++ b/lib/intel_compute.c
> > @@ -52,7 +52,7 @@
> >   * WMTP - Walker Mid Thread Preemption
> >   */
> >  #define TGP_long_kernel_loop_count		10
> > -#define WMTP_long_kernel_loop_count		1000000
> > +#define WMTP_long_kernel_loop_count		1500000
> 
> I am still observing sporadic issues in compute-preempt-many with this new
> value, but not with 2000000.
Will send it with 2000000, but not sure if this can increase the Ci run time.
> 
> Francois
> 
> >
> >  struct bo_dict_entry {
> >  	uint64_t addr;
> > --
> > 2.34.1
> >


More information about the igt-dev mailing list