On Tue, 2017-02-28 at 06:25 +0000, Yang, Rong R wrote: > We suggest GLK could disable pooled EUs for 2x6 configurations too. This patch does that. I tried to say that in the commit message but now that I re-read I see it doesn't really say it. > As I understand, 2x6 a pool must consist of a complete subslice, 12EUs, right? That's my understanding too. Ander > If so, only 64K SLM are valid, I am afraid it may affect some case's performance. > > > -----Original Message----- > > From: Conselvan De Oliveira, Ander > > Sent: Friday, February 24, 2017 21:13 > > To: intel-gfx@xxxxxxxxxxxxxxxxxxxxx > > Cc: Conselvan De Oliveira, Ander <ander.conselvan.de.oliveira@xxxxxxxxx>; > > Arun Siluvery <arun.siluvery@xxxxxxxxx>; Kuoppala, Mika > > <mika.kuoppala@xxxxxxxxx>; Ursulin, Tvrtko <tvrtko.ursulin@xxxxxxxxx>; > > Yang, Rong R <rong.r.yang@xxxxxxxxx> > > Subject: [PATCH] drm/915/glk: Enable pooled EUs for Geminilake > > > > Geminilake also supports pooled EUs. Enable it. > > > > It is unclear if the recommendation to disable it for 2x6 configurations from > > commit e015dd69b2cf ("drm/i915/bxt: Add WaEnablePooledEuFor2x6") > > should also apply to GLK, but the only userspace that uses this only cares > > about the 3x6 configuration. See Beignet's commit 6901899ec90a > > ("Runtime: set the sub slice according to kernel pooled EU configure."). > > > > Cc: Arun Siluvery <arun.siluvery@xxxxxxxxx> > > Cc: Mika Kuoppala <mika.kuoppala@xxxxxxxxx> > > Cc: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx> > > Cc: Yang Rong <rong.r.yang@xxxxxxxxx> > > Signed-off-by: Ander Conselvan de Oliveira > > <ander.conselvan.de.oliveira@xxxxxxxxx> > > --- > > drivers/gpu/drm/i915/intel_device_info.c | 9 +++++---- > > 1 file changed, 5 insertions(+), 4 deletions(-) > > > > diff --git a/drivers/gpu/drm/i915/intel_device_info.c > > b/drivers/gpu/drm/i915/intel_device_info.c > > index 2e1fd85..198752d 100644 > > --- a/drivers/gpu/drm/i915/intel_device_info.c > > +++ b/drivers/gpu/drm/i915/intel_device_info.c > > @@ -195,8 +195,10 @@ static void gen9_sseu_info_init(struct > > drm_i915_private *dev_priv) > > IS_GEN9_LP(dev_priv) && sseu_subslice_total(sseu) > 1; > > sseu->has_eu_pg = sseu->eu_per_subslice > 2; > > > > - if (IS_BROXTON(dev_priv)) { > > + if (IS_GEN9_LP(dev_priv)) { > > #define IS_SS_DISABLED(ss) (!(sseu->subslice_mask & BIT(ss))) > > + info->has_pooled_eu = hweight8(sseu->subslice_mask) == 3; > > + > > /* > > * There is a HW issue in 2x6 fused down parts that requires > > * Pooled EU to be enabled as a WA. The pool configuration > > @@ -204,9 +206,8 @@ static void gen9_sseu_info_init(struct > > drm_i915_private *dev_priv) > > * doesn't affect if the device has all 3 subslices enabled. > > */ > > /* WaEnablePooledEuFor2x6:bxt */ > > - info->has_pooled_eu = ((hweight8(sseu->subslice_mask) == > > 3) || > > - (hweight8(sseu->subslice_mask) == 2 && > > - INTEL_REVID(dev_priv) < > > BXT_REVID_C0)); > > + info->has_pooled_eu |= (hweight8(sseu->subslice_mask) == > > 2 && > > + IS_BXT_REVID(dev_priv, 0, > > BXT_REVID_B_LAST)); > > > > sseu->min_eu_in_pool = 0; > > if (info->has_pooled_eu) { > > -- > > 2.9.3 > > _______________________________________________ > Intel-gfx mailing list > Intel-gfx@xxxxxxxxxxxxxxxxxxxxx > https://lists.freedesktop.org/mailman/listinfo/intel-gfx _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx