Re: [PATCH 05/13] drm/i915/intel_cdclk: Add vdsc with bigjoiner constraints on min_cdlck

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, May 15, 2023 at 05:44:51PM +0300, Ville Syrjälä wrote:
> On Fri, May 12, 2023 at 11:54:09AM +0530, Ankit Nautiyal wrote:
> > As per Bsepc:49259, Bigjoiner BW check puts restriction on the
> > compressed bpp for a given CDCLK, pixelclock in cases where
> > Bigjoiner + DSC are used.
> > 
> > Currently compressed bpp is computed first, and it is ensured that
> > the bpp will work at least with the max CDCLK freq.
> > 
> > Since the CDCLK is computed later, lets account for Bigjoiner BW
> > check while calculating Min CDCLK.
> > 
> > Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@xxxxxxxxx>
> > ---
> >  drivers/gpu/drm/i915/display/intel_cdclk.c | 49 ++++++++++++++++++----
> >  1 file changed, 42 insertions(+), 7 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/display/intel_cdclk.c b/drivers/gpu/drm/i915/display/intel_cdclk.c
> > index 6bed75f1541a..3532640c5027 100644
> > --- a/drivers/gpu/drm/i915/display/intel_cdclk.c
> > +++ b/drivers/gpu/drm/i915/display/intel_cdclk.c
> > @@ -2520,6 +2520,46 @@ static int intel_planes_min_cdclk(const struct intel_crtc_state *crtc_state)
> >  	return min_cdclk;
> >  }
> >  
> > +static int intel_vdsc_min_cdclk(const struct intel_crtc_state *crtc_state)
> > +{
> > +	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
> > +	struct drm_i915_private *i915 = to_i915(crtc->base.dev);
> > +	int min_cdclk = 0;
> > +
> > +	/*
> > +	 * When we decide to use only one VDSC engine, since
> > +	 * each VDSC operates with 1 ppc throughput, pixel clock
> > +	 * cannot be higher than the VDSC clock (cdclk)
> > +	 */
> > +	if (!crtc_state->dsc.dsc_split)
> > +		min_cdclk = max(min_cdclk, (int)crtc_state->pixel_rate);
> > +
> > +	if (crtc_state->bigjoiner_pipes) {
> > +		/*
> > +		 * According to Bigjoiner bw check:
> > +		 * compressed_bpp <= PPC * CDCLK * Big joiner Interface bits / Pixel clock
> > +		 *
> > +		 * We have already computed compressed_bpp, so now compute the min CDCLK that
> > +		 * is required to support this compressed_bpp.
> > +		 *
> > +		 * => CDCLK >= compressed_bpp * Pixel clock / (PPC * Bigjoiner Interface bits)
> > +		 *
> > +		 * Since Num of pipes joined = 2, and PPC = 2 with bigjoiner
> > +		 * => CDCLK >= compressed_bpp * pixel_rate  / Bigjoiner Interface bits
> > +		 *
> > +		 * #TODO Bspec mentions to account for FEC overhead while using pixel clock.
> > +		 * Check if we need to use FEC overhead in the above calculations.
> > +		 */
> > +		int bigjoiner_interface_bits = DISPLAY_VER(i915) > 13 ? 36 : 24;
> > +		int min_cdclk_bj = crtc_state->dsc.compressed_bpp * crtc_state->pixel_rate /
> > +				   bigjoiner_interface_bits;
> 
> pixel_rate is the downscale adjusted thing, so it doesn't seem
> like the correct thing to use here.
> 
> Hmm. Assuming that the single VDSC engine really throttles the entire
> pipe to 1 PPC then we should probably account for the 1 vs. 2 PPC
> difference in *_plane_min_cdclk() and intel_pixel_rate_to_cdclk()
> directly. Currently all of those assume 2 PPC.

Main thing is to properly align that one you propose above with that check,
where we decide how many VDSC engines to use:

        /*
         * VDSC engine operates at 1 Pixel per clock, so if peak pixel rate
         * is greater than the maximum Cdclock and if slice count is even
         * then we need to use 2 VDSC instances.
         */
        if (adjusted_mode->crtc_clock > dev_priv->max_cdclk_freq) {
                if (pipe_config->dsc.slice_count > 1) {
                        pipe_config->dsc.dsc_split = true;
                } else {
                        drm_dbg_kms(&dev_priv->drm,
                                    "Cannot split stream to use 2 VDSC instances\n");
                        return -EINVAL;
                }
        }

Otherwise I agree that we should do that check preferrably in *_plane_min_cdclk
and use plane data rate which is adjusted after scaling is applied(I think we even have correspondent function there)
It is strange that scaling wasn't mentioned in BSpec formula.
I would also say that we should account for number of slices(i.e VDSC engines) now only in Bigjoiner case, but always, as I understand that number can be different not only for Bigjoiner cases.

Stan


> 
> > +
> > +		min_cdclk = max(min_cdclk, min_cdclk_bj);
> > +	}
> > +
> > +	return min_cdclk;
> > +}
> > +
> >  int intel_crtc_compute_min_cdclk(const struct intel_crtc_state *crtc_state)
> >  {
> >  	struct drm_i915_private *dev_priv =
> > @@ -2591,13 +2631,8 @@ int intel_crtc_compute_min_cdclk(const struct intel_crtc_state *crtc_state)
> >  	/* Account for additional needs from the planes */
> >  	min_cdclk = max(intel_planes_min_cdclk(crtc_state), min_cdclk);
> >  
> > -	/*
> > -	 * When we decide to use only one VDSC engine, since
> > -	 * each VDSC operates with 1 ppc throughput, pixel clock
> > -	 * cannot be higher than the VDSC clock (cdclk)
> > -	 */
> > -	if (crtc_state->dsc.compression_enable && !crtc_state->dsc.dsc_split)
> > -		min_cdclk = max(min_cdclk, (int)crtc_state->pixel_rate);
> > +	if (crtc_state->dsc.compression_enable)
> > +		min_cdclk = max(min_cdclk, intel_vdsc_min_cdclk(crtc_state));
> >  
> >  	/*
> >  	 * HACK. Currently for TGL/DG2 platforms we calculate
> > -- 
> > 2.25.1
> 
> -- 
> Ville Syrjälä
> Intel



[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux