All of lore.kernel.org
 help / color / mirror / Atom feed
From: brichang <Brian.Chang@amd.com>
To: <amd-gfx@lists.freedesktop.org>
Cc: stylon.wang@amd.com, Brian Chang <Brian.Chang@amd.com>,
	Dillon Varone <Dillon.Varone@amd.com>,
	Sunpeng.Li@amd.com, Harry.Wentland@amd.com,
	qingqing.zhuo@amd.com, Rodrigo.Siqueira@amd.com,
	roman.li@amd.com, solomon.chiu@amd.com, Aurabindo.Pillai@amd.com,
	Alvin Lee <Alvin.Lee2@amd.com>,
	wayne.lin@amd.com, Jun Lei <Jun.Lei@amd.com>,
	Bhawanpreet.Lakha@amd.com, agustin.gutierrez@amd.com,
	pavle.kotarac@amd.com
Subject: [PATCH 09/22] drm/amd/display: Add check for DET fetch latency hiding for dcn32
Date: Fri, 18 Nov 2022 20:59:22 +0800	[thread overview]
Message-ID: <20221118125935.4013669-10-Brian.Chang@amd.com> (raw)
In-Reply-To: <20221118125935.4013669-1-Brian.Chang@amd.com>

From: Dillon Varone <Dillon.Varone@amd.com>

[WHY?]
Some configurations are constructed with very marginal DET buffers relative to
the worst possible time required to fetch a swath.

[HOW?]
Add a check to see that the DET buffer allocated for each pipe can hide the
latency for all pipes to fetch at least one swath.

Reviewed-by: Alvin Lee <Alvin.Lee2@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Dillon Varone <Dillon.Varone@amd.com>
---
 .../dc/dml/dcn32/display_mode_vba_32.c        | 39 +++++++++++
 .../dc/dml/dcn32/display_mode_vba_util_32.c   | 69 +++++++++++++++++++
 .../dc/dml/dcn32/display_mode_vba_util_32.h   | 18 +++++
 .../drm/amd/display/dc/dml/display_mode_vba.h |  2 +
 4 files changed, 128 insertions(+)

diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c
index e5c8f6a71b5b..cb801eace8f4 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c
@@ -670,6 +670,25 @@ static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerforman
 		v->cursor_bw[k] = mode_lib->vba.NumberOfCursors[k] * mode_lib->vba.CursorWidth[k][0] * mode_lib->vba.CursorBPP[k][0] / 8 / (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * mode_lib->vba.VRatio[k];
 	}
 
+	v->NotEnoughDETSwathFillLatencyHiding = dml32_CalculateDETSwathFillLatencyHiding(
+						mode_lib->vba.NumberOfActiveSurfaces,
+						mode_lib->vba.ReturnBW,
+						v->UrgentLatency,
+						mode_lib->vba.SwathHeightY,
+						mode_lib->vba.SwathHeightC,
+						v->swath_width_luma_ub,
+						v->swath_width_chroma_ub,
+						v->BytePerPixelDETY,
+						v->BytePerPixelDETC,
+						mode_lib->vba.DETBufferSizeY,
+						mode_lib->vba.DETBufferSizeC,
+						mode_lib->vba.DPPPerPlane,
+						mode_lib->vba.HTotal,
+						mode_lib->vba.PixelClock,
+						mode_lib->vba.VRatio,
+						mode_lib->vba.VRatioChroma,
+						mode_lib->vba.UsesMALLForPStateChange);
+
 	for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) {
 		v->MaxVStartupLines[k] = ((mode_lib->vba.Interlace[k] &&
 				!mode_lib->vba.ProgressiveToInterlaceUnitInOPP) ?
@@ -1664,6 +1683,7 @@ static void mode_support_configuration(struct vba_vars_st *v,
 				&& mode_lib->vba.PTEBufferSizeNotExceeded[i][j] == true
 				&& mode_lib->vba.DCCMetaBufferSizeNotExceeded[i][j] == true
 				&& mode_lib->vba.NonsupportedDSCInputBPC == false
+				&& mode_lib->vba.NotEnoughDETSwathFillLatencyHidingPerState[i][j] == false
 				&& !mode_lib->vba.ExceededMALLSize
 				&& ((mode_lib->vba.HostVMEnable == false
 				&& !mode_lib->vba.ImmediateFlipRequiredFinal)
@@ -3158,6 +3178,25 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 					mode_lib->vba.UrgentBurstFactorChroma,
 					mode_lib->vba.UrgentBurstFactorCursor);
 
+			mode_lib->vba.NotEnoughDETSwathFillLatencyHidingPerState[i][j] = dml32_CalculateDETSwathFillLatencyHiding(
+					mode_lib->vba.NumberOfActiveSurfaces,
+					mode_lib->vba.ReturnBWPerState[i][j],
+					mode_lib->vba.UrgLatency[i],
+					mode_lib->vba.SwathHeightYThisState,
+					mode_lib->vba.SwathHeightCThisState,
+					mode_lib->vba.swath_width_luma_ub_this_state,
+					mode_lib->vba.swath_width_chroma_ub_this_state,
+					mode_lib->vba.BytePerPixelInDETY,
+					mode_lib->vba.BytePerPixelInDETC,
+					mode_lib->vba.DETBufferSizeYThisState,
+					mode_lib->vba.DETBufferSizeCThisState,
+					mode_lib->vba.NoOfDPPThisState,
+					mode_lib->vba.HTotal,
+					mode_lib->vba.PixelClock,
+					mode_lib->vba.VRatio,
+					mode_lib->vba.VRatioChroma,
+					mode_lib->vba.UsesMALLForPStateChange);
+
 			v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.VMDataOnlyReturnBWPerState = dml32_get_return_bw_mbps_vm_only(&mode_lib->vba.soc, i,
 					mode_lib->vba.DCFCLKState[i][j], mode_lib->vba.FabricClockPerState[i],
 					mode_lib->vba.DRAMSpeedPerState[i]);
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c
index debe46b24a3e..5af601cff1a0 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c
@@ -6228,3 +6228,72 @@ void dml32_CalculateImmediateFlipBandwithSupport(unsigned int NumberOfActiveSurf
 	*ImmediateFlipBandwidthSupport = (*TotalBandwidth <= ReturnBW);
 	*FractionOfUrgentBandwidth = *TotalBandwidth / ReturnBW;
 }
+
+bool dml32_CalculateDETSwathFillLatencyHiding(unsigned int NumberOfActiveSurfaces,
+		double ReturnBW,
+		double UrgentLatency,
+		unsigned int SwathHeightY[],
+		unsigned int SwathHeightC[],
+		unsigned int SwathWidthY[],
+		unsigned int SwathWidthC[],
+		double  BytePerPixelInDETY[],
+		double  BytePerPixelInDETC[],
+		unsigned int    DETBufferSizeY[],
+		unsigned int    DETBufferSizeC[],
+		unsigned int	NumOfDPP[],
+		unsigned int	HTotal[],
+		double	PixelClock[],
+		double	VRatioY[],
+		double	VRatioC[],
+		enum dm_use_mall_for_pstate_change_mode UsesMALLForPStateChange[DC__NUM_DPP__MAX])
+{
+	int k;
+	double SwathSizeAllSurfaces = 0;
+	double SwathSizeAllSurfacesInFetchTimeUs;
+	double DETSwathLatencyHidingUs;
+	double DETSwathLatencyHidingYUs;
+	double DETSwathLatencyHidingCUs;
+	double SwathSizePerSurfaceY[DC__NUM_DPP__MAX];
+	double SwathSizePerSurfaceC[DC__NUM_DPP__MAX];
+	bool NotEnoughDETSwathFillLatencyHiding = false;
+
+	/* calculate sum of single swath size for all pipes in bytes*/
+	for (k = 0; k < NumberOfActiveSurfaces; k++) {
+		SwathSizePerSurfaceY[k] += SwathHeightY[k] * SwathWidthY[k] * BytePerPixelInDETY[k] * NumOfDPP[k];
+
+		if (SwathHeightC[k] != 0)
+			SwathSizePerSurfaceC[k] += SwathHeightC[k] * SwathWidthC[k] * BytePerPixelInDETC[k] * NumOfDPP[k];
+		else
+			SwathSizePerSurfaceC[k] = 0;
+
+		SwathSizeAllSurfaces += SwathSizePerSurfaceY[k] + SwathSizePerSurfaceC[k];
+	}
+
+	SwathSizeAllSurfacesInFetchTimeUs = SwathSizeAllSurfaces / ReturnBW + UrgentLatency;
+
+	/* ensure all DET - 1 swath can hide a fetch for all surfaces */
+	for (k = 0; k < NumberOfActiveSurfaces; k++) {
+		double LineTime = HTotal[k] / PixelClock[k];
+
+		/* only care if surface is not phantom */
+		if (UsesMALLForPStateChange[k] != dm_use_mall_pstate_change_phantom_pipe) {
+			DETSwathLatencyHidingYUs = (dml_floor(DETBufferSizeY[k] / BytePerPixelInDETY[k] / SwathWidthY[k], 1.0) - SwathHeightY[k]) / VRatioY[k] * LineTime;
+
+			if (SwathHeightC[k] != 0) {
+				DETSwathLatencyHidingCUs = (dml_floor(DETBufferSizeC[k] / BytePerPixelInDETC[k] / SwathWidthC[k], 1.0) - SwathHeightC[k]) / VRatioC[k] * LineTime;
+
+				DETSwathLatencyHidingUs = dml_min(DETSwathLatencyHidingYUs, DETSwathLatencyHidingCUs);
+			} else {
+				DETSwathLatencyHidingUs = DETSwathLatencyHidingYUs;
+			}
+
+			/* DET must be able to hide time to fetch 1 swath for each surface */
+			if (DETSwathLatencyHidingUs < SwathSizeAllSurfacesInFetchTimeUs) {
+				NotEnoughDETSwathFillLatencyHiding = true;
+				break;
+			}
+		}
+	}
+
+	return NotEnoughDETSwathFillLatencyHiding;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h
index 3989c2a28fae..779c6805f599 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h
@@ -1141,4 +1141,22 @@ void dml32_CalculateImmediateFlipBandwithSupport(unsigned int NumberOfActiveSurf
 		double  *FractionOfUrgentBandwidth,
 		bool *ImmediateFlipBandwidthSupport);
 
+bool dml32_CalculateDETSwathFillLatencyHiding(unsigned int NumberOfActiveSurfaces,
+		double ReturnBW,
+		double UrgentLatency,
+		unsigned int SwathHeightY[],
+		unsigned int SwathHeightC[],
+		unsigned int SwathWidthY[],
+		unsigned int SwathWidthC[],
+		double  BytePerPixelInDETY[],
+		double  BytePerPixelInDETC[],
+		unsigned int    DETBufferSizeY[],
+		unsigned int    DETBufferSizeC[],
+		unsigned int	NumOfDPP[],
+		unsigned int	HTotal[],
+		double	PixelClock[],
+		double	VRatioY[],
+		double	VRatioC[],
+		enum dm_use_mall_for_pstate_change_mode UsesMALLForPStateChange[DC__NUM_DPP__MAX]);
+
 #endif
diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h
index d46aa4817e70..d8a6b8e22504 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h
+++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h
@@ -1050,6 +1050,7 @@ struct vba_vars_st {
 	double MinFullDETBufferingTime;
 	double AverageReadBandwidthGBytePerSecond;
 	bool   FirstMainPlane;
+	bool NotEnoughDETSwathFillLatencyHiding;
 
 	unsigned int ViewportWidthChroma[DC__NUM_DPP__MAX];
 	unsigned int ViewportHeightChroma[DC__NUM_DPP__MAX];
@@ -1233,6 +1234,7 @@ struct vba_vars_st {
 	unsigned int BlockWidthC[DC__NUM_DPP__MAX];
 	unsigned int SubViewportLinesNeededInMALL[DC__NUM_DPP__MAX];
 	bool VActiveBandwithSupport[DC__VOLTAGE_STATES][2];
+	bool NotEnoughDETSwathFillLatencyHidingPerState[DC__VOLTAGE_STATES][2];
 	struct dummy_vars dummy_vars;
 };
 
-- 
2.25.1


  parent reply	other threads:[~2022-11-18 17:00 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-18 12:59 [PATCH 00/22] DC Patches November 19, 2022 brichang
2022-11-18 12:59 ` [PATCH 01/22] drm/amd/display: new ABM config 2 brichang
2022-11-18 12:59 ` [PATCH 02/22] drm/amd/display: Add margin on DRR vblank start for subvp brichang
2022-11-18 12:59 ` [PATCH 03/22] drm/amd/display: No display after resume from WB/CB brichang
2022-11-18 12:59 ` [PATCH 04/22] drm/amd/display: Limit HW cursor size of >= 4k brichang
2022-11-18 12:59 ` [PATCH 05/22] drm/amd/display: Update Z8 watermarks for DCN314 brichang
2022-11-18 12:59 ` [PATCH 06/22] drm/amd/display: Add Z8 allow states to z-state support list brichang
2022-11-18 12:59 ` [PATCH 07/22] drm/amd/display: Update soc bounding box for dcn32/dcn321 brichang
2022-11-18 12:59 ` [PATCH 08/22] drm/amd/display: Use dummy pstate latency for subvp when needed on dcn32 brichang
2022-11-18 12:59 ` brichang [this message]
2022-11-18 12:59 ` [PATCH 10/22] drm/amd/display: Check if PSR enabled when entering MALL brichang
2022-11-18 12:59 ` [PATCH 11/22] drm/amd/display: Use viewport height for subvp mall allocation size brichang
2022-11-18 12:59 ` [PATCH 12/22] drm/amd/display: Add YCBCR2020 coefficients to CSC matrix brichang
2022-11-18 12:59 ` [PATCH 13/22] drm/amd/display: Phase 1 Add Bw Allocation source and header files brichang
2022-11-18 12:59 ` [PATCH 14/22] drm/amd/display: No display after resume from WB/CB[modify] brichang
2022-11-18 12:59 ` [PATCH 15/22] drm/amd/display: Add debug options for increasing phantom lines brichang
2022-11-18 12:59 ` [PATCH 16/22] drm/amd/display: Retain phantom plane/stream if validation fails brichang
2022-11-18 12:59 ` [PATCH 17/22] drm/amd/display: Fix display corruption w/ VSR enable brichang
2022-11-18 12:59 ` [PATCH 18/22] drm/amd/display: Avoid setting pixel rate divider to N/A brichang
2022-11-18 12:59 ` [PATCH 19/22] drm/amd/display: Use new num clk levels struct for max mclk index brichang
2022-11-18 12:59 ` [PATCH 20/22] drm/amd/display: Revert check for phantom BPP brichang
2022-11-18 12:59 ` [PATCH 21/22] drm/amd/display: Fix rotated cursor offset calculation brichang
2022-11-18 12:59 ` [PATCH 22/22] SWDEV-1 - dc: 3.2.213 brichang
2022-11-18 20:12   ` Alex Deucher
2022-11-19  1:55     ` Chang, Brian

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20221118125935.4013669-10-Brian.Chang@amd.com \
    --to=brian.chang@amd.com \
    --cc=Alvin.Lee2@amd.com \
    --cc=Aurabindo.Pillai@amd.com \
    --cc=Bhawanpreet.Lakha@amd.com \
    --cc=Dillon.Varone@amd.com \
    --cc=Harry.Wentland@amd.com \
    --cc=Jun.Lei@amd.com \
    --cc=Rodrigo.Siqueira@amd.com \
    --cc=Sunpeng.Li@amd.com \
    --cc=agustin.gutierrez@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=pavle.kotarac@amd.com \
    --cc=qingqing.zhuo@amd.com \
    --cc=roman.li@amd.com \
    --cc=solomon.chiu@amd.com \
    --cc=stylon.wang@amd.com \
    --cc=wayne.lin@amd.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.