From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org [79.124.17.100]) by master.gitmailbox.com (Postfix) with ESMTP id AF9B340D3F for ; Sun, 14 Apr 2024 12:49:25 +0000 (UTC) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 675A768D308; Sun, 14 Apr 2024 15:49:22 +0300 (EEST) Received: from w4.tutanota.de (w4.tutanota.de [81.3.6.165]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 51E7668D2E4 for ; Sun, 14 Apr 2024 15:49:15 +0300 (EEST) Received: from tutadb.w10.tutanota.de (unknown [192.168.1.10]) by w4.tutanota.de (Postfix) with ESMTP id 6CFCE1060136 for ; Sun, 14 Apr 2024 12:49:15 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; t=1713098954; s=s1; d=lynne.ee; h=From:From:To:To:Subject:Subject:Content-Description:Content-ID:Content-Type:Content-Type:Content-Transfer-Encoding:Cc:Date:Date:In-Reply-To:MIME-Version:MIME-Version:Message-ID:Message-ID:Reply-To:References:Sender; bh=pc5Akq0y2Ms2y3iHpypKhajA4Qw0IJFXRZCGvQsfjtY=; b=oeeebZ15yv3IvtD26fQ7NQETCR8/39C1g05Xgg0WSoqmyLxbw/M+U00wfmTclHRg f782b/8hF5vaJu2KEYgEqv8GSImz69exqjdCmFLFyu3MFh9wAMbzML9VT76u/XcFrH7 EE6LZNYqbzLPmh4WpleZqGIGp4E8adcskFkVd0DpFHCuDNSS7RxfRMliswe4QhrMYtX 5vvJKaLll0gR/8ozZOjT6sy2G/ottjHqiQHQWc/cv24iYxr4cYdYGsXEgA5qRn6OFKe NSrg3VBt2ljaQB91kLi0nw1pddxhleuaNN+DYbUiYd9Rtvyh87Sj/gJ/ju5J440bCJu woLOntk4YQ== Date: Sun, 14 Apr 2024 14:49:14 +0200 (CEST) From: Lynne To: Ffmpeg Devel Message-ID: MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="----=_Part_97537_886252062.1713098955095" Subject: [FFmpeg-devel] [PATCH] vulkan_av1: add workaround for NVIDIA drivers tested on broken CTS X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: FFmpeg development discussions and patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: FFmpeg development discussions and patches Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" Archived-At: List-Archive: List-Post: ------=_Part_97537_886252062.1713098955095 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit The first release of the CTS for AV1 decoding had incorrect offsets for the OrderHints values. The CTS will be fixed, and eventually, the drivers will be updated to the proper spec-conforming behaviour, but we still need to add a workaround as this will take months. Only NVIDIA use these values at all, so limit the workaround to only NVIDIA. Meant to be applied on top of jkqxz's previous 2 patches. Patch attached. ------=_Part_97537_886252062.1713098955095 Content-Type: text/x-diff; charset=us-ascii; name=0001-vulkan_av1-add-workaround-for-NVIDIA-drivers-tested-.patch Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename=0001-vulkan_av1-add-workaround-for-NVIDIA-drivers-tested-.patch >From bbd2cc90206e59098accced3ff3a8896e0bfa269 Mon Sep 17 00:00:00 2001 From: Lynne Date: Sun, 14 Apr 2024 14:41:26 +0200 Subject: [PATCH] vulkan_av1: add workaround for NVIDIA drivers tested on broken CTS The first release of the CTS for AV1 decoding had incorrect offsets for the OrderHints values. The CTS will be fixed, and eventually, the drivers will be updated to the proper spec-conforming behaviour, but we still need to add a workaround as this will take months. Only NVIDIA use these values at all, so limit the workaround to only NVIDIA. --- libavcodec/vulkan_av1.c | 20 +++++++++++++++----- libavcodec/vulkan_decode.c | 8 ++++++++ libavcodec/vulkan_decode.h | 4 ++++ 3 files changed, 27 insertions(+), 5 deletions(-) diff --git a/libavcodec/vulkan_av1.c b/libavcodec/vulkan_av1.c index 8d532445a1..ff663c8cec 100644 --- a/libavcodec/vulkan_av1.c +++ b/libavcodec/vulkan_av1.c @@ -97,9 +97,15 @@ static int vk_av1_fill_pict(AVCodecContext *avctx, const AV1Frame **ref_src, .RefFrameSignBias = hp->ref_frame_sign_bias_mask, }; - if (saved_order_hints) - for (int i = 0; i < AV1_TOTAL_REFS_PER_FRAME; i++) - vkav1_std_ref->SavedOrderHints[i] = saved_order_hints[i]; + if (saved_order_hints) { + if (dec->quirk_av1_offset) { + for (int i = 1; i < AV1_TOTAL_REFS_PER_FRAME; i++) + vkav1_std_ref->SavedOrderHints[i - 1] = saved_order_hints[i]; + } else { + for (int i = 0; i < AV1_TOTAL_REFS_PER_FRAME; i++) + vkav1_std_ref->SavedOrderHints[i] = saved_order_hints[i]; + } + } *vkav1_ref = (VkVideoDecodeAV1DpbSlotInfoKHR) { .sType = VK_STRUCTURE_TYPE_VIDEO_DECODE_AV1_DPB_SLOT_INFO_KHR, @@ -490,8 +496,12 @@ static int vk_av1_start_frame(AVCodecContext *avctx, } } - for (int i = 0; i < STD_VIDEO_AV1_NUM_REF_FRAMES; i++) { - ap->std_pic_info.OrderHints[i] = pic->order_hints[i]; + if (dec->quirk_av1_offset) { + for (int i = 1; i < STD_VIDEO_AV1_NUM_REF_FRAMES; i++) + ap->std_pic_info.OrderHints[i - 1] = pic->order_hints[i]; + } else { + for (int i = 0; i < STD_VIDEO_AV1_NUM_REF_FRAMES; i++) + ap->std_pic_info.OrderHints[i] = pic->order_hints[i]; } for (int i = 0; i < STD_VIDEO_AV1_TOTAL_REFS_PER_FRAME; i++) diff --git a/libavcodec/vulkan_decode.c b/libavcodec/vulkan_decode.c index 9c6c2d4efb..5cb328d8ca 100644 --- a/libavcodec/vulkan_decode.c +++ b/libavcodec/vulkan_decode.c @@ -1115,6 +1115,7 @@ int ff_vk_decode_init(AVCodecContext *avctx) FFVulkanFunctions *vk; const VkVideoProfileInfoKHR *profile; const FFVulkanDecodeDescriptor *vk_desc; + const VkPhysicalDeviceDriverProperties *driver_props; VkVideoDecodeH264SessionParametersCreateInfoKHR h264_params = { .sType = VK_STRUCTURE_TYPE_VIDEO_DECODE_H264_SESSION_PARAMETERS_CREATE_INFO_KHR, @@ -1276,6 +1277,13 @@ int ff_vk_decode_init(AVCodecContext *avctx) return AVERROR_EXTERNAL; } + driver_props = &dec->shared_ctx->s.driver_props; + if (driver_props->driverID == VK_DRIVER_ID_NVIDIA_PROPRIETARY && + driver_props->conformanceVersion.major == 1 && + driver_props->conformanceVersion.minor == 3 && + driver_props->conformanceVersion.subminor == 8) + dec->quirk_av1_offset = 1; + ff_vk_decode_flush(avctx); av_log(avctx, AV_LOG_VERBOSE, "Vulkan decoder initialization sucessful\n"); diff --git a/libavcodec/vulkan_decode.h b/libavcodec/vulkan_decode.h index 7ba8b239cb..076af93499 100644 --- a/libavcodec/vulkan_decode.h +++ b/libavcodec/vulkan_decode.h @@ -72,6 +72,10 @@ typedef struct FFVulkanDecodeContext { int external_fg; /* Oddity #2 - hardware can't apply film grain */ uint32_t frame_id_alloc_mask; /* For AV1 only */ + /* Workaround for NVIDIA drivers tested with CTS version 1.3.8 for AV1. + * The tests were incorrect as the OrderHints were offset by 1. */ + int quirk_av1_offset; + /* Thread-local state below */ struct HEVCHeaderSet *hevc_headers; size_t hevc_headers_size; -- 2.43.0.381.gb435a96ce8 ------=_Part_97537_886252062.1713098955095 Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe". ------=_Part_97537_886252062.1713098955095--