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 59AF640ED2 for ; Wed, 10 Aug 2022 20:49:03 +0000 (UTC) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id DCBF268B93D; Wed, 10 Aug 2022 23:47:40 +0300 (EEST) Received: from btbn.de (btbn.de [136.243.74.85]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 238B268B89B for ; Wed, 10 Aug 2022 23:47:31 +0300 (EEST) Received: from [authenticated] by btbn.de (Postfix) with ESMTPSA id 0E3182F260C; Wed, 10 Aug 2022 22:47:25 +0200 (CEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=rothenpieler.org; s=mail; t=1660164445; bh=H2csBUjFdQ2Aipfsh/Ovz+x8Wry9D0rysCl51yVY7JY=; h=From:To:Cc:Subject:Date:In-Reply-To:References; b=azJMuu5n+i/JHawR/wGptM9rWi4Eq0znBGWWPVxDPGD4OD71SPqzrgmBAvVqKfpjQ J1+5lDXBbb4xsFZWgj+aQQY17rBxpf5ZtCM960bChQ12xD3dE3gAxhFr4Vr4mLu00R wZhwHt33ZKahh2vuifPpCr+HxbHsyNZ6cp19/v0KeYTt8oXExg0JUWTP7xrvzB5tHE c7vWyG9r2gj1bSYLGrjfxwLCtxJ0C2O0+EHu+gsbxbZfed0aZObvWl+ZXl0FSVqjqw KPFyS4qdN91Tcac0G3+CDp8jwSzEl/wQJUAUq7XLTuCHV37KWPDwiEXl3EkfeVWt/G c2f4yyOqqjLUw== From: Timo Rothenpieler To: ffmpeg-devel@ffmpeg.org Date: Wed, 10 Aug 2022 22:47:07 +0200 Message-Id: <20220810204712.3123-6-timo@rothenpieler.org> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20220810204712.3123-1-timo@rothenpieler.org> References: <20220810204712.3123-1-timo@rothenpieler.org> MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 06/11] avutil/half2float: adjust conversion of NaN 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 Cc: Timo Rothenpieler Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" Archived-At: List-Archive: List-Post: IEEE-754 differentiates two different kind of NaNs. Quiet and Signaling ones. They are differentiated by the MSB of the mantissa. For whatever reason, actual hardware conversion of half to single always sets the signaling bit to 1 if the mantissa is != 0, and to 0 if it's 0. So our code has to follow suite or fate-testing hardware float16 will be impossible. --- libavcodec/exr.c | 2 +- libavcodec/pnm.h | 2 +- libavutil/half2float.h | 5 +++++ tests/ref/fate/exr-rgb-scanline-zip-half-0x0-0xFFFF | 2 +- 4 files changed, 8 insertions(+), 3 deletions(-) diff --git a/libavcodec/exr.c b/libavcodec/exr.c index 5c6ca9adbf..47f4786491 100644 --- a/libavcodec/exr.c +++ b/libavcodec/exr.c @@ -191,7 +191,7 @@ typedef struct EXRContext { float gamma; union av_intfloat32 gamma_table[65536]; - uint32_t mantissatable[2048]; + uint32_t mantissatable[3072]; uint32_t exponenttable[64]; uint16_t offsettable[64]; } EXRContext; diff --git a/libavcodec/pnm.h b/libavcodec/pnm.h index 5bf2eaa4d9..7e5445f529 100644 --- a/libavcodec/pnm.h +++ b/libavcodec/pnm.h @@ -34,7 +34,7 @@ typedef struct PNMContext { int half; float scale; - uint32_t mantissatable[2048]; + uint32_t mantissatable[3072]; uint32_t exponenttable[64]; uint16_t offsettable[64]; } PNMContext; diff --git a/libavutil/half2float.h b/libavutil/half2float.h index 1f6deade07..5af4690cfe 100644 --- a/libavutil/half2float.h +++ b/libavutil/half2float.h @@ -45,6 +45,9 @@ static void half2float_table(uint32_t *mantissatable, uint32_t *exponenttable, mantissatable[i] = convertmantissa(i); for (int i = 1024; i < 2048; i++) mantissatable[i] = 0x38000000UL + ((i - 1024) << 13UL); + for (int i = 2048; i < 3072; i++) + mantissatable[i] = mantissatable[i - 1024] | 0x400000UL; + mantissatable[2048] = mantissatable[1024]; exponenttable[0] = 0; for (int i = 1; i < 31; i++) @@ -58,7 +61,9 @@ static void half2float_table(uint32_t *mantissatable, uint32_t *exponenttable, offsettable[0] = 0; for (int i = 1; i < 64; i++) offsettable[i] = 1024; + offsettable[31] = 2048; offsettable[32] = 0; + offsettable[63] = 2048; } static uint32_t half2float(uint16_t h, const uint32_t *mantissatable, const uint32_t *exponenttable, diff --git a/tests/ref/fate/exr-rgb-scanline-zip-half-0x0-0xFFFF b/tests/ref/fate/exr-rgb-scanline-zip-half-0x0-0xFFFF index b6201116fe..e45a40b498 100644 --- a/tests/ref/fate/exr-rgb-scanline-zip-half-0x0-0xFFFF +++ b/tests/ref/fate/exr-rgb-scanline-zip-half-0x0-0xFFFF @@ -3,4 +3,4 @@ #codec_id 0: rawvideo #dimensions 0: 256x256 #sar 0: 1/1 -0, 0, 0, 1, 786432, 0x1445e411 +0, 0, 0, 1, 786432, 0xce9be2be -- 2.34.1 _______________________________________________ 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".