{"id":26212,"url":"https://patchwork.libcamera.org/api/patches/26212/?format=json","web_url":"https://patchwork.libcamera.org/patch/26212/","project":{"id":1,"url":"https://patchwork.libcamera.org/api/projects/1/?format=json","name":"libcamera","link_name":"libcamera","list_id":"libcamera_core","list_email":"libcamera-devel@lists.libcamera.org","web_url":"","scm_url":"","webscm_url":""},"msgid":"<20260219-kbingham-quantizers-v8-15-2b6ff68ead26@ideasonboard.com>","date":"2026-02-19T15:05:16","name":"[v8,15/15] ipa: libipa: fixedpoint: Move float conversion inline","commit_ref":"8064c4f33ad0fd47631877d85ea138b556b09424","pull_url":null,"state":"accepted","archived":false,"hash":"91e70494dc1730fe5da70f6db99c79ae70b627d8","submitter":{"id":4,"url":"https://patchwork.libcamera.org/api/people/4/?format=json","name":"Kieran Bingham","email":"kieran.bingham@ideasonboard.com"},"delegate":null,"mbox":"https://patchwork.libcamera.org/patch/26212/mbox/","series":[{"id":5802,"url":"https://patchwork.libcamera.org/api/series/5802/?format=json","web_url":"https://patchwork.libcamera.org/project/libcamera/list/?series=5802","date":"2026-02-19T15:05:01","name":"libipa: Introduce a Quantized type","version":8,"mbox":"https://patchwork.libcamera.org/series/5802/mbox/"}],"comments":"https://patchwork.libcamera.org/api/patches/26212/comments/","check":"pending","checks":"https://patchwork.libcamera.org/api/patches/26212/checks/","tags":{},"headers":{"Return-Path":"","X-Original-To":"parsemail@patchwork.libcamera.org","Delivered-To":"parsemail@patchwork.libcamera.org","Received":["from lancelot.ideasonboard.com (lancelot.ideasonboard.com\n\t[92.243.16.209])\n\tby patchwork.libcamera.org (Postfix) with ESMTPS id 11207C32F4\n\tfor ;\n\tThu, 19 Feb 2026 15:05:29 +0000 (UTC)","from lancelot.ideasonboard.com (localhost [IPv6:::1])\n\tby lancelot.ideasonboard.com (Postfix) with ESMTP id EED00622A2;\n\tThu, 19 Feb 2026 16:05:26 +0100 (CET)","from perceval.ideasonboard.com (perceval.ideasonboard.com\n\t[IPv6:2001:4b98:dc2:55:216:3eff:fef7:d647])\n\tby lancelot.ideasonboard.com (Postfix) with ESMTPS id EB3DD62275\n\tfor ;\n\tThu, 19 Feb 2026 16:05:12 +0100 (CET)","from ping.linuxembedded.co.uk\n\t(cpc89244-aztw30-2-0-cust6594.18-1.cable.virginm.net [86.31.185.195])\n\tby perceval.ideasonboard.com (Postfix) with ESMTPSA id B9E5B27FE;\n\tThu, 19 Feb 2026 16:04:19 +0100 (CET)"],"Authentication-Results":"lancelot.ideasonboard.com; dkim=pass (1024-bit key;\n\tunprotected) header.d=ideasonboard.com header.i=@ideasonboard.com\n\theader.b=\"LY55XWJa\"; dkim-atps=neutral","DKIM-Signature":"v=1; a=rsa-sha256; c=relaxed/simple; d=ideasonboard.com;\n\ts=mail; t=1771513459;\n\tbh=ZS4mVLpItJiKWId3mQLO8b9HUpOybF6GBGG9uLiV+uw=;\n\th=From:Date:Subject:References:In-Reply-To:To:Cc:From;\n\tb=LY55XWJawpY/k0jkdMpi4fX3QTd9BM3UT5m5ziu4AIG9dodSYkUBPrUF3z/bKSdMa\n\tATKUjP41hlz6EzbFRmbHa4YZ4mPFz9gYb7812NpjxxFYheZpZu4gIHqlvpZzOn1gCE\n\tAxDwXFaSWqrbNTLXzqNwWV8Z83JdEdCgT4kZ/dqY=","From":"Kieran Bingham ","Date":"Thu, 19 Feb 2026 15:05:16 +0000","Subject":"[PATCH v8 15/15] ipa: libipa: fixedpoint: Move float conversion\n\tinline","MIME-Version":"1.0","Content-Type":"text/plain; charset=\"utf-8\"","Content-Transfer-Encoding":"7bit","Message-Id":"<20260219-kbingham-quantizers-v8-15-2b6ff68ead26@ideasonboard.com>","References":"<20260219-kbingham-quantizers-v8-0-2b6ff68ead26@ideasonboard.com>","In-Reply-To":"<20260219-kbingham-quantizers-v8-0-2b6ff68ead26@ideasonboard.com>","To":"libcamera-devel@lists.libcamera.org","Cc":"Kieran Bingham , \n\tIsaac Scott , \n\tLaurent Pinchart , \n\tPaul Elder ","X-Mailer":"b4 0.14.3","X-Developer-Signature":"v=1; a=ed25519-sha256; t=1771513507; l=5788;\n\ti=kieran.bingham@ideasonboard.com; s=20260207;\n\th=from:subject:message-id; \n\tbh=ZS4mVLpItJiKWId3mQLO8b9HUpOybF6GBGG9uLiV+uw=;\n\tb=yYz+JGj+J4LPtbtHlUpILFVPkU7ywdFWBuUlsOfqnRq4E29F0K3xHDHYA3yMm8WwOfXK5/GRt\n\t/JV6cVnrEdoAfya7/h9DRBq1eOdBGeQa5fJGD1iEfoV4+k0qeWaadY+","X-Developer-Key":"i=kieran.bingham@ideasonboard.com; a=ed25519;\n\tpk=FVXKN7YuwHc6UtbRUeTMAmranfsQomA+vnilfglWdaY=","X-BeenThere":"libcamera-devel@lists.libcamera.org","X-Mailman-Version":"2.1.29","Precedence":"list","List-Id":"","List-Unsubscribe":",\n\t","List-Archive":"","List-Post":"","List-Help":"","List-Subscribe":",\n\t","Errors-To":"libcamera-devel-bounces@lists.libcamera.org","Sender":"\"libcamera-devel\" "},"content":"With all users of the floatingToFixedPoint and fixedToFloatingPoint\ncalls converted to use the FixedPoint Quantized types, remove the calls\nand documentation and move the implementation inline in the\nFixedPointTraits implementation.\n\nReviewed-by: Isaac Scott \nReviewed-by: Laurent Pinchart \nReviewed-by: Paul Elder \nSigned-off-by: Kieran Bingham \n---\n src/ipa/libipa/fixedpoint.cpp | 22 -------------\n src/ipa/libipa/fixedpoint.h | 75 +++++++++++++------------------------------\n 2 files changed, 22 insertions(+), 75 deletions(-)","diff":"diff --git a/src/ipa/libipa/fixedpoint.cpp b/src/ipa/libipa/fixedpoint.cpp\nindex b581ca8453ef..b06ad8d26719 100644\n--- a/src/ipa/libipa/fixedpoint.cpp\n+++ b/src/ipa/libipa/fixedpoint.cpp\n@@ -15,28 +15,6 @@ namespace libcamera {\n \n namespace ipa {\n \n-/**\n- * \\fn R floatingToFixedPoint(T number)\n- * \\brief Convert a floating point number to a fixed-point representation\n- * \\tparam I Bit width of the integer part of the fixed-point\n- * \\tparam F Bit width of the fractional part of the fixed-point\n- * \\tparam R Return type of the fixed-point representation\n- * \\tparam T Input type of the floating point representation\n- * \\param number The floating point number to convert to fixed point\n- * \\return The converted value\n- */\n-\n-/**\n- * \\fn R fixedToFloatingPoint(T number)\n- * \\brief Convert a fixed-point number to a floating point representation\n- * \\tparam I Bit width of the integer part of the fixed-point\n- * \\tparam F Bit width of the fractional part of the fixed-point\n- * \\tparam R Return type of the floating point representation\n- * \\tparam T Input type of the fixed-point representation\n- * \\param number The fixed point number to convert to floating point\n- * \\return The converted value\n- */\n-\n /**\n * \\struct libcamera::ipa::FixedPointQTraits\n * \\brief Traits type implementing fixed-point quantization conversions\ndiff --git a/src/ipa/libipa/fixedpoint.h b/src/ipa/libipa/fixedpoint.h\nindex df685e852bff..ed8aca9f150f 100644\n--- a/src/ipa/libipa/fixedpoint.h\n+++ b/src/ipa/libipa/fixedpoint.h\n@@ -16,55 +16,6 @@ namespace libcamera {\n \n namespace ipa {\n \n-#ifndef __DOXYGEN__\n-template &&\n-\t\t\t std::is_floating_point_v> * = nullptr>\n-#else\n-template\n-#endif\n-constexpr R floatingToFixedPoint(T number)\n-{\n-\tstatic_assert(sizeof(int) >= sizeof(R));\n-\tstatic_assert(I + F <= sizeof(R) * 8);\n-\n-\t/*\n-\t * The intermediate cast to int is needed on arm platforms to properly\n-\t * cast negative values. See\n-\t * https://embeddeduse.com/2013/08/25/casting-a-negative-float-to-an-unsigned-int/\n-\t */\n-\tR mask = (1 << (F + I)) - 1;\n-\tR frac = static_cast(static_cast(std::round(number * (1 << F)))) & mask;\n-\n-\treturn frac;\n-}\n-\n-#ifndef __DOXYGEN__\n-template &&\n-\t\t\t std::is_integral_v> * = nullptr>\n-#else\n-template\n-#endif\n-constexpr R fixedToFloatingPoint(T number)\n-{\n-\tstatic_assert(sizeof(int) >= sizeof(T));\n-\tstatic_assert(I + F <= sizeof(T) * 8);\n-\n-\tif constexpr (std::is_unsigned_v)\n-\t\treturn static_cast(number) / static_cast(T{ 1 } << F);\n-\n-\t/*\n-\t * Recreate the upper bits in case of a negative number by shifting the sign\n-\t * bit from the fixed point to the first bit of the unsigned and then right shifting\n-\t * by the same amount which keeps the sign bit in place.\n-\t * This can be optimized by the compiler quite well.\n-\t */\n-\tint remaining_bits = sizeof(int) * 8 - (I + F);\n-\tint t = static_cast(static_cast(number) << remaining_bits) >> remaining_bits;\n-\treturn static_cast(t) / static_cast(1 << F);\n-}\n-\n template\n struct FixedPointQTraits {\n private:\n@@ -97,11 +48,23 @@ public:\n \n \tstatic constexpr float toFloat(QuantizedType q)\n \t{\n-\t\treturn fixedToFloatingPoint(q);\n+\t\tif constexpr (std::is_unsigned_v)\n+\t\t\treturn static_cast(q) / static_cast(UT{ 1 } << F);\n+\n+\t\t/*\n+\t\t * Recreate the upper bits in case of a negative number by\n+\t\t * shifting the sign bit from the fixed point to the first bit\n+\t\t * of the unsigned and then right shifting by the same amount\n+\t\t * which keeps the sign bit in place. This can be optimized by\n+\t\t * the compiler quite well.\n+\t\t */\n+\t\tunsigned int remaining_bits = sizeof(UT) * 8 - (I + F);\n+\t\tT t = static_cast(static_cast(q) << remaining_bits) >> remaining_bits;\n+\t\treturn static_cast(t) / static_cast(UT{ 1 } << F);\n \t}\n \n-\tstatic constexpr float min = fixedToFloatingPoint(static_cast(qMin));\n-\tstatic constexpr float max = fixedToFloatingPoint(static_cast(qMax));\n+\tstatic constexpr float min = toFloat(qMin);\n+\tstatic constexpr float max = toFloat(qMax);\n \n \tstatic_assert(min < max, \"FixedPointQTraits: Minimum must be less than maximum\");\n \n@@ -109,7 +72,13 @@ public:\n \tstatic QuantizedType fromFloat(float v)\n \t{\n \t\tv = std::clamp(v, min, max);\n-\t\treturn floatingToFixedPoint(v);\n+\n+\t\t/*\n+\t\t * The intermediate cast to int is needed on arm platforms to\n+\t\t * properly cast negative values. See\n+\t\t * https://embeddeduse.com/2013/08/25/casting-a-negative-float-to-an-unsigned-int/\n+\t\t */\n+\t\treturn static_cast(static_cast(std::round(v * (1 << F)))) & bitMask;\n \t}\n };\n \n","prefixes":["v8","15/15"]}