{"id":27026,"url":"https://patchwork.libcamera.org/api/covers/27026/?format=json","web_url":"https://patchwork.libcamera.org/cover/27026/","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":"<20260624085849.873784-1-bryan.odonoghue@linaro.org>","date":"2026-06-24T08:58:39","name":"[00/10] libcamera: software_isp: gpu: Add go faster stripes","submitter":{"id":175,"url":"https://patchwork.libcamera.org/api/people/175/?format=json","name":"Bryan O'Donoghue","email":"bryan.odonoghue@linaro.org"},"mbox":"https://patchwork.libcamera.org/cover/27026/mbox/","series":[{"id":6015,"url":"https://patchwork.libcamera.org/api/series/6015/?format=json","web_url":"https://patchwork.libcamera.org/project/libcamera/list/?series=6015","date":"2026-06-24T08:58:39","name":"libcamera: software_isp: gpu: Add go faster stripes","version":1,"mbox":"https://patchwork.libcamera.org/series/6015/mbox/"}],"comments":"https://patchwork.libcamera.org/api/covers/27026/comments/","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 B9411C3306\n\tfor ;\n\tWed, 24 Jun 2026 08:59:03 +0000 (UTC)","from lancelot.ideasonboard.com (localhost [IPv6:::1])\n\tby lancelot.ideasonboard.com (Postfix) with ESMTP id 3CF8565883;\n\tWed, 24 Jun 2026 10:59:02 +0200 (CEST)","from mail-wm1-x32f.google.com (mail-wm1-x32f.google.com\n\t[IPv6:2a00:1450:4864:20::32f])\n\tby lancelot.ideasonboard.com (Postfix) with ESMTPS id EE01565870\n\tfor ;\n\tWed, 24 Jun 2026 10:58:59 +0200 (CEST)","by mail-wm1-x32f.google.com with SMTP id\n\t5b1f17b1804b1-490b613a17bso7035105e9.3\n\tfor ;\n\tWed, 24 Jun 2026 01:58:59 -0700 (PDT)","from inspiron14p-linux ([109.76.100.231])\n\tby smtp.gmail.com with ESMTPSA id\n\t5b1f17b1804b1-49261063d6esm25128375e9.2.2026.06.24.01.58.58\n\t(version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256);\n\tWed, 24 Jun 2026 01:58:58 -0700 (PDT)"],"Authentication-Results":"lancelot.ideasonboard.com; dkim=pass (2048-bit key;\n\tunprotected) header.d=linaro.org header.i=@linaro.org\n\theader.b=\"YXCEWsgw\"; dkim-atps=neutral","DKIM-Signature":"v=1; a=rsa-sha256; c=relaxed/relaxed;\n\td=linaro.org; s=google; t=1782291539; x=1782896339;\n\tdarn=lists.libcamera.org; \n\th=content-transfer-encoding:mime-version:message-id:date:subject:cc\n\t:to:from:from:to:cc:subject:date:message-id:reply-to;\n\tbh=9gFcpaqvSRqNN016whMBLoM9egWOmGYEXP5jdt0ONMk=;\n\tb=YXCEWsgwfgoWDh8OuNyltb/6h4J2taBwrwR255h1mqBX2qHtSx1o+xxoq5q/sD2IoH\n\trppZPEZdZf47VWYVmFjPNV8j0q7xLNHnBBTZpE/2AZ/7Ya6dR3SpJfI6Mo6ScApt5MaS\n\t7UdzMVPpNVuLnelYwucnKkORNsEZTq73jGpJm0YbM1YRl6z8+Z0KAHEIaZEZMzUoJ1Nj\n\tYrxmQAhQN85BJyb4wWwsKcGea3NtjvRI5sQ7AWSoZGyPJTtCfj5h03fLPnNRv2wmb9K0\n\t+IKv31a//y44jqEW7KqkKDRZtvr33rJEslBLuuEtCqoH9zapXd0Jz75LpznIaiWHy9hl\n\tWfJg==","X-Google-DKIM-Signature":"v=1; a=rsa-sha256; c=relaxed/relaxed;\n\td=1e100.net; s=20251104; t=1782291539; x=1782896339;\n\th=content-transfer-encoding:mime-version:message-id:date:subject:cc\n\t:to:from:x-gm-gg:x-gm-message-state:from:to:cc:subject:date\n\t:message-id:reply-to;\n\tbh=9gFcpaqvSRqNN016whMBLoM9egWOmGYEXP5jdt0ONMk=;\n\tb=YkSpTgJZruxHZI8TOOxk5wO1ahZQHSeqV418jHbmMBdYadyrF/5ADkdH1WeeqqfakO\n\tQVRtMuvr3iNdwb2X2vIhWn6/o/K2wlMZczUZXwMU1tM/ILXQiB5oGQ3sDS6EWovx/eHW\n\tnaAoQ0i9nrFHdbW62QvrMBb5SwbtA1rYDctu3WcgxdiHmoI0wIU/Xw8FuXsqbWRdYKC6\n\t29gfMwglEEeTwFtJkI7RQKHHSCDY9OIS2FAdoIlMtvLoTS3OpBAGz/4fErQTflV8sa+h\n\tEePvGoVoeYs0FV5+mt3Rfny0TxElRkcop+VYpuYoD9skzgY3bbpPB+boZ/Nt3S7+4izf\n\tjTPg==","X-Gm-Message-State":"AOJu0YyOERQqkllRyNcP1asXmzFTe4BEz2l9v1u6w+oozZLJR4x3riXG\n\tOB+hbqs4leB/h9bnkSkabAzxv7zcAqX9RtxjdmfAAPmpDhijQQoKu5ZAjyd3wSpWeV7WW4C5vpk\n\t5ZrUEg1k=","X-Gm-Gg":"AfdE7cnCg+hrGY/nbrEj4LqwLEkmbOCKCG1sHbJDaFm/Y0aeS+zG51+TcZEJnpf1jqH\n\t1P6SgJY14kdPo9R8jAP9yd7KI5vr8ootBPbrq/bh7oMv74Wucw32eUYKZfcvJwODrI9kneR8AlL\n\tgUWuMyAdhCLtNaWxjWbnQYqz0fnHSv9wZA8SCis7rs386OdD2M/Yij8haCd2ytF26fV7OEBX5J0\n\ti05sr+8BB/Y5c5fzcnBo/Pm+pQ49zrzW/F9WMkZdLdgXFehXe5GlnuNH4BR6wlwDcJYt+gbXWPI\n\t9njc/6E7+qMLgFhOs9k41ytWWLmYhv0CAojTA/ECY94sxuICg6KX6qd8LGrheaN1AaXdVjAEEFz\n\tvCyLjzq6opssN/eXM3qnAlTRtbWskfeqdqlQVo+O5qWvrj3qCNIJfWfeAVJHts+1TjFl9wgy/kH\n\tyPgF9+sb/aWioUKp2UgxXaIZsYUpT9V1kkaGvfpbQ=","X-Received":"by 2002:a05:600c:6990:b0:490:b189:212d with SMTP id\n\t5b1f17b1804b1-492490b7ea3mr248052905e9.33.1782291539400; \n\tWed, 24 Jun 2026 01:58:59 -0700 (PDT)","From":"Bryan O'Donoghue ","To":"libcamera-devel@lists.libcamera.org","Cc":"bryan.odonoghue@linaro.org,\n\tpavel@ucw.cz","Subject":"[PATCH 00/10] libcamera: software_isp: gpu: Add go faster stripes","Date":"Wed, 24 Jun 2026 09:58:39 +0100","Message-ID":"<20260624085849.873784-1-bryan.odonoghue@linaro.org>","X-Mailer":"git-send-email 2.54.0","MIME-Version":"1.0","Content-Transfer-Encoding":"8bit","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":"Following on from Robert Mader's ask to bring forward GPUISP multi-pass\ncache operations to mainline first, I've done some work to enable that.\n\nThis series implements an input/output texture caching scheme which results\nin an overall absolute 4.5 millisecond - roughly improvement in performance\nper processed frame. This is a bit of an odd result as I was expecting to\nshave a particular % off of each frame in the order of 20% or so. My best\nguess is that particular paths we are optimising here are around texture\ngeneration and these are \"fixed costs\" on the CPU side.\n\nOne very welcome outcome of this series is genuine zero-copy on dma-buf\npaths we can support it on i.e. on paths where the CSI2 and GPU strides\nagree.\n\nThe numbers cited in my example are for non dma-buf handle upload and\ndma-buf handle render-to-texture on the output so, in fact I'd expect to\nsee a much larger improvement on systems where dma-buf handle is used on\nboth paths.\n\nAs it is on my test reference systems I have a 50% improvement per frame\nfor one system, a 20% improvement on another system or we could view it as\na fixed 4.5 millisecond improvement on both.\n\nI implemented the cache around V4L2BufferCache as per Barnabas' suggestion,\nincluding a fixed cache size.\n\nBryan O'Donoghue (10):\n libcamera: v4l2_videodevice: Output cache hit as a parameter\n libcamera: egl: Add new helper attachTextureToFBO\n libcamera: egl: Extend eGL::createTexture2D to understand floats\n libcamera: egl: Rename createTexture2D to createInputTexture2D\n libcamera: egl: Add createOutputTexture2D\n libcamera: egl: Add updateInputTexture2D\n libcamera: egl: Add activateBindTexture\n libcamera: software_isp: debayer_egl: Pass eglImage as parameter to\n setShaderVariables\n libcamera: software_isp: debayer_egl: Flag dmabuf use once per session\n not for every frame\n libcamera: software_isp: debayer_egl: Implement input/output frame\n caching mechanism\n\n include/libcamera/internal/egl.h | 7 +-\n include/libcamera/internal/v4l2_videodevice.h | 2 +-\n src/libcamera/egl.cpp | 126 +++++++++++++++---\n src/libcamera/software_isp/debayer_egl.cpp | 119 +++++++++++++----\n src/libcamera/software_isp/debayer_egl.h | 16 ++-\n src/libcamera/v4l2_videodevice.cpp | 9 +-\n 6 files changed, 228 insertions(+), 51 deletions(-)"}