Recherche avancée

Sur d’autres sites (9932)

• swscale/aarch64 : use multiply accumulate and increase vector factor to 4

  17 novembre 2019, par Sebastian Pop

  swscale/aarch64 : use multiply accumulate and increase vector factor to 4
  This patch implements ff_hscale_8_to_15_neon with NEON fused multiply accumulate
  
  and bumps the vectorization factor from 2 to 4.
  
  The speedup is of 25% on Graviton1 A1 instances based on A-72 cpus :
  $ ffmpeg -nostats -f lavfi -i testsrc2=4k:d=2 -vf bench=start,scale=1024x1024,bench=stop -f null -
  
  before : t:0.040303 avg:0.040287 max:0.040371 min:0.039214
  
  after :  t:0.032168 avg:0.032215 max:0.033081 min:0.032146
  The speedup is of 39% on Graviton2 m6g instances based on Neoverse-N1 cpus :
  
  $ ffmpeg -nostats -f lavfi -i testsrc2=4k:d=2 -vf bench=start,scale=1024x1024,bench=stop -f null -
  
  before : t:0.019446 avg:0.019423 max:0.019493 min:0.019181
  
  after :  t:0.014015 avg:0.014096 max:0.015018 min:0.013971
  Tested with `make check` on aarch64-linux.
  Signed-off-by : Sebastian Pop <spop@amazon.com>
  
  Reviewed-by : Jean-Baptiste Kempf <jb@videolan.org>
  
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libswscale/aarch64/hscale.S

• Reason for write EPIPE error in my implementation ?

  25 mai 2019, par Chrisl446

  I currently have a small node.js express application for uploading images using sharp & thumbnails created from .mp4 videos using simple-thumbnail package/ffmpeg.

  The app works perfectly when uploading an image, the file is uploaded sharp processes it, then passes it along and it ends up in my amazon s3 bucket as expected. No errors what so ever.

  However, when I upload an mp4 video that I use to create and upload a thumbnail from by using simple-thumbnails genThumbnail() function, which uses ffmpeg child process, the thumbnail uploads successully to my s3 bucket, HOWEVER my app returns an EPIPE write error and NOT the url of the uploaded files url on S3.

  What is causing this and how can I fix it, considering it’s pretty much working aside from the EPIPE error that is being returned ? Thanks ahead !

  The packages of concern used are as follows :

  aws-sdk

  multer

  multer-s3 <- this one with the transform option, not the standard multer-s3 package

  simple-thumbnail

  const express = require('express');
  
  const app = express();
  
  
  
  const aws = require('aws-sdk');
  
  const multer = require('multer');
  
  const multerS3 = require('multer-s3'); //github:gmenih341/multer-s3 version of multer-s3 with transform option
  
  const sharp = require('sharp');
  
  const genThumbnail = require('simple-thumbnail');
  
  
  
  app.use((req, res, next) => {
  
      res.header('Access-Control-Allow-Origin', '*');
  
      res.header('Access-Control-Allow-Headers', 'Orgin, X-Requested-With, Content-Type, Accept, Authorization');
  
      if (req.method === 'OPTIONS') {
  
          res.header('Access-Control-Allow-Methods', 'POST');
  
          return res.status(200).json({});
  
      }
  
      next();
  
  });
  
  
  
  let uniqueFileName;
  
  let s3BucketName = 'bucketname';
  
  
  
  let s3 = new aws.S3({
  
      accessKeyId: ACCESS_KEY,
  
      secretAccessKey: SECRET_KEY,
  
      Bucket: s3BucketName
  
  });
  
  
  
  let upload = multer({
  
      storage: multerS3({
  
          s3: s3,
  
          bucket: s3BucketName,
  
          acl: 'public-read',
  
          cacheControl: 'max-age=31536000',
  
          contentType: multerS3.AUTO_CONTENT_TYPE,
  
          shouldTransform: true,
  
          transforms: [{
  
              id: 'thumbnail',
  
              key: function (req, file, cb) {
  
                  uniqueFileName = Date.now().toString();
  
                  cb(null, uniqueFileName + '.jpg')
  
              },
  
              transform: function (req, file, cb) {
  
                  if (file.mimetype == 'video/mp4') {
  
                      //When using simple-thumbnails' getThumbnail() on an mp4 video it uploads succesfully to S3 but node returns EPIPE write error
  
                      cb(null, genThumbnail(null, null, '250x?'))
  
                  } else {
  
                      //When using sharp to resize an image this works perfectly and retuns the JSON below with the files S3 URL
  
                      cb(null, sharp().jpeg())
  
                  }
  
  
  
              }
  
          }]
  
      })
  
  });
  
  
  
  app.post('/upload', upload.array('theFile'), (req, res) => {
  
      res.json({
  
          fileS3Url: 'https://s3.amazonaws.com/'+ s3BucketName +'/' + uniqueFileName
  
      });
  
  });
  
  
  
  app.use((req, res, next) => {
  
      const error = new Error('Not found');
  
      error.status = 404;
  
      next(error);
  
  });
  
  
  
  app.use((error, req, res, next) => {
  
      res.status(error.status || 500);
  
      res.json({
  
          error: {
  
              message: error.message
  
          }
  
      });
  
  });
  
  
  
  module.exports = app;

• swscale/aarch64 : use multiply accumulate and shift-right narrow

  9 décembre 2019, par Sebastian Pop

  swscale/aarch64 : use multiply accumulate and shift-right narrow
  This patch rewrites the innermost loop of ff_yuv2planeX_8_neon to avoid zips and
  
  horizontal adds by using fused multiply adds. The patch also uses ld1r to load
  
  one element and replicate it across all lanes of the vector. The patch also
  
  improves the clipping code by removing the shift right instructions and
  
  performing the shift with the shift-right narrow instructions.
  I see 8% difference on an m6g instance with neoverse-n1 CPUs :
  
  $ ffmpeg -nostats -f lavfi -i testsrc2=4k:d=2 -vf bench=start,scale=1024x1024,bench=stop -f null -
  
  before : t:0.014015 avg:0.014096 max:0.015018 min:0.013971
  
  after :  t:0.012985 avg:0.013013 max:0.013996 min:0.012818
  Tested with `make check` on aarch64-linux.
  Signed-off-by : Sebastian Pop <spop@amazon.com>
  
  Reviewed-by : Clément Bœsch <u@pkh.me>
  
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libswscale/aarch64/output.S