Recherche avancée

Sur d’autres sites (7269)

• Parallel transcoding with FFmpeg on M1 Mac

  27 août 2021, par Pushkar

  I'm trying to determine how effective an M1 Mac would be for transcoding video using FFmpeg (specifically resizing and adjusting bitrate). I can do single transcodes using a command like this :

  


  ffmpeg -I in.mp4 -nostdin -c:v h264_videotoolbox -c:a copy -vf scale=1280:720 -b:v 8000k out.mp4


  


  Now, when I run this, I can see the process in Activity Monitor, but it shows significant CPU use but 0% GPU use, although it's certainly using some form of hardware acceleration (using libx264 instead of h264_videotoolbox is much slower).

  


  When I try running multiple processes concurrently, the timings suggest little evidence of parallel execution :

  


  





  


  


  


  


  



  


  


  


  


  


  


  


  


  


  


  


  


  Test	Timing
  convert single test file	6.8s
  convert file 5 times sequentially	33.6s
  convert file 5 times in parallel	31.5s

  


  


  Since the M1 chip is supposed to have 7 or 8 GPUs inside it, I expected to see quite good parallelism, so are there options which I'm missing which would :
  

  


  

  1. ensure that the transcoding is actually running on the GPU ?

  


  2. allow parallel execution across the multiple GPUs ?

  


  


• Logic and lawyers

  22 mai 2013, par Mans — Law and liberty

  Reading about various patent litigation cases, I am struck by the frequency with which common logical fallacies such as the Appeal to Consequences are committed. We shall look at a couple of recent examples.

  In conjunction with the Federal Circuit ruling in CLS Bank v. Alice Corp., Judge Moore, joined by three others, filed a dissenting opinion wherein we find the following :

  I am concerned that the current interpretation of § 101, and in particular the abstract idea exception, is causing a free fall in the patent system. [...] And let’s be clear : if all of these claims, including the system claims, are not patent-eligible, this case is the death of hundreds of thousands of patents [...].

  A footnote adds :

  If the reasoning of Judge Lourie’s opinion were adopted, it would decimate the electronics and software industries. [...] There has never been a case which could do more damage to the patent system than this one.

  From the above, I get the impression Moore is primarily concerned with protecting the system, maintaining the status quo, less with ruling in line with the logical consequences of statute and case law. Furthermore, her argument rests on the premise that a weaker patent system would “decimate the industries,” a notion supported by little evidence, yet presented by Moore as an obvious truth. In fact, research exists suggesting that many important innovations are never actually patented. Let us also not overlook the fact that European companies do not appear to be suffering from the much weaker patent protection for software afforded there.

  Judge Moore’s reasoning can be summarised in three steps :

  1. Ruling this way could be disruptive to the patent system.
  2. The industry relies on patents.
  3. Therefore we must not rule this way.

  Not only does she commit the aforementioned logical fallacy, she does so by way of invalid arguments.

  The second example of such fallacious reasoning comes from the Supreme Court ruling in Bowman v. Monsanto :

  We have always drawn the boundaries of the exhaustion doctrine to exclude that activity [copying], so that the patentee retains an undiminished right to prohibit others from making the thing his patent protects. [...] That is because, once again, if simple copying were a protected use, a patent would plummet in value after the first sale of the first item containing the invention. The undiluted patent monopoly, it might be said, would extend not for 20 years (as the Patent Act promises), but for only one transaction. And that would result in less incentive for innovation than Congress wanted. Hence our repeated insistence that exhaustion applies only to the particular item sold, and not to reproductions.

  Here we find the same pattern repeated. The aim of the court appears to have been ensuring the continued validity of this class of patents, not reaching a logical conclusion regarding the question of infringement. Once again, we can break the reasoning down into three steps :

  1. A non-infringement ruling would weaken the patent.
  2. Weaker patents would provide less incentive for innovation.
  3. Therefore we must rule infringement.

  As in the first example, the argument presented in step two is at best questionable, and no supporting evidence is provided.

  These are, unfortunately, not the only examples of such fallacies ; one might even describe them as ubiquitous. Does a law education not include any material on logical reasoning ? Ought it not ? While we can never hope to find any kind of universal truth on which to base our laws, we should at least strive to make our system logically consistent. If we do not, notions such as fairness and justice lose their meanings.

• Anomalie #3418 : Les tables des plugins ne s’installent pas

  5 avril 2015, par Franck Dalot

  Nouvelle série de test :

  SPIP 3.0.18-dev [21967]
  Installation en MySQL

  Le plugin était Agenda avec le Mini Calendrier
  Les tables étaient bien présente à chaque fois
  spip_evenements
  spip_evenements_participants

  https://www.ovh.com/fr/g1207.configurer-php-web
  5.6.6 en mode prod = ok
  5.5.22 en mode prod = ok
  5.4.38 en mode prod = ok
  5.3.29 en mode prod = ok

  5.6.6 en mode dev = ok
  5.5.22 en mode dev = ok
  5.4.38 en mode dev = ok
  5.3.29 en mode dev = ok

  Ce qui est bizarre, c’est que b_b semble avoir eu un problème en php 5.4 et prod, j’ai peur que cela soit un bug tordu...