• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Re: Distributed Builds
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Distributed Builds


  • Subject: Re: Distributed Builds
  • From: Thomas Davie <email@hidden>
  • Date: Mon, 14 Mar 2005 17:39:37 +0000


On Mar 14, 2005, at 4:32 PM, Dix Lorenz wrote:


On 14.03.2005, at 16:57, Thomas Davie wrote:
But if your machine can't run two compiles in a row faster than running one compile and shipping the other one out, then you don't want it to be forced to run two compiles in a row – that negates your assertion that it's faster to run both compiles on your local machine.

I'm not sure I understand what you mean. In this scenario there were quite a few files to build so every machine got its part of the work and now the local machine is finished and waiting for some remote machine. All I am saying is that at this point the remote machine(s) has/have already started compiling the missing file(s) and assuming the machines have comparable speeds there is no point in starting local too because the local machine has no way of catching up, even counting that the remote result has to transfer back to the local machine.


Just as an aside: Actually there is a bug in gcc which some part of my code hits which leads to abnormal compile times for 1 certain file (10 mins) and in that case my local machine would be able to catch up. But the solution here is of course to fix the compiler...

I don't care so much about when compiling lots of files, then the fastest machine will get the most work done anyway. It's when I am compiling one (or two) files that local compiling would be faster. It saves a few MB of network traffic, the remote machine might be busy doing something else... I would think most of the time and in most setups if you are compiling only one or two files it is faster to do it locally and as it also won't hurt if you compile a lot of files I'd say the local machine should get first shot. Also it needs almost no change, just some way of setting the order in which machines are used...
That's exactly my point - if you're compiling lots of files then you don't care, it doesn't matter if you spend a little bit of CPU duplicating effort at the end. What does matter is if you have only 2 files to compile, and you would normally say, I'll build the first one, the second one gets shipped off. You make the assertion that it would be faster to compile both files locally rather than ship one off. If this is the case, then we can ship one off, and compile one locally, when that finishes, compile the one we shipped off locally... If it finishes first, then woo yay we've cut compile time, if it doesn't finish first then we would have been faster distributing it anyway.

Bob

_______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list      (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden


  • Follow-Ups:
    • Re: Distributed Builds
      • From: Dix Lorenz <email@hidden>
References: 
 >Distributed Builds (From: Dix Lorenz <email@hidden>)
 >Re: Distributed Builds (From: Rob Frohne <email@hidden>)
 >Re: Distributed Builds (From: Thomas Davie <email@hidden>)
 >Re: Distributed Builds (From: Dix Lorenz <email@hidden>)
 >Re: Distributed Builds (From: Thomas Davie <email@hidden>)
 >Re: Distributed Builds (From: Dix Lorenz <email@hidden>)

  • Prev by Date: Re: "Hardware" System Preferences
  • Next by Date: Overriding built-in dylibs?
  • Previous by thread: Re: Distributed Builds
  • Next by thread: Re: Distributed Builds
  • Index(es):
    • Date
    • Thread