Metal and OpenCL are micro-architecture programming languages; one cannot be faster unless one provides more access to the hardware than the other.
A perfect example is provided by Quartz Composer, in that it uses a subset of OpenGL (OpenGL ES)— not the full language. As a consequence, it is the slowest fastest markup language that one could use for GPU programming. But, for every function it shares with its parent, and and any similar functions it shares with other languages using the same hardware, the speed is the same. The distinction between Metal and OpenCL will never amount to anything more significant and language design specifications. Metal is for game programming; OpenCL is a generic, all-encompassing open-source language. You can achieve the same thing with both, but it would take a lot of kludging with Metal to make it do anything but what it was designed for—games. It is Apple's answer to a similarly purposed language by Google, which is extremely popular among Android developers. So, as you can see, it's a left-field question to think of Quartz Composer has a part in any of that. It's a graphics tool that belongs to a package of many other such tools, and as useful as a pocket calculator. Small, fast, convenient, essential—but you put it away once you've made your calculations, right before you start building the real thing. Sent from my iPhone > On Jun 18, 2015, at 12:00 PM, quartzcomposer-dev-requ...@lists.apple.com > wrote: > > Send Quartzcomposer-dev mailing list submissions to > quartzcomposer-dev@lists.apple.com > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.apple.com/mailman/listinfo/quartzcomposer-dev > or, via email, send a message with subject or body 'help' to > quartzcomposer-dev-requ...@lists.apple.com > > You can reach the person managing the list at > quartzcomposer-dev-ow...@lists.apple.com > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Quartzcomposer-dev digest..." > > > Today's Topics: > > 1. Re: Quartzcomposer-dev Digest, Vol 119, Issue 1 (Roger Bolton) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Thu, 18 Jun 2015 05:55:59 +0700 > From: Roger Bolton <ro...@eskatonia.net> > To: James Bush <theokn...@icloud.com> > Cc: "quartzcomposer-dev@lists.apple.com" > <quartzcomposer-dev@lists.apple.com> > Subject: Re: Quartzcomposer-dev Digest, Vol 119, Issue 1 > Message-ID: <12b3d9bb-aff6-4721-9fac-4b40ac0b9...@eskatonia.net> > Content-Type: text/plain; charset=utf-8 > > >> On 18 Jun 2015, at 01:23, James Bush <theokn...@icloud.com> wrote: >> >> This is sort of an uninformed question, in that you don't seem to know the >> difference between Core Image, OpenGL (which you didn't mention, but >> undoubtedly meant). and Metal (there is no Quartz Composer API). >> >> Anyway, if you're looking for the absolute fastest performance from a Quartz >> Composition, use OpenGL (correctly) within a Core Image patch and/or OpenCL >> with its respective patch. >> >> There is no faster than either—not even Metal. > > Gee thanks for teaching me to suck eggs. There is a Quartz Composer API, the > API to create new nodes and patches in QC thats what I meant. Programmers > more knowledgeable than me seem to think that actually Metal may allow some > optimisations even faster than OpenCL. Maybe they are all wrong and you are > right? Let’s see. > > Roger > CoreMelt > > > ------------------------------ > > _______________________________________________ > Quartzcomposer-dev mailing list > Quartzcomposer-dev@lists.apple.com > https://lists.apple.com/mailman/listinfo/quartzcomposer-dev > > > End of Quartzcomposer-dev Digest, Vol 119, Issue 3 > **************************************************
_______________________________________________ Do not post admin requests to the list. They will be ignored. Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com