Comments on: Supero: Making Haskell Faster http://enfranchisedmind.com/blog/posts/supero-making-haskell-faster/ programming, politics, & other religious issues Fri, 12 Mar 2010 09:19:33 +0000 hourly 1 By: Upped The Recent Post/Popular Post Widget Count | Enfranchised Mind http://enfranchisedmind.com/blog/posts/supero-making-haskell-faster/#comment-33662 Upped The Recent Post/Popular Post Widget Count | Enfranchised Mind Tue, 12 Aug 2008 19:31:30 +0000 http://enfranchisedmind.com/blog/2007/12/09/supero-making-haskell-faster/#comment-33662 [...] Supero: Making Haskell Faster [...] [...] Supero: Making Haskell Faster [...]

]]> By: Enfranchised Mind » Supero Followup http://enfranchisedmind.com/blog/posts/supero-making-haskell-faster/#comment-32022 Enfranchised Mind » Supero Followup Wed, 12 Dec 2007 23:04:06 +0000 http://enfranchisedmind.com/blog/2007/12/09/supero-making-haskell-faster/#comment-32022 [...] CandidateDevelopment Acceleration: The Second Derivative of FunctionalityCertification? Please, no.Supero: Making Haskell FasterCompassionate [...] [...] CandidateDevelopment Acceleration: The Second Derivative of FunctionalityCertification? Please, no.Supero: Making Haskell FasterCompassionate [...]

]]> By: Neil Mitchell http://enfranchisedmind.com/blog/posts/supero-making-haskell-faster/#comment-31997 Neil Mitchell Tue, 11 Dec 2007 10:40:51 +0000 http://enfranchisedmind.com/blog/2007/12/09/supero-making-haskell-faster/#comment-31997 I wasn't going to post up my new draft paper, but following all the interest, I think it makes sense if everyone can see the latest and greatest supero stuff. That paper does cover more of the details - the original paper was written with a more limited knowledge of existing supercompilation, the revised paper builds on existing work so is more powerful. I wasn’t going to post up my new draft paper, but following all the interest, I think it makes sense if everyone can see the latest and greatest supero stuff. That paper does cover more of the details – the original paper was written with a more limited knowledge of existing supercompilation, the revised paper builds on existing work so is more powerful.

]]> By: augustss http://enfranchisedmind.com/blog/posts/supero-making-haskell-faster/#comment-31996 augustss Tue, 11 Dec 2007 10:23:09 +0000 http://enfranchisedmind.com/blog/2007/12/09/supero-making-haskell-faster/#comment-31996 By saying it has been done before I didn't mean to belittle Neil's work. The main idea might be old, but the devil is in the detail. Doing it efficiently and well for a real language is not easy, and I hope Neil will succeed. By saying it has been done before I didn’t mean to belittle Neil’s work. The main idea might be old, but the devil is in the detail. Doing it efficiently and well for a real language is not easy, and I hope Neil will succeed.

]]> By: Neil Mitchell http://enfranchisedmind.com/blog/posts/supero-making-haskell-faster/#comment-31987 Neil Mitchell Mon, 10 Dec 2007 18:09:49 +0000 http://enfranchisedmind.com/blog/2007/12/09/supero-making-haskell-faster/#comment-31987 As augustss says, it has been done before, but its a while since it's been applied to a full scale functional language. It is indeed positive supercompilation. The original paper that is on my website (http://www-users.cs.york.ac.uk/~ndm/supero/) follows the ideas of supercompilation relatively closely. The new paper which I've just submitted makes a few more changes from the traditional supercompilation approach, which are needed in a larger programming language. It is quite different from C program optimisation and even optimisation in a GHC/Clean style manner. Those make small incremental changes to the program, moving from one program to another very similar but slightly faster program - then these changes are iterated a lot. Supercompilation starts at one program, and ends at another program, without any kind of step-by-step process. It's nice to see someone is interested :) As augustss says, it has been done before, but its a while since it’s been applied to a full scale functional language. It is indeed positive supercompilation. The original paper that is on my website (http://www-users.cs.york.ac.uk/~ndm/supero/) follows the ideas of supercompilation relatively closely. The new paper which I’ve just submitted makes a few more changes from the traditional supercompilation approach, which are needed in a larger programming language.

It is quite different from C program optimisation and even optimisation in a GHC/Clean style manner. Those make small incremental changes to the program, moving from one program to another very similar but slightly faster program – then these changes are iterated a lot. Supercompilation starts at one program, and ends at another program, without any kind of step-by-step process.

It’s nice to see someone is interested :)

]]>