Re: Algorithm Optimization

"Derek M. Jones" <derek@_NOSPAM_knosof.co.uk>
Wed, 16 Sep 2020 20:11:04 +0100

          From comp.compilers

Related articles
Algorithm Optimization rick.c.hodgin@gmail.com (Rick C. Hodgin) (2020-09-13)
Re: Algorithm Optimization elronnd@elronnd.net (Elijah Stone) (2020-09-14)
Re: Algorithm Optimization rick.c.hodgin@gmail.com (Rick C. Hodgin) (2020-09-15)
Re: Algorithm Optimization derek@_NOSPAM_knosof.co.uk (Derek M. Jones) (2020-09-15)
Re: Algorithm Optimization gah4@u.washington.edu (gah4) (2020-09-15)
Re: Algorithm Optimization mwmarkland@gmail.com (mwmarkland@gmail.com) (2020-09-16)
Re: Algorithm Optimization rick.c.hodgin@gmail.com (Rick C. Hodgin) (2020-09-16)
Re: Algorithm Optimization derek@_NOSPAM_knosof.co.uk (Derek M. Jones) (2020-09-16)
Re: Algorithm Optimization gah4@u.washington.edu (gah4) (2020-09-16)
Re: Algorithm Optimization richard.nospam@gmail.com (Richard Harnden) (2020-09-16)
Re: Algorithm Optimization DrDiettrich1@netscape.net (Hans-Peter Diettrich) (2020-09-17)
Re: Algorithm Optimization tkoenig@netcologne.de (Thomas Koenig) (2020-09-17)
Re: Algorithm Optimization minforth@arcor.de (A. K.) (2020-09-21)
| List of all articles for this month |

From: "Derek M. Jones" <derek@_NOSPAM_knosof.co.uk>
Newsgroups: comp.compilers
Date: Wed, 16 Sep 2020 20:11:04 +0100
Organization: virginmedia.com
References: 20-09-032 20-09-035 20-09-036
Injection-Info: gal.iecc.com; posting-host="news.iecc.com:2001:470:1f07:1126:0:676f:7373:6970"; logging-data="77785"; mail-complaints-to="abuse@iecc.com"
Keywords: optimize, performance
Posted-Date: 16 Sep 2020 17:34:22 EDT
In-Reply-To: 20-09-036
Content-Language: en-US

On 16/09/2020 06:25, gah4 wrote:


> Now, say someone is doing their CS project for class, where they are
> supposed to write, and time, bubblesort?


I don't think student projects should be a factor in implementation of
industrial compilers.


> I suppose you can find a Chebyshev series that closely approximates
> the series coded, but takes fewer terms.
>
> But what about the person who wants to compare two series'?
> If you replace one or both, then the comparison will be wrong.


More edge cases.


I'm sure you know about Herbie: https://herbie.uwplse.org/


> Not that there haven't been problems since the beginning of
> optimizing compilers, where the results were different than
> expected.


There probably won't be too many places where savings can be made.
The compiler could highlight them.


There are always people willing to pay for faster code, and there will
be compiler writers willing to implement go faster stuff.


> [Back when people cared about Whetstone and Dyrystone benchmarks,
> compilers recognized code sequences from those benchmarks for, uh,
> special processing. But it doesn't generalize very well. -John]


Intel and ??? have both been caught doing this. See chapter 13 oof
this book: http://www.knosof.co.uk/ESEUR


Benchmarking these days has gotten to be very unreliable:
https://shape-of-code.coding-guidelines.com/2015/02/24/hardware-variability-may-be-greater-than-algorithmic-improvement/


http://shape-of-code.coding-guidelines.com/2020/01/05/performance-variation-in-2386-identical-processors/


--
Derek M. Jones
blog:shape-of-code.coding-guidelines.com


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.