Thanks to Knuth SKA's Merging Stage 1 is 90% faster

[MarillionFan]

and breathe Atw. Breathe.

[AtW]

I understand most sorting algorithms: quicksort, mergesort, heapsort, radix sort. Yet I ALWAYS use a library.

What library? Its not like I am rewriting quicksort of anything - and I can and do benchmark before/after changes on realistic set of data, so with all due respect I know for fact if my code runs faster than that I used before.

I don't want to depend on anybody's library in such a key field for me - knowing how sorting works allows me to plan overall algorithms better and I am certainly going to get better performance than any free library out there, and I certainly ain't a sucker to pay through the nose for stuff that I can implement easily.

[Jabberwocky]

Jubber goes for the Jagular

What library? Its not like I am rewriting quicksort of anything - and I can and do benchmark before/after changes on realistic set of data, so with all due respect I know for fact if my code runs faster than that I used before.

I don't want to depend on anybody's library in such a key field for me - knowing how sorting works allows me to plan overall algorithms better and I am certainly going to get better performance than any free library out there, and I certainly ain't a sucker to pay through the nose for stuff that I can implement easily.

Really, is your code peer reviewed ? Better performance as measured by whom ? What sort of testing do you employ ?

The last time you pasted some code on here it was a crock of sh*te, as I pointed out at the time. Just because you bought three books doesn't make it any better.

[Fungus]

Really, is your code peer reviewed ? Better performance as measured by whom ? What sort of testing do you employ ?

The last time you pasted some code on here it was a crock of sh*te, as I pointed out at the time. Just because you bought three books doesn't make it any better.

But you must admit that they look good on his bookshelf.

A considerable amount of work goes into developing algorithms and optimising libraries. I used to use the NAG libraries as they were much better than anything I could develop. The secret is knowing a) which algorithm to use and b) knowing the limitations of the library. The problem with free libraries is that they are often not worth the money you pay for them.

I did once write a LZW compression routine. It certainly did compress, but was not as reversible as I would have liked.

I also use the STL libraries. They work well, though for ultimate speed it's best to code from scratch. And I do my own socket coding as it is better than any library I've yet come across.

[AtW]

Really, is your code peer reviewed ? Better performance as measured by whom ? What sort of testing do you employ ?

Its measured by me - I have so much data that server can run for weeks crunching it, so when run time drops from an hour to 6 minutes I know for fact that it was improved.

Best of all I am the person whose opinion matters.

Fungus - show me library that efficiently sorts billions of custom data types that may have their own unique to them comparison algorithms.

[AtW]

And I do my own socket coding as it is better than any library I've yet come across.

You use UDP?

[ControlG]

Knuth

As Bill Gates, himself, said : -

"If you think you're a really good programmer. . . read [Knuth's] Art of Computer Programming.... You should definitely send me a resume if you can read the whole thing."

Standard university texts when I was there (1980-1983). Stuck on my bookshelf ever since.

[Fungus]

You use UDP?

TCPIP.

[Churchill]

Indeed - they are my kind of people, pioneers who may have ultimately failed but they played their role and will always be remembered as pioneers.

And what you have achieved today?

AtW, you obviously don't know the history of PSION, they're a cunch of arrogant bunts. Not only that, it's only themselves that think they're any good.

[AtW]

AtW, you obviously don't know the history of PSION

I know that they pioneered palmtops, made mistakes and got overtaken by another pioneer - Palm, who in turn got whacked and now shadow of itself.

They may have all failed but they were also pioneers recorded in history of computing.