Code optimization chat

[xyzzy]

https://twitter.com/pomeranian99/status/858856994438094848

[agavi...@gmail.com]

Pragmatism or laziness.

[dotsla...@gmail.com]

I like how they calculated the rate of leak against maximum in-air time and added enough additional physical memory to support it.

Yo, old-tymers - wasn't memory pretty pricey back then?

Cheers.

[xyzzy]

Yeah but it was military spending so cheaper than a toilet seat.

[xyzzy]

On Monday, May 1, 2017 at 2:10:27 PM UTC-4, the_andr...@yahoo.com wrote:
> Pragmatism or laziness.

Since when are those different things?

[Con Reeder, unhyphenated American]

On 2017-05-01, xyzzy <xyzzy...@gmail.com> wrote:
> https://twitter.com/pomeranian99/status/858856994438094848

I guess you don't want this guy writing a daemon....

--
Traffic is like a bad dog. It isn't important to look both ways when
crossing the street. It's more important to not show fear.
-- P. J. O'Rourke

[Michael Press]

In article <5e3da08d-bbb9-42ce...@googlegroups.com>,

xyzzy <xyzzy...@gmail.com> wrote:

> https://twitter.com/pomeranian99/status/858856994438094848

The ideal of a general purpose memory allocator is unworkable.
The nihilistic scheme described is practicable and ugly.
There has to be a better way.

Memory leaks occur when some task is constantly allocating and
freeing memory. One slip and ssssst! Also you get memory fragmentation.

Divide the system into logical units. Each unit works out how
much memory it needs and allocates it all at the beginning. Then
it writes its own special purpose allocator.

--
Michael Press

[Some dued]

It'd be awfully wasteful for every application to claim its maximum possible memory requirement immediately upon launch. Also tools like web browsers are unpredictable as they depend on the sites, number of tabs etc...

[Michael Press]

In article <5afb59bd-da15-4da8...@googlegroups.com>,

Some dued <theodo...@gmail.com> wrote:

> It'd be awfully wasteful for every application to claim its maximum possible memory requirement immediately upon launch. Also tools like web browsers are unpredictable as they depend on the sites, number of tabs etc...

First of all it is virtual.
Second, a unit has other constraints besides memory
that give a good idea of how much memory it will
need. If a logical unit does not know how much memory
it will need, how will anybody else?

You have a point with web browsers. I'm out of ideas there.

--
Michael Press

[Some dued]

Virtual memory only means it spills onto the hard drive, its really slow. Every time I open word would it reserve enough space for 10000 highly formatted pages? And Excel enough for a 64kX64k grid of cells with formulas and formatted text?

[xyzzy]

On Tuesday, May 2, 2017 at 1:01:37 AM UTC-4, Michael Press wrote:
> In article <5afb59bd-da15-4da8...@googlegroups.com>,
> Some dued <theodo...@gmail.com> wrote:
>
> > It'd be awfully wasteful for every application to claim its maximum possible memory requirement immediately upon launch. Also tools like web browsers are unpredictable as they depend on the sites, number of tabs etc...
>
> First of all it is virtual.

You are saying virtual like it means cost-free. If that was the case there would be no need to pre-allocate memory, because none of the later allocations would fail.

> Second, a unit has other constraints besides memory
> that give a good idea of how much memory it will
> need. If a logical unit does not know how much memory
> it will need, how will anybody else?

A networking stack (which I work on) can have widely varying memory needs based on traffic (including what the web browsers are doing). We do preallocate chunks at a time and dole them out to components in pieces, but not all at once.

And memory leaks suck. The only thing harder to debug is overlays.