Back to home page

DOS ain't dead

Forum index page

Log in | Register

Search:  
Back to the forum
Board view  Mix view

Re: UIDE Unloading And Cache-Sizes. (Announce)

posted by Rugxulo Homepage, Usono, 08.06.2010, 01:35

> > However, I still think 15 MB used of 32 MB is too much, which you
> > suggested once; heck, I even tested it, couldn't find any speedup,
> > so I stick to 5 MB on that old clunker.
>
> "The theory" of UIDE's caches is that, if possible, you should set a
> cache TWICE the size of your largest data file, PLUS extra memory so
> a reasonable amount of DOS directories can remain "in cache" and not
> get discarded, when more new data arrives. Keeping DOS directories
> "in cache" is the GREATEST advantage of using UIDE, as DOS otherwise
> does only SLOW "designed in 1981" single-sector directory handling!
The problem is that backends for GCC 3.4.4 (which I use there) are 3.5 MB unpacked. (Latest DJGPP GCC 4.4.2 is 8.5 MB for CC1.EXE, yikes.) The 2.04 libc itself is 800k. And the assembler and linker (AS, LD) are also pretty big (800k, 600k). The internal Pentium cache is ridiculously small.

(EDIT: Looks like 3.4.4 CC1.EXE is actually 4.2 MB, I forgot I must've recompiled GCC 3.4.4 in blind hopes of speedup a while back.)

Anyways, I just now tried a bit messing with GPC. Seems that the linking stage takes the longest, so copying the relevant libs to RAM disk and using -L. speeds it up more than UIDE /S15 did. Hence, I'll stick to /S5 for now. (Plus I usually need the RAM for other stuff. And this is also why I like TDSK to be able to resize, as by default I use 5 MB for it too.)

> Using a 5-MB cache, if you have no files larger than about 2-MB, you
> can copy such files using 4-MB of cache (2 for input, 2 for output),
> and you still have about 1-MB left for directories. When DOS wants
> to copy another file or load a new program, it loses NO speed, since
> the directories were not discarded due to too-much "new" cache data.
Well, choice of software affects this, and you can indeed blame me for using either a "low RAM" cpu or bloated compiler or whatnot. I know TP55 and TMT are probably faster, but they aren't as good for some (most?) things. Haven't benchmarked FPC on that clunker yet (have to pare it down first, been busy with other things). They always claim faster compilation, so it would be nice to see if that's true. ;-)

Post reply
 

Complete thread:

Back to the forum
Board view  Mix view
22781 Postings in 2123 Threads, 402 registered users (0 online)
DOS ain't dead | Admin contact
RSS Feed
powered by my little forum