begin  quoting Gabriel Sechan as of Mon, Jul 11, 2005 at 05:52:07PM -0500:
> 
> >From: Stewart Stremler <[EMAIL PROTECTED]>
[snip]
> >Where do we see performance issues, anyway? UI responsiveness. Program
> >loading. Program processing. IO throughput. What else? Can we nail down
> >all the bits of percieved performance?
> >
> I do get complaints about UI responsiveness, but it tends not to eb when 
> they're typing-  it tends to be when they just hit save, or are doing 
> something else computationally or IOwise in the background.

Why should those things affect UI responsiveness?  Sure, when you save,
you don't want the user to save _again_ before you've finished. But that
can be done by greying out the Save... and Save As... menu-items and
remapping <mod>-S to <beep> -- there's no need to lock up the rest of 
the app while a save is underway, is there?

Same for processing... So I _do_ something computationally intensive.
Why should that affect the UI? The processor's idle time goes from 99%
to 1%, say, but that's a poor reason for my mouse to start moving 
jerkily.

> I also hear a lot of complaints about network lag.  The problem is that 
> when displaying website X, they get the lag to download it, parse and 
> render it, and parse and render any images and java/flash.  This is a huge 
> one.
>
> I'd say for the most part it has to do with startup of an IO or 
> computationally expensive process, in combination with other work.

I find startup lags to be annoying when there's no feedback.  And that's
software design, really.

[snip]
> No, but I don't think 2-3 year old PCs are excessively old.  Not many 
> people are willing to put out a 1K investment for a less than 3 year 
> return.  Even as a gamer I expect 2 years between upgrades, I refuse to do 
> it more often.

Yup.

My main machine is a 500MHz Blade 100.  It works fine.  I just couldn't
handle the typical PC upgrade cycle. (Plus, the high turnover rate
needed to keep up the cashflow to PC vendors is REALLY BAD for the
environment, AFAIAC.)

[snip]
> >So nothing ever is "fast enough" for you? :)
> >
> >Computers got "fast enough" for me a long time ago. Of course, with the
> >way things are going, I'm going to have to chase the bleeding edge, else
> >I'll once again be able to type faster than my word-processing
> >application....
> 
> Its possible to be fast enough, but bytecode interpreting kills way too 
> much performance.  Virtualizing the processor would do the same.

Virtualizing shouldn't be _that_ much of a hit; it would be drowned out
by the time-slicing effects anyway.  Assuming equal loading, you'd turn
a 2GHz machine into two something-less-than-1GHz machines... then again,
as most of the time machines are not running at top speed, that should
not be THAT noticable.

> As for fast enough- depends on your needs.  For email, web (non-flash), and 
> word processing, it was fast enough 7 or 8 years ago.  Of course, these 

Twenty years ago I couldn't out-type my computer....

> days a computer like that would have problems web surfing (I saw a web page 
> with a 2GHZ processor requirement the other day.  I didn't follow the link, 
> I barely made specs).

Bleah. I would studiously avoide any such web-page.

>                         If you do multitasking- email open at all times, 
> web open at all times (multiple sites),  mp3s  playing, compiling 
> something- you want everything to be snappy.  A reasonable PC can do that 
> these days, 

#include <std.amiga.story>

A 7Mhz Amiga could multitask like that.  Granted, it didn't have any
memory protection, but we've got machines that are what, two-and-a-half 
ORDERS of magnitude faster now?   It's obscene how much more processing
power we have, with not much to show for it....

>             but when you talk about 10% slow down to feature creep here, 

....I think it's more than 10% to feature creep. :-/

> and 10% due to language there, and 10% to virtualization here, you're 
> suddenly finding yourself slowed way down.

Nah, not suddenly. You can watch it. I used to play a lot of XGalaga
on my 266MHz AXP multia. (It was a fast machine in its day. Don't laugh.
Well, not too hard.)  I ran RedHat, as they seemed the best thing going.
Each version of RedHat (and I bought the point-version upgrades!) was
slower than the previous version, until the day came that XGalaga, under
TWM, was too slow and jerky to be playable.

I started reconsidering what 'progress' meant about then.

>                                             Thats why my current PC has 
> more problems word processing than my old Apple 2e did.  Even on 5" 
> floppies it loaded the program in comparable time to OO and Office, and it 
> never slowed to hell when I hit save.

I think it's a matter of user expectations.  The average user will put
up with a certain amount of pain, and so that's what we get.  If they'd
put up with any more, we'd get that instead.

If we go to a heavily virtualizaed and/or VM-based infrastructure, it
won't be long before the hardware catches up and the applications are
again "good enough" -- not too much pain, but certainly not painless.

-Stewart

Attachment: pgpvHnY2UU69c.pgp
Description: PGP signature

-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to