flat assembler
Message board for the users of flat assembler.

Index > Heap > i need an operating system

Goto page Previous  1, 2, 3, 4, 5, 6, 7, 8  Next
Author
Thread Post new topic Reply to topic
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17279
Location: In your JS exploiting you and your system
revolution
tom tobias wrote:
The time comparison which most impressed me, from the data above, was the one comparing Crunchbang 32, with 64. Why is the 32 bit version faster than the 64 bit version???? Why isn't the 64 bit version nearly twice as fast as the 32 bit version?
Can I assume that one expects a 64bit machine to be double the speed of a 32bit machine? If so, then the opposite should be expected. Very little of the OS can benefit from 64bit registers. All devices still have to be initialised in the same way, 64bit won't help there. And 64bit OSes require double size data structures for all the OS tables and things, so setting them up will take twice as much memory traffic.
tom tobias wrote:
Across machines, why isn't the 32 bit Crunchbang running on a 1Ghz machine three times slower than the 32 bit Crunchbang running on a 3 GHz machine?
Only the CPU runs at 3GHz and most of the boot up time is not CPU bound, it is device bound. The CPU speed advantage can only be noticed when data computations are executed. Normal things like HDD access have no advantage from a faster CPU.
tom tobias wrote:
I think revolution's quote may play a role here, in explaining the answer to some of these questions....
Not really actually. The quote only applies to comparisons of old systems (both hardware and software, as a bundle, compared old to new), not what you are comparing here (just old hardware to newer hardware but the same software).
Post 10 Oct 2009, 10:51
View user's profile Send private message Visit poster's website Reply with quote
kohlrak



Joined: 21 Jul 2006
Posts: 1421
Location: Uncle Sam's Pad
kohlrak
Quote:
Can I assume that one expects a 64bit machine to be double the speed of a 32bit machine? If so, then the opposite should be expected. Very little of the OS can benefit from 64bit registers. All devices still have to be initialised in the same way, 64bit won't help there. And 64bit OSes require double size data structures for all the OS tables and things, so setting them up will take twice as much memory traffic.


The added registers should, however, give an overall speed bonus due to the ability now to lower the amount of memory operations necessary, however in the long run. I would dis-agree that it would take twice as much, as part of the idea of 64 over 32 is that the bus-size is doubled. It'd take twice as much memory, but not as much traffic, no? I will, however, agree that it's not double speed, but i will say that it helps. Of course, with compilers getting ever worse at this sort of thing (because makers are getting lazy assuming that faster cpu means they can slack off more) and people predicting the speed benefits, this'll all really go down the drain anyway.

Quote:
Only the CPU runs at 3GHz and most of the boot up time is not CPU bound, it is device bound. The CPU speed advantage can only be noticed when data computations are executed. Normal things like HDD access have no advantage form a faster CPU.


Let's not forget memory. IIRC, overgrown caches is what has given intel lots of advantages over AMD speed wise, but it's only good for benchmarks. You'd be surprised, though, how often these nifty processor optimizations are more harm than help... I won't say who i'm quoting when i say this (though not me of course), but the x86 is optimizing itself for bloat. That's why the 64bit processor is coming out in the first place. 16 bit had issues with memory access (more memory was necessary than could be accessed without switching banks), but truthfully, the 32bit stuff has't hit those kinds of problems yet. We've hit about 1 gig of RAM just to run Vista, so we're thinking ahead and getting 64bit stuff out so we're ready when the 32bit processor can't address it all without some sort of bank switching.
Post 10 Oct 2009, 11:02
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger MSN Messenger Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17279
Location: In your JS exploiting you and your system
revolution
kohlrak: The memory bus size is still 64bits for both 32bit and 64bit modern x86 CPUs. The CPU is still MUCH faster than the external memory bus so the internal 32bit writes to cache are not the limiting factor in most cases. Double the memory usage still means double the memory bus bandwidth required.
Post 10 Oct 2009, 11:09
View user's profile Send private message Visit poster's website Reply with quote
kohlrak



Joined: 21 Jul 2006
Posts: 1421
Location: Uncle Sam's Pad
kohlrak
Quote:
The memory bus size is still 64bits for both 32bit and 64bit modern x86 CPUs. The CPU is still MUCH faster than the external memory bus so the internal 32bit writes to cache are not the limiting factor in most cases. Double the memory usage still means double the memory bus bandwidth required.


And in the case of the AMD where it's supposedly built in the CPU (i don't know how it's possible myself, but i'm told they build their own controller)?
Post 10 Oct 2009, 11:20
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger MSN Messenger Reply with quote
ManOfSteel



Joined: 02 Feb 2005
Posts: 1154
ManOfSteel
tom tobias wrote:
Puppy = Aqualung; XFCE = gxine?

I think Puppy Linux uses gxine (xine).
Xfce is a DE, not a GNU/Linux distro. You can install it on virtually any unix-like system. Its default media player is xfmedia (A.K.A. xfce4-media) which is based on/uses xine.


tom tobias wrote:
some of them however, failed to connect to the internet

Sometimes it's not because it doesn't support your hardware, but simply because it needs to be configured. Sometimes it's as easy as telling it to use DHCP...


tom tobias wrote:
but were unable to play any music. These programs asked, dumbly, which application to use with the downloaded OGG file

Do you have a media player installed?
Does the media player support OGG?
Do you have the required libraries (e.g. libvorbis, libogg, etc.)?
Do OGG files work when you open them from inside the media player instead of double clicking on them?
Was MIME data registered successfully?


tom tobias wrote:
those responsible for creating these many seriously flawed distributions, i.e. the aforementioned 25 coasters, which I now possess.)

Are *BSD included in the "25 coasters"?
1) BSD systems are not distributions and they're not GNU/Linux.
2) They're not flawed. They're simply not designed for desktop use, which is why they don't even come with X installed. But this doesn't mean you can't use them as desktop systems. On the contrary! They are so flexible and give you so much freedom (they don't impose any third-party software), you can do whatever you want with them.
3) They work *very* well, even for desktop use, when you know how to configure them, take the time to configure them (for this you only need time and RTFM-ing), and install the appropriate third-party application.
They're complete *production* operating systems, not Linux kernels bundled with hundreds of GNU applications in one desktop-oriented distribution.

In any case, I recommend you install mplayer with the gnome-mplayer interface.
mplayer is very stable and plays all popular audio and video formats (if you have the appropriate libraries).
The GNOME interface is a very minimalist, non-skinned, light interface.


tom tobias wrote:
ICEWM, failed to connect to an audio player

What do you mean by "connect"? ICEWM is just a window manager, i.e. it only manages windows. It has nothing to do with audio players and whatnot.
Maybe your "audio player connection" problem is caused by the file manager or incomplete MIME data and file association...


tom tobias wrote:
I was surprised, really, at how slowly Linux starts up.

One reason is that the Linux developers constantly want to add new not-so-useful "features" without first making the already existing ones better, more efficient, less buggy, etc., resulting in the bloat witnessed in recent years.
The other and most important reason is that years ago most companies behind the GNU/Linux project(s) decided to make their software "more accessible" and "user friendly". The result was that the entire set of servers is loaded at startup. So now, everyone can self-appoint him/herself as sysadmin since there's nothing to configure, load manually, etc. ... just like good ol' Windows!


tom tobias wrote:
how to make a "very clean" win98 start up

Like all Windows, it's "clean" the first 6 months of use, then, as the registry grows to horrible proportions and the whole system gets corrupted by system malfunction, malware, bad setups and badly designed and implemented software, it becomes impossible to use and you're back at square one, i.e. formatting and reinstalling.
Besides, it's *more than* outdated. Many new programs (e.g. Winamp?) will fail to work. And like all 9x systems, it'll constantly complain about missing DLLs.
If you want a light Windows, get W2K. It's more advanced and stable than 9x and lighter than XP.


tom tobias wrote:
after weeks and weeks of reading about how much "leaner" XFCE is, compared with KDE or Gnome, to observe that it takes three times as long to shut down as does Puppy (Jwm manager) or CrunchBang (OpenBox manager).

On a modern computer (PIV and up), it should load in <15 seconds and unload in 5 or something. But of course JWM and *box are much lighter since they're just WMs, not DEs, and are Xlib/Xt-based while Xfce is GTK2-based.
Besides, you're confusing the time it takes to boot/shutdown the operating system with the time it takes to load/unload Xorg and the WM/DE, since most Linuces load WM/DEs automatically on startup.
Start them manually using xinit and time everything again.


tom tobias wrote:
Why isn't the 64 bit version nearly twice as fast as the 32 bit version?

Because even with very well written software, it doesn't work like that...
Post 10 Oct 2009, 14:03
View user's profile Send private message Reply with quote
tom tobias



Joined: 09 Sep 2003
Posts: 1320
Location: usa
tom tobias
revolution wrote:

Can I assume that one expects a 64bit machine to be double the speed of a 32bit machine?
No, I did not anticipate double the speed, but something more than what was measured....
Boot + home + listen to music time = 41 seconds for 32 bit XP, so I would have thought that 64 bit linux would be perhaps 30 seconds, i.e. an improvement of about 30%, compared with code written ten years earlier, based on a 32 bit architecture. Instead, in the best case, I observed an insignificant improvement: 39 seconds versus 41 seconds, numbers consistent with statistical noise...If I had reformatted the hard drive ten times, and reinstalled the OS ten times, and remeasured the tasks ten times, I suppose that there would be at least a 3-4 second range of times... I haven't done that, yet....Maybe I should, before predicting the results, hehe, maybe I am just guessing--not a wise course of action!!!!!

revolution wrote:

The memory bus size is still 64bits for both 32bit and 64bit modern x86 CPUs.
Precisely. As with real estate: memory, memory, memory....
The properly designed 32 bit OS, employing the same 64 bit memory controller, can outperform a bloated 64 bit OS--> the data confirms this sad truth....Most Linux/UNIX distros are just bloatware. Crunchbang and Puppy are the exceptions, but neither one represents a significant improvement compared with plain vanilla (i.e. 32 bit) M$, on any of the three machines I tested.

revolution wrote:

Very little of the OS can benefit from 64bit registers.
I believe you, but, I am sufficiently long in the tooth to recall the transition from 16 bit to 32 bit Operating Systems, and therefore, your assertion, perhaps completely valid as we move from 32 bit to 64 bit OS, seems, to me at best, counterintuitive. Certainly, without any quantitative data to support my claim, but nevertheless without any doubt whatsoever, I recall marvelling at the improvement between the 16 bit world and the new 32 bit universe from that era of bygone days. Ah, nostalgia.

I don't see that dramatic shift, i.e. that dramatic improvement in execution times, with the 64 bit code, maybe because the new code is so cluttered carrying around BIOS relics, or maybe because of the newly introduced "security software", or for some other reason....I am just writing, that, based upon these data, one would be hard pressed to justify purchasing a new 64 bit cpu/motherboard, with 4 Gbytes of dual channel DDRII ram, instead of going to the junkyard to haul home an old socket 370 board with a 1 GHz cpu, and half a gigabyte of SDRAM.

yes, the new 64 bit board is faster, and quite a bit faster. Much faster. Zoom. It just isn't ten times faster, and the cost is about 100x greater. Ok, maybe not 100x, but it isn't two or three times more expensive, it is significantly more expensive than the older motherboards/cpu/memory combos, and the performance, even with XFCE, is dreadful, compared with win 98 running on an old board....I have not got the foggiest idea why the Intel engineers targeted win98 for extinction, and eliminated it as a choice for installation on the newer boards.... Remarkable.

revolution wrote:
All devices still have to be initialised in the same way, 64bit won't help there.
Hmm. Ok, you may be absolutely correct, I don't know, I just do know that the situation thirty years ago, in making the transition from the 16 bit world to a 32 bit instruction size represented an earth shaking event, quite different from the present trend....
I acknowledge that most hardware is only 16 or 32 bits in data path, even some older devices, only 8 bits, so the initialization steps could well be running equally slowly on 32 bit cpu's as 64 bit devices.

revolution wrote:
And 64bit OSes require double size data structures for all the OS tables and things, so setting them up will take twice as much memory traffic.
Gosh, I hate to be so contentious, ok, that's not correct. Umm. Well, in my opinion, how's that? In my opinion, the 64 bit data path to and from memory, OUGHT to provide no loss of time compared with 32 bit data structures, so I am not buying this argument. I need some data explaining it. Why should it require more time to establish 64 bit structures using a 64 bit data path, than establishing 32 bit structures using the same memory cycle? The 32 bit cycle simply wastes 32 bits on each transfer, in creating these empty containers designed to hold 32 bit data.....

revolution wrote:
Normal things like HDD access have no advantage from a faster CPU.
Now, see, here's the difference. Back when I was a young lad, we used the DMA controller to access the disk. Yes, it ran at a fixed 5 MHz frequency, i.e. much slower than the 16 MHz cpu clock, but still offloaded the work, freeing up the cpu.

Smile
Post 10 Oct 2009, 14:27
View user's profile Send private message Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17279
Location: In your JS exploiting you and your system
revolution
tom tobias wrote:
revolution wrote:
And 64bit OSes require double size data structures for all the OS tables and things, so setting them up will take twice as much memory traffic.
Gosh, I hate to be so contentious, ok, that's not correct. Umm. Well, in my opinion, how's that? In my opinion, the 64 bit data path to and from memory, OUGHT to provide no loss of time compared with 32 bit data structures, so I am not buying this argument. I need some data explaining it. Why should it require more time to establish 64 bit structures using a 64 bit data path, than establishing 32 bit structures using the same memory cycle? The 32 bit cycle simply wastes 32 bits on each transfer, in creating these empty containers designed to hold 32 bit data.....
The data structures are twice the size so there needs to be double the amount of data sent to/from memory. The 32bit internal accesses vs 64bit internal accesses is not the reason, since smaller accesses are coalesced into a single larger external access. The reason is that there is double the data to be transferred on the same size/speed external bus.

[edit] The old 16bit systems used 16bit buses and the upgrade to 32bit gave corresponding 32bit buses hence the speed upgrade was noticeable because the data bus was double and the instruction set was more efficient. Plus no segments gave another performance boost. But now the 32bit CPUs have 64bit buses and there is no equivalent widening when going to a 64bit CPU. And the 64bit instruction set is not any more efficient than 32bit, most instructions are still 32bit by default but the memory addresses are double size with no associated advantage.
Post 10 Oct 2009, 14:38
View user's profile Send private message Visit poster's website Reply with quote
Azu



Joined: 16 Dec 2008
Posts: 1160
Azu
tom tobias wrote:
edfed wrote:

when very clean, win98 start (mp3 player at startup)in less than 16 seconds, XP in 30 seconds
Thanks. Yes, I was surprised, really, at how slowly Linux starts up.
I would profit from your explaining how to make a "very clean" win98 start up. I don't really know what "very clean" means, nor how to achieve it.

The newest motherboard, in the data above, one which employs the Intel chip set G41, will not permit installation of win98. I don't know why.

I am a little surprised that XP is so much slower to turn off, than win98. In that regard, I was also astonished, frankly, after weeks and weeks of reading about how much "leaner" XFCE is, compared with KDE or Gnome, to observe that it takes three times as long to shut down as does Puppy (Jwm manager) or CrunchBang (OpenBox manager).

It is rather clear, looking at these data, that very little effort has been expended, to clean up the code from decades and decades of poor programming, in the case of most of these Linux distros. The emphasis seems to be instead on acquiring pretty images for the user to look at....

The time comparison which most impressed me, from the data above, was the one comparing Crunchbang 32, with 64. Why is the 32 bit version faster than the 64 bit version???? Why isn't the 64 bit version nearly twice as fast as the 32 bit version? Across machines, why isn't the 32 bit Crunchbang running on a 1Ghz machine three times slower than the 32 bit Crunchbang running on a 3 GHz machine? I think revolution's quote may play a role here, in explaining the answer to some of these questions....

I do think there may be something useful to be learned from studying how these different operating systems commence and shut down, in attempting to construct one's own operating system....

Smile
*nix-based OSs are generally very stable so they focus more on runtime performance than on how long they take to start-up/shut-down, which is really only important for unstable OSs.



kohlrak wrote:
Quote:
The memory bus size is still 64bits for both 32bit and 64bit modern x86 CPUs. The CPU is still MUCH faster than the external memory bus so the internal 32bit writes to cache are not the limiting factor in most cases. Double the memory usage still means double the memory bus bandwidth required.


And in the case of the AMD where it's supposedly built in the CPU (i don't know how it's possible myself, but i'm told they build their own controller)?
Yes AMD have had integrated memory controllers for a very long time. Since like Athlon 64, I think. Starting with Nehalem Intel does too.

_________________


Last edited by Azu on 12 Oct 2009, 22:51; edited 2 times in total
Post 10 Oct 2009, 15:07
View user's profile Send private message Send e-mail AIM Address Yahoo Messenger MSN Messenger ICQ Number Reply with quote
tom tobias



Joined: 09 Sep 2003
Posts: 1320
Location: usa
tom tobias
ManOfSteel wrote:

I think Puppy Linux uses gxine (xine).
nope
ManOfSteel wrote:
Sometimes it's as easy as telling it to use DHCP...
Yes, for example, Puppy Linux requires several such configuration steps, including, among may other hand waving steps, precisely this manoeuvre: invoking DHCP.... The difference is that, when finished, Puppy linux works. Some of my two dozen coasters do not, even after following configuration instructions, if some are provided--often the distro just stares at the user, as though the user understood what it was that the engineers wanted the user to do. Mepis takes the prize for the most obscure, most obnoxious, most user unfriendly distro: They require a password to boot up the "live distro". What is the password? Hahaha. Figure it out babe!! Good luck... Mepis is the one distro which I will absolutely never again try....

Remarkably, the configuration of Crunchbang is zero. It does everything but cook breakfast for the user. Excellent work Crunchbang!!!!

ManOfSteel wrote:
Do you have a media player installed?
Does the media player support OGG?


OOPS, we pause this commercial, for station identification.
n.b. (That is a latin abbreviation, it means, nota bene, or PAY ATTENTION)
it is NOT the user's responsibility to "install" anything. The user inserts the cdrom, and the OS ought to appear, end of story.
If there is no media player installed, then the distro is WORTHLESS. Hello. Do you understand?

OGG: If the media player installed, does not support the open architecture format, I am not interested in the Distro. Hello, Can you understand?

I better write it again, for clarity:
it is NOT the user's responsibility to "install" anything. The distro works, or it is a coaster. Slackware, Puppy, and Crunchbang are distros. The rest are coasters.

ManOfSteel wrote:
Was MIME data registered successfully?
Good heavens. What in the world is MIME? Here is what I know of Mime: yeah, I watched Marcel Marceau in Paris, in the 70's. Terrific skill.

ManOfSteel wrote:
They (BSD) work *very* well, even for desktop use, when you know how to configure them,...
I take your word for it....Where did you find the book or document that explains how to "configure" them? Oh, the MAN pages????
Holy Cow.

Well written software does not require the encyclopedia Britannica to figure out how to use it. Crunchbang, Puppy and Slackware all require some user manipulation. None of them are as simple as they ought to be. All of them require some experience. None of them will work when an Octogenarian takes them ought of the box and plugs them in. None of them are as simple or uncluttered as installing win98....

But, at least those three distros DO FUNCTION, without requiring attendance at the UNIX club of Cal Tech. Teach me HOW to find the information needed to operate BSD, and I will retrieve my disk from the coaster collection, and try again.....
n.b. That's latin, anyway, n.b. there are some of the BSD versions with catchy names, like PC, or desktop, or some such marketing scam....

Worthless.

ManOfSteel wrote:
They're complete *production* operating systems, not Linux kernels bundled with hundreds of GNU applications in one desktop-oriented distribution.

Oops, my mistake. I should have explained at the outset:
I am NOT representing myself as the head of a multimillion dollar corporation, anticipating thousands of employees all using BSD.....

I am only installing an OS on a desktop computer.....

I should have indicated that at the outset. My fault....

ManOfSteel wrote:
The GNOME interface is a very minimalist, non-skinned, light interface.
Balderdash.
What rubbish.
What nonsense.
Where's your data contradicting mine?
Holy Cow. I don't have any data with GNOME, why not???? Gee, do you suppose it is because NONE OF MY "distros", i.e. Coasters, with GNOME, could complete the task? Not the infamous Ubuntu, nor Suse, nor Fedora--> what a piece of garbage that is: TWO HOURS I wasted watching that fossil download, only to have it choke and die on rebooting!!!) Nope. Gnome was installed on two of the three computers, but I never could get connected, else, in the best two cases, did connect, but could not play music....I measured the times: GNOME is NOT faster than XFCE, it is faster than the dinosaur KDE. Gnome, in my hands, is utterly worthless. It certainly is NOT "minimalist", nor light. It is LOURDDD. Ask Edfed for a translation: Smile

ManOfSteel wrote:
So now, everyone can self-appoint him/herself as sysadmin since there's nothing to configure, load manually, etc. ... just like good ol' Windows!
I should hope so. What, you want me to call up Linus Torvalds to ask permission to appoint myself sysadmin of my own computer?

I bet you are one of those guys whol likes passwords, so that I don't accidently erase all my files by typing in the wrong command, too? Hmm. What can I say? Nothing. It is best to write nothing....

Hopeless...

ManOfSteel wrote:
Like all Windows, it's "clean" the first 6 months of use,...
I believe you have misunderstood Edfed:
He, in my opinion, is referring to a clean install, which has then been modified by him, to boot up more rapidly, by removing certain processes, probably....

In any event, the data above, represent brand new installations, starting with erasing the entire disk "Killdisk", then running Ranish, then installing win98, testing, then xp, testing, then Linux, testing.

ManOfSteel wrote:
But of course JWM and *box are much lighter since they're just WMs, not DEs, and are Xlib/Xt-based while Xfce is GTK2-based.
Besides, you're confusing the time it takes to boot/shutdown the operating system with the time it takes to load/unload Xorg and the WM/DE, since most Linuces load WM/DEs automatically on startup.
Start them manually using xinit and time everything again.
What is a DE?
I may indeed be confusing many things, since I am by nature, a confused person, however, I think that my testing scenario is relatively straight forward: I put in the distro, push the reboot button, start the stop watch, and when the distro is installed, I record the time elapsed. Ditto for the configuration process. For Boot up, I start the stop watch while pushing the power button on, i stop it when the screen appears, in the case of Crunchbang, when the little icon is visible on the lower right, indicating that a connection to the internet is possible. For the two other measurements: Home, I click the mouse on the icon for the browser, and start the watch. I stop when the SeaMonkey lower left indicates "done". For the music, since the "home" page is the music page previously described: http://www.listenlive.eu/classical.html
I click on the 256 OGG to start, and turn the watch off upon hearing music.

Smile
Post 10 Oct 2009, 15:17
View user's profile Send private message Reply with quote
Borsuc



Joined: 29 Dec 2005
Posts: 2466
Location: Bucharest, Romania
Borsuc
tom tobias wrote:
The 32 bit cycle simply wastes 32 bits on each transfer, in creating these empty containers designed to hold 32 bit data.....
Not necessarily. You'll have to disassemble Nehalem's integrated memory controller with HyperThreading to confirm that. Razz

Most CPUs are designed to use, if needed, the stuff that isn't used. I don't know about memory bus, but they sure do that with execution units -- that's what HyperThreading actually does.

_________________
Previously known as The_Grey_Beast
Post 10 Oct 2009, 15:19
View user's profile Send private message Reply with quote
ManOfSteel



Joined: 02 Feb 2005
Posts: 1154
ManOfSteel
tom tobias wrote:
I have not got the foggiest idea why the Intel engineers targeted win98 for extinction, and eliminated it as a choice for installation on the newer boards.

Maybe because none uses it anymore?

More than 60% of Windows users use Windows XP. 98 is a decade old and its support by Microsoft was dropped some time ago, because as you know, support costs money and allocating resources for useless tasks is nothing but waste. (Intel are MS's best friends BTW).

Win9x system are all outdated and their place is in museums.
Post 10 Oct 2009, 15:28
View user's profile Send private message Reply with quote
Azu



Joined: 16 Dec 2008
Posts: 1160
Azu
n.b. OMG!!! I think hell just froze over; tobias has actually started caring about performance and memory usage! Are these the end times?? WHAT'S GOING ON HERE!?
Post 10 Oct 2009, 15:32
View user's profile Send private message Send e-mail AIM Address Yahoo Messenger MSN Messenger ICQ Number Reply with quote
edfed



Joined: 20 Feb 2006
Posts: 4237
Location: 2018
edfed
to make a very clean 98:

install it.
erase a lot of files.
set desktop as a desktop, not a webpage
desactivate the client for m$ network
erase a lot of files again.

set display params to a simple sheme.
view as list
don't show explorer bar.

erase files again.

Ccleaner
and win98 boots in 16 seconds only 10 times. Very Happy
Post 10 Oct 2009, 15:42
View user's profile Send private message Visit poster's website Reply with quote
tom tobias



Joined: 09 Sep 2003
Posts: 1320
Location: usa
tom tobias
ManOfSteel wrote:
Win9x system are all outdated and their place is in museums.
you may be correct. Sliced bread is the only way to go. Hmm. I still like the good old French Baguette. I guess I am just an old fuddy duddy....

still, one would like to see some data, to support such a strong opinion: If you were from Mars, and not Marseilles, i.e. knowing nothing about these different operating systems, would you draw the same conclusion?
hint: look at the time in the last column, comparing win98 with any of the others.......

I am unpersuaded that Win98 is obsolete.

A better test, than the one I am using, is performance over time: i.e. how long can the internet music play, without interruption, on a 1 Ghz cpu? Just anectdotally, i.e. without actually measuring this, I can argue, that in the case of the slower cpu, i.e. 1 GHz, versus the 1.4 GHz Tualatin, WinXP skips frequently, i.e. once every 40-50 seconds, and win98 once every half hour, and Crunchbang, not at all.

On a faster cpu, for example, this problem is not observed...The improvement in hardware masks the software inefficiency. But, at least for me, based upon this casual observation, not a rigorous test, as should be done, I would say, XP, not win98, is the less desirable, between those two choices. Obviously Crunchbang is superior to either one, from a listening perspective, as determined ad hoc, and not rigorously....

Mother............Time....Time..Time...Time...Time....Time.....boot + home
Board....O.S....Install.Config.Boot....P.Off...Home...Music....+ music
......................min.....min....sec.....sec......sec.......sec.......sec

Bio-....Win98.....40.......30.....50.......3.......9............14.......73
star.....XP.........47........30.....50......12......7............11.......68
..........Puppy.....8..........2......50......11......6............20.......76
...Crunchbang....12........0......60.......14.....5.............5........70

DFI....Win98......30.......30.....38.......4.......3.............14......55
...........XP.........40.......30.....50.......12.....7.............12.......69
.........Puppy......8..........2......42.......9.......7.............18......67
...Crunchbang....16........0......52.......15.....7..............5.......64
Post 10 Oct 2009, 15:47
View user's profile Send private message Reply with quote
Borsuc



Joined: 29 Dec 2005
Posts: 2466
Location: Bucharest, Romania
Borsuc
It's only "obsolete" because money dictates it in a capitalist world, not necessarily quality.
Post 10 Oct 2009, 16:55
View user's profile Send private message Reply with quote
Azu



Joined: 16 Dec 2008
Posts: 1160
Azu
Borsuc wrote:
It's only "obsolete" because money dictates it in a capitalist world, not necessarily quality.
Obsolete doesn't imply lack of quality anyways. Just that it's old and has been replaced (for better or for worse).
Post 10 Oct 2009, 16:59
View user's profile Send private message Send e-mail AIM Address Yahoo Messenger MSN Messenger ICQ Number Reply with quote
kohlrak



Joined: 21 Jul 2006
Posts: 1421
Location: Uncle Sam's Pad
kohlrak
Quote:
No, I did not anticipate double the speed, but something more than what was measured....
Boot + home + listen to music time = 41 seconds for 32 bit XP, so I would have thought that 64 bit linux would be perhaps 30 seconds, i.e. an improvement of about 30%, compared with code written ten years earlier, based on a 32 bit architecture. Instead, in the best case, I observed an insignificant improvement: 39 seconds versus 41 seconds, numbers consistent with statistical noise...If I had reformatted the hard drive ten times, and reinstalled the OS ten times, and remeasured the tasks ten times, I suppose that there would be at least a 3-4 second range of times... I haven't done that, yet....Maybe I should, before predicting the results, hehe, maybe I am just guessing--not a wise course of action!!!!!


Speed up your harddrive itself to get added bonus. For more fun, compare the startup times of livecds.

Quote:
Precisely. As with real estate: memory, memory, memory....
The properly designed 32 bit OS, employing the same 64 bit memory controller, can outperform a bloated 64 bit OS--> the data confirms this sad truth....Most Linux/UNIX distros are just bloatware. Crunchbang and Puppy are the exceptions, but neither one represents a significant improvement compared with plain vanilla (i.e. 32 bit) M$, on any of the three machines I tested.


I have, myself, noticed slower boot times, however, the actual running is significantly faster. Let's not forget that the filesystem (especially on newer distros where they have "transparent encryption" on ext4) is also a factor to consider. Though, one look at MS's company architecture, you'd realize they're really good programmers actually (most), but they're under a heavy staff. This makes for a small, but unstanble system. They can't implement the features they want (thus why the system is as fast as it is).

Also, let us all not forget that the 64bit instructions themselves are larger than 32bit equivalents, and therefore a problem as well. Some people (and compilers) use 64bit regiters when entirely unnecessary. Remember that ops done to a 32bit register zero-extend to the 64bit register automatically.

revolution wrote:
tom tobias wrote:
revolution wrote:
And 64bit OSes require double size data structures for all the OS tables and things, so setting them up will take twice as much memory traffic.
Gosh, I hate to be so contentious, ok, that's not correct. Umm. Well, in my opinion, how's that? In my opinion, the 64 bit data path to and from memory, OUGHT to provide no loss of time compared with 32 bit data structures, so I am not buying this argument. I need some data explaining it. Why should it require more time to establish 64 bit structures using a 64 bit data path, than establishing 32 bit structures using the same memory cycle? The 32 bit cycle simply wastes 32 bits on each transfer, in creating these empty containers designed to hold 32 bit data.....
The data structures are twice the size so there needs to be double the amount of data sent to/from memory. The 32bit internal accesses vs 64bit internal accesses is not the reason, since smaller accesses are coalesced into a single larger external access. The reason is that there is double the data to be transferred on the same size/speed external bus.


Take, for example, things like GDT. It not actually double, but to cover the whole 64bits of the spectrum, you need some degree of improvement.

Quote:

Now, see, here's the difference. Back when I was a young lad, we used the DMA controller to access the disk. Yes, it ran at a fixed 5 MHz frequency, i.e. much slower than the 16 MHz cpu clock, but still offloaded the work, freeing up the cpu.


I'm told it goes 1 byte at a time, too. =p

Quote:

Yes AMD have had integrated memory controllers for a very long time. Since like Athlon 64, I think. Starting with Nehalem Intel does too.


As young as athlon? I feel even more fortunate, now, that my lappy has an athlon. I assume it's still clocked seperately though...

Quote:
Balderdash.
What rubbish.
What nonsense.
Where's your data contradicting mine?
Holy Cow. I don't have any data with GNOME, why not???? Gee, do you suppose it is because NONE OF MY "distros", i.e. Coasters, with GNOME, could complete the task? Not the infamous Ubuntu, nor Suse, nor Fedora--> what a piece of garbage that is: TWO HOURS I wasted watching that fossil download, only to have it choke and die on rebooting!!!) Nope. Gnome was installed on two of the three computers, but I never could get connected, else, in the best two cases, did connect, but could not play music....I measured the times: GNOME is NOT faster than XFCE, it is faster than the dinosaur KDE. Gnome, in my hands, is utterly worthless. It certainly is NOT "minimalist", nor light. It is LOURDDD. Ask Edfed for a translation: Smile


Take a look at the gnome optimization manual to see how true what you're saying is. It's a real joke. If you think alot of those comments are funny, look into the fact that gnome (and unfortunately many other APIs) have a measure that could be seen as anti-assembly at the cost of performance. Yes, i'm not crazy when i say this, but they have runtime type checking. It's not mandatory, but it's stupid...

Quote:
n.b. OMG!!! I think hell just froze over; tobias has actually started caring about performance and memory usage! Are these the end times?? WHAT'S GOING ON HERE!?


That's nothing, MS committed code to the linux kernel.

Quote:
It's only "obsolete" because money dictates it in a capitalist world, not necessarily quality.


Capitalism stands up for quality, so long as the consumer does. The consumer hasn't complained about low quality, they've just taken the garbage that what goes wrong is their fault (not entirely untrue, nor entirely true), so people continue marketting crapware. You should hear what they say in schools. It doesn't matter if it's the right answer, as long as the user thinks it is and blames himself it isn't.
Post 10 Oct 2009, 20:18
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger MSN Messenger Reply with quote
sleepsleep



Joined: 05 Oct 2006
Posts: 8903
Location: ˛                             ⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣Posts: 334455
sleepsleep
i actually love using windows 95. been so long i didn't hear the tada sound.... long live windows 95!!!!!!!!!!!!! the fastest faked multitasking OS on earth!! muah hahahaha
Post 10 Oct 2009, 21:03
View user's profile Send private message Reply with quote
ManOfSteel



Joined: 02 Feb 2005
Posts: 1154
ManOfSteel
Azu wrote:
Wtf? I thought only Windows used that registry nonsense. It has spread to *nix distributions too now?

Read the text you quoted again. I was talking about Windows 98.
Post 10 Oct 2009, 22:10
View user's profile Send private message Reply with quote
ManOfSteel



Joined: 02 Feb 2005
Posts: 1154
ManOfSteel
tom tobias wrote:
nope

The wikipedia page is here. I can't find anything about Puppy Linux. But you may be right, I don't know for sure.


tom tobias wrote:
it is NOT the user's responsibility to "install" anything.

Sorry, but no. Aside from the "exclusively desktop" distros, many of these systems are *production* systems, mainly used as HTTP, SMTP, FTP, etc. servers, or as firewalls, or as repair/maintenance live systems. These don't need X and even less media players.
Besides, don't you ever install software under Windows? Your C:\ partition must be pretty much... empty.


tom tobias wrote:
If there is no media player installed, then the distro is WORTHLESS

Wow, that's an interesting point of view.


tom tobias wrote:
Hello. Do you understand?

Um no, come again. Rolling Eyes


tom tobias wrote:
OGG: If the media player installed, does not support the open architecture format, I am not interested in the Distro.

Just like under Windows all media players don't support all formats by default without the proper plugins, media players under unix-like systems need the proper libraries to understand one format or another. The media player is nothing more than an interface. It usually can't decode formats by itself. Actually it can't even send the decoded sound to the sound card drivers. Audio output plugins/libraries do that.


tom tobias wrote:
Hello, Can you understand?

It's getting really tiring... *yawn*


tom tobias wrote:
it is NOT the user's responsibility to "install" anything. The distro works, or it is a coaster. Slackware, Puppy, and Crunchbang are distros. The rest are coasters.

So by your logic, FreeBSD -- which powers Yahoo! and was used to generate the Matrix FX -- is a coaster, since by default it comes with no media player at all (only an audio CD player). Hehe, nice!


tom tobias wrote:
Good heavens. What in the world is MIME?

Open Windows Explorer, and go to Tools > Folder Options > File Types. This is what I'm talking about.
In simple terms, MIME's audio/ogg entry says "ogg" is an audio format and it describes it. Another database sets file associations and says audio/ogg is associated with, say, mplayer. So, every time I use an application that supports and understands MIME types (e.g. a file manager), I'll see a translation of "Ogg Audio" according to my locale settings under the file type column, and mplayer will open it when I double-click it.


tom tobias wrote:
Where did you find the book or document that explains how to "configure" them? Oh, the MAN pages????

No. Divine intervention, as everything else in my life.


tom tobias wrote:
Well written software does not require the encyclopedia Britannica to figure out how to use it.

Yes, there are Windows-like software that don't require much (or any) reading, are easy to use (but hell to administer and keep in good shape) and are as mediocre, "uncustomizable", lack as many features and flexibility and are as unstable. Why? Because the "designers" began by making the interface and wasted all their time on it, then grafted the inner function and removed some of the planned features because there was not enough place on one graphical screen to include them all.
And then, there are the well-designed and implemented software, made by ___s for ___s, with a lot of features but with non-"wow kewl!" interfaces, that require some reading and configuration but will never break when you need them.
But even with Windows, you'd have to read a lot on *everything* to *really* be able to admin a machine running it. But gosh, I forgot there's almost no documentation other than the ridiculous "help" files than contain nothing more than the obvious!


tom tobias wrote:
Teach me HOW to find the information needed to operate BSD, and I will retrieve my disk from the coaster collection, and try again.....

Sure, read the relevent section of the handbook. And if you still have problems, ask the community and mention your hardware specs, the operation you made and the exact error/problem you encountered.
But really, if you're looking for a user-friendly system (whatever that could mean for you), and you abhor having to read documentation and learn new things, *please* stick with Windows and never look back!


tom tobias wrote:
there are some of the BSD versions with catchy names, like PC, or desktop, or some such marketing scam

They're not marketing scams. They're desktop-oriented derived products. The people behind them worked hard to make them and some are one-person projects. Absolutely not the type of "marketing scams" you'd find in the highly commercial software industry.


tom tobias wrote:

I am NOT representing myself as the head of a multimillion dollar corporation, anticipating thousands of employees all using BSD.....

I am only installing an OS on a desktop computer.....

As I said, they *can* be made into desktops. It just needs some reading and configuration.


tom tobias wrote:

Balderdash.
What rubbish.
What nonsense.

Stop being insulting for fun and behaving like a 10-years old, will you? This may give you broken bones outside the virtual world.


tom tobias wrote:
Gnome was installed on two of the three computers, but I never could get connected [...]

The gnome-mplayer interface for the mplayer media player is not synonymous with the GNOME DE. You do not even need to use GNOME as your DE to use gnome-mplayer. It only requires 1 or 2 GNOME dependencies (along with GTK), which is nothing.
I won't change anything I said. It is a light software. On my system it uses less than 13MB of reserved memory. Winamp 5.5 uses 20MB on Windows XP. Beat that!


tom tobias wrote:
It is LOURDDD. Ask Edfed for a translation

Pourquoi redoubler une lettre muette? Pour insister, j'aurais plutôt redoublé le u.


tom tobias wrote:
I should hope so. What, you want me to call up Linus Torvalds to ask permission to appoint myself sysadmin of my own computer?

You misunderstood the idea. Most Windows sysadmins in the corporate world are as knowledgeable of Windows as the average user, i.e. they are self-appointed ignorants and incompetents. Most users cannot fix even the simplest problems with the "user-friendly" Windows so why should/how could they be knowledgeable of systems originally designed by computer scientists for computer scientists to run scientific and military machines and networks.
Let the user-friendly Linuces be more user-friendly if they can. But this will come at a great cost...


tom tobias wrote:
What is a DE?

A desktop environment.


tom tobias wrote:
I think that my testing scenario is relatively straight forward

GNU/Linux and other unix-like systems being formed of distinguishable parts working together (the base system, the X windowing system implementation, the DM, the WM/DE), you can theoretically make a system work slower or faster by choosing one combination of parts or another.
The methodology you used to time distros mixes all of these parts for every distro you tried, so you can't know what system/setup/configuration is the most time-efficient, what WM/DE loads faster, etc.


Last edited by ManOfSteel on 10 Oct 2009, 22:52; edited 1 time in total
Post 10 Oct 2009, 22:16
View user's profile Send private message Reply with quote
Display posts from previous:
Post new topic Reply to topic

Jump to:  
Goto page Previous  1, 2, 3, 4, 5, 6, 7, 8  Next

< Last Thread | Next Thread >
Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum


Copyright © 1999-2020, Tomasz Grysztar. Also on YouTube, Twitter.

Website powered by rwasa.