Re: Windows 2000 is The Borg!

Maybe a peculiarity with AMD chips? I ran Win95 for *many* years on a

400MHz Pentium II. And I'm now running Win2K on the same machine... actually *not* unbearably slow... although I think I'll soon relegate the machine to the grandkids.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

If you ran revision two (with USB support I believe that had the bug corrected), but perhaps it was only with the AMD chips. That's all I run. :)

:) When it first appeared, it was only met with the AMD chips (it was a timing problem in the IDE handler, and they were the only chips fast enough at the time to cause it - their integer arithmetic was massively faster than the Intel chips at the same clock rate). At 600MHz, it appeared on the Xeons (if you wanted to run W95 on a Xeon!), and it took the PIII, at 800Mhz, before it became a problem on the Intel chips. By the time these had appeared, the fix was 'standard' (included in the latter W9x releases).

Best Wishes

If you had only half as much knowledge of computers than you have f-words, you wouldn't have any problems mastering them.

Michael

get

The big problem with Linux is that it is very difficult to get into from the beginning. Once sorted and working it really is smooth, reliable and a pleasure to use but the little things, that the knowleageable people who write the software have forgotten, make it hard. For example, where does software get installed to, nothing tells you and if you don't know what the executable is called you can't search for it and different software installs in different places. It's bad practise to use Root for anyting other than system maintainance but when I use smbmount to mount a Windows PC only root has access to the files and when I copy them into a users home directory as root then the user doesn't have the right permissions. There are loads of little things like that which prevent Linux from taking off and it really is a great pity. They are all small things that can be answered by searching on the net or asking on usenet, Linux users are really helpful when it comes to sorting people out but it takes a lot of time and often if the task is only a small one then the time involved by using Linux is prohibitive, particularly if it's work stuff.

I now have to add that for years I have tried using open office / star office but always had to go back to Windows and Word because everyone uses it, but the latest version of Open Office, both on Windows and Linux is a dream and most important the import and export filters for Word are excellent which means I can now use it, as I want, but still satisfy everyone I need to send files to. My only whinge is the initial opening time, it's a bit slow (dinosaur for a PC here) but everyone should give it a go.

Actually, it wasn't integer arithmetic that was so much faster. Unless you're saying that NOP is integer arithmetic. ;-) WinBlows had a NOP timing loop. The K6 had a zero cycle NOP instruction, where the P5 took one cycle for the NOP. I believe Athlon went back to a one-cycle NOP, perhaps for this reason.

Yes, since the timing loop underran at these speeds. Note that the K6 integer unit was *not* as fast as a Xeon/600 or PIII/800.

Yes.

--
  Keith

A classic example of the unbelievably complex interaction between chip architects and programming tradition/practice that takes place behind the scenes where most of us never see.

Nah -

I think he's got the f-words mastered! :-)

And a classic example of the proliferation of "bad programming practice". We've all done it though. It's not as bad as some, I suppose, but asynchronous timing should never be done using a method that is reliant on clock speed - especially when you know the clock speed is not fixed.

I guess this is the reason behind "bogoMIPS" on Linux and probably other software.

Ben

--
I'm not just a number. To many, I'm known as a String...

For whatever reason, the only way I got a serial link up was to turn Flow Control OFF. You'll be happier with NIC's and a crossover cable.

--
_______________________________________
John E. Todd         jtodd@island.net

Note: Ensure correct polarity prior to connection.

rpm -qa rpm -ql package rpm -qf file

tels you everything ever wanted to know about installations but were afraid to ask. Even dependencies etc. But most of the time you don't need to know where something is installed, with rpm -e you can remove it without knowing anything.

locate find and [TAB][TAB] in bash tells you which exec's are where

It's bad practise to use Root for anyting other than

Put the smbmount in fstab with the "user" option and anyone can mount it, you also have uid, guid, read/write/exec options in fstab.

well, this should all be on a linux newsgroup, not an electronics design newsgroup, sorry for that, my experience is that indeed I had to invest a lot of time in learning, but the non-crashing machines saves me a lot of time in my work.

need

afraid

of

really

searching

comes

I wasn't looking for answers I was trying to point out that it is difficult but only because of the little things. All the answers are available is you search and if you post a specific question you will get specific answers fairly quickly but it is the need for some understanding in order to make it work that is currently stopping it from being used by everyone. It is however definately making progress and provided there is no odd hardware a blank install can be done and will work. There is also the problem of people not knowing, a few years ago the thought of someone owning a computer and not knowing anything about it was alien but now there are a lot of people who know absolutely nothing about how, they can only just turn it on and off but they are succesful in getting it to do what they want (for the most part).

Cars used to be the same i.e. people who had them knew how they worked, my grandma for example once mended the throttle cable with a hair clip when we were miles away from anywhere, I can't see my wife doing anything that involves opening anything but the petrol cap. It is now normal and accepted that you can own and use a car and not have any understanding of it at all.

Computers are going the same way but Linux is struggling coming to terms with that maybe. It really is unfortunate because it is so much better than Microcrap at every level, problems get fixed, properly, difficulties are understood by thousands of people who will bend over backwards to help other people get into it. Money and economies play no part in the development of the stable system and contributors are genuinly horrified when problems are proven to exist with lots of midnight oil burning to fix them.

You are absolutely right a non-crashing machine is worth the effort and in the long run you will live longer because of the stress that Windows forces on you.

Windows has some amazing features about it though, for example, how do they make the severity of the crash inversely related to the time you have to complete the task?

***WARNING- Extreme Rant Follows*** All I care about is if it lets me program PICs and access hardware at low level. Which is why I use DOS 6.22 on my laptop. No problem. 2K is just a royal pain when it goes wrong, though it is fairly stable (as long as you only run M$ apps and use approved drivers on it). Otherwise it *will* fail eventually.

XP is worse ! Much worse. I mean, it shows how much of a stranglehold M$ have that I cannot buy a machine without XP preinstalled (and the dreaded "restore disk") from the local vendor, ******** ********** FYI (deleted to save lawsuit) Plus its every hacker and virus writer's dream OS due to all the horrible critical security holes (chasms) that you have to download hundreds of updates to fix. Pity those on a slow Net connection, they haven't a snowball's chance in hell!

If more companies wrote a DOS or Linux version of their software they would make much more money, especially from the less well off people who have for various reasons opted out of the Windows upgrade ladder. As far as I am concerned Windows software that won't allow low level hardware access that WORKS is a no-no. (Are you listening, Gates et al) ?????!

BTW, Rich . Windows 2000 is *NOT* the Borg . The Borg have proper networking support! :):):)

-A

...

I had a flash memory stick that worked the first time I plugged it in, then gradually stopped working. No amount of reading HOWTOs and messing with fstab and the other stuff could get it to work.

Had the same problem with the sound card.

I used Linux (LM81) for months but eventually gave up because of the

5-10% of stuff I could never get fully and permanently working. I learned from the experience and will use it again, but only once hardware support is a given. It's a shame because it does so many things well and logically.

The need to understand 90% of the picture before you can fix the 10% that's not working is a big turnoff. In my experience the graphical configurators are only half baked (cos no serious Linux head ever uses them, so they never get properly tested) and so you *have* to manually edit. In my case I probably messed up, and there was no recovery once devfsd had messed with it further.

The other problem is you'll always need two machines. I tried dual booting. I had 3 time consuming MBR corruptions (probably caused by MS, not Linux) and I won't attempt it again. Nobody in the engineering field can economically do everything in Linux, due to the sheer volume of Win32 stuff that works just fine. Wine takes ages to set up, doesn't look right, runs like treacle. I wish it weren't so but this is my experience.

I reckon on rebooting Windows once or twice a day. It takes about 30s. I go and make a coffee. It would be nice if this didn't happen, but at least I get to use the computer 99% of the time for nearly everything I need, and I rarely lose data (unlike the KDE office apps I tried, which regularly crashed and trashed my files).

Regards, Mike.

--
Mike Page BEng(Hons) MIEE           www.eclectic-web.co.uk

boy that's scary.. if i code in a nop, it's because i need that timing delay.

Not much of a programmer but I am left wondering why you would want a 0 cycle NOP, what other uses does it have other than a bit of time?

It is also useful for "zeroing out" code space where actual instructions used to be (or to safely fill out a block of code space).

Well, accessing hardware 'trivially' is great if you're a hacker -- it's a nightmare if you're a regular user trying to protect yourself from poorly written programs and virii.

All operating systems will fail if you just willy-nilly add half-baked drivers. I won't get into whether or not Linux or Windows is inherently more stable, but I will say I've seen both Linux and Windows machines crash, as well as both staying up for months at a time (and only being taking down for maintenance, not crashes!).

If Linux were anywhere near as popular as Windows, you can bet all the virus writers would be targeting it with just as much enthusiasm. And while I'm again not qualified to perform a security analysis of either OS, I think it's safe to say that Linux has security holes as well... and all it takes is one...

The Windows Update approach to patching is a good idea, IMO. Although I agree that those without a high-speed Internet connection are at a disadvantage, be aware that Windows will slowly download patches in the background for those with a dial-up connection, so assuming the user spends some hours per week connected to the Internet, they'll still get all the patches they good.

Last time I tried Linux (about a year and a half ago now), it didn't have automatic updating available -- it still required user interaction to install each patch. And the typical argument that Linux programmers will more quickly fix a found security risk may be true (but it's certainly debatable!), but unless those patches can be pushed to users effectively, it's all for naught.

DOS is far too limiting for modern prorams. You can do sophisticated things in it, but realistically 99% of all the cool programs that run in DOS have already been written. I agree that more companies writing Linux software would generate revenue for those companies -- but the question is whether or not it's _enough_ revenue so as to pay for their added development costs!

You do have a point, but keep in mind that someone buying a new Dell or Gateway PC every 3-5 years is effectively paying all of about $50 for the operating system (bundled with the system). It's very challenging these days to build your own PC for the same price that Dell, et al. can if you actually intend to go out and purchase legal copies of Windows, Office, etc. from a retailer. XP Pro. brand new is $299 retail! :-(

Linux, Open Office, etc. is a wonderful solution for people who can't really even afford, say, $500-$1000 for a PC. The groups that use 'last generatrion' (e.g., 2+ year old PCs) hardware and package them up with free software and either give away the machines or sell them for next to nothing are performing a valuable service.

Ever heard of a device driver? You don't just willy-nilly bang on the hardware if you expect your driver to obtain Microsoft certification, but if all you want to do is hack around, there are several drivers out there that will take your "peek, poke, inport, or outport" request and just stick it on the hardware.

The Windows driver model is a Good Thing -- the 'layering' approach that allows the guy who creates the USB oscilloscope (ob. electronics group reference!) to not have to know anything at all about the USB chipset on the motherboard in question is a huge convenience vs. having to do it in DOS. Although there were some pseudo-layering standards in DOS (e.g., packet interfaces for network cards), the lack of a real OS is why your DOS PC doesn't support USB CD ROM drives, flash ROM cards, etc.

The big annoyances I had with Linux were:

-- Lack of program installation standards. Depending on which desktop you were running, there were variegated methods to get a program connected to AN ICON ON THE START MENU. Very few programs were smart enough to do this themselves -- much less to add themselves to a standard 'add/remove program' database. This is all completely standard in Windows. Of course, there's no reason Linux couldn't do this -- it's just a lack of standards.

-- Lack of common keyboard control standards. I really despised how some programs would use, e.g., Alt+X to cut whereas others used Ctrl+X! Yes, of course these could (typically) be 'fixed,' but I really don't want to spend my time doing that. Standadization is a good thing. Customizability is great too -- but I want the base product to have the same defaults as all the rest. The keys Windows uses for routine tasks are so common these days that many keyboards have them printed on the front sides of the keys.

---Joel Kolstad

I doubt that 'zeroing out' occurs that often, but 'safely filling...' would be used, for instance, in order to align the next chunk of code on a cache line boundary. Of course, this is prone to break on other processors as well -- I would consider using NOPs for alignment purposes or timing purposes a 'reasonable' programming approach only if your code has checked what kind of CPU it's running on and has a 'safe' (but slower) alternative method if it isn't 100% positive it has the right target.

ObAnnoyance: The Microsoft Windows 2000 IDE drivers will run on Intel, Via, and SiS chipsets, albeit with non-optimal performance for the later two. The Via driver will CRASH if started on a non-Via chipset. :-( Score one for Microsoft.

---Joel Kolstad