Is AMD Unreliable?

[Surreal Killer]

Yes, because it was Microsoft's bug, and not AMD's bug.

Hehehe.. I found it interesting that while, the fix was "free" you
had to pay Micro$oft's $35/per incident support line to receive it:

"A patch for the bug developed by Microsoft was posted late last week.
To get the patch, users are told to contact Microsoft's support lines
and request the "hotfix for Windows 95." The call is free but
customers will be charged $35 for the support incident, according to
an update on AMD's Web site."

Typical Gates greed. :-\

I'm *almost* out of the buggy WIntel loop. I run all AMD, and am
just getting used to Linux. As soon as I'm comfortable with Linux
apps, the FAT32 Windoze system partitios all get reformatted to blank
ext2.

 

[Stacey]

Typical Gates greed. :-\

Yep. I read in the recent PC world about people complaining to them about
system lags after applying SP1. When they contacted MS, PC world was
instructed to tell these people to call MS's toll support lines for the
"hot fix" and then were told there was no fix for their problem!

I'm *almost* out of the buggy WIntel loop. I run all AMD, and am
just getting used to Linux. As soon as I'm comfortable with Linux
apps, the FAT32 Windoze system partitios all get reformatted to blank
ext2.

I only use windows for video editing and wish someone would port a good
video app to linux so I could get rid of windows all together.

 

[Adam Selene]

Greetings...

Many people say AMD processors heat too much too fast compared to
Intel products and thus they cause computers to crash. Is that true?

Short answer - No... It's not a problem...

How much more heat does it generates compared to an equivalent spec.
Intel processor?

It's irrelevant.

Is it true that professionals avoid AMD processors for these reasons
and also because many people think that they may not be compatible
with some software.

No. The only professionals that do that are Professional Intel Fan boys.. Most
professionals (I.E. They get paid for what they do) build what sells, and AMD systems are
typically less expensive than a similar performing Intel system. I personally know of at
least 2 local shops that USED to say, "we only build Intel systems" who now build ONLY AMD
systems...

I have NEVER heard of any commercial software with which AMD's were incompatible.

I have been using Cyrix and AMD processors since some time and i am
very happy with them. Excellent quality / price.

Well, there you go, You answered all of your own questions....

 

[Ancra]

Hmmm when all Intel CPU's did have this problem and ran windows
without any bugs, then I would say it was a buggy AMD CPU not windows.
Hence Windows released a software patch so these AMD CPU's would
work correctly with windows.

That's a ridiculous stand!
First of all, timing bugs in software have not been entirely rare. And
Intels cpus have. of course, been equally afflicted.
So you are wrong when you assume they haven't.

Secondly, that old Win95 does not handle PIII, PIIIe or any P4
correctly either, and not only for timing reasons.
So you are also wrong in assuming that "all Intel CPU's" could run
that windows.
I wouldn't be surprised if it failed on late PII as well.
That makes all those Intel cpus "buggy"?

Thirdly, thankfully, it's not up to 'brand fanatics' to define what's
a bug or not. It's much more simple. If it adheres correctly to the
software engineering specs, then it works correctly and there is no
bug. According to both AMD and MS, IIRC, - the K6 did work correctly,
Win95 didn't.

Fourthly, MS has continuosly patched it's various Windows OS'es, to
work correctly with all kinds of previously unaccommodated hardware
improvements. New chipsets, new interfaces, faster hd's, bigger
caches, etc. You desire to interpret that as "buggy" hardware? So MS
is forced to patch Windows to make that buggy hardware "work correctly
with windows"? >:-/


ancra

 

[Stacey]

Ancra wrote:

Thirdly, thankfully, it's not up to 'brand fanatics' to define what's
a bug or not. It's much more simple. If it adheres correctly to the
software engineering specs, then it works correctly and there is no
bug. According to both AMD and MS, IIRC, - the K6 did work correctly,
Win95 didn't.

http://hypermail.idiosynkrasia.net/linux-kernel/archived/2001/week47/0124.html


Looks like other OS's found weirdnesses with them as well? Kinda like
blaming soundblaster that their sound cards cause problems on Via boards
when these cards work fine on anything else..

 

[Strontium]

-
Ancra stood up, at show-n-tell in (e-mail address removed) and
said:

That's a ridiculous stand!
First of all, timing bugs in software have not been entirely rare. And
Intels cpus have. of course, been equally afflicted.
So you are wrong when you assume they haven't.

Secondly, that old Win95 does not handle PIII, PIIIe or any P4
correctly either, and not only for timing reasons.
So you are also wrong in assuming that "all Intel CPU's" could run
that windows.
I wouldn't be surprised if it failed on late PII as well.
That makes all those Intel cpus "buggy"?

Thirdly, thankfully, it's not up to 'brand fanatics' to define what's
a bug or not. It's much more simple. If it adheres correctly to the
software engineering specs, then it works correctly and there is no
bug. According to both AMD and MS, IIRC, - the K6 did work correctly,
Win95 didn't.

Fourthly, MS has continuosly patched it's various Windows OS'es, to
work correctly with all kinds of previously unaccommodated hardware
improvements. New chipsets, new interfaces, faster hd's, bigger
caches, etc. You desire to interpret that as "buggy" hardware? So MS
is forced to patch Windows to make that buggy hardware "work correctly
with windows"? >:-/

Microsoft cannot seem to get the 'security' thing right, without people
exploiting it's weaknesses and throwing it in their face first, either....

/shrug

 

[Ancra]

Ancra wrote:

Looks like other OS's found weirdnesses with them as well?

I'm not sure what you think is weird. Or why it is 'weirder' than any
other of zillions distinctions between different processor models and
'steppings'.

I find the document less than transparent.
I'm certainly not an expert on K6'es, I was completely uninterested in
AMD at the time, but as I see it; this is a bug in some Linux-kernel
code that failed to handle the different 'Write Handling Control
Registers' formats of K6'es correctly.
(which are involved with L1 cache handling)

There are two different K6 WHCR formats. Apparently, original code
failed to correctly identify some K6 cpu model 8 steppings.
K6 models 6, 7, and 8 with steppings 0, 1,..8 use one format.
K6 models 8 with steppings 9,10,..F, and model 9, use another.

Sorry, but this look to me like a Linux bug. Not a cpu bug.
But I know as well as you that cpu-bugs do exists. But it's not an AMD
domain. Intel have had their fair share of them.

Kinda like
blaming soundblaster that their sound cards cause problems on Via boards
when these cards work fine on anything else..

- Ah, come on. Are you serious? I'm not claiming that blame or
responsibility can always be assigned to one party. My assumption is
that problems often occure due to two different "undocumented
features" clashing.


ancra

 

[Ancra]

Ancra stood up, at show-n-tell in (e-mail address removed) and
said:

Microsoft cannot seem to get the 'security' thing right, without people
exploiting it's weaknesses and throwing it in their face first, either....

Well, it's not exactly easy. But MS cetainly seem to be rather lax,..
Just consider ActiveX, and executable contents in email... etc.

What really also bugs me is the vast amount of websites, that now
think they're going to run ActiveX controls on my PC. >:-(


ancra

 

[Stacey]

Sorry, but this look to me like a Linux bug. Not a cpu bug.

So 2 different OS's (the main 2?) have had problems with this chip yet it's
not a bug with the chip?

But I know as well as you that cpu-bugs do exists. But it's not an AMD
domain. Intel have had their fair share of them.

So it's a CPU bug when it happens to an intel chip but an OS bug when it
happens to an AMD?

 

[Stacey]

Well, it's not exactly easy. But MS cetainly seem to be rather lax,..
Just consider ActiveX, and executable contents in email... etc.

Bingo and that ANY software or user can access ANY files/folder and
read/write to any of them.. The above is why no one has sucessfully written
a linux virus. Sure people can individually hack into a linux system but
widespread code that transfers automatically isn't going to happen. You'd
think M$ would learn from this unless they are making money from AV
software companies to not secure their OS's. You really have to wonder...

 

[Ancra]

Ancra wrote:

So 2 different OS's (the main 2?) have had problems with this chip yet it's
not a bug with the chip?

Yes. As I said, I found the piece of patch-code not very transparent.
But that's how I read it.

Apparently the original code authors prescient gifts failed him, and
he missed that some future generations of model 8 K6-2 would change
WHCR format to the same the model 9 K6-3 uses.

You really do have to take notice of that old OS'es don't handle later
Intel CPU's correctly either. Both Windows and Linux have had to
change in order to accomodate new cpu's.
If you would make a career out of singling out AMD cases, in order to
'support' some notion of AMD inferiority, - It would make me cringe to
see you degrade yourself like that.

So it's a CPU bug when it happens to an intel chip but an OS bug when it
happens to an AMD?

? No. I haven't said anything like that. Are you going to get childish
now?
What I would accept responsibility for, is that I've argued there
isn't any foundation for any view of AMD as 'unreliable' in relation
to Intel.
There was the case of instability with >32MB on the first K6 batch.
But there were also the cases of the Pentium FPU inaccuracy, the
Pentiums that in conjunction with chipsets from the very same
manufacturer and >64MB wouldn't cache pagedescriptors, resulting in a
massive performance loss, and the whole string of memory performance
problems with early PIII's. So what I'm saying is,- what the hell are
we trying to prove here?


ancra

 

[JAD]

When / IF linux ever gets as popular as windows, there will be plenty of
malicious code written just for it.......

 

[Stacey]

You really do have to take notice of that old OS'es don't handle later
Intel CPU's correctly either. Both Windows and Linux have had to
change in order to accomodate new cpu's.

Like? 98SE runs a P4 fine (as well as an athlon) without any added "patches"
required. What has changed with windows (or linux?) to accomodate new
CPU's? Maybe some optimazations for stuff like SSE etc but nothing that
caused them to not boot or corupt data. Do you really believe it was smart
of AMD to release a chip that wouldn't run right on the premier OS of the
time and then expect them to come up with a patch so it would? That isn't a
bug?

If you would make a career out of singling out AMD cases, in order to
'support' some notion of AMD inferiority,

??? AMD K6-2's WERE inferiour in lots of uses, made tons of heat and AMD
stopped making them because they came out with something MUCH better. If
they hadn't, I doubt AMD would still be around. Until the athlon AMD's
were seen as a cheap solution with poor FPU. That was then not today.

? No. I haven't said anything like that. Are you going to get childish
now?

Lighten up dude... The AMD K6's had obvious problems. That has NOTHING to do
with the AMD chips of today. Of course Intel has screwed up before and will
in the future. You AMD zealots act like everything AMD has ever produced
was gold!

 

[Stacey]

When / IF linux ever gets as popular as windows, there will be plenty of
malicious code written just for it.......

Given it (unix) is used in many very secure environments and servers, you'd
think in 20+ years someone would have come up with ONE virus by now just to
prove it can be done? As big a hard on as M$ has to destroy linux they
would if they could that I am sure of. If you had a clue as to how the
linux filesystems works you'd realize how difficult trying to get
self-installing malicious code to destroy computers would be.. The WORST it
could do is ruin the files in one users home directory.

So if M$ doesn't want viruses to spead, whey enable active-X in email and
the preview pane that defaults to open EXE files etc stuff that is totally
unneeded except to spead viruses?

 

[Ancra]

Like?

- Sigh... - Like the sheduler needs to save off the additional
registers, introduced with PIII...

98SE runs a P4 fine (as well as an athlon) without any added >"patches"
required. What has changed with windows (or linux?) to accomodate new
CPU's? Maybe some optimazations for stuff like SSE etc but nothing that
caused them to not boot or corupt data. Do you really believe it was smart
of AMD to release a chip that wouldn't run right on the premier OS of the
time and then expect them to come up with a patch so it would? That isn't a
bug?

? - As 'smart' as releasing chips that don't fit on old boards, but
require mobo manufacturers to come up with new ones!
Meaning: - Yes I think it's smart!

The Win95 timing bug would probably have popped up eventually, even
without the K6.

Maybe there's a good chance an OS treating a cpu entirely as a '386
would work on anything, barring timing bugs. I'm not sure. But then it
wouldn't take advantage of improved features. And a problem will
surely pop up if it assigns the wrong feature to the cpu.

Let's assume my interpretation of the patch-code is correct, I'm not
entirely sure.
Then: Original cpu identification and classification failed to assign
correct cache handling to some (late) K6-2s. The code handles an
intentional design feature of the CPU wrong.Thus it's no cpu bug. Also
seems like real cpu bugs can't be fixed with software patches.

Seems sloppy, not to make sure _all_ OS builders are informed of
changes well in advance. That's normally how it works, I guess. But it
might have been Linux wasn't a towering presence in AMD consciousness
at the time. Might also have been the information really was there,
but Linus slipped up. I don't really care. Explanations for exactly
why Win95 (timing bug) and Linux needed retrospective patches, are
irrelevant.

Really relevant is that Linux 'missed' the timing bug and Win9x
'missed' the cache write handle bug. How do you explain that, from
your stand that the K6 is to blame?

Lighten up dude... The AMD K6's had obvious problems. That has NOTHING to do
with the AMD chips of today. Of course Intel has screwed up before and will
in the future. You AMD zealots act like everything AMD has ever produced
was gold!

I don't see any evidence of the K6 not actually functioning correctly.
Which is what this discussion was about.
(it's fun to be dude, though. :-D)


ancra

 

[Ralph Wade Phillips]

Howdy!

Given it (unix) is used in many very secure environments and servers, you'd
think in 20+ years someone would have come up with ONE virus by now just to
prove it can be done? As big a hard on as M$ has to destroy linux they

<cough> Morris Worm.

There HAVE been Unix worms, and there are even more exploits being
tried on Linux now than on Windows.

There's enough reason to bitch at Microsoft - there's no need to
outright lie about Windows vunerabilities.

would if they could that I am sure of. If you had a clue as to how the
linux filesystems works you'd realize how difficult trying to get
self-installing malicious code to destroy computers would be.. The WORST it
could do is ruin the files in one users home directory.

So if M$ doesn't want viruses to spead, whey enable active-X in email and
the preview pane that defaults to open EXE files etc stuff that is totally
unneeded except to spead viruses?

Naw, there's reasons for it. it's just that Windows is too stupid
in handling it.

RwP

 

[Conrad Edwards]

Greetings...



Short answer - No... It's not a problem...


It's irrelevant.

I believe AMD don't have thermal overload protection on a lot of their
cpu....so if your fan fails or your heatsink isn't properly seated,
the cpu can overheat and destroy itself.
Intel do have thermal overload protection.

That's the only drawback i see....and I've run AMD ever since my last
Intel cpu which was an overclocked Celeron

 

[spodosaurus]

I believe AMD don't have thermal overload protection on a lot of their
cpu....so if your fan fails or your heatsink isn't properly seated,
the cpu can overheat and destroy itself.
Intel do have thermal overload protection.

So do AMD. The motherboard manufacturers haven't bothered to add the
protection until recently. Don't blame AMD for that.

Ari (who has only ever had thermal problems with an Intel Pentium CPU)

That's the only drawback i see....and I've run AMD ever since my last
Intel cpu which was an overclocked Celeron

--

Are you registered as a bone marrow donor? You regenerate what you
donate. You are offered the chance to donate only if you match a person
on the recipient list. Call your local Red Cross and ask about
registering to be a bone marrow donor.

spam trap: replace shyah_right! with hotmail when replying

 

[Ed Light]

The latest motherboards have a too-hot cpu cutout in the bios.

You can use motherboard monitor and a freeware system shutdown program to
try to get a clean shutdown at a lower temperature.

For instance, 70C in bios, 55C in MBM.