Felix Miata wrote:
> Lyvim Xaphir wrote:
>
>
>>On Sat, 2002-03-23 at 16:49, Felix Miata wrote:
>>>I currently have four machines with AMD chips and one (486) with Intel.
>>>If I had to buy a motherboard today, it would probably be one that uses
>>>an Intel CPU. Most AMD motherboards use performance crippled VIA
>>>chipsets. How good the CPU is matters little when the primary bottleneck
>>>is the I/O bus.
>>>
>>overclocked to 964 mhz. If you've got more URL's backing up what you
>>
>
> I don't overclock anything. I want bulletproof.
>
I'd like to hear your definition of bulletproof. I don't think
you can get bulletproof in any variation of todays PC architecture.
But it doesn't sound like you want to spend the money for bulletproof.
>
> I spent quite some time looking for the URL I provided. I started first
> at Tom's Hardware, thinking that's where I had seen it. My mistake.
> Nowhere there could I find a relevant comparo of chipset performance at
> Tom's. There was a chipset link there, but quite ancient. I spent a
> bunch more time at Google extracting what I was after to show the list.
>
I spent alot of time reading it too. And I still don't understand
where you get the 'handicapped' from. There doesn't seem to be any
evidence that there are any bugs or design flaws in the gates of the
ASICS. There only seems to be bugs in how the software drivers
initialize the registers. The patches all seem to fix the problem
easy enough. Calling this a 'handicap' is like those who call the
VIA 4 in 1 drivers 'Patches for windows'
On top of that, while it did seem that this might be the cause of
some of the problems many people have been seeing when using some
PCI cards in a VIA mother board, it seemed that this performance
bug really only affects large burst transfers. Now I have no idea
what you're using your machines for, but that shouldn't show up very
often. Actually the fact that it hasn't been more widely reported
leads me to think that hardly anyone will really hit this driver
bug.
> Maybe you missed the point "if . . . today". Most VIA boards in retail
> inventory today either require new RAM (DDR or PC2100 or whatever) or
> have the handicapped chipsets (most). If I was buying at today's higher
> than when I bought memory prices, I'd want to use my PC133 memory, not
> spend just as much for (my current amount of) RAM as for the motherboard
> and CPU.
>
It may be true that there are no AMD chipsets out that still take PC133.
I don't know, I haven't looked. As PC2700 takes off PC2100 should come
down in price. (You can't stop the progress of technology,) though I
don't think you can expect to see prices again like we saw this past
summer and fall. There was a lot of dumping going on that brought prices
artifically low, and I believe some companies are in rough shape from
it. Be happy you got what you did.
> As it happens I did some research on exactly this, as my sister's Tyan
> 1846 (Intel BX) has eaten four HD's in 26 months and I was investigating
> what to replace it with without needing to buy RAM to replace the 768 Mb
> we paid $69 for in November. AMD chipsets need faster RAM, as do the
> non-handicapped VIA. SiS support doesn't seem to be mature enough
> lately, so that leaves only Intel vs. handicapped.
>
There are far more (what's wrong with ALi? memory?) different brands
and models to choose from in the AMD market than there are on the Intel
end of things. And once you find that's acceptable to you, you'll get
much better performance from an AMD system. Also with the added
competition you'll probably find something you can use that is even
cheaper.
--------
Now on something totally different. I think this grew from the thread
that started with someone complaining that the P4 that they either had
or were going to get wouldn't run Win95.
I can't beleive that this was attributed to some conspiracy between
Intel and Microsoft. Yeah right (There are other things they're in bed
on -- but this ain't one of them.) It's more likely Intel just not
wanting to keep spending the large sum of money it takes to continue
the backward compatibilty that Win95 would need. The number of people
out there that will decide not to buy a P4 because of this is so small
that it doesn't justify the expense. It's Capitalism, not Conspiracy.
I don't think people realize how much money, time, and effort, it takes
to design, test, support, etc this type of thing. These chips are very
complex. The machines they end up a part of are even more so. It takes
an extraordinary effort to truely make sure of compatibility.
This is also why I don't understand how anyone can say a motherboard,
chipset, or OS (like Mandrake Linux) is totally crap and won't work
for anyone just becuase it didn't work for them. If microsoft with
their vast budget can't test every hardware combination, How does anyone
expect Mandrake to? At least Microsoft has either inside knowledge
when writing drivers, or the hardware vendor itself writes the windows
drivers.
Linux has to rely on volunteers. Who many times don't have the chipset
documentation, or when they do, it contains errors and/or omissions, and
they can't just call or walk down the hall to the ASIC designer and ask
a question.
As high as we all think MS's prices are. They spend pretty much every
dollar they take in. Where do you think it all goes? Not all of it
goes to testing but a big chunk does. Even with all of us sending
contributions to RedHat or Mandrake it still wouldn't be a drop in the
bucket compared to to the resources MS has.
Mandrake, Redhat, etc dont have the money to go out and buy the HW to
put together every motherboard with every sound card with every video
board and every network card. Every one of those things (and more) is a
variable in an extremely complex system. And if you change even just
one thing it can cause changes in other places that aren't readily
obvious. Basically if you've ever gotten a distribution that worked
flawlessly, you got lucky. You practically won the lottery. The odds
are probably that great.
Obviously the people who develop and distribute Linux put alot of
work into it (I don't want to downplay the effort these people put into
the community,) so on popular combinations it's hard work and design,
not luck, that it works alot of the time. The luck is in the fact
that you happen to own one of the HW combinations that the developers
owned and tested.
This is why the Linux distributions rely on the end users to test beta
releases and report bugs. If you like your distribution of Linux, and
want to make sure the next release will continue to work on your
machines, then you need to try it out. You may be the only person out
there with that CDROM plugged into that port on that mother board
alongside that Harddrive. If you don't test it who is going to??
If you can't afford to wipe your working system to load something that
will probably have bugs the first time around, then go buy a second
hardrive, or dust one off that you were going to throw out. Pick a
weekend. Put it in, load it up, try out the things you do everyday.
Take notes on the bugs and problems.
When the weekend's over, put the original drive back in. Go back to
your current stable release. Lastly make sure you write up your findings
ans send them in. Keep that second drive around though. If a later
release is suposed to fix your problems, then put it back in, load up
the changes and try it out.
I don't want to talk anyone out of paying for the Mandrake CD's or
sending in a contribution, but it may just be that $50 or $60 bucks
spent on a 'testing' Harddrive might actually be a better investment
for the linux community as a whole.
-Kyle
--
_
-------------------------------ooO( )Ooo-------------------------------
Kyle J. McDonald (o o) Systems Support Engineer
Sun Microsystems Inc. |||||
Enterprise Server Products [EMAIL PROTECTED]
1 Network Drive BUR03-4630 \\\// voice: (781) 442-2184
Burlington, MA 01803 (o o) fax: (781) 442-1542
-------------------------------ooO(_)Ooo-------------------------------
Want to buy your Pack or Services from MandrakeSoft?
Go to http://www.mandrakestore.com
