Motherboard Forums


Reply
Thread Tools Display Modes

Adding 2GB to A7v266-E ?

 
 
Gary Fritz
Guest
Posts: n/a
 
      05-06-2006, 05:02 AM
I have an A7V266-E with an Athlon 1800+. I'm running Windows 2000. The
mobo currently has 512MB on it (not sure but probably PC2100).

I need to run VMware on this system, running 2-3 virtual PCs in addition to
the base OS. This is going to take major amounts of RAM. I want to add at
least 2GB.

Problem: finding PC2100 RAM is like finding parts for a model T.

Questions:

What speeds of RAM will work in this mobo? I've looked on the Asus site
(pretty useless) and looked at several RAM mfg sites and haven't found a
definite answer.

Will fast (e.g. PC3200) RAM work alongside the existing slow (PC2100) 512MB
stick? Do different speeds of memory play together OK, or would I be
better off to pitch the 512MB stick and get several GB of the same speed?

Will I need to make any BIOS changes (or whatever) to make faster RAM work?

Would 1 2GB stick be better than 2 1GB sticks, other than the obvious
difference of only using one slot?

Thanks!
Gary
 
Reply With Quote
 
 
 
 
Paul
Guest
Posts: n/a
 
      05-06-2006, 07:34 PM
In article <Xns97BAEA52A5BFDfritzfriicom@216.168.3.50>, Gary Fritz
<(E-Mail Removed)> wrote:

> I have an A7V266-E with an Athlon 1800+. I'm running Windows 2000. The
> mobo currently has 512MB on it (not sure but probably PC2100).
>
> I need to run VMware on this system, running 2-3 virtual PCs in addition to
> the base OS. This is going to take major amounts of RAM. I want to add at
> least 2GB.
>
> Problem: finding PC2100 RAM is like finding parts for a model T.
>
> Questions:
>
> What speeds of RAM will work in this mobo? I've looked on the Asus site
> (pretty useless) and looked at several RAM mfg sites and haven't found a
> definite answer.
>
> Will fast (e.g. PC3200) RAM work alongside the existing slow (PC2100) 512MB
> stick? Do different speeds of memory play together OK, or would I be
> better off to pitch the 512MB stick and get several GB of the same speed?
>
> Will I need to make any BIOS changes (or whatever) to make faster RAM work?
>
> Would 1 2GB stick be better than 2 1GB sticks, other than the obvious
> difference of only using one slot?
>
> Thanks!
> Gary


You can see here, that Crucial thinks it can run with just about
anything.

http://www.crucial.com/store/listpar...66-E&submit=Go

PC3200 memory, can run at DDR400, DDR333, DDR266, DDR200. It is
backward compatible.

With a well designed BIOS, you can mix memory types, and the
BIOS will select the timing of the slowest stick and use that
timing for all of them. So, yes, you could mix it.

I would suggest you buy 2x1GB memory, with a chip type of
64Mx8 chips. There is plenty of cheap memory using 128Mx4 chips,
and I do not recommend buying it. Ebay would be a good source
of the 128Mx4 bad memory, so stick with branded memory from a
reputable source. The Crucial memory should be safe to buy.
Pull the previous memory, and just install the two new
sticks in slot 1 and slot 3.

To see the caveats with the cheaper 128Mx4 memory, read this
advert. It lists the bad RAM and the good RAM, in the same
ad. Notice the restricted list of chipsets known to work
with 128Mx4 chips. That is why I recommend avoiding the
cheap, Ebay-style stuff.

http://www.portatech.com/catalog/memory.asp?ID=285

One reason I would suggest PC3200 memory, and getting 2x1GB
sticks, is if you find your machine bogs down to a crawl running
that applications environment, you can reuse the sticks on
a new Athlon64 motherboard. Then the memory won't be a complete
loss, and can be reused on a new motherboard. But DDR motherboards
won't last forever, so to execute that plan, you'd want to do that
some time this year.

Paul
 
Reply With Quote
 
 
 
 
Gary Fritz
Guest
Posts: n/a
 
      05-07-2006, 04:32 AM
(E-Mail Removed) (Paul) wrote:
> I would suggest you buy 2x1GB memory, with a chip type of
> 64Mx8 chips.


OK, I was looking at some Kingston memory, but the Crucial PC3200 is only
a few bucks more. Probably the Crucial CT12864Z40B for $84 at newegg:
http://www.newegg.com/Product/Produc...82E16820146545

> Pull the previous memory, and just install the two new
> sticks in slot 1 and slot 3.


Why slots 1 and 3?

What happens if I leave the old slower RAM in slot 1 and put the new
Crucial in 2 and 3? I hate to throw away half a gig. I suspect any
speed gains I'd get by running at the faster RAM's speed would be far
outweighed by the increased likelihood of thrashing. (And would my
266MHz FSB mobo run the new PC3200 RAM any faster than the existing RAM
anyway?? Isn't the mobo itself limited to PC2100 speeds?)

> To see the caveats with the cheaper 128Mx4 memory, read this advert.


Sorry, all I see there is a listing of 3 different RAMs.

> One reason I would suggest PC3200 memory, and getting 2x1GB
> sticks, is if you find your machine bogs down to a crawl running
> that applications environment, you can reuse the sticks on
> a new Athlon64 motherboard.


Since PC3200 is actually cheaper than most of the slower RAM I see,
there's no reason not to get it. I just wanted to make sure I wouldn't
have any compatibility problems.

Thanks Paul!
Gary
 
Reply With Quote
 
Paul
Guest
Posts: n/a
 
      05-07-2006, 06:43 AM
In article <Xns97BBE554DA214fritzfriicom@216.168.3.50>, Gary Fritz
<(E-Mail Removed)> wrote:

> (E-Mail Removed) (Paul) wrote:
> > I would suggest you buy 2x1GB memory, with a chip type of
> > 64Mx8 chips.

>
> OK, I was looking at some Kingston memory, but the Crucial PC3200 is only
> a few bucks more. Probably the Crucial CT12864Z40B for $84 at newegg:
> http://www.newegg.com/Product/Produc...82E16820146545
>
> > Pull the previous memory, and just install the two new
> > sticks in slot 1 and slot 3.

>
> Why slots 1 and 3?


In transmission line theory, you can compute a rough transmission
line impedance based on capacitive loading. The more loading per
inch of track, the lower the track impedance. By spacing the two
DIMMs apart, it helps to raise the impedance a little bit (closer
to the unloaded value). The difference in stability between using
slots 2 and 3, versus using slots 1 and 3, would be pretty small,
but it won't stop me from suggesting 1 and 3 to people, for a two
stick install.

>
> What happens if I leave the old slower RAM in slot 1 and put the new
> Crucial in 2 and 3? I hate to throw away half a gig. I suspect any
> speed gains I'd get by running at the faster RAM's speed would be far
> outweighed by the increased likelihood of thrashing. (And would my
> 266MHz FSB mobo run the new PC3200 RAM any faster than the existing RAM
> anyway?? Isn't the mobo itself limited to PC2100 speeds?)


Yes, it is probably limited to DDR266. By all means, try all three
sticks if you want. I'm not stopping you. Test with memtest86+
(www.memtest.org) and see if all three sticks are stable. If
you are having trouble, you'll end up trying different combinations
anyway, so start with three, and work from there. Allow the pass
counter on memtest86+ to do two full passes, error free, before you
try booting into Windows. That is to help prevent your Windows install
from getting corrupted.

>
> > To see the caveats with the cheaper 128Mx4 memory, read this advert.

>
> Sorry, all I see there is a listing of 3 different RAMs.


The advert says, for the right-most (cheap) RAM:

"(128x64 Module - 128x4 DRAM)

* Generic / Unbranded 128x4 DRAM Chips
* 1 Year Warranty

*** Verify that you have a motherboard with a VIA PT800,
KT600, P4X266A chipset or an a SIS 648, 648FX, 746FX
chipset before purchase"

Those chipsets are the ones that support 128Mx4 chips. Notice
there are no Intel chipsets mentioned there. And it also
doesn't say whether any of those chipsets can drive three
of those modules with any stability. In the Mushkin testing
of "stacked ram", they only found one chipset that could
drive multiple sticks like that. Unfortunately, Mushkin has
seen fit to take down all of its nice memory test results
pages, so I cannot give you a URL.

>
> > One reason I would suggest PC3200 memory, and getting 2x1GB
> > sticks, is if you find your machine bogs down to a crawl running
> > that applications environment, you can reuse the sticks on
> > a new Athlon64 motherboard.

>
> Since PC3200 is actually cheaper than most of the slower RAM I see,
> there's no reason not to get it. I just wanted to make sure I wouldn't
> have any compatibility problems.
>
> Thanks Paul!
> Gary


HTH,
Paul
 
Reply With Quote
 
Gary Fritz
Guest
Posts: n/a
 
      05-07-2006, 02:27 PM
(E-Mail Removed) (Paul) wrote:
> In transmission line theory, you can compute a rough transmission
> line impedance based on capacitive loading. The more loading per
> inch of track, the lower the track impedance. By spacing the two
> DIMMs apart, it helps to raise the impedance a little bit (closer
> to the unloaded value).


I see. (I had most of an EE degree decades ago, so I follow what you're
saying reasonably well.) Sounds like if I want to continue using the old
512MB stick, I'll need to get that memtest86+ tool you mentioned, and
verify whatever config I use works OK.

> Unfortunately, Mushkin has seen fit to take down all of its
> nice memory test results pages, so I cannot give you a URL.


For future use, maybe you could find the page you want in the Wayback
Machine? http://www.archive.org/web/web.php

Thanks for the excellent advice!
Gary
 
Reply With Quote
 
Paul
Guest
Posts: n/a
 
      05-07-2006, 06:41 PM
In article <Xns97BC55FFA6E9Bfritzfriicom@216.168.3.50>, Gary Fritz
<(E-Mail Removed)> wrote:

> (E-Mail Removed) (Paul) wrote:
> > In transmission line theory, you can compute a rough transmission
> > line impedance based on capacitive loading. The more loading per
> > inch of track, the lower the track impedance. By spacing the two
> > DIMMs apart, it helps to raise the impedance a little bit (closer
> > to the unloaded value).

>
> I see. (I had most of an EE degree decades ago, so I follow what you're
> saying reasonably well.) Sounds like if I want to continue using the old
> 512MB stick, I'll need to get that memtest86+ tool you mentioned, and
> verify whatever config I use works OK.


The correction to Z0 is to divide by SQRT(1+Cd/C0) where Cd is
the distributed capacitance in farads per inch or similar units.
(Just make sure Cd and C0 use the same units, as they cancel.)
The more capacitance, the lower the equivalent impedance, as an
approximation. When Cd is zero (no DIMMs installed), the correction
factor becomes SQRT(1). The terminations on a memory bus have
to be optimized for one value of impedance, so divergence from
the design point reduces signal quality.

http://www.allbusiness.com/periodica.../444403-1.html

>
> > Unfortunately, Mushkin has seen fit to take down all of its
> > nice memory test results pages, so I cannot give you a URL.

>
> For future use, maybe you could find the page you want in the Wayback
> Machine? http://www.archive.org/web/web.php
>
> Thanks for the excellent advice!
> Gary


Mushkin used an "inline cookie" method, which prevents a
robot from traversing the site. While archive.org would have
been able to archive the top level of the site, they cannot
go down into stuff below the top level. (I know this, because
we couldn't quote a URL while the web pages still existed, and
have them work for people. You had to tell people to hack the
URL, by replacing a certain section of the URL, with a number
randomly generated by the Mushkin website as each person visited.
That is what I mean by an "inline cookie". Very frustrating,
when you are trying to help someone, to have to teach them
how to hack links as well. In a way, maybe it is better that
Mushkin just buried that stuff, if that is the best they can
do in terms of web site design.)

Paul
 
Reply With Quote
 
Gary Fritz
Guest
Posts: n/a
 
      05-08-2006, 02:49 AM
(E-Mail Removed) (Paul) wrote::
> The correction to Z0 is to divide by SQRT(1+Cd/C0) where Cd is
> the distributed capacitance in farads per inch or similar units. ...


OK, you have officially lost me. :-)

Gary
 
Reply With Quote
 
Gary Fritz
Guest
Posts: n/a
 
      05-10-2006, 10:34 PM
(E-Mail Removed) (Paul) wrote:
> Yes, it is probably limited to DDR266. By all means, try all three
> sticks if you want. I'm not stopping you. Test with memtest86+
> (www.memtest.org) and see if all three sticks are stable. If
> you are having trouble, you'll end up trying different combinations
> anyway, so start with three, and work from there. Allow the pass
> counter on memtest86+ to do two full passes, error free, before you
> try booting into Windows.


Yow. I've got the 2GB + 512MB sticks in there now and I'm running
Memtest86+. It looks like it will take 15-16 hours per pass! I will let
it run overnight and complete one pass, but I can't lose my computer for
2 days. I've got work to do.

I put the two 1GB sticks in slots 1 & 2, and the 512MB in slot 3. Dunno
if it makes much difference where?

Memtest86+ says the RAM access speed is 389 MB/sec. I forgot to note the
speed before I put the 2GB in but I suspect it was the same. That's
probably about the max rate for a 266MHz FSB, yes?

Gary
 
Reply With Quote
 
Paul
Guest
Posts: n/a
 
      05-11-2006, 03:07 AM
In article <Xns97BFA8A39A611fritzfriicom@216.168.3.50>, Gary Fritz
<(E-Mail Removed)> wrote:

> (E-Mail Removed) (Paul) wrote:
> > Yes, it is probably limited to DDR266. By all means, try all three
> > sticks if you want. I'm not stopping you. Test with memtest86+
> > (www.memtest.org) and see if all three sticks are stable. If
> > you are having trouble, you'll end up trying different combinations
> > anyway, so start with three, and work from there. Allow the pass
> > counter on memtest86+ to do two full passes, error free, before you
> > try booting into Windows.

>
> Yow. I've got the 2GB + 512MB sticks in there now and I'm running
> Memtest86+. It looks like it will take 15-16 hours per pass! I will let
> it run overnight and complete one pass, but I can't lose my computer for
> 2 days. I've got work to do.
>
> I put the two 1GB sticks in slots 1 & 2, and the 512MB in slot 3. Dunno
> if it makes much difference where?
>
> Memtest86+ says the RAM access speed is 389 MB/sec. I forgot to note the
> speed before I put the 2GB in but I suspect it was the same. That's
> probably about the max rate for a 266MHz FSB, yes?
>
> Gary


With three sticks, I wouldn't expect the ordering to make a
difference. Sometimes, when a motherboard has a particularly
bad BIOS design, changing the order may allow the BIOS to
detect all the sticks - so sometimes you have to change the
order, for the DIMMs to be seen and used. If you got all your
memory to work, then leave it as is.

A while back, looking at datasheets for memory chips, at least
one brand had the same loading numbers for their 32Mx8 and
64Mx8 chips, and that means electrically, the DIMMs should be
indistinguishable from one another.

I'd probably check with CPUZ, back in Windows, and see what
memory clock and timings are being used. You can do that after
you finish your two passes of testing. (www.cpuid.com/cpuz.php
has CPUZ.) Looking at timing values and clocks is a better way
to determine if everything is set up right or not.

With the smaller amount of RAM I've got, two memtest passes might
cost me less than an hour. You've got to expect a bit more test
time with 2.5GB.

In terms of numbers, a DIMM is 8 bytes wide. If the memory rate
was DDR266, that would be 8*266 = 2100MB/sec theoretical transfer
rate (assuming the memory never took a breather). I think in
memtest, my memory (in a single channel mode) gives about 28%
of the theoretical number, and 28% of 2100 is a bit more than
you are getting. But once you get back into Windows, it won't
take long to check the vital statistics with CPUZ.

Who knows. You may end up cranking the memory clock up a bit,
and have to repeat the testing :-)

Paul
 
Reply With Quote
 
Gary Fritz
Guest
Posts: n/a
 
      05-11-2006, 04:06 AM
I spoke too soon. The first 11% of a pass took about 90 minutes, but the
remaining 89% only took about another 90 minutes. In 6+ hours I
completed 2 passes with no errors. So everything is cool? NOT!! :-(

> I'd probably check with CPUZ, back in Windows, and see what
> memory clock and timings are being used.
> Looking at timing values and clocks is a better way
> to determine if everything is set up right or not.


OK, I exited Memtest86+, booted up into Windows, read your note,
downloaded CPUZ, and started to read off some data. First thing I
noticed was that the FSB speed was 136MHz or so -- not what I expected,
since the mobo's box says it's a 266MHz FSB. Should I be concerned?

Then I started looking at the RAM stats. First thing I noticed was that
CPUZ only reported 2GB -- even though the BIOS says it sees 2621440KB.
??? So I went to the System Information tool to see what Windows thought
it had, and... whammo. Sudden crash with vertical stripes across the
monitor. UH-oh.

So I rebooted. Now before it gets to the login page, it flashes a blue
screen (so fast I can't read any more than "STOP") and boots again.

Urrghghghghhhh...

Now what do I do!?
Gary
 
Reply With Quote
 
 
 
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
OCZ 2GB (2x1GB) EL DDR PC-4000 / 500MHz / 2GB Gold Edition / Dual Channel Good for AMD system ??/ Paul Mathews Overclocking 7 11-20-2005 01:04 AM
OCZ 2GB (2x1GB) EL DDR PC-4000 / 500MHz / 2GB Gold Edition / Dual Channel Good for AMD system ??/ Paul Mathews AMD Overclocking 3 11-16-2005 04:26 PM
=2GB RAM memory gives errors, <2gb ok , Is there a design flaw withon D865PERL? nobody Intel 12 12-12-2003 04:19 AM
IC7-G and 2GB Ram Colin Dyckes Abit 5 08-03-2003 01:13 PM
Re: IC7-G with 2Gb wont boot Colin Dyckes Abit 0 08-02-2003 11:28 AM


All times are GMT. The time now is 07:38 PM.


Welcome!
Welcome to Motherboard Point
 

Advertisment