Memory mixing

  • Thread starter Thread starter stevebuy
  • Start date Start date
S

stevebuy

Hi All, this is my first post to this board and I hope this is the
right one to post to.

I have a question about memory mixing and am wondering if anyone knows
the answer (or can point me in the right direction).

I have a PC with a single 256MB stick of RAMBUS 711-45 memory (and a
blank).
I have just got my hands on a single 256MB stick of RAMBUS 600-53
memory.

Can these be used together?

I know there in no problem mixing 600-45,711-45,800-45 sticks together
(the machine would use the lowest common denominator), but can you mix
the "45" and "53".

Many thanks in advance for any help on the matter.
 
Hi All, this is my first post to this board and I hope this is the
right one to post to.

I have a question about memory mixing and am wondering if anyone knows
the answer (or can point me in the right direction).

I have a PC with a single 256MB stick of RAMBUS 711-45 memory (and a
blank).
I have just got my hands on a single 256MB stick of RAMBUS 600-53
memory.

Can these be used together?

I don't know much about the i820 chipset, but it won't hurt to
try. If the modules don't work one way, try switching the slots
they're in. To avoid possible data corruption, download
Memtest86 from http://www.memtest86.com before you start
experimenting with the memory. Then if your computer boots
with the two modules, run Memtest86 for a few cycles before
you boot your OS. You should get zero errors.

-- Bob Day
http://bobday.vze.com
 
You may have a problem mixing the 2 sticks you mentioned, I don't know for
sure though. Are there any memory frequency settings in the BIOS?
Not only is the PC600 slower bus-wise but the chips are probably a lower
grade as well. If you do get it running, boot off a memtest86 disk and make
sure they're totally stable before even going into Windows.
I doubt you can hurt anything by testing it out so give it a try and post
back.
 
Appeciate the help guys. I'll test it out with memtest86 over the
weekend and post back early next week.
 
I installed the 600-53 and 711-45 together and rebooted using
memtest86. It ran for over an hour (at least 8 tests). It made a full
pass though the tests without error. I then booted the OS (Win2000) and
everything seemed fine.

Having passed memtest86, does that mean that the configuration is
stable and I wont get any more instability over the use of the machine
compared with using a matched pair ?
 
I installed the 600-53 and 711-45 together and rebooted using
memtest86. It ran for over an hour (at least 8 tests). It made a full
pass though the tests without error. I then booted the OS (Win2000) and
everything seemed fine.

Having passed memtest86, does that mean that the configuration is
stable and I wont get any more instability over the use of the machine
compared with using a matched pair ?

YMMV and no guarantees, but "the signs point to yes". I'd
recommend running Memtest86 for a couple of more passes.

-- Bob Day
http://bobday.vze.com
 
memtest86. It ran for over an hour (at least 8 tests). It made a full
pass though the tests without error. I then booted the OS (Win2000) and

To be sure you need to run memtest for at least 3 or 4 hours during the
hottest part of the day (for the room they're in).
--



"I don't cheat to survive. I cheat to LIVE!!"
- Alceryes
 
I installed the 600-53 and 711-45 together and rebooted using
memtest86. It ran for over an hour (at least 8 tests). It made a full
pass though the tests without error. I then booted the OS (Win2000) and
everything seemed fine.

Having passed memtest86, does that mean that the configuration is
stable and I wont get any more instability over the use of the machine
compared with using a matched pair ?

Memtest is best seen as a method to detect errors IF they're
occuring, not to prove there are none, ever. The longer you
test with memtest, the less likely there are errors, and if
you were to overclock the memory bus some and retest, you'd
even further increase the odds that you have no errors
because you've put some margin inbetween the memtest results
and the regular running state of the system.

To be sure you have no errors (at least as much as
reasonably possible) you need ECC memory. Most people don't
find it reasonably necessary to be that sure though, or to
put it another way, the OEMs decided to cut a few $$ off
their systems and everyone else just followed their lead. I
suspect if OEMs had continued using ECC memory, quite a few
homebuilders would've, too.
 
This is all very helpful guys - many thanks.

Okay, suppose I test using memtest for 24 hours and get no errors. And
suppose I am willing to assume the system is stable, can I assume that
the performance of 256M of 711-45 and 256M of 600-53 will be simpler to
that of 512M of 600-53 (or will there be some performance hit for
balancing 2 different units)?
 
This is all very helpful guys - many thanks.

Okay, suppose I test using memtest for 24 hours and get no errors. And
suppose I am willing to assume the system is stable, can I assume that
the performance of 256M of 711-45 and 256M of 600-53 will be simpler to
that of 512M of 600-53 (or will there be some performance hit for
balancing 2 different units)?

There is little point in assuming as much since it is
relatively easy to benchmark. Sisoft Sandra provides a
relatively simple benchmark, but an even simpler one is that
figure reported by Memtest86+ right when it first loads....
if the memory is running at same speed with both sticks you
should have very nearly (there is a minor deviation in
memtest's reported figures for this) same as with only one.

IMO, memtest86+ is only the beginning, the first test to run
before ever booting the OS. At that point it's useful to
run Prime95's Torture Test for a few hours, ZIP or RAR some
very large files and do CRC checks, and if it's a system
meant able to game, loop 3DMARK (whichever version is most
appropriate for your particular video *card*) for at least
an hour.
 
Back
Top