PC3200 in place of 2100? 400 MHz General question but
mm
it for 2 years and I don't know how old it was when I got it.
I have 512M** and I wanted to buy another 1G of memory, but mostly
what they sell now is pc3200 instead of pc2100. Should I hunt
further for 2100 or something in between?
Also, the ones I've seen for sale are 400Mhz. Is that too fast for
the mobo?
I really don't understand what makes something too fast for something
else, and someday I'd like to learn, but right now the more important
thing is to buy the right thing.
P.S. I'm not going to do any overclocking (whatever that is. :-) )
P.P.S. If I buy the 1 gig, should I move the 512 to the second slot
and put the 1 gig in the first slot? Or does it matter?
Thanks a lot.
If you are inclined to email me
for some reason, remove NOPSAM :-)
Paul
The user manual is here.
http://dlsvr02.asus.com/pub/ASUS/mb/socka/amd760/a7m266/a7m266-104.pdf
It is a two socket AMD761 board, which can use registered or unbuffered memory,
but not both types at the same time. If you want to reuse your current stick,
make sure both the old one and the new one are the same type. You can use
CPUZ from www.cpuid.com , to get some basic info about your system as it is now.
The speed thing is backward compatible. A DDR400 stick can be used on a
DDR266 motherboard, for example. In fact, if you look at a datasheet for
one of the memory chips, the memory chips themselves are actually rated
to run slower than DDR200 if necessary.
Assuming the chipset on the motherboard is density limited (number of address
bits on memory bus is barely enough to address a 1GB DIMM), you'd want
to use a branded 1GB stick. This Kingston module, uses (16) 64Mx8 chips,
and would be the right construction for the job. If you were thinking
of buying Ebay RAM, there is a lot of (16) 128Mx4 chip modules for sale there,
and the memory has no brand name on the module. (That is because a branded
memory company would be ashamed to make such a module. But Ebay has no trouble
selling them, no ethics problem there...)
(Example of a module that uses 64M x 8 bit chips)
http://www.valueram.com/datasheets/KVR400X64C3A_1G.pdf
The fact that it is a CAS3 stick at DDR400 shouldn't matter, because
when you operate the stick at DDR266, the CAS setting can be dropped
down to CAS2. And CAS2 is the fastest setting possible for CAS. So
even a CAS3 DDR400 stick would be good enough for the job. (Don't worry
about the timings on the first day - just plug it in and use it :-) That
is what the "Auto" setting in the BIOS is for. You can play with the
timing some day when you're bored.)
When you get the memory, you can try testing it by itself, using a
program like memtest86+ from memtest.org . If the stick by itself
tests OK, then add in the second stick and retest. The objective
is to see if the bus operation is upset at all, by having the two
sticks present. It shouldn't be, but it is a good idea to do a little
testing, before attempting to boot back into Windows. I've had at
least one poster, who followed the precaution of trying memtest86+, only
to have Windows corrupted by the new memory anyway. So memtest86+ is
not a guarantee that the memory is perfect -- it is just the best
test available right now, for the job.
Memtest86+ is a self-booting test program. You can place it on a
floppy disk or on a CDROM, and boot the computer using the
memtest86+ disk. Since memtest86+ never goes near your hard drive,
there is little chance of something happening to it.
This is what the memtest86+ screen looks like. The test will run
forever, unless you stop it. Pressing the Escape key (ESC) will
abort the test and cause the system to reboot. You'd want at least
a couple full test passes to run, which takes a couple hours. If there
are any errors, they should be listed in the middle of the screen.
No errors are acceptable for memory.
http://www.techpowerup.com/reviews/DFI/LPNF4Expert/images/b_memtest.jpg
Paul
mm
>mm wrote:
>> I have an A7M266 mobo and of course it is several years old. I've had
>> it for 2 years and I don't know how old it was when I got it.
>>
>> I have 512M** and I wanted to buy another 1G of memory, but mostly
>> what they sell now is pc3200 instead of pc2100. Should I hunt
>> further for 2100 or something in between?
>> ....
>>
>The user manual is here.
>http://dlsvr02.asus.com/pub/ASUS/mb/socka/amd760/a7m266/a7m266-104.pdf
Tbanks. I have the manual somewhere, but this was easier. (My house
is a mess right now. :) )
>It is a two socket AMD761 board, which can use registered or unbuffered memory,
>but not both types at the same time. If you want to reuse your current stick,
The unbuffered has 8 chips per side, and the other 9, right? I can
find out exactly what I bought the first time, but the one I bought
has 8.
>make sure both the old one and the new one are the same type. You can use
>CPUZ from www.cpuid.com , to get some basic info about your system as it is now.
Very nice. Thank you.
>The speed thing is backward compatible. A DDR400 stick can be used on a
>DDR266 motherboard, for example. In fact, if you look at a datasheet for
>one of the memory chips, the memory chips themselves are actually rated
>to run slower than DDR200 if necessary.
Great.
I see one for sale that has an "aluminum heat spreader for thermal
diffusion". Is this heat spreader ok in A7M266, or might it not fit
or somethign?
KHX3200A/1G Kingston Memory 1GB/184-Pin/DDR SD RAM/400
new in original box , tested works .
Kingston HyperX Features:
* Capacity: 1gb
* Aluminum heat spreader for thermal diffusion
* 184-pin Unbuffered DDR Modules
* 400MHz settings: 2-3-2-6-1 (CAS Latency 2)
* PC3200 PCB height: 1.250" (34.78mm)
Thanks a lot.
>
> Paul
Paul
>
> I see one for sale that has an "aluminum heat spreader for thermal
> diffusion". Is this heat spreader ok in A7M266, or might it not fit
> or somethign?
>
> KHX3200A/1G Kingston Memory 1GB/184-Pin/DDR SD RAM/400
> new in original box , tested works .
> Kingston HyperX Features:
>
> * Capacity: 1gb
> * Aluminum heat spreader for thermal diffusion
> * 184-pin Unbuffered DDR Modules
> * 400MHz settings: 2-3-2-6-1 (CAS Latency 2)
> * PC3200 PCB height: 1.250" (34.78mm)
>
>
That is a pretty fancy stick.
http://www.newegg.com/Product/Product.aspx?Item=N82E16820134040
When they put a heat spreader on a stick, they are supposed
to make it thin enough to fit standard slot spacings. So it
should still fit. It is OK if one heat spreader touches another.
A heat spreader helps, if you are abusing the memory. Like
when an overclocker turns up the memory voltage and the memory
runs hotter.
If you are buying the memory second hand, make sure the seller
is willing to do the right thing, if testing it uncovers
problems. Test tools you can use are: memtest86+ (memtest.org)
on a floppy, before you boot Windows. And perhaps Orthos or
Prime95, once you get into Windows.
http://sp2004.fre3.com/beta/beta2.htm (Orthos for Windows)
Memtest86+ tests all the memory (all except a tiny amount of
reserved space). Orthos or Prime95, only tests the memory not
occupied by the Windows OS, so while Orthos or Prime95 are a
more stressful test, they cannot test the entire memory. That
is why using both tools is recommended, to get the best
coverage you can get for free.
The only memory that really needed a heat spreader, was
RDRAM. On RDRAM, one memory chip could get hotter than the
others, and the chunk of metal helped spread out the heat.
DDR doesn't work that way, and all chips run at the same
temperature. The memory chips have a significant surface area
on their own, and when the heat spreader is missing, there is a
tiny bit more room for air to circulate. Which is why, I feel
that a heat spread on DDR memory, is a mixed blessing. For the
unscrupulous, it offers an opportunity to hide the brand and
quality of memory chipset being used (like UTT chips). Usually
the warranty will be voided, if the heatspreader is removed,
which means what is underneath can remain a secret.
Paul
Helob
On Mon, 24 Sep 2007 04:30:01 -0400, Paul <nos...@needed.com> wrote:
Paul,
I would appreciate if you can give me advise in my case.
I have been running 2x512MB (DDR400) Apacer RAM with mem Timing of
2.5-3-3-8 on my Asus P4P800SE WinXP Pro system with Mem Frequency set
at auto in BIOS and timing at 2.5-3-3-8.
memtest86 test all passed.
Recently I bought an addition 2x512MB (DDR400) Kingston RAM with mem
Timing of 3.0-3-3-8 and added to the system.
However on running memtest86, the mem frequency was drop to 166
instead of 200. All tests passed.
I reconfig Men Frequency to manual and set it at 200 and rerun test.
All tests passed.
In both cases, Win XP boot and operate as normal, both showing 2GB of
memory.
Questions:
1. What is the best BIOS Memory timing I should set in my case with
mix pairs of CAS memory modules timing?
2. Why BIOS default to mem frequency of 166 even both pairs are rated
at DDR400? Any harm in maually setting it at 200?
3. I have a problem with Kernel Pacnic running OS X system with the
added memory. What is your advise on this?
Thank you for your time.
Paul
>
> Paul,
> I would appreciate if you can give me advise in my case.
> I have been running 2x512MB (DDR400) Apacer RAM with mem Timing of
> 2.5-3-3-8 on my Asus P4P800SE WinXP Pro system with Mem Frequency set
> at auto in BIOS and timing at 2.5-3-3-8.
> memtest86 test all passed.
>
> Recently I bought an addition 2x512MB (DDR400) Kingston RAM with mem
> Timing of 3.0-3-3-8 and added to the system.
> However on running memtest86, the mem frequency was drop to 166
> instead of 200. All tests passed.
> I reconfig Men Frequency to manual and set it at 200 and rerun test.
> All tests passed.
> In both cases, Win XP boot and operate as normal, both showing 2GB of
> memory.
> Questions:
> 1. What is the best BIOS Memory timing I should set in my case with
> mix pairs of CAS memory modules timing?
> 2. Why BIOS default to mem frequency of 166 even both pairs are rated
> at DDR400? Any harm in maually setting it at 200?
> 3. I have a problem with Kernel Pacnic running OS X system with the
> added memory. What is your advise on this?
>
> Thank you for your time.
When you have two matched pairs with different timing, on a
dual channel board with four slots, you should use the slowest
timings of the two sets of timings. In your case, that would be 3.0-3-3-8,
as 3.0 is slower than 2.5 .
The BIOS used a conservative clock setting, to give a stable result.
Adjusting the clock manually, then testing the memory, is a good
approach to tuning up the system. The memory is rated for DDR400, so
the DDR400 setting is safe from that perspective.
For DDR, there will be a bus loading effect, from using four sticks.
If you were an overclocker, it means you might get to DDR460 when
using four sticks, versus making it to DDR480 or DDR500 with only
two sticks. So there is some impact from using the four sticks.
The DDR400 setting should be pretty safe, as your present settings
are not seriously stressing the memory.
If you are having just the odd error with the current setup, see
if there is a Vdimm setting in the BIOS, and set it to 2.7V. That
may provide additional stability.
To test the stability, boot into Windows, then use Orthos. It should
run for hours, without reporting an error. Once Orthos runs without
error, then you can go back to trying MacOSX if you want.
http://sp2004.fre3.com/beta/beta2.htm (Orthos - a test based on Prime95)
The default setting of "blended", will cause Orthos to make a large
memory allocation. Eventually, that memory will get tested by the
various sized tests run by the program. An unstable system will
throw errors in a matter of a minute or two. You want your system
to be able to run that program for hours, without any complaints or
errors seen in the log on the screen. The program stops when it
detects a single error.
Paul
Helob
Thank you for the indepth and useful explaination and advise.
I used Orthos to test the 2GB with timing set to 2.5-3-3-8 as well as
3.0-3-3-8 and both passed after running for more than 4 hours each.
However I am still getting Kernel Panic on my dual boot Mac OSX 10.4.9
during startup
' panic(CPU 0 caller 0x003A8D2F): Unable to find driver for this
platform:"ACPI" '
OSX started up Ok if I removed the added 2x512MB pair.
What should I do now?
Thanks
Paul
>
> Thank you for the indepth and useful explaination and advise.
> I used Orthos to test the 2GB with timing set to 2.5-3-3-8 as well as
> 3.0-3-3-8 and both passed after running for more than 4 hours each.
> However I am still getting Kernel Panic on my dual boot Mac OSX 10.4.9
> during startup
> ' panic(CPU 0 caller 0x003A8D2F): Unable to find driver for this
> platform:"ACPI" '
> OSX started up Ok if I removed the added 2x512MB pair.
> What should I do now?
> Thanks
All I can suggest at this point, is there is something in the BIOS
tables that MacOSX doesn't like. The BIOS passes tables to the OS.
This is a snipping of a portion of the output of "dmesg" from a
Linux platform. "e820" is a standard for specifying what parts
of memory are reserved. The ACPI portion would presumably be used
by any ACPI software that the system has (and a Mac should have
suspend to RAM, just like other computers). So something in the
information exchange is being upset by the existence of the
extra RAM. You'd need to compare the info shown below, in the
two cases, to begin to figure out why. And since the thing
crashes when 2GB is present, it is going to be difficult to
do that.
*******
Linux version 2.6.6-2-k7 (wli@tisifone) (gcc version 3.3.3 (Debian 20040422)) #1 Wed Jun 16 02:28:18 PDT 2004
BIOS-provided physical RAM map:
BIOS-e820: 0000000000000000 - 00000000000a0000 (usable)
BIOS-e820: 00000000000f0000 - 0000000000100000 (reserved)
BIOS-e820: 0000000000100000 - 000000000fff0000 (usable)
BIOS-e820: 000000000fff0000 - 000000000fff3000 (ACPI NVS)
BIOS-e820: 000000000fff3000 - 0000000010000000 (ACPI data)
BIOS-e820: 00000000ffff0000 - 0000000100000000 (reserved)
0MB HIGHMEM available.
255MB LOWMEM available.
On node 0 totalpages: 65520
DMA zone: 4096 pages, LIFO batch:1
Normal zone: 61424 pages, LIFO batch:14
HighMem zone: 0 pages, LIFO batch:1
DMI 2.2 present.
ACPI: RSDP (v000 VIA694 ) @ 0x000f7610
ACPI: RSDT (v001 VIA694 MSI ACPI 0x42302e31 AWRD 0x00000000) @ 0x0fff3000
ACPI: FADT (v001 VIA694 MSI ACPI 0x42302e31 AWRD 0x00000000) @ 0x0fff3040
ACPI: DSDT (v001 VIA694 AWRDACPI 0x00001000 MSFT 0x0100000c) @ 0x00000000
ACPI: PM-Timer IO Port: 0x4008
*******
OK, maybe you can put just 1GB of memory in your computer, boot into
MacOSX, then run "dmesg" and examine the info.
Paul
Helob
Thanks
The error msg also contain the followings:
Debugger called :<panic>
Backtrace, Format-Frame: Return Address (4 Potential args on stack)
0x24abbdf8 : 0128d1f (0x3c9728 0x24abbe1c 0x131ed0 0x0)
0x24abbe38: 03a8d2f (0x3ed10c 0x32ce7b0 0x32e1a04 0x3356d40)
0x24abbe58: 038a61c (0x334b880 0x32b3c8d 0x1 0x3356d40)
0x24abbeb8: .......................................
0x24abbf38: ................................
0x24abbf88: .................................
0x24abbfc8: 0x19a96c (0x32ce6b0 0x0 0x19d0b5 0x32ccb70)
Backtrace terminated - Invalid frame pointer 0x0
Kernel Version
Darwin Kernel Version 8.8.1: Sat Dec 9 22:18:27 2006;
What does above means?
I will "dmesg" as you suggested and share findings with you.
Paul
:-) You have me mistaken for a hexidecimal wizard.
Generally, I like the "text strings" in things, as they give
better hints about what is broken. The objective here, is
to get enough contextual information about the crash, to
feed to a search engine, for similar symptoms. To actually
debug a problem like this, you'd need a version of the system
with symbols still in place, access to a kernel and applications
debugging programs etc. I've dabbled in this, with various Unix
computers, but never long enough to learn anything. Since the problem
is a kernel panic, you need a kernel debugger program to work on it.
I think I had more fun, playing with Macsbug, which was the debugger
for the previous MacOS. It even had inline documentation.
Paul