General said:
Greetings.
Using Everest diagnostic software, under DMI [Memory Controller] indicates:
Memory Controller Properties:
Error Detection Method None
Supported Memory Speeds 70ns, 60ns, 50ns
Supported Memory Types DIMM, SDRAM
Supported Memory Voltages 3.3V
Maximum Memory Module Size 512 MB
Memory Slots 3
I have a 512MB in slot 1 and nothing in slots 2 or 3.
Is the Maximum Memory Module Size meaning per slot, or is it the most the
motherboard can handle?
I did experiment by placing a 256MB in slot 2 and 128MB in slot 3. It
recognized the sum of all 3, but the display would go blank for 1-2 seconds
every minute and ultimately the system (or at least the graphics) would
freeze and I would have to do a hard shut down. Am I looking at RAM that's
likely bad, or is the system not able to handle over 512MB?
If it helps, my system is
Computer:
Operating System Microsoft Windows XP Professional
OS Service Pack Service Pack 3
DirectX 4.09.00.0904 (DirectX 9.0c)
Computer Name HP8705 (hp8705)
User Name Kevin
Motherboard:
CPU Type Intel Pentium IIIE, 800 MHz (6 x 133)
Motherboard Name Asus Pegasus
Motherboard Chipset VIA VT82C694X Apollo Pro133A
System Memory 512 MB (PC133 SDRAM)
BIOS Type Award Medallion (07/26/00)
Thanks for you reply in advance.
B.rgds,
Kevin
HP8705 with Apollo Pro 133A chipset. The chipset is here, and
appears to support up to four slots of 512MB each, for a total
of 2GB. Generally, if a designer installed four slots, the
memory speed would be limited to PC100. By using 3 slots only,
they can handle PC133. So they "threw away one slot" on purpose.
That means you can have 3*512MB for a total of 1536MB.
http://www.via.com.tw/en/products/chipsets/legacy/pro133a/index.jsp
I recommend testing each DIMM separately. And you should be using
memtest86+, before booting any OS.
Install the 128MB by itself. Run memtest86+ at 133MHz memory speed.
(Do two complete passes. The pass counter is on the upper right
hand side of the screen. Since these are quick tests, just do two
passes. If you're in a real hurry, you can just run Test #5.)
Install the 256MB by itself. Run memtest86+ at 133MHz memory speed.
When installing a single stick, place it furthest from the processor,
in slot 3.
Next, test two sticks. Install them in slot 1 and slot 3. If it
doesn't work, swap the sticks between those two slots. Now, run
memtest86+ again, on the 384MB total memory.
Next, test the 512MB stick with one of the other sticks. Try
slots 1 and 3 again. Use your memtest86+ floppy.
Now, depending on how well things have gone, now fill the
empty slot 2 with the remaining stick. And retest again
with memtest86+.
The above test sequence, is intended to highlight how "nervous"
your memory bus is. If you encounter resistance at some point,
then maybe it isn't happy. You can play with memory timings,
or change the bus speed (that is, if the BIOS allows it). On
some of the older motherboards, this was easy, as you had
jumpers or DIP switches for overriding the FSB BSEL bits, or
for (attempting) to change the multiplier. Most processors
of that era will be locked (all mine were), so the multiplier
switches never did anything for me. But I did play with FSB
choices, as part of testing.
In your case, you want as much memory, as tests without errors.
If it simply won't run three slots at 133MHz, then just use
the two sticks (512MB + 256MB).
In any case, don't boot into Windows with an untested memory
configuration. You should always do the memtest86+ thing first,
to reduce the risk of corruption in WinXP.
http://www.memtest.org
The floppy package, includes a tool to write the executable to
a floppy. There will be no file system visible on the floppy, once the
installer is finished. The memtest program runs with no OS, and
basically takes control of the machine when it runs. The program
is designed this way, so that as much of the memory as possible
can be tested. Less than 1MB of memory will be untested, because
memtest86+ won't write to any areas marked "reserved for the BIOS".
But it will test the rest of the memory. Which is better than
running a memory tester from within Windows.
http://www.memtest.org/download/4.00/memtest86+-4.00.floppy.zip
A second (stress) tester program, is Prime95. It is for usage, once
you've booted into Windows again. It is a more sensitive test of
memory problems, but because it relies on the OS, it can't be run
first, before booting into Windows. I run that program for four
hours, and if no errors are detected, then I'm finished with testing.
(Note - web site is flaky, and isn't always reachable. Be patient.)
http://www.mersenne.org/freesoft
When using that program, indicate you're "stress testing only". You
don't have to join GIMPs to use the program. The program will run
a thread on each core of a multicore processor. You can select
a smaller amount of memory to test, than the default. I sometimes
do that, if I'm web surfing and testing at the same time.
I guess the current version is 25.11. I have 25.5 and 25.6 here.
http://mersenneforum.org/gimps/p95v2511.zip
The older versions are here.
ftp://mersenne.org/gimps_archive/archived_executables/
ftp://mersenne.org/gimps_archive/archived_executables/p95v256.zip
My old 440BX board has four SDRAM slots (for PC100). If I install
2*256MB, it is as stable as can be. As soon as I install the third
stick, the graphics card operation becomes extremely unstable. But
with only two sticks (doesn't matter what slots), I've run Prime95
for 16 hours on it, without error. I never did find out whether
that was a chipset bug or whatever. The chipset was from Intel.
The behavior is reproducible in both Windows and Linux, so it
isn't an OS problem.
HTH,
Paul
