GIGABYTE GA-MA78G-DS3H memory question
 

When I built my system, I bought 4GB of G.SKILL 240-Pin DDR2 800 RAM
(2x2GB chips).

http://www.newegg.com/Product/Produc...82E16820231111

My day job is doing Motion Graphics using After Effects and Cinema 4D
which both run without any problems. I never have any isssues with
corrupted images or corrupted video files.

During my "off time", I play World of Warcraft and I have been having
issues since the new expansion came out with graphics image files
getting corrupted and the game crashing on me.

I have done everything that Blizzard has asked and worked with a couple
of different techs and the final conclusion THEY came to is that I am
not using memory listed on the "approved memory" list for the
motherboard manual, so therefore my memory (even though it passes all
memory tests) is causing the corruption and not their crappy software.

Is there ANY possibility that this is true? Especially considering that
I run 30 hour renders and do high end 3D rendering on a regular basis
with no issues.

Is it worth it for me to swap out the G.SKILL memory for something on
Gigabyte's list?

GIGABYTE GA-MA78G-DS3H memory question
 

DaveN wrote:
When I built my system, I bought 4GB of G.SKILL 240-Pin DDR2 800 RAM
(2x2GB chips).

http://www.newegg.com/Product/Produc...82E16820231111

My day job is doing Motion Graphics using After Effects and Cinema 4D
which both run without any problems. I never have any isssues with
corrupted images or corrupted video files.

During my "off time", I play World of Warcraft and I have been having
issues since the new expansion came out with graphics image files
getting corrupted and the game crashing on me.

I have done everything that Blizzard has asked and worked with a couple
of different techs and the final conclusion THEY came to is that I am
not using memory listed on the "approved memory" list for the
motherboard manual, so therefore my memory (even though it passes all
memory tests) is causing the corruption and not their crappy software.

Is there ANY possibility that this is true? Especially considering that
I run 30 hour renders and do high end 3D rendering on a regular basis
with no issues.

Is it worth it for me to swap out the G.SKILL memory for something on
Gigabyte's list?

The Newegg reviews look pretty good, in terms of the number of issues
people are having. The memory is loose, at 6-6-6-18 DDR2-800 and 1.9V,
meaning they've slacked off on the specs to make it easier to make.
I have a pack of CAS5 memory and a pack of CAS4 memory, and there are
even some CAS3 DDR2-800 out there. So CAS6 is pretty loose (easy to
make timing).

I don't give any credence to approved memory lists. First of
all, they may not use statistically significant numbers
of DIMMs for testing. Testing a single package of DIMMs
to make one entry in the table, is not real testing. Some
of the Asus lists, may show "fail" when a single stick is
used and "pass" when four sticks are used.

I've occasionally seen, on enthusiast sites, that "brand x" memory
chips don't like "brand y" chipsets. So there can be issues which
have not been addressed by the people who write the BIOS. The chips
have adjustments, such as clock skew, drive strength settings
and the like, and not all of those may be exposed in the BIOS.
And even if they were exposed, without a storage scope to look at,
it may be hard to dial the thing in. Trying values in an eight bit
register randomly, isn't going to lead to success quickly.

On my own cheesy motherboard, 1GB sticks work fine, but 2GB sticks
do not. And I've seen mention before, that the tuning for those
is a bit different. Even though, if you look at some
chip specs, a few of the specs appear to be identical.

When I test my board, I can run Prime95 for a while without
an error. Then, if I leave enough spare memory to run a game,
I'll find that one Prime95 thread will die with an error,
after the game starts. So in that example, adding the game
does seem to make a difference to my test results. You can
get Prime95 multithreaded here. You can test with Prime95
by itself. Then "stop" and "exit", start WOW, alt-tab out,
start Prime95 setting the custom memory setting to a
reasonable value (don't use all the remaining memory), and
then go back to WOW. See how long Prime95 runs under those
conditions, and compare to how long Prime95 runs by itself.

http://majorgeeks.com/Prime95_d4363.html

On my Asrock motherboard with DDR2 RAM, I'm basically
using the same settings I started with the first day
I got the board. Because it isn't truly stable with
anything else.

For another experimental test case, you can drop down
to a single 2GB stick, and see if that helps. If it
does help, that lets Blizzard Tech Support off the
hook :-)

Since you have an AMD processor, with the memory
controller actually inside the processor, the
chipset should not influence memory issues. It is
the processor's ability to drive the memory bus, that
is at issue. Sometimes, a little extra Vdimm will help.
Or changing Command Rate to 2T, if it isn't there already.

Paul

GIGABYTE GA-MA78G-DS3H memory question
 

I don't give any credence to approved memory lists.

Those lists are also not fixed in stone. It is the approved memory list at
the time the manual was made or the web page was updated. Those lists
change all the time but there is no way they could test every single type of
memory, memory that would run fine on the board.

--g

GIGABYTE GA-MA78G-DS3H memory question
 

"geoff" wrote in message
m...
I don't give any credence to approved memory lists.

Those lists are also not fixed in stone. It is the approved memory list
at the time the manual was made or the web page was updated. Those lists
change all the time but there is no way they could test every single type
of memory, memory that would run fine on the board.

--g



As above.

What Blizzard is saying is code for "we don't have a clue"

Why would the original run fine and not the expansion?

GIGABYTE GA-MA78G-DS3H memory question
 

"geoff" wrote in message
m...
I don't give any credence to approved memory lists.

Those lists are also not fixed in stone. It is the approved memory list
at
the time the manual was made or the web page was updated. Those lists
change all the time but there is no way they could test every single type
of
memory, memory that would run fine on the board.

--g


Does that make us beta testers for them, then we complain to mfgr they take
it off list if there are too many complaints... lol and we keep buying new
gear

P

GIGABYTE GA-MA78G-DS3H memory question
 

"pokey man" wrote in message
...

"geoff" wrote in message
m...
I don't give any credence to approved memory lists.

Those lists are also not fixed in stone. It is the approved memory list
at
the time the manual was made or the web page was updated. Those lists
change all the time but there is no way they could test every single type
of
memory, memory that would run fine on the board.

--g


Does that make us beta testers for them, then we complain to mfgr they
take
it off list if there are too many complaints... lol and we keep buying new
gear

P



If you're that concerned, then you buy from the list. MS do the same thing,
but I expect it may cost the manufacturer to have your product "certified".

I've never had a problem, other than bad memory, with buying cheap memory
modules in any board.
I could be wrong but I think a few of these memory manufactureres build
outside the specs.