all Gskill users please coming in if you have problem for support

[kakaroto]

i see, yeah shipping all the rma memory around the world for rma would be expensive.
are you responsible for the european market then kakaroto?
i hope you can get your mail server working again soon, if you need any help let me know

I'm here on XS to help. Any question to me are most welcome!

European users can mail their shops for RMA. I am not the person for RMA. Technical assistance can be done by Kevin, Gskill HQ and I can do some help too.

The easiest and cheapest way to RMA if return the memory to your (local)shop where you bought it. Then they will return that memory to Gskill.

Just what Kevin said about the RMA that almost 90% of the Ram isn't 100% faulty. The main cause is most of the time the memorycontroller that controls the memory or the settings are wrong.
This is a difficult issue because Gskill RAM requires a overclocked system.

Again make sure there is no other bottleneck.
I will write a guide for Dual channel platform (DFI nF4 and MSI neo4 how to optimize Gskill memory on these boards)

[bias_hjorth]

Okay now its my turn.
I have 1 gb kit LE with 437 tccd. I cant get them to run anywere near stock speeds. I am on this rather old Epox k8da3+ and I tried pretty much every timings / setting. At 2t timings I can boot at 280 but thats about max with these modules. At t1 they are stable around 260mhz 2.83v.
Please let me know what could be the issue. I´ve been talking back and forth with kakaroto on msn a couple of month ago but that didnt gain me nothing.
And its not a htt issue as I can run this board fine all the way up to 350.
The cpu is a a64@3700 2.6 edition.

[saaya]

I will write a guide for Dual channel platform (DFI nF4 and MSI neo4 how to optimize Gskill memory on these boards)

yeah, i recommended that to kevin as well

[clockedOut]

Im next.....
2x256 GSkill DDR600 LA 440, I have never been able to get these sticks to run at 1T. I have tried them with 4 different winchesters and 1 newcastle and 3 different mobo's
AN8 Fatal1ty, DFI Ultra-D, MSI Neo2.

HELP...... The last test was the newcastle core at stock default timings & speed at 1T Prime & Super PI would fail instantly.

What to do....?

EDIT: I'm just gonna RMA these sticks to NewEg and get replacements.

[Damien]

I just ran memtest86+ today. At default speed (400mhz), default timings, and default voltage I got 3638 errors and not even getting through test 6 on the FIRST run. Would this be considered RMA'able?

Pic here

Edit : restocking fee will be $22! I can't believe they charge you even when it's defective. Think I can get them to wave the fee if I give them a call?

[WolfF]

I spent time trying to contact for support.

http://xtremesystems.org/forums/showthread.php?t=54156

LA's doesn't run 300MHz 2T Dual Channel 3-4-4-8

[Damien]

Seems like they work 100% error-free when by themselves. But since it came in a dual channel pack, isn't it still "defective"?

[twiggy]

Seems like they work 100% error-free when by themselves. But since it came in a dual channel pack, isn't it still "defective"?

Sounds like a mem controller prob to me.

[Damien]

Even not o/c'ed at all?

[mad mikee]

The fact that reps from places like Gskill, OCZ, ATI, etc. 'Live' here @ these forums and work so extensively w/ all of us is one of the main reasons that this is my favorite set of forums to visit (And you don't even have to be part of a company to be xtremely helpful around here - information and help just oozes out of the html )

Now for some 'obvious' suggestions as to what constitutes buggy memory. (Most of us know this but I am trying to be explicit for any visiting noobs). Alot of this stuff can also be used to push your memory further if it is okay.

Patience is a virtue here (and no I don't have any either but I muddle through somehow )

1. Sticks should be tested initially at 'normal' speeds and settings (NOT AUTO, put them in!) and voltages (Same as settings).
Need baseline to know if they are getting better .
Run the full memtest, not just 5 (once for starters, just to make sure).
Unless they generate 1000s of errors, you should then 'acclimate' them to your system for 1-2 days @ low (normal) speeds + 1 step up on the volts (you can drop it back when testing later) (burnin has saved me many times)

Of course if you are the lucky recipient of damaged goods(1000s of errors @ lower speeds), stop here and check for stupids, then get tech support

If you have 'borderline' problems - test sticks individually

2. At rated speeds or below, if you get reproducable memtest errors (same addresses, even though maybe different tests) that follow the memory when moved to another slot - buggy memory.

3. Also different sticks usually have a 'favored' slot (drive strength = distance from CPU), Try swapping sticks to see if the # errors decrease and if some errors move w/ the stick.

4. Memory chip settings change for different speeds and I have even had to swap sticks when OCing @ the edge.

I have one stick for 0437 that is a POS(was borderline when I remove the stickers to zap local heating issues, then some bits failed altogether but no stickee - no RMA (OPB, any suggestions? ))

OTOH you see in my sig what I have now (and they are SOLID) Partly because of OPBs, etc settings. etc. (And I am leaving the labels on this time )

I have 3 3200+, the others don't get near what this one does. I also can't get much over 278 on my neo2. So remember that these speeds require matching hardware (chips and MBs) that are up to the job. We forget sometimes that not all hardware is equal (cough CBBID 05.. cough) etc and that most of this stuff is based on reference designs that are based on FSB standards of 200 Mhz .... We're just the lucky ones that look and find the good stuff (well mostly)

[kakaroto]

Im next.....
2x256 GSkill DDR600 LA 440, I have never been able to get these sticks to run at 1T. I have tried them with 4 different winchesters and 1 newcastle and 3 different mobo's
AN8 Fatal1ty, DFI Ultra-D, MSI Neo2.

HELP...... The last test was the newcastle core at stock default timings & speed at 1T Prime & Super PI would fail instantly.

What to do....?

EDIT: I'm just gonna RMA these sticks to NewEg and get replacements.

Have you tested the modules on 1T seperately?

I just ran memtest86+ today. At default speed (400mhz), default timings, and default voltage I got 3638 errors and not even getting through test 6 on the FIRST run. Would this be considered RMA'able?

Pic here

Edit : restocking fee will be $22! I can't believe they charge you even when it's defective. Think I can get them to wave the fee if I give them a call?

What have you tryed?
Use bank 1 & 2 or 1 & 3. Set everything to AUTO and voltage to 2.6v. What are the results? Important that Drive strength is low (AUTO)
And what are results with only one module in bank 1?

I spent time trying to contact for support.

http://xtremesystems.org/forums/showthread.php?t=54156

LA's doesn't run 300MHz 2T Dual Channel 3-4-4-8

Hey dude, sorry that the problem isn't solved yet.

Tryed each stick separately, both run on Single Channel at 310 T1 2.5-4-4-8.
The cause is most of the time the memorycontroller that has difficult with adressing two modules. On a DC platform this is rear, especialy with a FX-55 processor.

OR the board is crappy. This happend to me with my two crappy MSI Neo4 Platinum boards.

Have you tryed the BOXED cooler? The memorycontroller may fail under zero temperature, unlikely but worth to try.

Let use know.

[kakaroto]

Okay now its my turn.
I have 1 gb kit LE with 437 tccd. I cant get them to run anywere near stock speeds. I am on this rather old Epox k8da3+ and I tried pretty much every timings / setting. At 2t timings I can boot at 280 but thats about max with these modules. At t1 they are stable around 260mhz 2.83v.
Please let me know what could be the issue. I´ve been talking back and forth with kakaroto on msn a couple of month ago but that didnt gain me nothing.
And its not a htt issue as I can run this board fine all the way up to 350.
The cpu is a a64@3700 2.6 edition.

If the memory is tested seperately and is still faulty, then it might be the controller that is holding you back.
Have you tested them seperately and what are the results? If they do DDR600 or close to DDR600 1T then your memory is fine.

Even though is it a revision 2.6, there may be still problems with adressing more then one modules.
You can see this at the Command Rate you try to run. Why is 2T running better then 1T on my system.

[Dumo]

I just tried memtest with GskillLA/tccd440@1T and passed 100%..with Ultra-D/bios2.17 and its 24/7 setting....Other than these settings, memtest errored at 7%...Hope this will help....

[kakaroto]

Running 2x 512 at speeds close to DDR600.

Tip:

Run preamble latency at 8ns and Max async. Latency at 5 ~ 6ns.

[Damien]

The ram works perfectly by themselves in every bank but slot 3 (i don't use slot 3 anyways). I've tried every possible combination in all the slots, they just will NOT work together without giving thousands of errors. I honestly think it's either PSU related or Motherboard related though.

[kakaroto]

The ram works perfectly by themselves in every bank but slot 3 (i don't use slot 3 anyways). I've tried every possible combination in all the slots, they just will NOT work together without giving thousands of errors. I honestly think it's either PSU related or Motherboard related though.

which version CPU do you got? or from what week?

This looks like a memorycontroller issue.

[Damien]

AMD64 3000+ DTR s754 week 01 of 2005.

[Skip]

how about this guys, i'm thoroughly stumped.

Memtest passes fine for my gskill, even at 320HTT. Yet i can't get into windows with BSOD.

So i take out some centon value sticks, and everything works fine. this sticks can only run 200htt at 3-4-4-7, and won't OC at all. but yet i'm totally stable in windows and won't BSOD will booting windows.

How is it that i can go into memtest at 320HTT 1:1 with memory and pass, yet not boot into windows without bsod?

also when running the centon memory i'm doing 333HTT at 1:2 memory. So i think the 333 would eliminate the board and the memory controller.

Should i go and get my RMA through newegg for the GSKILL? or should i investigate this issue further. I will try the gskill at the same settings as this centon. if it fails, then i know for sure.

Edit: ran gskill at same settings as centon, and it failed. it was like 166 2-3-2.

[neokenzo]

My LA cant run at 2.5-3-3-7 after 245HTT. Any higher then that, I have to change the timing to 2.5-4-3-7 at 1T. Is this normal?

[JNav89GT]

had similar problem. I had to set my cas, trp, trcp, and tras manually, set everything else to auto, then work backwards on timings to see what held you up.

how about this guys, i'm thoroughly stumped.

Memtest passes fine for my gskill, even at 320HTT. Yet i can't get into windows with BSOD.

So i take out some centon value sticks, and everything works fine. this sticks can only run 200htt at 3-4-4-7, and won't OC at all. but yet i'm totally stable in windows and won't BSOD will booting windows.

How is it that i can go into memtest at 320HTT 1:1 with memory and pass, yet not boot into windows without bsod?

also when running the centon memory i'm doing 333HTT at 1:2 memory. So i think the 333 would eliminate the board and the memory controller.

Should i go and get my RMA through newegg for the GSKILL? or should i investigate this issue further. I will try the gskill at the same settings as this centon. if it fails, then i know for sure.

Edit: ran gskill at same settings as centon, and it failed. it was like 166 2-3-2.

[jagzzzzzzz]

Have you tryed the BOXED cooler? The memorycontroller may fail under zero temperature, unlikely but worth to try.

What is a boxed cooler? I have get TONS of errors when at 250 FSB (P4) and I think I have read that memtest will error out with butt loads of errors when there is a heat issue. Is this true?

[jagzzzzzzz]

Oh and by the way, this should be a sticky because this thread is going to grow!

[Skip]

i'll give it one more shot. i had only cas-rascas-trp-trfc-trc set manually, everything was at auto. and i just simply can't boot into windows. memtest will run without an error, but it won't go into windows.

[twiggy]

Got my sticks back today and OPB is the man! Thanks a ton OPB, damn good support.

[fareastgq]

I think I have read that memtest will error out with butt loads of errors when there is a heat issue. Is this true?

Yes, heat is not your freind, example, pump 3.4 volts into some ram running at 200 2 2 2 timings ie: UTT, I run about 20 passes of memtest 5 error free, after that I start getting errors more frequently, most likely cause = heat. and yes I did test that hypothesis with a big square fan cooling my pc, and presto, no memory errors If heat is your problem, you are lucky, it's not really a problem, just a heat issue, if you can't even run memtest error free at stock volts and speeds, you have different problems. I was thinking of swapping my nice window panel for a huge fan, it's about the size of my case and cools, hehe. I could bolt it on.