G.Skill Memory

i can't seem to get my gskill 4400LE's (rated to run 2.5-3-3-7) prime stable at 275 1:1 even at 2.85v. but if i loosen them up to 2.5-4-3-7 2.75v prime doesn't error out.
Do you think G.Skill is a little overzealous with their speed ratings? it certainly seems so. now it will run memtest at 275 2.5-3-3-7 but not prime.

Anyone know any G.Skill resellers in Canada?

There is no resellers at this time in Canada...

As for Prime...I wouldnt put too much faith in prime...if you can run 3d marks 01 - 03 - 05 and super pi and past pifast...then you are stable...Prime has some issues with some winnies and the memory controller...

Mine wont past prime even when underclocked...no matter what ram...

glock19owner said:

There is no resellers at this time in Canada...

Click to expand...

Darn.

well some of us old schoolers still use and believe in it. i have no problems priming at stock or at 250x9 1:1 1.5v. i'm currently priming at 275x9 1:1, 1.60vcore, 2.75vdimm.
the problem with winnies and the memory controller is pure speculation and extremely random. some saying it pertained to older cpu's. well mine is a 0436XPBW OEM. I'll still argue the fact that if u can't prime for atleast 3+ hours then ur o/c isn't 100% stable. the last thing i want is a BSOD after three hours of gaming. sure lots of guys post screenies of crazy o/c's with sandra benchmarks.

Scar said:

well some of us old schoolers still use and believe in it. i have no problems priming at stock or at 250x9 1:1 1.5v. i'm currently priming at 275x9 1:1, 1.60vcore, 2.75vdimm.
the problem with winnies and the memory controller is pure speculation and extremely random. some saying it pertained to older cpu's.

Click to expand...

Actually its not...this has been proven at other forums that the memory controller or Prime software is messing up...

Now if going on what you stated...you just said my 3200 is trash since it cant run prime95 at 1.8Ghz...yet I can run every benchmark and game for hours without a single error at 2.78Ghz and can pass memtest up to 300FSB/HT at 8-4-4-2.5...

I understand about old school...been doing this for 20+ yrs...but you also have to remember...that older programs have issues with newer CPU's...if this wasnt the case...then we should all still using win 95 and playing doom3 with the old solitary engine ...

glock19owner said:

I understand about old school...been doing this for 20+ yrs...but you also have to remember...that older programs have issues with newer CPU's...if this wasnt the case...then we should all still using win 95 and playing doom3 with the old solitary engine ...

Click to expand...

I love old guys. Just like John Wayne toilet paper. Don't take crap off anyone. I have to agree with glock I have had prime fail and all other proggys run fine but do like to see prime stable.

My new Gskill LE's wouldn't do 2.5,3,3,7 at 275 at first either but like glock after a week they are doing much much better. In fact I memtested them errror free (single channel) at their spd of 2.5,4,4,8 at 295 and am running them Dual channel now at 285fsb 2.5,4,4,8 with 6300 mb's in Sandra on an Abit 865pe board.

hey, i'm from Missouri...Show Me where it's been Proven and not just Speculated. Real documentation and not a bunch of random posts in a 15 page thread. there's nothing magical about P95, all it does is crunch numbers with a predermined outcome. maybe you had your memory timings too tight at stock speed. could be any number of things.
back to the point, is it possible gskill has been a little overzealous with their speed timings. after all, Who else sells samsung TCCD's or any pc4400 that can run 2.5-3-3 at 275mhz? No One

Scar said:

after all, Who else sells samsung TCCD's or any pc4400 that can run 2.5-3-3 at 275mhz? No One

Click to expand...

I'm still new to this, but I think the PCB has to do with how well the memory OC's too.

9mmCensor said:

Anyone know any G.Skill resellers in Canada?

Click to expand...

You dont need to buy hi speed ram to use with an A64 anyway. You can run pc3200 cas2.5 medicore ram using the divider and you'll be just as fast as a guy running TCCD 1/1 at 290. The point that everyone seems to be missing with A64 is it's not a P4. Memory bandwidth on A64 is not important after you get to pc3200 levels, its more important to maximize your CPU core speed. Just keep whatever ram you have, run it as fast as possible, then max your CPU while keeping the ram running its fastest using the memory divider. There is no penalty with memory dividers on A64, its not a P4. Using some decent 5-2-2-2 1t memory at 290 with a 5/4 divider (230Mhz) you'll be just as fast as G.Skill running 1/1 at 290.....

EvilAerosolCan said:

I'm still new to this, but I think the PCB has to do with how well the memory OC's too.

Click to expand...

That is a key point...G. Skill is one of the few companys that redesigned the first PCB brain which helps the chips run at higher and better timings and in some cases at lower voltages...also 1 or 2 (max) out of 10 chips make it too LE standards...and even then some still dont even make the LD standards...

As for the show me...I can not find the thread that had the link to the issues with certain A64's...and a new update was supposed to be in the works...

As for my own proof...the G. SKills have no issues running prime95 on my 2500 barton...yet like I stated above...my 3200 fails even when underclocked...so this is not a memory issue but either a memory controller or software issue...and with more and more people reporting that P95 errors out with winnies...it is point more to software then hardware...I would seriously doubt a few million 939's would be bad without some type of statement or recall from AMD...especially since this seems to be the only software they all fail...

If you still think your G. Skill is the issue(s)...then contact Jack at G. Skills forum...he is very detailed with his answers and might know of any known issues or settings that might have to deal with your issue(s) you are having with them

If I worked for them and had the info they have advailible to them...I would be glad to help you farther...Jack can go in more detail on other questions you might/will have also

If you can't hit what it's rated at, that is reason for RMA. I tend to agree with Glock, but if Prime95 is important to you, and you feel you've tried everything and ruled everything out but the memory, RMA the sticks. How do they do in memtest @ 275? When i first got mine, they were a little slow to hit anything over 275, now i've been benching them at 295+

I'd like to add....

I just purchased the LA's which are supposedly the best of the best. They don't do their 300mhz rated speeds. This is NOT because of the sticks though, it is because of my memory controller. Seems that my patriot and these max out at the exact same speed...the g.skill holds tighter timings though. Are you sure it's not your mem controller?

-Collin-

Prime can stress out your mem controller easier than those sticks. Try memtest then see. If you ramped up your dimm voltage, and you still got the same amount of errors, then it your cpu. If you ramped up dimm voltage and got less errors then you can rma those sticks.

well, i'm pretty sure i have a definative answer now. it IS the memory controller.
memory passes memtest at 275x7 at amazingly 2.65v
prime fails at 1:1 275x9,275x8 & 275x7 at any vcore and vdimm voltage.
set the mem to max clock of 166/ fsb to 275 and it's been priming for almost two hours now.
do you think a newer processor of the 044x variety will get me to my 275+ 1:1 action?

If you think it is your memory controller...run super pi at 32mb and pifast...if it can pass both of those test...then run CPU stability test 6.0...set the option for trouble shooting...and then keep hitting start until the prime test starts to run...if your CPU can handle the prime test...then it is not the memory controller...but P95 that is failing...

Doing this will show you (if your controller truelly isnt bad)...how Prime95 is past its prime with the A64's newer memory controllers...

Thanks dude, i'll use those tools as additional stability testers but i still rely on prime95. there's nothing wrong with the software. just cuz ur +3200 overclocked to 2.75ghz fails prime (almost instantly im sure) doesn't mean it's the program. if it makes you feel better to blame it on the software then that's certainly your prerogative. here's another screenie of a 10h prime session without failure. i'm certain at this point that the mem controller just can't handle the sustained load. thanks!

Once again...it fails even when underclocked...so this is not a failed OC or a failed memory controller...and a lot of newer winnies are failing even when set at default...

Also, I can prime for hours on ends with other prime programs...primed for over 19 hours with CPU Test 6.0 running 2.75Ghz...and it runs the same prime test as P95 (just not to the million power)...just dont get paid for it if you do happen to hit a prime number

But if me and a few other thousand people have the same issues...I dont think its just my CPU then ...sounds like you might of lucked out...

I am getting my LEs today and I will have a run with my 3000+ Winnie 0449GPEW. Now mine is newer but we will see what happens.

Scar, this is a link to a review that talks about some of the same issues being discussed here.
http://www.madshrimps.be/?action=getarticle&number=5&artpage=953&articID=244
I'm from the old school too and I believe in passing Prime, but it seems that this discssion always ignites a heated controversy. Perhaps, we can just agree to disagree rather than trying to bash the other person into thinking the "correct" way. There should be plenty of room in this forum for an honest disagreement.
Eclipse- I'd like to here how you make out with your new Gskill's as I'm shopping for the same. Take care everyone and chill.