Kingmax PC2700 RAM and Memtest 86 errors

[RalfHutter]

Im running 2 x 256Mb of Kingmax PC2700 5ns RAM at 333Mhz (125 FSB with my 1.8A on an Asus P4B266 board).

I ran Memtest 86 v2.9 to check out my RAM. I get an occaisional error on tests #5 and #8. If I loop the tests I don't get an error every time, just maybe once out of 5-6 tests. Here's what I've tried (that hasn't helped): lowered FSB in increments so RAM eventually was running at only 250Mhz. Changed from CAS2 to CAS2.5. Bumped Vram up to 2.6. Ran only one stick at a time. Swapped sticks. None of this has helped.

I've ran Prime95 many times for many hours at the most aggressive timings (333Mhz, Cas2) and NEVER had any type of error. I've ran "DocMem" and never had any errors.

Any suggestions? Anybody else tried running Memtest on their Kingmax?

BTW - I swapped RAM and put in one of my Crucial 512Mb sticks (with Infineon chips) and set the speed to 305Mhz, CAS2.5, Stock Vram and ran Memtest on it. I looped it for a few hours and got no errors at all.

 

[sean2002]

My Crucial PC2100 DDR @ 143FSB does the same thing will pass all the tests sometines, and then get an error or two on another run. My system is rock stable, so I would take the results with a grain of salt. BTW I've seen people getting 10,000's of errors and still be rock stable

 

[RalfHutter]

<< memtest86 is no better than it should be >>



Huh?

Does this mean I should ignore the errors and just use the RAM?

 

[sharkeeper]

IF you get ONE or TWO errors at EXACTLY the same spot it COULD mean defective ram. I say COULD because I've seen this, but then it goes away with less aggressive memory timings! Go figure! Windows NT is pretty picky about memory anyways. You'll know if you're having memory troubles for sure. One thing I have noted about MT86 is I get thousands of errors scrolling like crazy (this is a SERVER at that!) but if ECC (scrub) is enabled in the BIOS, I get none letting the test run over a long weekend. The server was running with a month of uptime with ECC OFF with no errors btw...

Prime95 torture test will also detect obvious errors...sometimes errors that won't bring Windows down either!

This is why I always turn a PC over to a client with very conservative settings. I remember building several 512MB (256 x 2) PC's with Crucial Memory that would run at 160+ at 222 and they left here at 133 333. I'm sure these client's won't be able to distinguish CAS levels when running Front Page, Excel, Power Point, etc.

Cheers!

 

[RalfHutter]

Well, go figure...

I exchanged my Kingmax RAM for 2 new sticks and took it home and ran Memtest 86 on it. Now instead of getting a few (usually 2-8 errors) I'm getting THOUSANDS. Apparently these sticks are even worse than the first two.

Now what??? Should I get Corsair or Mushkin instead? Everyone raves about Kingmax, could there be an issue with this memory and my Asus board? It must be something to do with the memory because the Crucial runs fine. Last night I left Memtest running in a loop for 9hrs with one stick of 512Mb Crucial set at 311Mhz, 2.5V, CAS2.5 and it ran without one error.

 

[Duvie]

First of all kingmax 5ns pc2700ddr is only rated cas2.5....I get no benefit score wise by going to cas2 on my asus p4s333...so I left it at default....I still get 2525/2526 in my sandra scores at 333mhz...I have ran prime solid and no errors....


I think you are putting to much stock in the memtest program....I too have seen some real anamolies when it has comes to reporting memory problems...real wacky stuff...

have you ran the kingmax at cas 2.5 and still got errors??? I have prime tested my memory at 390mhz (no voltage boost) with 3:5 ratio at 118fsb and got no errors for several hours...If I had issues at 333mhz this would have fished them out....

 

[RalfHutter]

Do you think Memtest 86 has issues with this Kingmax RAM but not other RAM? It runs my Crucial 512Mb fine. I've run it on my Dell PIII laptop with 384Mb of Crucial RAM w/o any errors. I've run it on my dad's P4 1.8 Willy with 2 x 256 Crucial RAM (at default settings - 266Mhz) with no errors. Hmm...maybe it just likes Crucial??

I tried the Kingmax at 250Mhz (well under rated speed) and still get errors.

 

[Ryu]

Sup guys i just ran this test last night cuz im getting random reboot like crazy since the weekend. I got 50,000+ errors on my 2x 256 megs of Crucial DDR. I don't know if the test is accurate or not, but i just pulled out the older stick (i got the 2 sticks at different times). Any one know of a more accurate mememory tester?

I really want to know if it is my memory or if its something else (nothings over clocked...)

 

[blacktalon]

I use Memtest 86 to test my RAM. I've heard that it's one of the best memory testing programs. I've used DocMemory and several others but a lot of the time they say the RAM is fine even though I've been having trouble. I'll run the Memtest 86 and it shows errors. Get new/different RAM and it work fine, check it with Memtest and it passes. It just seems that Memtest is more accurate. It is a more thorough test than the others.

 

[RalfHutter]

I'm taking this RAM back to Kingmax so their tech support guy can check it out, but I'm not feeling too confident.

Does my MoBo (Asus P4B266 + 1.8A)even support this RAM? Is anybody out there in Anandtechville using Kingmax PC2700 on their P4B266? Has anybody run Memtest 86 with this combo?

I'm thinking of trying a stick of Corsair 512Mb PC2700 CAS2 RAM. Does anyone have any experience with this RAM (the 512Mb is brand new, has anyone used the 256Mb PC2700 Corsair?) on an Asus P4B266 board? Any thoughts on Corsair in general?

 

[LostHiWay]

I'm running 3 sticks of kingmax pc2700 ram at 166fsb (CAS2). I ran Memtest 86 v2.9 overnight and had no errors. Maybe you do have a bad stick

 

[RalfHutter]

Apparently I had 4 bad sticks of Kingmax!

Got myself a stick of the Corsair 521Mb PC2700 XMS RAM from Mwave. Ran Memtest 86 (the complete set of 11 tests, looping) with RAM set at 340Mhz, CAS2, 2.5 volts for 10hrs. Had exactly ZERO errors. Was getting 5000+ errors with my last 2 sticks of Kingmax at these same settings. Dunno why.

Everyone raves about the Kingmax PC2700 RAM. Maybe it just doesn't like something about my setup?

Remember, I've also ran Memtest 86 on Crucial (2 x 512Mb, Infineon chips) @ 311Mhz for 9 hours with no errors either.

 

[flexy]

i think memtest86 is pretty good (?)...and (face it...) if it gives erros...(even thousands...) then you have a bad ram stick !

Never had any problems running my crucial 2100 (cas2.5, 133fsb) fully maxed out at cas2, interleave 4 and 144 FSB (even higher!)...no error with memtest86 !

Should maybe give some hints what memory to recommend and what memory to avoid !

I just think it's "funny" that the number of crucial users with complaints seems a lot smaller than the number of users who spent GOOD money on 2700 ram....2700 ram...well DUH !...if you have to run it at cas2.5 or cas3....who gives....maybe crucial would even do 150 if i'd run it at cas2.5 and tweak all timings down

ddr 2700 seems to be a rip-off then..anyone agree ?

 

[pm]

DDR PC2700 SDRAM is not a standard and the specifications have not been ratified. What DDR PC2700 SDRAM actually is has never been defined, so manufacturers are selling their "best guess". If their "best guess" doesn't match up exactly with the "best guess" of the chipset manufacturer then you end up with signal integrity issues.

Motherboard signalling is not simply a matter of "if it's fast enough, it will work", there are a lot more factors involved in spec'ing memory than merely picking really fast memory parts off of the wafer, testing them at 166MHz and labelling them as such. The signals need to be synchronized with each other and with respect to the clock and this where specifications make sure that everything works correctly. As it is, once you get beyond 133MHz, you have moved into the "wild west" where each manufacturer things that he is right and everyone should be following their lead.

It's also worth mentioning that none of the motherboards available for the Pentium 4 are spec'd for operation higher than 100MHz. So you can have a CPU that can run at 133MHz, a chipset that can as well, and memory that is spec'd for 133MHz and not have the thing work because the routing through the motherboard PCB is not optimal for 133MHz operation.

Prime95 is not a good test of memory. It's memory footprint is very small. What you probably want to do is find some application that is self-checking (so you can make sure it works correctly) and requires all available memory to run, or, more ideally, you want something that systematically goes through memory row by row, column by column and tries every possible pattern on each bank of memory. This is what memtest is supposed to be doing. I've never used, so I could tell you if it's a programming error or not, but I'd definitely be concerned. There are several good memory testers for Linux. If you don't run Linux then this might be too big a hassle to consider but if you have a Linux partition lying around, then you might want to fire up one of those.

 

[MadRat]

You can only use one PC2700 stick of RAM at a time due to signal integrity issues according to Aceshardware's P4 review today. PC2100 was limited to two sticks in most cases, and PC1600 was for anything past three sticks.

 

[phamtastik]

I was playing with Memtest86 2.9 last night and my crucial on my P4B266 board can only go 142 fsb anything over that will get zillions of errors...
Maybe they went bad since I oc'd them at 3.0 voltage.

Tried a stick of my room mates ram still get errors at 148 FSB... but alot less.

Personally I think its a fairly good tester.
I'm going to get a nice stick of PC2400 or PC2700 to test it out further.

 

[rbV5]

<< You can only use one PC2700 stick of RAM at a time due to signal integrity issues according to Aceshardware's P4 review today >>



No issue with 2 sticks on my rig, 188FSB cas2 12hours looping memtest86 no errors.

 

[Slacker]

you have kingmax rbv?