Should RAM timings have to be set manually?

[Larc]

| If the board runs stable with the other memory at certain
| timings, but can't run stabiliy with that memory at same
| timings, and those timings are what the Kingston was spec'd
| to use, it implicates the memory.

Bingo!

Larc



§§§ - Change planet to earth to reply by email - §§§

 

[Rod Speed]

If the board runs stable with the other memory at certain
timings, but can't run stabiliy with that memory at same
timings, and those timings are what the Kingston was
spec'd to use, it implicates the memory.

I just dont believe that Kingston is shipping ram with anything
like that percentage BAD at the top. Bad in the sense that you
get that percentage in every system. That is just plain silly, it
must have a problem with just some chipsets, like I said.

 

[Rod Speed]

Bingo!

Nope, its just not believable that Kingston is actually shipping
some ram that 11 out of 12 bad in all systems its used in.

No one's quality control is THAT bad.

It'll work fine in some other chipsets/bios so aint actually BAD at all.

 

[larry moe 'n curly]

I just dont believe that Kingston is shipping ram with anything
like that percentage BAD at the top. Bad in the sense that you
get that percentage in every system.

Not in every system. I counted a module as bad if it failed in at
least one of those systems when it was run at its SPD settings.

That is just plain silly, it must have a problem with just some chipsets, like I said.

Then why didn't any of my Corsair or PNY modules fail in any of the
systems, even when overclocked, and why did the Kingstons with the
Fxxxxxx chips work much better than other Kingstons with the same SPD
information?

 

[kony]

Then why didn't any of my Corsair or PNY modules fail in any of the
systems, even when overclocked, and why did the Kingstons with the
Fxxxxxx chips work much better than other Kingstons with the same SPD
information?

Remember who you are arguing with. There is no end.

 

[Rod Speed]

Rod Speed wrote

Not in every system. I counted a module as bad if it failed in at
least one of those systems when it was run at its SPD settings.

Thats nothing like bad ram, just that system not using the spd values properly.

Then why didn't any of my Corsair or PNY modules fail in any of the systems,

Different timing detail that the bios can handle better.

even when overclocked, and why did the Kingstons with the Fxxxxxx chips
work much better than other Kingstons with the same SPD information?

Again, different timing detail that the bios can handle better.

It would only be bad ram if you got that 8 out of 11 or 12 result
in all systems that specify that that particular ram is what it can
handle and I dont believe Kingston ships ram like that.

It isnt bad ram unless its not possible to find any timing specs which give
an error free result with say memtest86 and the Prime95 ramfucker test.

 

[larry moe 'n curly]

It would only be bad ram if you got that 8 out of 11 or 12 result
in all systems that specify that particular ram is what it can
handle and I dont believe Kingston ships ram like that.

I thought the standard of quality should be that the memory works
perfectly in all computers for which it's specifed, not just in one of
them.

It isnt bad ram unless its not possible to find any timing specs which give
an error free result with say memtest86 and the Prime95 ramfucker test.

I had some PC2100 modules that always failed testing at their rated 266
MHz bus speed, regardless of how slow the other timings were, but they
worked fine at 200 MHz. So by your reasoning, those modules weren't
bad?

 

[Rod Speed]

Rod Speed wrote

I thought the standard of quality should be that the memory
works perfectly in all computers for which it's specifed,

Not all bios can setup the chipset properly with some ram.

We're currently seeing a considerable problem with DDR2 ram
in some gigabyte motherboards, with later bios doing better.

We have also seen some DDR ram thats
know to not work well with all chipsets etc.

not just in one of them.

You havent established that it only works fine in one, just
one of the ones you have. I just dont believe that Kingston
is shipping ram that fails at that 8 out of 11 or 12 rate in
the vast bulk of systems that can take that type of ram.

I had some PC2100 modules that always failed testing at their rated 266
MHz bus speed, regardless of how slow the other timings were, but they
worked fine at 200 MHz. So by your reasoning, those modules weren't bad?

Only if you get that result in all systems that can take PC2100 modules
and I just dont believe that Kingston would be shipping ram like that.

 

[larry moe 'n curly]

larry moe 'n curly <[email protected]> wrote

It would only be bad ram if you got that 8 out of 11 or 12
result in all systems that specify that particular ram is what
it can handle and I dont believe Kingston ships ram like that.


Not all bios can setup the chipset properly with some ram.

That doesn't explain why some brands of modules were much less
troublesome than others or why I've never had an error during testing
with any module containing chips that had the chip manufacturer's full
markings on them.

It isnt bad ram unless its not possible to find any timing specs which give
an error free result with say memtest86 and the Prime95 ramfucker test.


Only if you get that result in all systems that can take PC2100 modules
and I just dont believe that Kingston would be shipping ram like that.

I still don't buy your line that if a module works correctly in even
one computer, then it's fine, especially because some module makers
claim that they use more than one type of mobo to test their memory.
Why would they do that if one was enough? Also Kingston ValueRAM has
always been the least overclockable I've ever tested, not that I
consider a module bad if it won't overclock.

 

[Rod Speed]

Rod Speed wrote

That doesn't explain why some brands of modules
were much less troublesome than others

Corse it does, they have different detail in the spd that the bios can handle.

or why I've never had an error during testing
with any module containing chips that had the
chip manufacturer's full markings on them.

Plenty of others have, and currently with DDR2 ram like that which wont
even allow the system to boot at all, let alone just produce some errors.

I still don't buy your line that if a module works
correctly in even one computer, then it's fine,

Its clearly not BAD RAM. THAT was your original silly claim.

especially because some module makers claim that they
use more than one type of mobo to test their memory.

You dont know that that ram you saw such an obscene failure
rate with doesnt work fine in a number of types of motherboard.

I just dont believe that Kingston would be shipping ram that gets
the result you got in all except one particular motherboard.

Why would they do that if one was enough?

I never ever said that one was enough. I JUST said that you dont
have any evidence to support your silly claim that its BAD RAM.

Also Kingston ValueRAM has always been
the least overclockable I've ever tested,

Irrelevant to whether its BAD RAM.

ALL that indicates is that it isnt as conservatively rated.

not that I consider a module bad if it won't overclock.

Its irrelevant to what is being discussed, whether those modules
you saw that obscene failure rate with are actually BAD RAM.

 

[tanstafl]

Indeed; it shows the values being used. I read that CPU-Z works in DOS, but
it turns out it only runs in a dos window, not pure DOS.


Aye, thanks. I did d/l Bart's PE the other night, gonna have to put it
together and do some poking around with it.

In the meantime, I ordered a pair of the Crucial Ballistix (512 x 2) sticks,
so I'll see how those do when they arrive.

You might try SPD Tool @
http://www.benchmarkhq.ru/fclick/fclick.php?fid=178 . Or google for
spd_tool12.zip - returns one FTP download site. I haven't used it.
Judging from the limited info, it seem a bit long in the tooth, but
you might be able to read the raw SPD data under DOS.

I've used that Ballistix ram for a year now on my Abit board - runs
solid, but I did have one stick fail a few months ago. The Crucial
lifetime warranty is gold. One quirk with Ballistix is that they will
only replace the pair - but they'll cross ship so you can keep
trucking (limping?) on one stick until it arrives in a couple days.
Be sure to retain the email order confirmations to complete the RMA
forms if you ever have a failure.

 

[kony]

Its clearly not BAD RAM. THAT was your original silly claim.

"bad" is too non-descript. I'd call it memory with lesser
stability margin, low grade memory chips that were spec'd
too aggressively or put onto a module with SPD programmed
too aggressively.

 

[Rod Speed]

"bad" is too non-descript.

Its just plain wrong. The ram isnt actually bad at all.

I'd call it memory with lesser stability margin, low grade
memory chips that were spec'd too aggressively or put
onto a module with SPD programmed too aggressively.

You dont even know that that is the case in his case.
You dont know that it isnt just a bios that doesnt handle
the values in the spd as well for whatever reason.

 

[kony]

Its just plain wrong. The ram isnt actually bad at all.


You dont even know that that is the case in his case.
You dont know that it isnt just a bios that doesnt handle
the values in the spd as well for whatever reason.

True, but I've also had Kingston valueram that barely had
any margin of stability at all, the only memory worse that
wasn't fradulently mislabeled or had damaged cells was some
of the old FAR PC133 stuff... and I can't complain much
about it when it's free.

 

[Rod Speed]

True, but I've also had Kingston valueram
that barely had any margin of stability at all,

What exactly do you mean by that ? What were
you varying manually that saw it stop being stable ?

 

[larry moe 'n curly]

larry moe 'n curly <[email protected]> wrote

Not all bios can setup the chipset properly with some ram.


Corse it does, they have different detail in the spd that the bios can handle.

It also doesn't explain why those 512MB PC3200 Kingstons with the
Fxxxxxx chips tested much better than same model Kingstons with the
completely unmarked chips and the same SPD information.

Plenty of others have, and currently with DDR2 ram like that which wont
even allow the system to boot at all, let alone just produce some errors.

It's clearly not BAD RAM.

Irrelevant to whether its BAD RAM.

It's just an amazing coincidence, over and over.

 

[kony]

What exactly do you mean by that ? What were
you varying manually that saw it stop being stable ?

Memory clockgen speed.
Timings remained manually set to same as programmed into
SPD.

 

[Rod Speed]

Memory clockgen speed.
Timings remained manually set to same as programmed into SPD.

You cant just fiddle with the clock speed alone
and expect the spd values to be fine.

 

[Rod Speed]

Rod Speed wrote

It also doesn't explain why those 512MB PC3200 Kingstons with
the Fxxxxxx chips tested much better than same model Kingstons
with the completely unmarked chips and the same SPD information.

I dont believe the spd data was identical and if it was, all
that proves is that the bios setup timing details that suited the
Fxxxxxx chips better, NOT that the unmarked chips are BAD.

I say again, I just dont believe that most systems would
get that obscene result you got with the unmarked chips.

It's clearly not BAD RAM.

Yep, but plenty claim it is, just as erroneously are you are doing.

It's just an amazing coincidence, over and over.

Not a coincidence at all. Hardly surprising that some ram is more overclockable than others.

Pity its irrelevant to whether its BAD RAM.

 

[kony]

You cant just fiddle with the clock speed alone
and expect the spd values to be fine.

That's just it, I can, and do on most memory and it is fine
running at the SPD values. The exception is memory with
less conservative ratings.