Ping kony

[dos]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

the memory company says I need to just send the one bad module in.Since this
was a dual kit, you think sending the one stick is ok?
TIA

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

On Mon, 28 Mar 2005 19:02:36 -0500, "doS"
<kobo65@hotREMOVETHISmail.com> wrote:

>the memory company says I need to just send the one bad module in.Since this
>was a dual kit, you think sending the one stick is ok?
>TIA
>

Did the errors _ALWAYS_ occur at the exact same addresses?
Then if you swapped the modules around, put each in same
slot the other hand used, but still running both for the
test, did the addresses change from the first set of tests
BUT then after modules were swapped, the errors continued to
remain at same addresses?

If the addresses remained the same over multiple restests
with modules remaining in same slot, that would tend to
indicate an actual failure in the bits of the chips. What
is more common is for the memory's spec'd timings to simply
not work on a motherboard, which was my inital impression of
what was happening with your system.

In other words, you likely need BOTH modules replaced,
probably can't have them send you the same make/model of
modules to replace them, the replacement would "most likely"
need to be different memory. Whether the vendor can
accomodate that and/or give I refund I can't say, but don't
let them try to pull a fast one and charge a restocking fee
claiming they "work OK", because they don't. Too often a
vendor will sell aggressively timed memory knowing it may
not work, then they just collect that return fee while still
having stock to resell, though of course sometimes it's by
accident but other times the situation keeps reoccuring with
vendor so they'd have to know about it already but kept
selling product.

In other words, the memory was spec'd for those timings and
can't run them, thus you do not have to pay ANYTHING more in
order to demand a full refund, or a replacement that works
(see their return policy but remember that you aren't
necessarily bound to "everything" they want to impose, the
product must be fit for it's spec'd, specified purpose. Do
not accept a replacement of exact same make/model memory if
you'd then have to pay return shipping again and again, else
you'll just end up in an endless loop of getting same memory
that won't work and having to pay to return it over and over
again.

But since this was a dual kit, either way I would not want
to start splitting it up. The two modules are ONE product
as the kit and should be handled as a purchased "product" as
a pair, not "they'll replace that one module". If the other
module really is ok, they then have it to do whatever they
want with it so they don't really have any argument towards
returning only the one module.

 

[dos]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

I ran memtest86 with both in, got the errors. I then ran the tests again,
with only one module at a time.One module no errors, the other got errors
consistently in tests 5 and 8.

"kony" <spam@spam.com> wrote in message
news:tuah41hhql4negjm4n1kfull9a3tqh9rnu@4ax.com...
> On Mon, 28 Mar 2005 19:02:36 -0500, "doS"
> <kobo65@hotREMOVETHISmail.com> wrote:
>
>>the memory company says I need to just send the one bad module in.Since
>>this
>>was a dual kit, you think sending the one stick is ok?
>>TIA
>>
>
> Did the errors _ALWAYS_ occur at the exact same addresses?
> Then if you swapped the modules around, put each in same
> slot the other hand used, but still running both for the
> test, did the addresses change from the first set of tests
> BUT then after modules were swapped, the errors continued to
> remain at same addresses?
>
> If the addresses remained the same over multiple restests
> with modules remaining in same slot, that would tend to
> indicate an actual failure in the bits of the chips. What
> is more common is for the memory's spec'd timings to simply
> not work on a motherboard, which was my inital impression of
> what was happening with your system.
>
> In other words, you likely need BOTH modules replaced,
> probably can't have them send you the same make/model of
> modules to replace them, the replacement would "most likely"
> need to be different memory. Whether the vendor can
> accomodate that and/or give I refund I can't say, but don't
> let them try to pull a fast one and charge a restocking fee
> claiming they "work OK", because they don't. Too often a
> vendor will sell aggressively timed memory knowing it may
> not work, then they just collect that return fee while still
> having stock to resell, though of course sometimes it's by
> accident but other times the situation keeps reoccuring with
> vendor so they'd have to know about it already but kept
> selling product.
>
> In other words, the memory was spec'd for those timings and
> can't run them, thus you do not have to pay ANYTHING more in
> order to demand a full refund, or a replacement that works
> (see their return policy but remember that you aren't
> necessarily bound to "everything" they want to impose, the
> product must be fit for it's spec'd, specified purpose. Do
> not accept a replacement of exact same make/model memory if
> you'd then have to pay return shipping again and again, else
> you'll just end up in an endless loop of getting same memory
> that won't work and having to pay to return it over and over
> again.
>
> But since this was a dual kit, either way I would not want
> to start splitting it up. The two modules are ONE product
> as the kit and should be handled as a purchased "product" as
> a pair, not "they'll replace that one module". If the other
> module really is ok, they then have it to do whatever they
> want with it so they don't really have any argument towards
> returning only the one module.

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

On Mon, 28 Mar 2005 20:37:31 -0500, "doS"
<kobo65@hotREMOVETHISmail.com> wrote:

>I ran memtest86 with both in, got the errors. I then ran the tests again,
>with only one module at a time.One module no errors, the other got errors
>consistently in tests 5 and 8.

Ok but that's not necessarily all the applicable info. It
may be significant whether it was ALWAYS the exact same
addresses. If it's not always exact same addresses, a
replacement module may have same problem (might anyway if
defective but odds of that are far lower?). It could even
be that if you ran tests long enough on the other seemingly
"good" module that it too would have errors. Suppose the
"good" module has an error rate of 1 error every 24 hours...
if you didn't test for over 12 hours the odds are you didn't
see any of the errors that would result. Granted the longer
it runs the lower the odds of any errors (ignoring cosmic
rays and the like as causes) but I would be skeptical of
that brand/model memory as suitable at all, if the errors
weren't always in exact same locations... not just the one
module but both and any of the same make/model they'd send
as replacements. IMO, your best bet is to return both and
get a different brand/model or a refund you can use towards
another (different make/model).

Personally I would treat the matched pair as a single
product, but if you need one module to keep the system going
then that might be motivation to return just the one. YMMV.

 

[Philo]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

"doS" <kobo65@hotREMOVETHISmail.com> wrote in message
newsW02e.11327$rm1.1630@fe06.lga...
> the memory company says I need to just send the one bad module in.Since
this
> was a dual kit, you think sending the one stick is ok?
> TIA
>
>

IIRC the microsft memory test will tell you specifically which module is
defective
http://oca.microsoft.com/en/windiag.asp

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

"kony" <spam@spam.com> wrote in message
news:cdhh4156e7rlbbdpd3s023b4l106gcocor@4ax.com...
> On Mon, 28 Mar 2005 20:37:31 -0500, "doS"
> <kobo65@hotREMOVETHISmail.com> wrote:
>
>>I ran memtest86 with both in, got the errors. I then ran the tests again,
>>with only one module at a time.One module no errors, the other got errors
>>consistently in tests 5 and 8.
>
> Ok but that's not necessarily all the applicable info. It
> may be significant whether it was ALWAYS the exact same
> addresses. If it's not always exact same addresses, a
> replacement module may have same problem (might anyway if
> defective but odds of that are far lower?). It could even
> be that if you ran tests long enough on the other seemingly
> "good" module that it too would have errors. Suppose the
> "good" module has an error rate of 1 error every 24 hours...
> if you didn't test for over 12 hours the odds are you didn't
> see any of the errors that would result. Granted the longer
> it runs the lower the odds of any errors (ignoring cosmic
> rays and the like as causes) but I would be skeptical of
> that brand/model memory as suitable at all, if the errors
> weren't always in exact same locations... not just the one
> module but both and any of the same make/model they'd send
> as replacements. IMO, your best bet is to return both and
> get a different brand/model or a refund you can use towards
> another (different make/model).
>
> Personally I would treat the matched pair as a single
> product, but if you need one module to keep the system going
> then that might be motivation to return just the one. YMMV.
>
One other thing to consider is that any premium you paid for the "matched"
set should be refunded since, if you return just one module, they have NOT
tested the pair to insure they work together at the advertised settings. I
sometimes wonder if they even do that anyway. A pair of modules that work on
their test bench does not mean they will work on your board anyway. I always
forget about these matched sets and just buy exact modules (same part#) and
save the premium of having them "matched". Just the fact that they told you
to return the one module should tell you something about these so-called
matched sets of modules.

Ed
>
>

 

[dos]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

I decided, due to your comments, to return both, as you and the others say,
I should be getting what I pay for. The vendor says they will exchange them,
so hopefully I will get a better replacement. Thanks for your input, it
makes sense.

"kony" <spam@spam.com> wrote in message
news:cdhh4156e7rlbbdpd3s023b4l106gcocor@4ax.com...
> On Mon, 28 Mar 2005 20:37:31 -0500, "doS"
> <kobo65@hotREMOVETHISmail.com> wrote:
>
>>I ran memtest86 with both in, got the errors. I then ran the tests again,
>>with only one module at a time.One module no errors, the other got errors
>>consistently in tests 5 and 8.
>
> Ok but that's not necessarily all the applicable info. It
> may be significant whether it was ALWAYS the exact same
> addresses. If it's not always exact same addresses, a
> replacement module may have same problem (might anyway if
> defective but odds of that are far lower?). It could even
> be that if you ran tests long enough on the other seemingly
> "good" module that it too would have errors. Suppose the
> "good" module has an error rate of 1 error every 24 hours...
> if you didn't test for over 12 hours the odds are you didn't
> see any of the errors that would result. Granted the longer
> it runs the lower the odds of any errors (ignoring cosmic
> rays and the like as causes) but I would be skeptical of
> that brand/model memory as suitable at all, if the errors
> weren't always in exact same locations... not just the one
> module but both and any of the same make/model they'd send
> as replacements. IMO, your best bet is to return both and
> get a different brand/model or a refund you can use towards
> another (different make/model).
>
> Personally I would treat the matched pair as a single
> product, but if you need one module to keep the system going
> then that might be motivation to return just the one. YMMV.
>
>
>

 

[dos]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Thanks for the reply, I too think maybe they just pick up 2 sticks at random
and put them in a pack. I did send both back.

"Ed Medlin" <ed@edmedlin.com> wrote in message
news:8Ya2e.15809$DW.13983@newssvr17.news.prodigy.com...
>
> "kony" <spam@spam.com> wrote in message
> news:cdhh4156e7rlbbdpd3s023b4l106gcocor@4ax.com...
>> On Mon, 28 Mar 2005 20:37:31 -0500, "doS"
>> <kobo65@hotREMOVETHISmail.com> wrote:
>>
>>>I ran memtest86 with both in, got the errors. I then ran the tests again,
>>>with only one module at a time.One module no errors, the other got errors
>>>consistently in tests 5 and 8.
>>
>> Ok but that's not necessarily all the applicable info. It
>> may be significant whether it was ALWAYS the exact same
>> addresses. If it's not always exact same addresses, a
>> replacement module may have same problem (might anyway if
>> defective but odds of that are far lower?). It could even
>> be that if you ran tests long enough on the other seemingly
>> "good" module that it too would have errors. Suppose the
>> "good" module has an error rate of 1 error every 24 hours...
>> if you didn't test for over 12 hours the odds are you didn't
>> see any of the errors that would result. Granted the longer
>> it runs the lower the odds of any errors (ignoring cosmic
>> rays and the like as causes) but I would be skeptical of
>> that brand/model memory as suitable at all, if the errors
>> weren't always in exact same locations... not just the one
>> module but both and any of the same make/model they'd send
>> as replacements. IMO, your best bet is to return both and
>> get a different brand/model or a refund you can use towards
>> another (different make/model).
>>
>> Personally I would treat the matched pair as a single
>> product, but if you need one module to keep the system going
>> then that might be motivation to return just the one. YMMV.
>>
> One other thing to consider is that any premium you paid for the "matched"
> set should be refunded since, if you return just one module, they have NOT
> tested the pair to insure they work together at the advertised settings. I
> sometimes wonder if they even do that anyway. A pair of modules that work
> on their test bench does not mean they will work on your board anyway. I
> always forget about these matched sets and just buy exact modules (same
> part#) and save the premium of having them "matched". Just the fact that
> they told you to return the one module should tell you something about
> these so-called matched sets of modules.
>
> Ed
>>
>>
>
>