Why would adding 2 new chips of memory cause Win XP Pro Not to Boot?

  • Thread starter Thread starter oscarc
  • Start date Start date
O

oscarc

I have a Gateway Athlon 700 with 584MB of memory. 1 256MB SDRAM chip
with ECC, 2 128MB chips, non ECC. Got 2 256MB SpecTek Non ECC chips
from my brother from his old computer. Replaced the 2 128 chips with
the 2 256MB chips, figuring no problems would occur. Lo and behold,
boot the system up and get:

Windows could not start because of an error in the software.
Please report this problem as :
load needed DLLs for kernel
Please contact your support person to report this problem.

Take the 2 chips out, problem fixed, boots up fine. What about these
chips would make the system fail? They should be compatible with my
system.

Any help would be appreciated

thanks,
Oscar
 
I have a Gateway Athlon 700 with 584MB of memory. 1 256MB SDRAM chip
with ECC, 2 128MB chips, non ECC. Got 2 256MB SpecTek Non ECC chips
from my brother from his old computer. Replaced the 2 128 chips with
the 2 256MB chips, figuring no problems would occur. Lo and behold,
boot the system up and get:

Windows could not start because of an error in the software.
Please report this problem as :
load needed DLLs for kernel
Please contact your support person to report this problem.

Take the 2 chips out, problem fixed, boots up fine. What about these
chips would make the system fail? They should be compatible with my
system.

Any help would be appreciated

thanks,
Oscar
Click to expand...

Unfortunately many boards aren't extensively tested with all memory
slot populated, or perhaps those that are, are not redesigned if
that's a problem. In other words, simply because your board has 3
slots, and can theoretically accept X amount of memory per slot, you
can't necessarily assume it would run stable with all slots maxed out.

You could try entering the BIOS and relaxing, rasing the memory timing
numbers, or choosing "slower" or similar wording, but even that may
not help enough. Test the memory with http://www.memtest86.com,
ALWAYS, for several hours before ever booting to the OS, to avoid file
corruption.

You might also try different combinations, see if the system would
even run stable with only the pair of "new" modules in it. By "run
stable" I again mean running memtest86, not booting to the OS till it
at least tests ok at that.
 
Gateway is particularly evil about providing memory facts.
Better to go to someone like www.crucible.com to learn which
types of 256 memory is required.

For example, same size memory in nine chip or two chip
configurations; and other differences. To you, difference
should be transparent. But computer hardware accesses
memories differently. BIOS should solve that identification
for you.

Memories have a single pin that tells computer what kind of
memory it is. However many Gateway BIOSes are not properly
programmed and will not be corrected by updates. Some Gateway
motherboards only understand nine chip memory. Even worse,
Gateway cannot be bothered to inform you which memory will and
will not work.

Look. Any decent computer vendor also provided
comprehensive diagnostics for each machine - for free.
Gateway does not. No diagnostics and their inability to even
report which kind of memory works in their machines are but
some reasons why Gateway is no longer on my recommended list.
Add to this - even their technical support does not understand
difference exists between two chip and nine chip memory!

Goto to crucible.com to learn which type is acceptable for
your machine. Don't even waste time with Gateway support or
their 'tech unfriendly' website - that does not even provide
diagnostics for your machine.

Customer support tech claimed he had access to more
information than I did. So I had him read from his screen.
From memory, I interrupted his sentences and finished them.
Gateway customer support is that technically ignorant. That
website contained little useful technical facts.
 
I have a Gateway Athlon 700 with 584MB of memory. 1 256MB SDRAM chip
with ECC, 2 128MB chips, non ECC. Got 2 256MB SpecTek Non ECC chips
from my brother from his old computer. Replaced the 2 128 chips with
the 2 256MB chips, figuring no problems would occur. Lo and behold,
boot the system up and get:

Windows could not start because of an error in the software.
Please report this problem as :
load needed DLLs for kernel
Please contact your support person to report this problem.

Take the 2 chips out, problem fixed, boots up fine. What about these
chips would make the system fail? They should be compatible with my
system.
Click to expand...

Review and possibly change your ECC memory settings in your BIOS.


Have a nice week...

Trent

Follow Joan Rivers' example --- get pre-embalmed!
 
I have a Gateway Athlon 700 with 584MB of memory. 1 256MB SDRAM chip
with ECC, 2 128MB chips, non ECC. Got 2 256MB SpecTek Non ECC chips
from my brother from his old computer. Replaced the 2 128 chips with
the 2 256MB chips, figuring no problems would occur. Lo and behold,
boot the system up and get: ... snip ...
Click to expand...

If you have any system with ECC capability, you should ensure ALL
your memory is ECC memory. The ECC corrections won't work
otherwise. The few extra dollars will give you a reliable memory
system.

BTW, the numbers you posted indicate you have 512 M mounted now,
not 584.
 
This morning I recieved a 256mb module of Spectek DDR333 (non-ECC), and
as soon as I got into Windows, the machine would crash. Sometimes it
didn't even make it into Windows. I changed the speed of it from 333mhz
down to 250mhz and it's running fine at the moment (I haven't had a
chance to test it extensively in games or run MemTest86 on it, yet).

I will be sending it back to the place I got it, though! Apparently
Spectek memory is made by Micron (see
http://www.spectek.com/menus/about.asp) and it's the stuff which has
failed testing for Crucial memory so they pass it off as this Spectek shit.

Before I send it back, though, I'll make sure it's not the fact that I'm
running 2 different makes of DDR333 alongside it (TwinMOS and Samsung)
so I'll take the TwinMOS and Samsung modules out and try it at 333mhz
with just the Spectek (I hate the name already).
 
This morning I recieved a 256mb module of Spectek DDR333 (non-ECC), and
as soon as I got into Windows, the machine would crash. Sometimes it
didn't even make it into Windows. I changed the speed of it from 333mhz
down to 250mhz and it's running fine at the moment (I haven't had a
chance to test it extensively in games or run MemTest86 on it, yet).

I will be sending it back to the place I got it, though! Apparently
Spectek memory is made by Micron (see
http://www.spectek.com/menus/about.asp) and it's the stuff which has
failed testing for Crucial memory so they pass it off as this Spectek shit.

Before I send it back, though, I'll make sure it's not the fact that I'm
running 2 different makes of DDR333 alongside it (TwinMOS and Samsung)
so I'll take the TwinMOS and Samsung modules out and try it at 333mhz
with just the Spectek (I hate the name already).
Click to expand...

No, it did not "fail testing for crucial", it's not "shit", it's
simply the lower grade memory, which can't run at the faster timings.
More often than not, when a motherboard won't run the lower grade
memory, it's a flaw with the motherboard, not the memory. Many boards
require higher-spec memory to run at lower spec settings to compensate
for that motherboard's design, especially when multiplie modules are
used.

You're running 3 modules... it was pretty likely you'd have problems
with most modules you try. Many boards required lowering memory bus
speed or relaxing the memory timings to run 3 modules. In other
words, it's "almost" always better to use the least number of modules
possible, though that's 2 if you're running a dual-channel board.
 
Unfortunately many boards aren't extensively tested with all memory
slot populated, or perhaps those that are, are not redesigned if
that's a problem. In other words, simply because your board has 3
slots, and can theoretically accept X amount of memory per slot, you
can't necessarily assume it would run stable with all slots maxed out.

You could try entering the BIOS and relaxing, rasing the memory timing
numbers, or choosing "slower" or similar wording, but even that may
not help enough. Test the memory with http://www.memtest86.com,
ALWAYS, for several hours before ever booting to the OS, to avoid file
corruption.

You might also try different combinations, see if the system would
even run stable with only the pair of "new" modules in it. By "run
stable" I again mean running memtest86, not booting to the OS till it
at least tests ok at that.
Click to expand...
Kony,

I have tested with just the new modules in it and I get the same
results. I took your advice and dowloaded memtest86. Both chips failed
miserably on test 4, the errors just kept incrementing. Now, does this
mean the chips are bad or that these chips just are not compatible on my
system? Thanks for your help.
 
Kony,

I have tested with just the new modules in it and I get the same
results. I took your advice and dowloaded memtest86. Both chips failed
miserably on test 4, the errors just kept incrementing. Now, does this
mean the chips are bad or that these chips just are not compatible on my
system? Thanks for your help.
Click to expand...

It could mean either, but it doesn't really matter... for your needs,
they're not suitable. You may as well try to get them replaced.
 
I have a Gateway Athlon 700 with 584MB of memory. 1 256MB SDRAM chip
with ECC, 2 128MB chips, non ECC. Got 2 256MB SpecTek Non ECC chips
from my brother from his old computer. Replaced the 2 128 chips with
the 2 256MB chips, figuring no problems would occur. Lo and behold,
boot the system up and get:

Windows could not start because of an error in the software.
Please report this problem as :
load needed DLLs for kernel
Please contact your support person to report this problem.

Take the 2 chips out, problem fixed, boots up fine. What about these
chips would make the system fail? They should be compatible with my
system.

Any help would be appreciated

thanks,
Oscar
Click to expand...

ALL of the Slot A Athlon mobos were very fussy about the
quality of the memory used in them. Most of these boards
will not support 3 sticks of memory. They also have some
weird chip configuration requirements: they wont take double
sided DIMMs or a certain number of memory chips, etc..

I have 3 Asus K7M Slot A mobos. One of them will only run
with Micron Crucial DIMMs. It will take 2 PC100/128 ECC
DIMMs or 1 PC100/256 and 1 PC100/128 ECC but nothing else
that I've tried.

Another K7M was running great with a mishmash of DIMMs until
this past week, a Corsair PC100/256 DIM, a Kingston
PC100/128 Value RAM DIMM and a Kingston PC133/128 Value RAM
DIMM.

All of a sudden it was posting 384MB instead of 512MB and I
was getting Windows Protection Errors. I spent hours
switching memory and testing with Memtest86 and DocMem. It
only runs with 1 stick now, so it's a mobo problem.

The 3rd board will only run 1 or 2 PC100/128 DIMMs.

I use Micron Crucial memory most of the time and
occasionally the best quality Muskin RAM. I got a "real
deal" on some Kingston ValueRam a while back. It tested fine
with Memtest86 and DocMem memory testing programs.

This weekend I tested all of my systems and found 3 of the
ValueRam DIMMs failed the memory tests! They have a
"lifetime" guarantee but still it's not worth the hassle!

BTW, the new version 3.0 of Memtest86 does a much better job
with Athlons. earlier versions had some problems with AMD
chipsets used on some Athlon mobos.

There is a lot of junk memory on the market. The memory
chips themselves are produced on 200mm (8") or 300mm (12")
silicon wafers. All of the chips are tested before the
wafers are cut up into "die" (chips).

Chips near the center of the wafers are generally the best
quality. To keep their yields per wafer up, chip makers
market finished memory chips in several quality levels. A
few chip makers like Micron, Samsung and Siemens produce
finished memory products but most chips are sold to memory
fabricators who assemble them onto the small printed circuit
boards called memory modules (SIMMs, DIMMs SODIMMs, etc.).

The best modules are made with premium quality matched
memory chips. The worst are made from the bottom of the
yield chips and are sometimes fudged to report that they are
better quality products. Cheap memory sometimes isn't even
tested by low end fabricators; it takes time and costs money
to run these tests - it's cheaper to just replace the
defective products !

Bad or intermittently defective memory is one of the biggest
causes of Windows stability problems.

Check out this site for a better explanation of chip and
memory module manufacturing:
http://www.crucial.com/library/manufacturing.asp

ALL of the newer CPUs, both Intel and AMD have memory
issues - and Power Supply problems. Athlon systems require a
minimum of 300W output from a premium quality power supply.
There have been PSs rated at 300W that actually put out only
195W !
 
kony said:
No, it did not "fail testing for crucial", it's not "shit", it's
simply the lower grade memory, which can't run at the faster timings.
More often than not, when a motherboard won't run the lower grade
memory, it's a flaw with the motherboard, not the memory. Many boards
require higher-spec memory to run at lower spec settings to compensate
for that motherboard's design, especially when multiplie modules are
used.

You're running 3 modules... it was pretty likely you'd have problems
with most modules you try. Many boards required lowering memory bus
speed or relaxing the memory timings to run 3 modules. In other
words, it's "almost" always better to use the least number of modules
possible, though that's 2 if you're running a dual-channel
Click to expand...
board.

Good points but there is some real junk, untested memory on
the market that wont work in fussy systems.

Three memory slots create real electrical problems for
motherboard engineers (Read they are too cheap to add some
additional Caps) !

Intel reduced the slots from 3 to 2 on some of their mobos
for this reason. Some of the early DIMM mobos only had 1
memory slot.
 
No, it did not "fail testing for crucial", it's not "shit", it's
simply the lower grade memory, which can't run at the faster timings.
More often than not, when a motherboard won't run the lower grade
memory, it's a flaw with the motherboard, not the memory. Many boards
require higher-spec memory to run at lower spec settings to compensate
for that motherboard's design, especially when multiplie modules are
used.
Click to expand...

I do not buy memory *rated* at DDR333 and then be happy with it not to
work at that speed. Quite simply, the memory is shit, with only that
module in the motherboard it did not even make the memory test on the
POST. I've run a Nanya memory chip (not exactly the most respected brand
in the world) quite happily alongside the existing TwinMOS and Samsung
modules without any hassles. It's a good motherboard and not one which
is fussy as far as memory is concerned, in my experience. I can honestly
concluse that Spectek honestly is "shit".
 
I do not buy memory *rated* at DDR333 and then be happy with it not to
work at that speed. Quite simply, the memory is shit, with only that
module in the motherboard it did not even make the memory test on the
POST. I've run a Nanya memory chip (not exactly the most respected brand
in the world) quite happily alongside the existing TwinMOS and Samsung
modules without any hassles. It's a good motherboard and not one which
is fussy as far as memory is concerned, in my experience. I can honestly
concluse that Spectek honestly is "shit".
Click to expand...

Calling memory "shit" because it doesn't work as well as top-shelf, much
more expensive memory, is a waste of time. You're trying to run 3 mixed
dimms, a challenge for many boards that are otherwise "good" unless the
memory is higher-spec.

What motherboard are you using? At the very least it would be useful for
others who have that board, to know about this potential issue, since many
boards can run Spectek memory fine. In case it isn't dawning on you, that
probably does make your board "fussy as far as memory is concerned". Are
you running synchronous or asynchronous memory bus?

Maybe this IS a poor lot of spectek memory... or one marginal module...
could happen, though I have seen Spectek PC2100 and 2700 modules run fine
in older boards, up to 185MHz... tested for over a day with memtest86, and
running windows... might've even ran stable faster but the tested systems
were, due to their age, not able to run over 185MHz anyway even with
higher spec memory.
 
I have tested with just the new modules in it and I get the same
results. I took your advice and dowloaded memtest86. Both chips failed
miserably on test 4, the errors just kept incrementing. Now, does this
mean the chips are bad or that these chips just are not compatible on my
system? Thanks for your help.
Click to expand...

Did you go into your BIOS and make sure the settings are compatible to
the memory that you've installed?


Have a nice week...

Trent

Follow Joan Rivers' example --- get pre-embalmed!
 
Calling memory "shit" because it doesn't work as well as top-shelf, much
more expensive memory, is a waste of time.
Click to expand...

The Spectek memory was as expensive excluding P&P as Crucial memory is
delivered, I only bought it because it was a special offer and came with
25 free CDRs. I am going to get a refund and buy Crucial instead!

You're trying to run 3 mixed
dimms, a challenge for many boards that are otherwise "good" unless the
memory is higher-spec.
Click to expand...

If you read my first post, you'd already know that with just the Spectek
module in my system, the system didn't make it past the POST at the
rated speed. However, when I'd lowered the speed, it would be fine. I
ran a MemTest86 at this lowered speed and it didn't detect any errors.
When I used it alongside the other two memory modules, it would crash as
soon as it reached Windows at 333mhz, but work fine at 250mhz.
Therefore, I can only conclude
What motherboard are you using? At the very least it would be useful for
others who have that board, to know about this potential issue, since many
boards can run Spectek memory fine. In case it isn't dawning on you, that
probably does make your board "fussy as far as memory is concerned". Are
you running synchronous or asynchronous memory bus?
Click to expand...

It's an Abit motherboard (BD7-II-RAID to be exact). As I have already
established above, the motherboard recognised and ran the memory OK,
albeit at a vastly reduced speed.

It is a simple matter of the memory not running at its rated speed. Period.
 
Anton said:
The Spectek memory was as expensive excluding P&P as Crucial memory is
delivered, I only bought it because it was a special offer and came
with 25 free CDRs. I am going to get a refund and buy Crucial instead!



If you read my first post, you'd already know that with just the
Spectek module in my system, the system didn't make it past the POST
at the
rated speed. However, when I'd lowered the speed, it would be fine. I
ran a MemTest86 at this lowered speed and it didn't detect any errors.
When I used it alongside the other two memory modules, it would crash
as soon as it reached Windows at 333mhz, but work fine at 250mhz.
Therefore, I can only conclude
Click to expand...

Please! Use actual FSB/memory speeds when discussing this, not DDR ratings.
DDR ratings are not measured in MHz! (Regardless of what the spin-doctors
would have us believe). Unless you are actually trying to run at 333MHz?
(What board would allow that speed?)
 
When memory was sold back in the kilobyte range, it was
common for some resellers to sell memory by claiming higher
speeds. IOW original IC manufacturer would rate memory for a
lower access time based upon statistical measurements - using
good engineering principals. Then reseller would sell that
memory by claiming faster access - only because it could
perform at higher speed when tested later in one condition.

For example, if memory rated at 250 could run in a tester at
333, then that reseller might sell it at 333. Original IC
manufacturer was not at fault. That memory would not run
every time at 333 when driving a full load (ie three SIMMs
attached to memory bus). It would only run at 333 when tested
in some conditions. Just because a memory can run at 333 in
one location is not good enough. Memory rated at 333 must run
at 333 under every condition - max load, max and min
temperature, low but still acceptable voltage, etc. Therefore
original IC manufacturer rated that memory at only 250 and
reseller sold it as 333.

Has Spectek played that game with your memory? Maybe. If
true, it is not about memory only running at lower speed. It
would be about a reseller trying to scam the public to increse
his profits. We can only speculate. But that 'change the
rated memory speed' game has been played by previous memory
resellers to increase profits - customer be damned.
 
I have a Gateway Athlon 700 with 584MB of memory. 1 256MB SDRAM chip
with ECC, 2 128MB chips, non ECC. Got 2 256MB SpecTek Non ECC chips
from my brother from his old computer. Replaced the 2 128 chips with
the 2 256MB chips, figuring no problems would occur. Lo and behold,
boot the system up and get:

Windows could not start because of an error in the software.
Please report this problem as :
load needed DLLs for kernel
Please contact your support person to report this problem.

Take the 2 chips out, problem fixed, boots up fine.
Click to expand...

Maybe the modules are just incompatible with one another, but my
experience last year with Spectec chips on RK-byte modules was pretty
bad. Even setting the BIOS to manual memory configuration and
choosing the slowest timings didn't help, but the errors disappeared
once I slowed the memory bus from 266 MHz to 200 MHz. I went through
several of these modules, and all showed errors, but I finally got a
module with Elixir brand chips, and it worked fine, even with all but
one of the BIOS timings at their fastest settings. Other RK-byte and
generic modules with Elixir chips have tested similarly.

It seems that many memory modules now don't have chip makers' prime
chips but instead have house markings or markings for sub-prime chips,
like Spectec (Micron) or Elixir (Nanya), and now I'm having as good
luck with generic memory from Fry's as with Kingston ValueRAM and
RK-byte. That's not to say that Fry's sells good memory but that the
other companies are putting out garbage'.

I test everything with both MemTest86 (www.memtest86.com) and Gold
Memory (www.goldmemory.cz) because they're two of the best free
diagnostics, and I've had cases where one found errors but the other
didn't. I usually run Gold Memory first because it's a lot faster
(twenty minutes for 256MB, verses hours for MemTest86), and for a long
time it detected every error that MemTest86 would eventually find.
But the last Kingston module I bought passed Gold Memory every time
but failed MemTest86.
 
If you read my first post, you'd already know that with just the Spectek
module in my system, the system didn't make it past the POST at the
rated speed. However, when I'd lowered the speed, it would be fine. I
ran a MemTest86 at this lowered speed and it didn't detect any errors.
When I used it alongside the other two memory modules, it would crash as
soon as it reached Windows at 333mhz, but work fine at 250mhz.
Therefore, I can only conclude


It's an Abit motherboard (BD7-II-RAID to be exact). As I have already
established above, the motherboard recognised and ran the memory OK,
albeit at a vastly reduced speed.

It is a simple matter of the memory not running at its rated speed. Period.
Click to expand...

Don't forget that you have one module, one motherboard... not a very
large sample size. Contrast that with hundreds of thousands (or more) of
the modules sold, and your results look pretty atypical. I myself have
had exposure to tens of dozens, maybe even hundreds of people who've
bought Spectek memory from deals posted in web forums in just this past
year alone, and there were very, very few reports of any problems,
actually I only recall a single user with a questionable motherboard
(early K7S5A) having a problem, except for a few PC133 deals where people
bought memory of higher density than their motherboard could support.

So now I know of two people, each having tried one module, of memory sold
in very large chain stores to hundreds of thousands of people... don't you
think we'd have heard about more problems if the memory was as bad as (the
conclusion you jumped to)?

At this point all we know for sure is that it doesn't work properly in
your specific motherboard... might be constructive to report that to
Spectek.
 
konywrote:
If you read my first post, you'd already know that with just the Spectek
module in my system, the system didn't make it past the POST at the
rated speed. However, when I'd lowered the speed, it would be fine. I
ran a MemTest86 at this lowered speed and it didn't detect any errors.
When I used it alongside the other two memory modules, it would crash as
soon as it reached Windows at 333mhz, but work fine at 250mhz.
Therefore, I can only conclude

What motherboard are you using? At the very least it would be useful for
others who have that board, to know about this potential issue, since many
boards can run Spectek memory fine. In case it isn't dawning on you, that
probably does make your board "fussy as far as memory is concerned". Are
you running synchronous or asynchronous memory bus?

Hello all. This is my first post here.

I am playing with Linux and am running Red Hat 9 on my very old
Click to expand...
Pentium class machine. In order to do this I have upgraded
everything but the case and floppy drive!
Hence my observation and request for information.

I am running an ASUS P/I-P55T2P4 mother board with 200 MHz Pentium.
Click to expand...
The mobd has an Intel 82430HX chipset.
I have purchased 4 identical sticks of EDO (Kingston) RAM of 64Mb. I
Click to expand...
can use any two of these quite happily. My problem occurs when I add
all 4 chips.
The machine boots and runs Win98 quite happily, until I open an
Click to expand...
application. I then get the blue screen of death. Linux does not boot
at all. I remove half the RAM and away we go again, no problems,
 
Back
Top