Puzzling Problems -- Pulling Hair

Paul wrote:
> Bill Anderson wrote:
>> Paul wrote:
>>> Bill Anderson wrote:
>>
>>>>
>>>> Still -- am I making a mistake trying to make my memory sticks run
>>>> at 1066? Crucial advertises that they will. Should I just give
>>>> that up? Ideas for me to try on Tuesday?
>>>>
>>>
>>> Is Vdimm raised, to help the memory work at DDR2-1066 ? I don't think
>>> any DDR2 wants to do that with only 1.8V applied. It probably takes
>>> more than that.
>>>
>>
>> I haven't run CPU-Z today. Here's what it looked like earlier this
>> month when I was having the original problems:
>>
>> http://paintrock.net/cpu-z_spd.jpg
>>
>> So, is that 1.8V or 2.2V?
>>
>>
>
> Cool. I've never seen an EPP displayed in CPUZ before. Yes,
> that looks like 2.2V to run DDR2-1066 (right most column).
> And the command rate is set to 2T as well (command/address
> cycles last for two clocks).
>
> Column 1 shows DDR2-667 with no sweat. Column 2 versus
> column 3 shows that added voltage allows the timing to
> be tightened to 4-4-4. A little looser timing in column 4,
> allows the clock to be raised to DDR2-1066.
>
> There is probably a manufacturer provided maximum voltage
> value, but that might not be shown in CPUZ. In other words,
> more than 2.2V may be allowed, but it would be a good idea
> to check what the value is first. For some companies, you
> have to download their datasheet to find out what the limit
> is.
>

Thanks, Paul. Now I hope you will translate that for me. Is cup-z
saying my system memory is powered by 1.8V or 2.2V? Or is it saying,
"If you give the memory 1.8V you'll get this, but if you give it 2.2V
you'll get that?" Meaning, it's not reading the actual voltage, but
merely examining the memory sticks and reporting on what they're capable
of doing if they're powered as described? Thanks.

--
Bill Anderson

I am the Mighty Favog

Bill Anderson wrote:
> Paul wrote:
>> Bill Anderson wrote:
>>> Paul wrote:
>>>> Bill Anderson wrote:
>>>
>>>>>
>>>>> Still -- am I making a mistake trying to make my memory sticks run
>>>>> at 1066? Crucial advertises that they will. Should I just give
>>>>> that up? Ideas for me to try on Tuesday?
>>>>>
>>>>
>>>> Is Vdimm raised, to help the memory work at DDR2-1066 ? I don't think
>>>> any DDR2 wants to do that with only 1.8V applied. It probably takes
>>>> more than that.
>>>>
>>>
>>> I haven't run CPU-Z today. Here's what it looked like earlier this
>>> month when I was having the original problems:
>>>
>>> http://paintrock.net/cpu-z_spd.jpg
>>>
>>> So, is that 1.8V or 2.2V?
>>>
>>>
>>
>> Cool. I've never seen an EPP displayed in CPUZ before. Yes,
>> that looks like 2.2V to run DDR2-1066 (right most column).
>> And the command rate is set to 2T as well (command/address
>> cycles last for two clocks).
>>
>> Column 1 shows DDR2-667 with no sweat. Column 2 versus
>> column 3 shows that added voltage allows the timing to
>> be tightened to 4-4-4. A little looser timing in column 4,
>> allows the clock to be raised to DDR2-1066.
>>
>> There is probably a manufacturer provided maximum voltage
>> value, but that might not be shown in CPUZ. In other words,
>> more than 2.2V may be allowed, but it would be a good idea
>> to check what the value is first. For some companies, you
>> have to download their datasheet to find out what the limit
>> is.
>>
>
> Thanks, Paul. Now I hope you will translate that for me. Is cup-z
> saying my system memory is powered by 1.8V or 2.2V? Or is it saying,
> "If you give the memory 1.8V you'll get this, but if you give it 2.2V
> you'll get that?" Meaning, it's not reading the actual voltage, but
> merely examining the memory sticks and reporting on what they're capable
> of doing if they're powered as described? Thanks.
>

SPD is a record of capability. Preferably, the capability should be
there by design (all chips can make it based on the label printed
on the chip), but the enthusiast business involves testing memory
outside JEDEC bounds and selling the stuff that passes.

I think "If you give the memory 1.8V you'll get this, but if you give it
2.2V you'll get that" is a good description of the situation.

With the wide variety of raw memory used to make modules, it should
not come as a surprise, that some of it needs more than 1.8V
for practically all settings. And that is the stuff that can get
stuck, when a builder tries to POST for the first time.

Paul

Paul wrote:


>
> I think "If you give the memory 1.8V you'll get this, but if you give it
> 2.2V you'll get that" is a good description of the situation.
>
> With the wide variety of raw memory used to make modules, it should
> not come as a surprise, that some of it needs more than 1.8V
> for practically all settings. And that is the stuff that can get
> stuck, when a builder tries to POST for the first time.
>

Paul, if you have some experience with an Asus P5K-Deluxe Wi-Fi Ap,
maybe you can explain something to me.

Remember, here are the basics:

Motherboard: ASUS P5K DELUXE/WIFI-AP LGA 775 Intel P35 ATX
Processor: Intel Core 2 Quad Q6600 Kentsfield 2.4GHz 2 x 4MB L2 Cache
LGA 775
Memory: Crucial 2GB kit (1GBx2), Ballistix 240-pin DIMM, DDR2
PC2-8500
Video Card: SAPPHIRE 100210L Radeon HD 2600XT 256MB 128-bit GDDR4 PCI
Express x16
Power Supply: PC Power and Cooling ULTRA-QUIET PSU: SILENCER(R) 610
EPS12V

The 1-gig stick I've had since I built the system in August 2007 still
works fine. At least Memtext found no problem with it on one full pass
earlier this evening. My other original stick went bad in early
December, I RMA'd it, and I ran the new one paired with the old one for
about three weeks with no problem.

But the new, RMA'd stick, went bad this past Saturday morning while I
was doing nothing more than typing a Usenet post in Thunderbird.
Nothing else was running. The system just froze and I couldn't reboot
-- couldn't even get a post beep.

I know the stick is bad because:

1) I removed the new stick and found that with just the old stick
installed the system runs fine. A little slow, but fine. Memtest86+
shows no errors.

2) But with just the new stick installed in the same DIMM socket I'd
used for the old stick, the computer won't even boot. These fancy
memory sticks have blinky LEDs on top. The dead stick won't even blink.
Can't run Memtest86+ if the system won't boot.

3) Just be sure, I switched out the sticks again. Old stick:
works/blinks. New Stick: doesn't work/doesn't blink.

OK, I'll be RMAing again tomorrow morning when Crucial reopens from the
holiday.

But what will I tell them? Is it possible I've been killing my memory
with bad BIOS settings? I've had all the memory settings but one set to
AUTO. The one setting I changed from default was the DRAM Frequency
item in BIOS. I set it to DDR2-1066MHz because page 2-13 of the manual
says, "If you install a DDR2-1066 memory module, make sure that you set
the DRAM Frequency item in BIOS to [DDR2-1066MHz]."

Now you have told me I should give the memory 2.2v. When I dug into
BIOS earlier this evening, I found the only way to force 2.2v would be
to look under AI Overclocking and activate N.O.S. THEN I see DRAM
voltage defaulted to AUTO and I'm offered the opportunity to set it
manually to 2.2v. But as best I can tell, this should not be necessary.
BIOS seems to indicate that with everything set to AUTO and N.O.S. not
specifically turned on (merely set to AUTO because AI Overclocking is
set to AUTO), the system will sense the increased voltage demand from
the memory set to 1066, and will adjust accordingly.

Is that right? Have I done nothing wrong? Or by not forcing 2.2v or by
making some other misjudgment have I been damaging the memory sticks
Crucial has sent me? Thanks.

--
Bill Anderson

I am the Mighty Favog

Bill Anderson wrote:
> Paul wrote:
>
>
>>
>> I think "If you give the memory 1.8V you'll get this, but if you give it
>> 2.2V you'll get that" is a good description of the situation.
>>
>> With the wide variety of raw memory used to make modules, it should
>> not come as a surprise, that some of it needs more than 1.8V
>> for practically all settings. And that is the stuff that can get
>> stuck, when a builder tries to POST for the first time.
>>
>
> Paul, if you have some experience with an Asus P5K-Deluxe Wi-Fi Ap,
> maybe you can explain something to me.
>
> Remember, here are the basics:
>
> Motherboard: ASUS P5K DELUXE/WIFI-AP LGA 775 Intel P35 ATX
> Processor: Intel Core 2 Quad Q6600 Kentsfield 2.4GHz 2 x 4MB L2 Cache
> LGA 775
> Memory: Crucial 2GB kit (1GBx2), Ballistix 240-pin DIMM, DDR2
> PC2-8500
> Video Card: SAPPHIRE 100210L Radeon HD 2600XT 256MB 128-bit GDDR4 PCI
> Express x16
> Power Supply: PC Power and Cooling ULTRA-QUIET PSU: SILENCER(R) 610
> EPS12V
>
> The 1-gig stick I've had since I built the system in August 2007 still
> works fine. At least Memtext found no problem with it on one full pass
> earlier this evening. My other original stick went bad in early
> December, I RMA'd it, and I ran the new one paired with the old one for
> about three weeks with no problem.
>
> But the new, RMA'd stick, went bad this past Saturday morning while I
> was doing nothing more than typing a Usenet post in Thunderbird. Nothing
> else was running. The system just froze and I couldn't reboot --
> couldn't even get a post beep.
>
> I know the stick is bad because:
>
> 1) I removed the new stick and found that with just the old stick
> installed the system runs fine. A little slow, but fine. Memtest86+
> shows no errors.
>
> 2) But with just the new stick installed in the same DIMM socket I'd
> used for the old stick, the computer won't even boot. These fancy
> memory sticks have blinky LEDs on top. The dead stick won't even blink.
> Can't run Memtest86+ if the system won't boot.
>
> 3) Just be sure, I switched out the sticks again. Old stick:
> works/blinks. New Stick: doesn't work/doesn't blink.
>
> OK, I'll be RMAing again tomorrow morning when Crucial reopens from the
> holiday.
>
> But what will I tell them? Is it possible I've been killing my memory
> with bad BIOS settings? I've had all the memory settings but one set to
> AUTO. The one setting I changed from default was the DRAM Frequency
> item in BIOS. I set it to DDR2-1066MHz because page 2-13 of the manual
> says, "If you install a DDR2-1066 memory module, make sure that you set
> the DRAM Frequency item in BIOS to [DDR2-1066MHz]."
>
> Now you have told me I should give the memory 2.2v. When I dug into
> BIOS earlier this evening, I found the only way to force 2.2v would be
> to look under AI Overclocking and activate N.O.S. THEN I see DRAM
> voltage defaulted to AUTO and I'm offered the opportunity to set it
> manually to 2.2v. But as best I can tell, this should not be necessary.
> BIOS seems to indicate that with everything set to AUTO and N.O.S. not
> specifically turned on (merely set to AUTO because AI Overclocking is
> set to AUTO), the system will sense the increased voltage demand from
> the memory set to 1066, and will adjust accordingly.
>
> Is that right? Have I done nothing wrong? Or by not forcing 2.2v or by
> making some other misjudgment have I been damaging the memory sticks
> Crucial has sent me? Thanks.
>

Sorry to reply to my own post, but I've realized there's something
misleading in my previous message.

Here's the description of my memory stick from the Crucial website:

Specs: DDR2 PC2-8500 . 5-5-5-15 . Unbuffered . NON-ECC . DDR2-1066 .
2.2v . 128Meg x 64

So it really is rated DDR2-1066 at 2.2v.

--
Bill Anderson

I am the Mighty Favog

Bill Anderson wrote:
> Paul wrote:
>
>
>>
>> I think "If you give the memory 1.8V you'll get this, but if you give it
>> 2.2V you'll get that" is a good description of the situation.
>>
>> With the wide variety of raw memory used to make modules, it should
>> not come as a surprise, that some of it needs more than 1.8V
>> for practically all settings. And that is the stuff that can get
>> stuck, when a builder tries to POST for the first time.
>>
>
> Paul, if you have some experience with an Asus P5K-Deluxe Wi-Fi Ap,
> maybe you can explain something to me.
>
> Remember, here are the basics:
>
> Motherboard: ASUS P5K DELUXE/WIFI-AP LGA 775 Intel P35 ATX
> Processor: Intel Core 2 Quad Q6600 Kentsfield 2.4GHz 2 x 4MB L2 Cache
> LGA 775
> Memory: Crucial 2GB kit (1GBx2), Ballistix 240-pin DIMM, DDR2
> PC2-8500
> Video Card: SAPPHIRE 100210L Radeon HD 2600XT 256MB 128-bit GDDR4 PCI
> Express x16
> Power Supply: PC Power and Cooling ULTRA-QUIET PSU: SILENCER(R) 610
> EPS12V
>
> The 1-gig stick I've had since I built the system in August 2007 still
> works fine. At least Memtext found no problem with it on one full pass
> earlier this evening. My other original stick went bad in early
> December, I RMA'd it, and I ran the new one paired with the old one for
> about three weeks with no problem.
>
> But the new, RMA'd stick, went bad this past Saturday morning while I
> was doing nothing more than typing a Usenet post in Thunderbird. Nothing
> else was running. The system just froze and I couldn't reboot --
> couldn't even get a post beep.
>
> I know the stick is bad because:
>
> 1) I removed the new stick and found that with just the old stick
> installed the system runs fine. A little slow, but fine. Memtest86+
> shows no errors.
>
> 2) But with just the new stick installed in the same DIMM socket I'd
> used for the old stick, the computer won't even boot. These fancy
> memory sticks have blinky LEDs on top. The dead stick won't even blink.
> Can't run Memtest86+ if the system won't boot.
>
> 3) Just be sure, I switched out the sticks again. Old stick:
> works/blinks. New Stick: doesn't work/doesn't blink.
>
> OK, I'll be RMAing again tomorrow morning when Crucial reopens from the
> holiday.
>
> But what will I tell them? Is it possible I've been killing my memory
> with bad BIOS settings? I've had all the memory settings but one set to
> AUTO. The one setting I changed from default was the DRAM Frequency
> item in BIOS. I set it to DDR2-1066MHz because page 2-13 of the manual
> says, "If you install a DDR2-1066 memory module, make sure that you set
> the DRAM Frequency item in BIOS to [DDR2-1066MHz]."
>
> Now you have told me I should give the memory 2.2v. When I dug into
> BIOS earlier this evening, I found the only way to force 2.2v would be
> to look under AI Overclocking and activate N.O.S. THEN I see DRAM
> voltage defaulted to AUTO and I'm offered the opportunity to set it
> manually to 2.2v. But as best I can tell, this should not be necessary.
> BIOS seems to indicate that with everything set to AUTO and N.O.S. not
> specifically turned on (merely set to AUTO because AI Overclocking is
> set to AUTO), the system will sense the increased voltage demand from
> the memory set to 1066, and will adjust accordingly.
>
> Is that right? Have I done nothing wrong? Or by not forcing 2.2v or by
> making some other misjudgment have I been damaging the memory sticks
> Crucial has sent me? Thanks.
>

OK, to start with, try to find customer reviews for the product. I don't know
if Newegg has Ballistix or not, but they do have customer reviews. If you
could find a web forum (I don't know of one off hand for Crucial), that
would be another source of feedback. Other sources of customer comments
may be too sparse to make for a consistent picture. I have seen memory
products listed on Newegg, where the product failed after a short period
of time, so it does happen. With Crucial, that is what you've got the
warranty for, and why you paid a few extra bucks.

And I have run into the odd posting in the groups here, where a board
consistently blows one stick of memory after another (taking several
months to do it to each one). That could be supply voltage, but it is
a bit difficult for the average owner to reverse engineer the thing
and check for that.

I checked three datasheets, and examined the absolute maximum table.
The Absolute Max for all three DDR2 memories was 2.3V for the supply.
The P35 chipset has a spec of 4.0V max for the same rail, so the chipset
should not be able to be blown (if you can believe that number - that
sounds more like the clamp diode rating than a supply rail).

In terms of getting at the voltage setting, I would *not* use N.O.S.
When I get a board, I set certain of the fields to [Manual]. That
exposes more of the useful settings. It also makes *me* in control
of the voltages used. For me, the less stuff going on behind my
back, the better.

The reason I say that, is there have been cases where a BIOS applied
the wrong voltage. And for any voltages which lack a hardware monitor
input, you cannot really look at the results and see what is going on.
I'm not going to conclude, based on your experience, that this is the
root cause of your problems. Philosophically, I prefer the user to
always be in control, and for the settings to be visible so you
can see them (and get help if you need it).

In terms of the actual circuit implementation, sometimes Asus uses
simple opamp circuits for voltage regulation purposes. Those are
not quite as accurate as say, a switching regulator chip with a
bandgap reference inside. Now, in your case, I think I see a switcher
near the top of this picture, above the four DIMM slots. So the
supply for the memory should be pretty good. Of course, if the
BIOS feeds a bogus setting to the circuit, that wouldn't help matters.

http://www.xbitlabs.com/images/mainb...luxe/board.jpg

The "EPP" feature I was marveling at, is used by some of the Nvidia
chipset boards. The Nvidia BIOS code would parse the extra information.
Based on that, it is possible for an EPP enabled BIOS, to adjust the
DIMM voltage according to the contents of the SPD on the DIMM. In the
case of your board, you have an Intel P35 Northbridge, so Nvidia isn't
going to write an EPP enabled module for you.

That is good news in a way, because it means it is more likely to
require you, the operator, to turn up the voltage when needed. And
especially if using the [Manual] setting - in that case you should
be in control of it.

Let's walk through a typical scenario, for a motherboard without an
EPP BIOS.

1) User selects DDR2-533. Board happens to be at 1.8V supply.
Board posts, user is happy.

2) User decides to dial to DDR2-1066, but doesn't know about voltage.
System POSTs, but gets stuck in a boot loop, maybe crashes when the
mouse moves. User posts to USENET. Respondents say "turn up voltage,
test with memtest86+".

3) User finds the manual voltage adjustment, bumps up Vdimm, and finally
the system is stable.

With an EPP BIOS and EPP memory plugged in, it would be:

1) User selects DDR2-533. EPP BIOS automatically sets 1.8V. Everything is OK.
2) User selects DDR2-1066. EPP BIOS automatically sets 2.2V. Everything is OK.

In your case, I see no reason for a guilt trip. It is the BIOS designer's
responsibility, to use reasonable judgment in selecting a Vdimm operating
voltage. The BIOS designer has access to the same datasheets that I just
looked at, so they know that 2.3V is Absolute Max. When the BIOS is placed
in Manual mode, then the user is in control, and the user has to use good
judgment (or at least be aware of the danger of going too far). This is
one reason, that more than one Asus board has had limited voltage range,
since they don't want people to go too far. There have been DFI boards
with tremendous adjustment ranges (like allowing 3.3V for a 2.5V memory),
so there are brands of motherboards known to provide more voltage. There
has been the odd accident with those boards, as you'd imagine.

Return the bad stick and try again. It could even be the SPD that
failed, and not a memory chip. (SPD connects to the SMBUS, a serial
bus that goes all over the place. A failure by any device on that bus,
can cause the bus to stop functioning.)

Paul

Paul wrote:

>
> Return the bad stick and try again. It could even be the SPD that
> failed, and not a memory chip. (SPD connects to the SMBUS, a serial
> bus that goes all over the place. A failure by any device on that bus,
> can cause the bus to stop functioning.)
>

I put the bad stick in the mail to Crucial about an hour ago. They're
shipping my new stick today.

When I explained the details to the nice lady at Crucial customer
support, she suggested I manually set the voltage to 2.2v. So with the
one good stick installed, I have just set AI Overclocking to MANUAL in
order to get access to DRAM voltage options. I've set DRAM Frequency to
DDR2-1066MHz and I've set the voltage to 2.2v. Everything seems stable,
and in fact even with just the one 1-gig stick of RAM installed, things
seem to be running a bit faster now.

She too saw no reason to think I had harmed my memory sticks through
improper BIOS settings. She assumes the AUTO settings should have
boosted the voltage properly when DRAM Frequency is set to DDR2-1066MHz,
but like you she believes the user should be in control.

She said my only problem may be that I have just had very bad luck. I
agreed that that is one possibility. We'll see what happens.

Paul, thank you once again for your very valuable assistance.

--
Bill Anderson

I am the Mighty Favog

Bill Anderson wrote:
> Paul wrote:
>
>>
>> Return the bad stick and try again. It could even be the SPD that
>> failed, and not a memory chip. (SPD connects to the SMBUS, a serial
>> bus that goes all over the place. A failure by any device on that bus,
>> can cause the bus to stop functioning.)
>>
>
> I put the bad stick in the mail to Crucial about an hour ago. They're
> shipping my new stick today.
>
> When I explained the details to the nice lady at Crucial customer
> support, she suggested I manually set the voltage to 2.2v. So with the
> one good stick installed, I have just set AI Overclocking to MANUAL in
> order to get access to DRAM voltage options. I've set DRAM Frequency to
> DDR2-1066MHz and I've set the voltage to 2.2v. Everything seems stable,
> and in fact even with just the one 1-gig stick of RAM installed, things
> seem to be running a bit faster now.
>
> She too saw no reason to think I had harmed my memory sticks through
> improper BIOS settings. She assumes the AUTO settings should have
> boosted the voltage properly when DRAM Frequency is set to DDR2-1066MHz,
> but like you she believes the user should be in control.
>
> She said my only problem may be that I have just had very bad luck. I
> agreed that that is one possibility. We'll see what happens.
>
> Paul, thank you once again for your very valuable assistance.
>


OK, the new stick arrived yesterday and I plugged it in and carefully
made sure I had BIOS set manually to feed both sticks 2.2v and that I'd
set DRAM frequency item to [DDR2-1066MHz] just as the manual prescribed.
The woman from Crucial had assured me that this is the way to go. And
indeed everything seemed to be working perfectly.

This morning before leaving for the office I started up Memtest86+ and
let it run all day. When I got home this afternoon, I found this:

http://paintrock.net/memtest080108b.jpg

So please, can anyone around here help me interpret? Should I RMA the
memory for a third time? Can I even tell from this page which of the
two memory sticks returned the 63 errors -- the old one in DIMM_A1 or
the new one in DIMM_B1? Maybe I need to remove one stick and test them
one at a time?

Or is this even something to worry about? Do all memory sticks return
occasional errors like this?

Please -- help again. And thanks.

--
Bill Anderson

I am the Mighty Favog

Bill Anderson wrote:
> Bill Anderson wrote:
>> Paul wrote:
>>
>>>
>>> Return the bad stick and try again. It could even be the SPD that
>>> failed, and not a memory chip. (SPD connects to the SMBUS, a serial
>>> bus that goes all over the place. A failure by any device on that bus,
>>> can cause the bus to stop functioning.)
>>>
>>
>> I put the bad stick in the mail to Crucial about an hour ago. They're
>> shipping my new stick today.
>>
>> When I explained the details to the nice lady at Crucial customer
>> support, she suggested I manually set the voltage to 2.2v. So with
>> the one good stick installed, I have just set AI Overclocking to
>> MANUAL in order to get access to DRAM voltage options. I've set DRAM
>> Frequency to DDR2-1066MHz and I've set the voltage to 2.2v.
>> Everything seems stable, and in fact even with just the one 1-gig
>> stick of RAM installed, things seem to be running a bit faster now.
>>
>> She too saw no reason to think I had harmed my memory sticks through
>> improper BIOS settings. She assumes the AUTO settings should have
>> boosted the voltage properly when DRAM Frequency is set to
>> DDR2-1066MHz, but like you she believes the user should be in control.
>>
>> She said my only problem may be that I have just had very bad luck. I
>> agreed that that is one possibility. We'll see what happens.
>>
>> Paul, thank you once again for your very valuable assistance.
>>
>
>
> OK, the new stick arrived yesterday and I plugged it in and carefully
> made sure I had BIOS set manually to feed both sticks 2.2v and that I'd
> set DRAM frequency item to [DDR2-1066MHz] just as the manual prescribed.
> The woman from Crucial had assured me that this is the way to go. And
> indeed everything seemed to be working perfectly.
>
> This morning before leaving for the office I started up Memtest86+ and
> let it run all day. When I got home this afternoon, I found this:
>
> http://paintrock.net/memtest080108b.jpg
>
> So please, can anyone around here help me interpret? Should I RMA the
> memory for a third time? Can I even tell from this page which of the
> two memory sticks returned the 63 errors -- the old one in DIMM_A1 or
> the new one in DIMM_B1? Maybe I need to remove one stick and test them
> one at a time?
>
> Or is this even something to worry about? Do all memory sticks return
> occasional errors like this?
>
> Please -- help again. And thanks.
>

There has got to be an answer here somewhere :-)

Take the working stick. Test by itself. Try in each socket of the
motherboard and run a couple passes of memtest86+ in each case.
Do all the sockets pass ?

Now, take the brand new stick you got from Crucial. Test in just
one slot. You know the slots work, so there is no need to try
all slots against the brand new stick.

If that passes, the next test case is two sticks. Install stick 1
in channel 1 and stick 2 in channel 2. (Maybe you already did
that test.) Now, install stick 1 in channel 2 and stick 2 in
channel 1. Does the error "follow the stick" or "stay with the
channel", or did the error disappear entirely ?

Either the fault is motherboard, or it is the DIMM. By running
enough test cases, you should be able to determine whether the
motherboard Northbridge has a bad driver, or the DIMM is
bad (yet again). Having one working (known good) stick is a
great benefit in this case, because you can use that to establish
whether the motherboard is flat out broke or not.

I take it, you've also done the requisite check of postings here.
Sometimes, the BIOS design is responsible for problems with
memory compatibility. See if there is evidence here, that the
memory should be working or not.

http://vip.asus.com/forum/topic.aspx...Language=en-us

Paul

Paul wrote:
> Bill Anderson wrote:
>> Bill Anderson wrote:
>>> Paul wrote:
>>>
>>>>
>>>> Return the bad stick and try again. It could even be the SPD that
>>>> failed, and not a memory chip. (SPD connects to the SMBUS, a serial
>>>> bus that goes all over the place. A failure by any device on that bus,
>>>> can cause the bus to stop functioning.)
>>>>
>>>
>>> I put the bad stick in the mail to Crucial about an hour ago.
>>> They're shipping my new stick today.
>>>
>>> When I explained the details to the nice lady at Crucial customer
>>> support, she suggested I manually set the voltage to 2.2v. So with
>>> the one good stick installed, I have just set AI Overclocking to
>>> MANUAL in order to get access to DRAM voltage options. I've set DRAM
>>> Frequency to DDR2-1066MHz and I've set the voltage to 2.2v.
>>> Everything seems stable, and in fact even with just the one 1-gig
>>> stick of RAM installed, things seem to be running a bit faster now.
>>>
>>> She too saw no reason to think I had harmed my memory sticks through
>>> improper BIOS settings. She assumes the AUTO settings should have
>>> boosted the voltage properly when DRAM Frequency is set to
>>> DDR2-1066MHz, but like you she believes the user should be in control.
>>>
>>> She said my only problem may be that I have just had very bad luck.
>>> I agreed that that is one possibility. We'll see what happens.
>>>
>>> Paul, thank you once again for your very valuable assistance.
>>>
>>
>>
>> OK, the new stick arrived yesterday and I plugged it in and carefully
>> made sure I had BIOS set manually to feed both sticks 2.2v and that
>> I'd set DRAM frequency item to [DDR2-1066MHz] just as the manual
>> prescribed. The woman from Crucial had assured me that this is the
>> way to go. And indeed everything seemed to be working perfectly.
>>
>> This morning before leaving for the office I started up Memtest86+ and
>> let it run all day. When I got home this afternoon, I found this:
>>
>> http://paintrock.net/memtest080108b.jpg
>>
>> So please, can anyone around here help me interpret? Should I RMA the
>> memory for a third time? Can I even tell from this page which of the
>> two memory sticks returned the 63 errors -- the old one in DIMM_A1 or
>> the new one in DIMM_B1? Maybe I need to remove one stick and test them
>> one at a time?
>>
>> Or is this even something to worry about? Do all memory sticks return
>> occasional errors like this?
>>
>> Please -- help again. And thanks.
>>
>
> There has got to be an answer here somewhere :-)
>
> Take the working stick. Test by itself. Try in each socket of the
> motherboard and run a couple passes of memtest86+ in each case.
> Do all the sockets pass ?
>
> Now, take the brand new stick you got from Crucial. Test in just
> one slot. You know the slots work, so there is no need to try
> all slots against the brand new stick.
>
> If that passes, the next test case is two sticks. Install stick 1
> in channel 1 and stick 2 in channel 2. (Maybe you already did
> that test.) Now, install stick 1 in channel 2 and stick 2 in
> channel 1. Does the error "follow the stick" or "stay with the
> channel", or did the error disappear entirely ?
>
> Either the fault is motherboard, or it is the DIMM. By running
> enough test cases, you should be able to determine whether the
> motherboard Northbridge has a bad driver, or the DIMM is
> bad (yet again). Having one working (known good) stick is a
> great benefit in this case, because you can use that to establish
> whether the motherboard is flat out broke or not.
>
> I take it, you've also done the requisite check of postings here.
> Sometimes, the BIOS design is responsible for problems with
> memory compatibility. See if there is evidence here, that the
> memory should be working or not.
>
> http://vip.asus.com/forum/topic.aspx...Language=en-us
>
>
> Paul

When I tested the new stick I got no errors. When I tested the old
stick (the original from August) I got this:

http://paintrock.net/memtest080109.jpg

I have had a nice long conversation this evening with a young man at
Crucial. He has read my detailed email and looked at the screen photos
and listened to my story and he has concluded I'm doing nothing wrong,
my temperatures aren't too high, and so forth and so on. We did find
that my timings (set to auto) weren't right. They were 5-5-5-18 rather
than 5-5-5-15, so I fixed that with a manual setting.

Now I am once again RMAing a stick of memory to Crucial. I'm also
buying two more identical sticks because he has told me that even if
32-bit XP and Vista won't see all if it, my BIOS will and I'll get the
added advantage. I also figure that this way when the next stick goes
bad I'll have three to fall back on.

I furthermore figure I'm an idiot.

--
Bill Anderson

I am the Mighty Favog

Bill Anderson wrote:
> Paul wrote:
>> Bill Anderson wrote:
>>> Bill Anderson wrote:
>>>> Paul wrote:
>>>>
>>>>>
>>>>> Return the bad stick and try again. It could even be the SPD that
>>>>> failed, and not a memory chip. (SPD connects to the SMBUS, a serial
>>>>> bus that goes all over the place. A failure by any device on that bus,
>>>>> can cause the bus to stop functioning.)
>>>>>
>>>>
>>>> I put the bad stick in the mail to Crucial about an hour ago.
>>>> They're shipping my new stick today.
>>>>
>>>> When I explained the details to the nice lady at Crucial customer
>>>> support, she suggested I manually set the voltage to 2.2v. So with
>>>> the one good stick installed, I have just set AI Overclocking to
>>>> MANUAL in order to get access to DRAM voltage options. I've set
>>>> DRAM Frequency to DDR2-1066MHz and I've set the voltage to 2.2v.
>>>> Everything seems stable, and in fact even with just the one 1-gig
>>>> stick of RAM installed, things seem to be running a bit faster now.
>>>>
>>>> She too saw no reason to think I had harmed my memory sticks through
>>>> improper BIOS settings. She assumes the AUTO settings should have
>>>> boosted the voltage properly when DRAM Frequency is set to
>>>> DDR2-1066MHz, but like you she believes the user should be in control.
>>>>
>>>> She said my only problem may be that I have just had very bad luck.
>>>> I agreed that that is one possibility. We'll see what happens.
>>>>
>>>> Paul, thank you once again for your very valuable assistance.
>>>>
>>>
>>>
>>> OK, the new stick arrived yesterday and I plugged it in and carefully
>>> made sure I had BIOS set manually to feed both sticks 2.2v and that
>>> I'd set DRAM frequency item to [DDR2-1066MHz] just as the manual
>>> prescribed. The woman from Crucial had assured me that this is the
>>> way to go. And indeed everything seemed to be working perfectly.
>>>
>>> This morning before leaving for the office I started up Memtest86+
>>> and let it run all day. When I got home this afternoon, I found this:
>>>
>>> http://paintrock.net/memtest080108b.jpg
>>>
>>> So please, can anyone around here help me interpret? Should I RMA
>>> the memory for a third time? Can I even tell from this page which of
>>> the two memory sticks returned the 63 errors -- the old one in
>>> DIMM_A1 or the new one in DIMM_B1? Maybe I need to remove one stick
>>> and test them one at a time?
>>>
>>> Or is this even something to worry about? Do all memory sticks
>>> return occasional errors like this?
>>>
>>> Please -- help again. And thanks.
>>>
>>
>> There has got to be an answer here somewhere :-)
>>
>> Take the working stick. Test by itself. Try in each socket of the
>> motherboard and run a couple passes of memtest86+ in each case.
>> Do all the sockets pass ?
>>
>> Now, take the brand new stick you got from Crucial. Test in just
>> one slot. You know the slots work, so there is no need to try
>> all slots against the brand new stick.
>>
>> If that passes, the next test case is two sticks. Install stick 1
>> in channel 1 and stick 2 in channel 2. (Maybe you already did
>> that test.) Now, install stick 1 in channel 2 and stick 2 in
>> channel 1. Does the error "follow the stick" or "stay with the
>> channel", or did the error disappear entirely ?
>>
>> Either the fault is motherboard, or it is the DIMM. By running
>> enough test cases, you should be able to determine whether the
>> motherboard Northbridge has a bad driver, or the DIMM is
>> bad (yet again). Having one working (known good) stick is a
>> great benefit in this case, because you can use that to establish
>> whether the motherboard is flat out broke or not.
>>
>> I take it, you've also done the requisite check of postings here.
>> Sometimes, the BIOS design is responsible for problems with
>> memory compatibility. See if there is evidence here, that the
>> memory should be working or not.
>>
>> http://vip.asus.com/forum/topic.aspx...Language=en-us
>>
>>
>> Paul
>
> When I tested the new stick I got no errors. When I tested the old
> stick (the original from August) I got this:
>
> http://paintrock.net/memtest080109.jpg
>
> I have had a nice long conversation this evening with a young man at
> Crucial. He has read my detailed email and looked at the screen photos
> and listened to my story and he has concluded I'm doing nothing wrong,
> my temperatures aren't too high, and so forth and so on. We did find
> that my timings (set to auto) weren't right. They were 5-5-5-18 rather
> than 5-5-5-15, so I fixed that with a manual setting.
>
> Now I am once again RMAing a stick of memory to Crucial. I'm also
> buying two more identical sticks because he has told me that even if
> 32-bit XP and Vista won't see all if it, my BIOS will and I'll get the
> added advantage. I also figure that this way when the next stick goes
> bad I'll have three to fall back on.
>
> I furthermore figure I'm an idiot.
>

Hmmm. Maybe with four sticks, you'll be able to scrape together two
working ones ??? I guess that is what the young man at Crucial was
thinking. [ I wonder at what point, they'd give you your money back ? ]

I just tried the vip.asus.com forums, and I'm getting a blank page
come up. So I cannot have a look through their results right now.

Another place to look, is the Corsair houseofhelp site. See if any
of the experiences here, match what you are seeing. I searched
for "P5K Deluxe" in this search. (I don't know if the searchid will
still be valid when you use it or not.) Some of Corsair's products
will use Micron chip, and you may see a pattern there.

http://www.houseofhelp.com/v3/search...archid=1375521

You might also want to try a site like this one. Their search engine
isn't working for me, so I use Altavista instead. The second link
is a search for "Ballistix PC2-8500" on the xtremesystems.org domain.

http://www.xtremesystems.org/forums/index.php

http://www.altavista.com/web/results...stix+pc2-8500&
aqp=&aqo=&aqn=&kgs=1&kls=0&dt=tmperiod&d2=0&dfr%5B d%5D=1&dfr%5Bm%5D=1&dfr%5By%5D=1980&
dto%5Bd%5D=9&dto%5Bm%5D=1&dto%5By%5D=2008&filetype =&rc=dmn&swd=xtremesystems.org&lh=&nbq=10

For example, in this thread, the OP had to bump the Northbridge
voltage.

http://www.xtremesystems.org/forums/...36#post2523236

Paul