PC Motherboard Chipsets and Parts Vendors

Mike Rivers wrote:
> Chel van Gennip wrote:
>
>> Some say, you don't make important recordings that often, so it won't
>> matter, others say a few $ to exclude a source of random errors is
>> worth the money.
>
> But all ECC does is correct a single-bit error. It doesn't guarantee a
> safe landing all the time. If a cosmic ray can hit one bit, it can hit
> two. There are probably real statistics about this, but I probably
> couldn't understand them even if I found them.
>

ECC memory does correct single bit errors and does detect multitbit
errors up to 4 bits (depending on the error). Cosmic radiation flipping
bits are small (much smaller than the size of a bit on a chip)
particles. The chances of 2 particles hitting the same word at the same
time are many orders of magnitude smaller than 1 particle hitting one
bit, but even those errors are detected and will not propagate to
serious errors in your data.

--
Chel van Gennip (chel vangennip nl)
Visit Serg van Gennip's site * w w w .serg.vangennip . com

"Soundhaspriority" wrote ...
> Here are some numbers for you. This paper was revised 2004, the most
> up to date source I could find on the web.
> w w w .tezzaron . com /about/papers/soft_errors_1_1_secure.pdf
>
> The Micron estimate in the table is 2-4 errors per year per 256 mb. In
> a computer with 4 gigabytes of ram, this translates to 32-64 errors in
> memory, per year, or approximately:
>
> ONE PER WEEK.
>
> Please consult your keychain before you reply.

LOL! :-)
Perhaps SER is what causes the Radium phenomenon.

Sorry, I remain unconvinced. Me and 100s of millions
of other users of "trashy consumer desktop" systems
will just have to live on the edge waiting for that Sword
of Damacles SER waiting to fall on us. I'm trembling
in fear.

Could it be that Micron, et.al. makes more profit on
ECC memory? No, they couldn't possibly have such a
crass motivation. They are just looking out for our good.

Buy ECC if you wish. I'm sure Micron will appreciate it.

"Richard Crowley" <rcrowley@xp7rt . net > wrote in message
news:lIOdndu6X68wUrfVnZ2dnUVZ_uadnZ2d@posted.pcez...
> "Soundhaspriority" wrote ...
>> Here are some numbers for you. This paper was revised 2004, the most up
>> to date source I could find on the web.
>> w w w .tezzaron . com /about/papers/soft_errors_1_1_secure.pdf
>>
>> The Micron estimate in the table is 2-4 errors per year per 256 mb. In a
>> computer with 4 gigabytes of ram, this translates to 32-64 errors in
>> memory, per year, or approximately:
>>
>> ONE PER WEEK.
>>
>> Please consult your keychain before you reply.
>
> LOL! :-)
> Perhaps SER is what causes the Radium phenomenon.
>
> Sorry, I remain unconvinced. Me and 100s of millions
> of other users of "trashy consumer desktop" systems
> will just have to live on the edge waiting for that Sword
> of Damacles SER waiting to fall on us. I'm trembling
> in fear.
>
> Could it be that Micron, et.al. makes more profit on
> ECC memory? No, they couldn't possibly have such a
> crass motivation. They are just looking out for our good.
>
> Buy ECC if you wish. I'm sure Micron will appreciate it.
Rich, you're not being very scientific about this. You're allowing yourself
to be ruled by personal prejudice and comfort factor.

Bob Morein
(310) 237-6511

"Soundhaspriority" wrote ...
> Rich, you're not being very scientific about this.

Nor are you. You're quoting press releases and marketing
stuff from companies who make products they are trying
to sell.

> You're allowing yourself to be ruled by personal prejudice
> and comfort factor.

Yes, I admit that I am prejudiced against people trying
to sell me things that will take my $$ and give me no
perceivable equitable benefit. It is called "prudence"
or even "frugality" if you wish.

OTOH, you personal prejudice appears to be that you
read press releases with industry rag headers around them
and think that it is the latest critical scientific information.

If you think ECC is so critical, by all means go out and buy
whatever you wish. Don't let my silly prejudice dissuade you.

The fact that the 100s of millions of "trashy consumer
desktop" systems seem to be muddling along without
ECC would appear to indicate that our reality isn't
quite in sync with their marketing plans.

Perhaps running >100 "trashy consumer desktop" systems
gives me a different perspective than someone who runs
only 2 or 3 of them.

"Richard Crowley" <rcrowley@xp7rt . net > wrote in message
news:cN-dnYahXpqwQrfVnZ2dnUVZ_uOdnZ2d@posted.pcez...
> "Soundhaspriority" wrote ...
>> Rich, you're not being very scientific about this.
>
> Nor are you. You're quoting press releases and marketing
> stuff from companies who make products they are trying
> to sell.
>
>> You're allowing yourself to be ruled by personal prejudice
>> and comfort factor.
>
> Yes, I admit that I am prejudiced against people trying to sell me things
> that will take my $$ and give me no perceivable equitable benefit. It is
> called "prudence"
> or even "frugality" if you wish.
>
In this case, you are misguided.

> OTOH, you personal prejudice appears to be that you
> read press releases with industry rag headers around them
> and think that it is the latest critical scientific information.
>
Actually, it is.

> If you think ECC is so critical, by all means go out and buy whatever you
> wish. Don't let my silly prejudice dissuade you.
>
I've been doing it for years.

> The fact that the 100s of millions of "trashy consumer desktop" systems
> seem to be muddling along without
> ECC would appear to indicate that our reality isn't
> quite in sync with their marketing plans.

They muddle, and they crash. The consumer experience isn't too good.

>
> Perhaps running >100 "trashy consumer desktop" systems
> gives me a different perspective than someone who runs only 2 or 3 of
> them.

Well, sure. And you've got your magic keychain.

Bob Morein
(310) 237-6511

"Soundhaspriority" wrote ...
> "Richard Crowley" wrote ...
>> "Soundhaspriority" wrote ...
>>> Rich, you're not being very scientific about this.
>>
>> Nor are you. You're quoting press releases and marketing
>> stuff from companies who make products they are trying
>> to sell.
>>
>>> You're allowing yourself to be ruled by personal prejudice
>>> and comfort factor.
>>
>> Yes, I admit that I am prejudiced against people trying to sell me
>> things that will take my $$ and give me no perceivable equitable
>> benefit. It is called "prudence" or even "frugality" if you wish.
>>
> In this case, you are misguided.

"The error rate in today's consumer-level memory is so low
so that for most everyday applications, adding ECC is pure
overkill. For standard DDR2 memory, the error rate is
something like 100 soft errors over 1 billion device hours.
If there are 16 memory devices or chips on a given module,
that translates to one soft error every 30 years. Even if you
only have two such DIMMs in a system, that's still less than
one error for more than the lifetime of the system as a whole."

From an interview with Mike Sanor, compatibility and
performance manager at Crucial Technology, a division
of DRAM manufacturer Micron Technology.
* searchwincomputing.techtarget . com /tip/0,289483,sid68_gci1251848,00.html
This is the reference posted by "JulienBH" elsewhere in this
discussion.

> They muddle, and they crash. The consumer experience isn't too good.

OTOH, my consumer experience and that of everyone
I know personally is rather good. Sorry about yours.

> Well, sure. And you've got your magic keychain.

Bob, "you're not being very scientific about this." :-)

"less than one error for more than the lifetime of the system"
"for most everyday applications, adding ECC is pure overkill"

I'm only doing everyday applications like editing audio and
video and graphics and text. The sky is not falling. SER
doesn't make my list of top 100 things to worry about.

You may have the last word. I'm returning to audio topics.

"Richard Crowley" <rcrowley@xp7rt . net > wrote in message
news:59-dnQM-7eI1crfVnZ2dnUVZ_uadnZ2d@posted.pcez...
> "Soundhaspriority" wrote ...
>> "Richard Crowley" wrote ...
>>> "Soundhaspriority" wrote ...
>>>> Rich, you're not being very scientific about this.
>>>
>>> Nor are you. You're quoting press releases and marketing
>>> stuff from companies who make products they are trying
>>> to sell.
>>>
>>>> You're allowing yourself to be ruled by personal prejudice
>>>> and comfort factor.
>>>
>>> Yes, I admit that I am prejudiced against people trying to sell me
>>> things that will take my $$ and give me no perceivable equitable
>>> benefit. It is called "prudence" or even "frugality" if you wish.
>>>
>> In this case, you are misguided.
>
> "The error rate in today's consumer-level memory is so low
> so that for most everyday applications, adding ECC is pure
> overkill. For standard DDR2 memory, the error rate is
> something like 100 soft errors over 1 billion device hours.
> If there are 16 memory devices or chips on a given module,
> that translates to one soft error every 30 years. Even if you
> only have two such DIMMs in a system, that's still less than
> one error for more than the lifetime of the system as a whole."
>
> From an interview with Mike Sanor, compatibility and
> performance manager at Crucial Technology, a division
> of DRAM manufacturer Micron Technology.
> * searchwincomputing.techtarget . com /tip/0,289483,sid68_gci1251848,00.html
> This is the reference posted by "JulienBH" elsewhere in this
> discussion.
>
>> They muddle, and they crash. The consumer experience isn't too good.
>
> OTOH, my consumer experience and that of everyone
> I know personally is rather good. Sorry about yours.
>
I don't have a problem, because I use ECC. But your expectations are too
low.

>> Well, sure. And you've got your magic keychain.
>
> Bob, "you're not being very scientific about this." :-)
>
> "less than one error for more than the lifetime of the system"
> "for most everyday applications, adding ECC is pure overkill"
>
Well, at least you've stopped referring to your keychain. This puts the
argument on a more level footing. On my side, I have three references. You
have a statement by Mike Sanor.

> I'm only doing everyday applications like editing audio and
> video and graphics and text. The sky is not falling. SER
> doesn't make my list of top 100 things to worry about.
>
I understand that, but anecdotal experience is not admissible. You lack the
credentials, other than your keychain. The fact that you are a computer
administrator is simply not good enough to be admissible. But you have
brought up an interesting statement. Actual published papers would be even
more interesting.

We can continue to argue, on this uncomfortably personal level, where you
have accused me of gullibility, and I have responded in kind. Or we can
leave the question open. I continue to recommend ECC to self-builders like
Mike Rivers, because of the nominal cost, and the established increase in
reliability. I respect your right to disagree. But I do request that you
cease carrying on this discussion on a personal level. If you do so,
consider my references to your keychain, etc., erased. I really didn't want
to go there, but felt provoked.


Bob Morein
(310) 237-6511

On May 14, 10:14 am, "Soundhaspriority" <nowh...@nowhere . com > wrote:
> "Richard Crowley" <rcrow...@xp7rt . net > wrote in message
>
> news:59-dnQM-7eI1crfVnZ2dnUVZ_uadnZ2d@posted.pcez...
>
> > "Soundhaspriority" wrote ...
> >> "Richard Crowley" wrote ...
> >>> "Soundhaspriority" wrote ...
> >>>> Rich, you're not being very scientific about this.
>
> >>> Nor are you. You're quoting press releases and marketing
> >>> stuff from companies who make products they are trying
> >>> to sell.
>
> >>>> You're allowing yourself to be ruled by personal prejudice
> >>>> and comfort factor.
>
> >>> Yes, I admit that I am prejudiced against people trying to sell me
> >>> things that will take my $$ and give me no perceivable equitable
> >>> benefit. It is called "prudence" or even "frugality" if you wish.
>
> >> In this case, you are misguided.
>
> > "The error rate in today's consumer-level memory is so low
> > so that for most everyday applications, adding ECC is pure
> > overkill. For standard DDR2 memory, the error rate is
> > something like 100 soft errors over 1 billion device hours.
> > If there are 16 memory devices or chips on a given module,
> > that translates to one soft error every 30 years. Even if you
> > only have two such DIMMs in a system, that's still less than
> > one error for more than the lifetime of the system as a whole."
>
> > From an interview with Mike Sanor, compatibility and
> > performance manager at Crucial Technology, a division
> > of DRAM manufacturer Micron Technology.
> > * searchwincomputing.techtarget . com /tip/0,289483,sid68_gci125184...
> > This is the reference posted by "JulienBH" elsewhere in this
> > discussion.
>
> >> They muddle, and they crash. The consumer experience isn't too good.
>
> > OTOH, my consumer experience and that of everyone
> > I know personally is rather good. Sorry about yours.
>
> I don't have a problem, because I use ECC. But your expectations are too
> low.
>
> >> Well, sure. And you've got your magic keychain.
>
> > Bob, "you're not being very scientific about this." :-)
>
> > "less than one error for more than the lifetime of the system"
> > "for most everyday applications, adding ECC is pure overkill"
>
> Well, at least you've stopped referring to your keychain. This puts the
> argument on a more level footing. On my side, I have three references. You
> have a statement by Mike Sanor.
>
> > I'm only doing everyday applications like editing audio and
> > video and graphics and text. The sky is not falling. SER
> > doesn't make my list of top 100 things to worry about.
>
> I understand that, but anecdotal experience is not admissible. You lack the
> credentials, other than your keychain. The fact that you are a computer
> administrator is simply not good enough to be admissible. But you have
> brought up an interesting statement. Actual published papers would be even
> more interesting.
>
> We can continue to argue, on this uncomfortably personal level, where you
> have accused me of gullibility, and I have responded in kind. Or we can
> leave the question open. I continue to recommend ECC to self-builders like
> Mike Rivers, because of the nominal cost, and the established increase in
> reliability. I respect your right to disagree. But I do request that you
> cease carrying on this discussion on a personal level. If you do so,
> consider my references to your keychain, etc., erased. I really didn't want
> to go there, but felt provoked.
>
> Bob Morein
> (310) 237-6511

Bob... Don't be so addicted to ECC that's all we're saying. It's not
the MUST HAVE it was a few years ago...
Richard, understand that ECC could be needed for a studio PC that
stays always ON.

"Julien BH" <julienbh@gmail . com > wrote in message
news:c4ee3a6b-969d-43c0-a9d6-edc346105d19@d1g2000hsg.googlegroups . com ...
> On May 14, 10:14 am, "Soundhaspriority" <nowh...@nowhere . com > wrote:
>> "Richard Crowley" <rcrow...@xp7rt . net > wrote in message
>>
>> news:59-dnQM-7eI1crfVnZ2dnUVZ_uadnZ2d@posted.pcez...
>>
>> > "Soundhaspriority" wrote ...
>> >> "Richard Crowley" wrote ...
>> >>> "Soundhaspriority" wrote ...
>> >>>> Rich, you're not being very scientific about this.
>>
>> >>> Nor are you. You're quoting press releases and marketing
>> >>> stuff from companies who make products they are trying
>> >>> to sell.
>>
>> >>>> You're allowing yourself to be ruled by personal prejudice
>> >>>> and comfort factor.
>>
>> >>> Yes, I admit that I am prejudiced against people trying to sell me
>> >>> things that will take my $$ and give me no perceivable equitable
>> >>> benefit. It is called "prudence" or even "frugality" if you wish.
>>
>> >> In this case, you are misguided.
>>
>> > "The error rate in today's consumer-level memory is so low
>> > so that for most everyday applications, adding ECC is pure
>> > overkill. For standard DDR2 memory, the error rate is
>> > something like 100 soft errors over 1 billion device hours.
>> > If there are 16 memory devices or chips on a given module,
>> > that translates to one soft error every 30 years. Even if you
>> > only have two such DIMMs in a system, that's still less than
>> > one error for more than the lifetime of the system as a whole."
>>
>> > From an interview with Mike Sanor, compatibility and
>> > performance manager at Crucial Technology, a division
>> > of DRAM manufacturer Micron Technology.
>> > * searchwincomputing.techtarget . com /tip/0,289483,sid68_gci125184...
>> > This is the reference posted by "JulienBH" elsewhere in this
>> > discussion.
>>
>> >> They muddle, and they crash. The consumer experience isn't too good.
>>
>> > OTOH, my consumer experience and that of everyone
>> > I know personally is rather good. Sorry about yours.
>>
>> I don't have a problem, because I use ECC. But your expectations are too
>> low.
>>
>> >> Well, sure. And you've got your magic keychain.
>>
>> > Bob, "you're not being very scientific about this." :-)
>>
>> > "less than one error for more than the lifetime of the system"
>> > "for most everyday applications, adding ECC is pure overkill"
>>
>> Well, at least you've stopped referring to your keychain. This puts the
>> argument on a more level footing. On my side, I have three references.
>> You
>> have a statement by Mike Sanor.
>>
>> > I'm only doing everyday applications like editing audio and
>> > video and graphics and text. The sky is not falling. SER
>> > doesn't make my list of top 100 things to worry about.
>>
>> I understand that, but anecdotal experience is not admissible. You lack
>> the
>> credentials, other than your keychain. The fact that you are a computer
>> administrator is simply not good enough to be admissible. But you have
>> brought up an interesting statement. Actual published papers would be
>> even
>> more interesting.
>>
>> We can continue to argue, on this uncomfortably personal level, where you
>> have accused me of gullibility, and I have responded in kind. Or we can
>> leave the question open. I continue to recommend ECC to self-builders
>> like
>> Mike Rivers, because of the nominal cost, and the established increase in
>> reliability. I respect your right to disagree. But I do request that you
>> cease carrying on this discussion on a personal level. If you do so,
>> consider my references to your keychain, etc., erased. I really didn't
>> want
>> to go there, but felt provoked.
>>
>> Bob Morein
>> (310) 237-6511
>
> Bob... Don't be so addicted to ECC that's all we're saying. It's not
> the MUST HAVE it was a few years ago...
> Richard, understand that ECC could be needed for a studio PC that
> stays always ON.

Julien,
Referring to my other post in this thread where I describe the
consequences of a soft error, some of effects can be cumulative. When a
project file is saved to disk, any errors in the RAM image are saved as
well. In this way, the errors accumulate proportional to the number of hours
spent on the project, not the number of hours the computer is on. Because a
wav sample can be manually fixed, this is not as devastating as with CAD and
video. On the other hand, corruption of a project control file has pretty
bad conseuqences.

Someone who has to buy a packaged computer, as Mike initially did with
Dell, has little choice. Now Mike is building a machine, and he can have ECC
for $25 - $50 additional. In my opinion, it's the logical choice.

The best way to carry on a usenet discussion is to present the evidence,
which in this case comes from third parties, not us, and let the reader make
the choice. We all benefit from the high quality publications on the web.
It's not about winning arguments. Whatever we find is a win-win for all of
us.

Bob Morein
(310) 237-6511

On May 14, 11:19 am, "Soundhaspriority" <nowh...@nowhere . com > wrote:
> "Julien BH" <julie...@gmail . com > wrote in message
>
> news:c4ee3a6b-969d-43c0-a9d6-edc346105d19@d1g2000hsg.googlegroups . com ...
>
>
>
> > On May 14, 10:14 am, "Soundhaspriority" <nowh...@nowhere . com > wrote:
> >> "Richard Crowley" <rcrow...@xp7rt . net > wrote in message
>
> >>news:59-dnQM-7eI1crfVnZ2dnUVZ_uadnZ2d@posted.pcez...
>
> >> > "Soundhaspriority" wrote ...
> >> >> "Richard Crowley" wrote ...
> >> >>> "Soundhaspriority" wrote ...
> >> >>>> Rich, you're not being very scientific about this.
>
> >> >>> Nor are you. You're quoting press releases and marketing
> >> >>> stuff from companies who make products they are trying
> >> >>> to sell.
>
> >> >>>> You're allowing yourself to be ruled by personal prejudice
> >> >>>> and comfort factor.
>
> >> >>> Yes, I admit that I am prejudiced against people trying to sell me
> >> >>> things that will take my $$ and give me no perceivable equitable
> >> >>> benefit. It is called "prudence" or even "frugality" if you wish.
>
> >> >> In this case, you are misguided.
>
> >> > "The error rate in today's consumer-level memory is so low
> >> > so that for most everyday applications, adding ECC is pure
> >> > overkill. For standard DDR2 memory, the error rate is
> >> > something like 100 soft errors over 1 billion device hours.
> >> > If there are 16 memory devices or chips on a given module,
> >> > that translates to one soft error every 30 years. Even if you
> >> > only have two such DIMMs in a system, that's still less than
> >> > one error for more than the lifetime of the system as a whole."
>
> >> > From an interview with Mike Sanor, compatibility and
> >> > performance manager at Crucial Technology, a division
> >> > of DRAM manufacturer Micron Technology.
> >> > * searchwincomputing.techtarget . com /tip/0,289483,sid68_gci125184...
> >> > This is the reference posted by "JulienBH" elsewhere in this
> >> > discussion.
>
> >> >> They muddle, and they crash. The consumer experience isn't too good.
>
> >> > OTOH, my consumer experience and that of everyone
> >> > I know personally is rather good. Sorry about yours.
>
> >> I don't have a problem, because I use ECC. But your expectations are too
> >> low.
>
> >> >> Well, sure. And you've got your magic keychain.
>
> >> > Bob, "you're not being very scientific about this." :-)
>
> >> > "less than one error for more than the lifetime of the system"
> >> > "for most everyday applications, adding ECC is pure overkill"
>
> >> Well, at least you've stopped referring to your keychain. This puts the
> >> argument on a more level footing. On my side, I have three references.
> >> You
> >> have a statement by Mike Sanor.
>
> >> > I'm only doing everyday applications like editing audio and
> >> > video and graphics and text. The sky is not falling. SER
> >> > doesn't make my list of top 100 things to worry about.
>
> >> I understand that, but anecdotal experience is not admissible. You lack
> >> the
> >> credentials, other than your keychain. The fact that you are a computer
> >> administrator is simply not good enough to be admissible. But you have
> >> brought up an interesting statement. Actual published papers would be
> >> even
> >> more interesting.
>
> >> We can continue to argue, on this uncomfortably personal level, where you
> >> have accused me of gullibility, and I have responded in kind. Or we can
> >> leave the question open. I continue to recommend ECC to self-builders
> >> like
> >> Mike Rivers, because of the nominal cost, and the established increase in
> >> reliability. I respect your right to disagree. But I do request that you
> >> cease carrying on this discussion on a personal level. If you do so,
> >> consider my references to your keychain, etc., erased. I really didn't
> >> want
> >> to go there, but felt provoked.
>
> >> Bob Morein
> >> (310) 237-6511
>
> > Bob... Don't be so addicted to ECC that's all we're saying. It's not
> > the MUST HAVE it was a few years ago...
> > Richard, understand that ECC could be needed for a studio PC that
> > stays always ON.
>
> Julien,
> Referring to my other post in this thread where I describe the
> consequences of a soft error, some of effects can be cumulative. When a
> project file is saved to disk, any errors in the RAM image are saved as
> well. In this way, the errors accumulate proportional to the number of hours
> spent on the project, not the number of hours the computer is on. Because a
> wav sample can be manually fixed, this is not as devastating as with CAD and
> video. On the other hand, corruption of a project control file has pretty
> bad conseuqences.
>
> Someone who has to buy a packaged computer, as Mike initially did with
> Dell, has little choice. Now Mike is building a machine, and he can have ECC
> for $25 - $50 additional. In my opinion, it's the logical choice.
>
> The best way to carry on a usenet discussion is to present the evidence,
> which in this case comes from third parties, not us, and let the reader make
> the choice. We all benefit from the high quality publications on the web.
> It's not about winning arguments. Whatever we find is a win-win for all of
> us.
>
> Bob Morein
> (310) 237-6511

Did you read at all the link I posted earlier? Probably not because
you wouldn't be trying to convince me now.
I already bought my machine 1 month ago and WITHOUT ECC. I have yet to
experience any problems.
If you have the 50-100$ more to spend on that go for it. I preferred
to pay a bit more for my motherboard and my power supply instead.
That's a choice those of us on a budget have to make. ECC is not
necessary, sorry to break your bubble.

"Julien BH" <julienbh@gmail . com > wrote in message
news:256cb1ec-e291-4f4d-8a74-866e53c75282@25g2000hsx.googlegroups . com ...
> On May 14, 11:19 am, "Soundhaspriority" <nowh...@nowhere . com > wrote:
>> "Julien BH" <julie...@gmail . com > wrote in message
>>
>> news:c4ee3a6b-969d-43c0-a9d6-edc346105d19@d1g2000hsg.googlegroups . com ...
>>
>>
>>
>> > On May 14, 10:14 am, "Soundhaspriority" <nowh...@nowhere . com > wrote:
>> >> "Richard Crowley" <rcrow...@xp7rt . net > wrote in message
>>
>> >>news:59-dnQM-7eI1crfVnZ2dnUVZ_uadnZ2d@posted.pcez...
>>
>> >> > "Soundhaspriority" wrote ...
>> >> >> "Richard Crowley" wrote ...
>> >> >>> "Soundhaspriority" wrote ...
>> >> >>>> Rich, you're not being very scientific about this.
>>
>> >> >>> Nor are you. You're quoting press releases and marketing
>> >> >>> stuff from companies who make products they are trying
>> >> >>> to sell.
>>
>> >> >>>> You're allowing yourself to be ruled by personal prejudice
>> >> >>>> and comfort factor.
>>
>> >> >>> Yes, I admit that I am prejudiced against people trying to sell me
>> >> >>> things that will take my $$ and give me no perceivable equitable
>> >> >>> benefit. It is called "prudence" or even "frugality" if you wish.
>>
>> >> >> In this case, you are misguided.
>>
>> >> > "The error rate in today's consumer-level memory is so low
>> >> > so that for most everyday applications, adding ECC is pure
>> >> > overkill. For standard DDR2 memory, the error rate is
>> >> > something like 100 soft errors over 1 billion device hours.
>> >> > If there are 16 memory devices or chips on a given module,
>> >> > that translates to one soft error every 30 years. Even if you
>> >> > only have two such DIMMs in a system, that's still less than
>> >> > one error for more than the lifetime of the system as a whole."
>>
>> >> > From an interview with Mike Sanor, compatibility and
>> >> > performance manager at Crucial Technology, a division
>> >> > of DRAM manufacturer Micron Technology.
>> >> > * searchwincomputing.techtarget . com /tip/0,289483,sid68_gci125184...
>> >> > This is the reference posted by "JulienBH" elsewhere in this
>> >> > discussion.
>>
>> >> >> They muddle, and they crash. The consumer experience isn't too
>> >> >> good.
>>
>> >> > OTOH, my consumer experience and that of everyone
>> >> > I know personally is rather good. Sorry about yours.
>>
>> >> I don't have a problem, because I use ECC. But your expectations are
>> >> too
>> >> low.
>>
>> >> >> Well, sure. And you've got your magic keychain.
>>
>> >> > Bob, "you're not being very scientific about this." :-)
>>
>> >> > "less than one error for more than the lifetime of the system"
>> >> > "for most everyday applications, adding ECC is pure overkill"
>>
>> >> Well, at least you've stopped referring to your keychain. This puts
>> >> the
>> >> argument on a more level footing. On my side, I have three references.
>> >> You
>> >> have a statement by Mike Sanor.
>>
>> >> > I'm only doing everyday applications like editing audio and
>> >> > video and graphics and text. The sky is not falling. SER
>> >> > doesn't make my list of top 100 things to worry about.
>>
>> >> I understand that, but anecdotal experience is not admissible. You
>> >> lack
>> >> the
>> >> credentials, other than your keychain. The fact that you are a
>> >> computer
>> >> administrator is simply not good enough to be admissible. But you
>> >> have
>> >> brought up an interesting statement. Actual published papers would be
>> >> even
>> >> more interesting.
>>
>> >> We can continue to argue, on this uncomfortably personal level, where
>> >> you
>> >> have accused me of gullibility, and I have responded in kind. Or we
>> >> can
>> >> leave the question open. I continue to recommend ECC to self-builders
>> >> like
>> >> Mike Rivers, because of the nominal cost, and the established increase
>> >> in
>> >> reliability. I respect your right to disagree. But I do request that
>> >> you
>> >> cease carrying on this discussion on a personal level. If you do so,
>> >> consider my references to your keychain, etc., erased. I really didn't
>> >> want
>> >> to go there, but felt provoked.
>>
>> >> Bob Morein
>> >> (310) 237-6511
>>
>> > Bob... Don't be so addicted to ECC that's all we're saying. It's not
>> > the MUST HAVE it was a few years ago...
>> > Richard, understand that ECC could be needed for a studio PC that
>> > stays always ON.
>>
>> Julien,
>> Referring to my other post in this thread where I describe the
>> consequences of a soft error, some of effects can be cumulative. When a
>> project file is saved to disk, any errors in the RAM image are saved as
>> well. In this way, the errors accumulate proportional to the number of
>> hours
>> spent on the project, not the number of hours the computer is on. Because
>> a
>> wav sample can be manually fixed, this is not as devastating as with CAD
>> and
>> video. On the other hand, corruption of a project control file has pretty
>> bad conseuqences.
>>
>> Someone who has to buy a packaged computer, as Mike initially did
>> with
>> Dell, has little choice. Now Mike is building a machine, and he can have
>> ECC
>> for $25 - $50 additional. In my opinion, it's the logical choice.
>>
>> The best way to carry on a usenet discussion is to present the
>> evidence,
>> which in this case comes from third parties, not us, and let the reader
>> make
>> the choice. We all benefit from the high quality publications on the web.
>> It's not about winning arguments. Whatever we find is a win-win for all
>> of
>> us.
>>
>> Bob Morein
>> (310) 237-6511
>
> Did you read at all the link I posted earlier? Probably not because
> you wouldn't be trying to convince me now.
> I already bought my machine 1 month ago and WITHOUT ECC. I have yet to
> experience any problems.
> If you have the 50-100$ more to spend on that go for it. I preferred
> to pay a bit more for my motherboard and my power supply instead.
> That's a choice those of us on a budget have to make. ECC is not
> necessary, sorry to break your bubble.

Julien,
I am not trying to convince you. When I post, I do so for the entire
viewing public. Congrats on your new computer.

Bob Morein
(310) 237-6511

Richard Crowley wrote:
> "Soundhaspriority" wrote ...
>> Here are some numbers for you. This paper was revised 2004, the most
>> up to date source I could find on the web.
>> w w w .tezzaron . com /about/papers/soft_errors_1_1_secure.pdf
>>
>> The Micron estimate in the table is 2-4 errors per year per 256 mb. In
>> a computer with 4 gigabytes of ram, this translates to 32-64 errors in
>> memory, per year, or approximately:
>>
>> ONE PER WEEK.
>>
>> Please consult your keychain before you reply.
>
> LOL! :-)
> Perhaps SER is what causes the Radium phenomenon.
>
> Sorry, I remain unconvinced. Me and 100s of millions
> of other users of "trashy consumer desktop" systems
> will just have to live on the edge waiting for that Sword
> of Damacles SER waiting to fall on us. I'm trembling
> in fear.
>
> Could it be that Micron, et.al. makes more profit on
> ECC memory? No, they couldn't possibly have such a
> crass motivation. They are just looking out for our good.
>
> Buy ECC if you wish. I'm sure Micron will appreciate it.

Certainly it's not a bad idea, if not too expensive. OTOH, using the
figures cited--one soft error a week--would seem to put the chance of
any significant problem at a vanishingly small percentage.

How many trillions of memory read/write events happen in the course of a
day, week, or month? I expect that memory read/write errors on magnetic
media are a much more common event, although I have no data to compare.

jak

"jakdedert" <jakdedert@bellsouth . net > wrote in message
news:lvBWj.4738$hv2.897@bignews5.bellsouth . net ...
> Richard Crowley wrote:
>> "Soundhaspriority" wrote ...
>>> Here are some numbers for you. This paper was revised 2004, the most up
>>> to date source I could find on the web.
>>> w w w .tezzaron . com /about/papers/soft_errors_1_1_secure.pdf
>>>
>>> The Micron estimate in the table is 2-4 errors per year per 256 mb. In a
>>> computer with 4 gigabytes of ram, this translates to 32-64 errors in
>>> memory, per year, or approximately:
>>>
>>> ONE PER WEEK.
>>>
>>> Please consult your keychain before you reply.
>>
>> LOL! :-)
>> Perhaps SER is what causes the Radium phenomenon.
>>
>> Sorry, I remain unconvinced. Me and 100s of millions
>> of other users of "trashy consumer desktop" systems
>> will just have to live on the edge waiting for that Sword
>> of Damacles SER waiting to fall on us. I'm trembling
>> in fear.
>>
>> Could it be that Micron, et.al. makes more profit on
>> ECC memory? No, they couldn't possibly have such a
>> crass motivation. They are just looking out for our good.
>>
>> Buy ECC if you wish. I'm sure Micron will appreciate it.
>
> Certainly it's not a bad idea, if not too expensive. OTOH, using the
> figures cited--one soft error a week--would seem to put the chance of any
> significant problem at a vanishingly small percentage.
>
The chances are quite significant.

> How many trillions of memory read/write events happen in the course of a
> day, week, or month?

The error rate of RAM is not related to the number of read/writes. It is
exactly proportional to power-on-hours.

I expect that memory read/write errors on magnetic
> media are a much more common event, although I have no data to compare.
>
> jak

Data on hard disc is ECC encoded, so there is no equivalent problem.

Bob Morein
(310) 237-6511

Julien BH <julienbh@gmail . com > said in response to my bewilderment

> There is a dedicated chip
> on the motherboard in the DS3R

cool ..
i'll look at this board closer.

Steve L. wrote:

> AMD likes to change socket platforms often.
> Socket 775 from intel is pretty versitle in it's choice of chips that will
> fit.
> A consideration if you want to upgrade your processor in the future.

I'm not sure I'll ever upgrade my processor, but I'm enough of a
traditionalist that I'd be inclined to stick with an Intel processor. (I
still have a Beta video recorder that still works, too)

--
If you e-mail me and it bounces, use your secret decoder ring and reach
me here:
double-m-eleven-double-zero at yahoo -- I'm really Mike Rivers
(mriv...@d-and-d . com )