Questions about DDR RAM

[Robert Redelmeier]

History would show that bad cheap drives out good.

.... expensive. Plenty of similar pseudo-examples like VHS
vs beta. Yet it is a canard. Quality is an attribute like
any other. Just like price. There is such a thing as
excessive quality, particularly if it comes at the expense
of some other desireable attribute like price.

I give you microchannel vrs ISA as an example.

Another canard. I used both in the day. Yes, ISA had fun
IRQ clashes, but those were easy enough to avoid. Not worse
than todays PCI-BM card shuffling. Microchannel certainly
was more elegant but had that rather tedious install diskette
process. In the end, both boxes were stable with good drivers.
IBM usually had better, but I ran Linux MC TR for years.

In fact the whole consumer PC market is an example.

When a new product is introduced, it is usually expensive
and aimed and very demanding customers. The quality almost
always is the maximum that can be achieved. As the product
gains acceptance and market size, both the price and the
quality should decrease because these new customers have
different values. Their values are indisputably theirs and
they have a right to pursue them.

With small margins, and no evidence that people walking in
walmart or best buy have any interest in paying a premium
for some nebulous reliability claim why should manufacturers
waste perfectly good bits.

Actually, the PC market is highly fragmented, with quite a
quality range. A desktop sells anywhere from $200 to $900+.
The upper end would surely like more to differentiate
themselves with.

If ECC was that big a reliability win, it would not be a
nebulous claim.

Servers are a different story.
Always.

Last time I talked to my buddy that tracks failures, it was
software first, then disks, then electronics. A little
research and a few calculations will tell you how often
there will be a memory error.

Electronics probably including electrolytic capacitors
which have to be at least half of all "electronic" failures.

How seriously you take it depends on how you feel about
errors and especially undetected errors.

Certainly. I would be most interested in ECC error log reports --
how many errors detected in how many GB over how many power-on
hours. Hard data like this makes a reasoned decision over ECC
possible. Otherwise, it's all anectdotal and worse.

-- Robert

 

[Frank McCoy]

In alt.comp.hardware.pc-homebuilt Robert Redelmeier

Plenty of similar pseudo-examples like VHS vs beta.

While Beta was a much better *recording media* than VHS ever will be,
the competition wasn't between the media but between the *recorders*;
and there Sony completely dropped the ball. People bought VHS, not
because VHS was better or cheaper; it wasn't on BOTH counts. Beta tapes
were better, produced far better pictures, were far less likely to jam
AND were slightly cheaper (because of the reduced size). People bought
VHS *recorders* because they offered far more features (like delayed and
programmed recording) long before Beta machines ever did. Since most
people don't give a shit about minor quality issues; tiny problems not
showing up until quite a while *after* the machine was bought, the minor
price-differential (in favor of Beta, BTW), they both had similar
prerecorded movies out, and the prices of the machines were almost
identical, what made VHS "win the war" was the fact that Sony thought it
owned the market; and saw no reason to upgrade or add features to their
machines. The competition did.

I still have one of the last Beta machines around.
The only "programming" you can do is set up a recording-time pressing
buttons on the front of the machine. You can't set up multiple records,
sets of records, or even change stations from one station to another
while recording except by manually going up to the machine and pressing
another station-button. All the VHS machines by then could do FAR more;
most having on-screen programming for at least a week of recording.

Pure idiocy (in my opinion anyway) on the part of Sony management.

 

[Frank McCoy]

In alt.comp.hardware.pc-homebuilt Robert Redelmeier

If ECC was that big a reliability win, it would not be a
nebulous claim.

Actually, it IS.
That's why almost every SERVER gets ECC memory.
What ECC doesn't have, is anybody SELLING the advantage.
For most common PC users (and retailers) nobody gives a damn; and those
few people who *do* know the advantage, usually don't speak to the
customers and TELL them.

How is anybody supposed to know the difference, if they'r not educated?
Besides, all most lusers see on the sticker is the hard-drive size, the
monitor-size, the CPU speed, and the memory-size. Since NONE of the
packaged products comes default-with ECC, how are they going to know
it's even an option?

Nobody mentions it to them. Nobody *suggests* it as an option. Nobody
sees it in any on-shelf PC. Only those who *know* about it would even
ask! How many are that? Less than 1%?

If it ain't being sold, *of course* it ain't going to sell!

Now most professionals, people who build server-boxes, probably wouldn't
be caught dead putting non-ECC memory into a system. But THEY know what
it is and is-for, what it does, and how it improves reliability.

However geeks like that don't talk to home-PC customers.
I'll bet most put ECC memory in their own systems at home though.
Just like they buy decent power-supplies and motherboards; not the crap
that Dell sells.

 

[daytripper]

Now most professionals, people who build server-boxes, probably wouldn't
be caught dead putting non-ECC memory into a system. But THEY know what
it is and is-for, what it does, and how it improves reliability.

Well, no. The *customer* has a check list, and ECC is on it. Otherwise the
"people who build server boxes" would put non-ECC memory in it and pocket the
cost differential.

However geeks like that don't talk to home-PC customers.
I'll bet most put ECC memory in their own systems at home though.

So, how many desktop chipsets actually even support ECC these days?

/daytripper

 

[Frank McCoy]

In alt.comp.hardware.pc-homebuilt daytripper

Well, no. The *customer* has a check list, and ECC is on it. Otherwise the
"people who build server boxes" would put non-ECC memory in it and pocket the
cost differential.


So, how many desktop chipsets actually even support ECC these days?

ALL of them do.
It's on the DIMM, not the motherboard.
Completely transparent. The mobo doesn't even know it's there.
The DIMM is just a bit bigger; and has extra chips on it.
Fits in the same slot.
You just buy PC2100 ECC memory instead of PC2100 non-ECC memory.
They meet the exact same specifications.

The only difference is the ECC memory doesn't have errors.

 

[daytripper]

In alt.comp.hardware.pc-homebuilt daytripper

ALL of them do.
It's on the DIMM, not the motherboard.
Completely transparent. The mobo doesn't even know it's there.
The DIMM is just a bit bigger; and has extra chips on it.
Fits in the same slot.
You just buy PC2100 ECC memory instead of PC2100 non-ECC memory.
They meet the exact same specifications.

The only difference is the ECC memory doesn't have errors.

Ummm....No. And might I add, you really stepped in the dog poo this time...

/daytripper (sic' im, Keith ;-)

 

[Nate Edel]

price-differential (in favor of Beta, BTW), they both had similar
prerecorded movies out, and the prices of the machines were almost
identical, what made VHS "win the war" was the fact that Sony thought it
owned the market; and saw no reason to upgrade or add features to their
machines. The competition did. [...]
Pure idiocy (in my opinion anyway) on the part of Sony management.

It's not clear that they HAD similar prerecorded movies: one "fact" (no idea
if it's correct, or an urban legend) about beta which is often repeated:
Early on, Sony didn't permit porn being distributed on Beta (although it
clearly either did later, or couldn't stop it - I've a porn tapes on Beta
that I picked up on clearance well into the 90s, and saw a few on the
shelves in Mexico in the late 80s where rental tapes in general - not
specific to porn - were easier to find on Beta.)

 

[Frank McCoy]

In alt.comp.hardware.pc-homebuilt daytripper

In alt.comp.hardware.pc-homebuilt daytripper

[blah blah ...]
Now most professionals, people who build server-boxes, probably wouldn't
be caught dead putting non-ECC memory into a system. But THEY know what
it is and is-for, what it does, and how it improves reliability.

Well, no. The *customer* has a check list, and ECC is on it. Otherwise the
"people who build server boxes" would put non-ECC memory in it and pocket the
cost differential.

However geeks like that don't talk to home-PC customers.
I'll bet most put ECC memory in their own systems at home though.

So, how many desktop chipsets actually even support ECC these days?

ALL of them do.
It's on the DIMM, not the motherboard.
Completely transparent. The mobo doesn't even know it's there.
The DIMM is just a bit bigger; and has extra chips on it.
Fits in the same slot.
You just buy PC2100 ECC memory instead of PC2100 non-ECC memory.
They meet the exact same specifications.

The only difference is the ECC memory doesn't have errors.

Ummm....No. And might I add, you really stepped in the dog poo this time...

Well ... I plan on buying some ECC memory shortly, and putting it in my
computer. PC2100 ECC memory, instead of PC100 non-ECC memory.
We'll see if it fits.

 

[Frank McCoy]

price-differential (in favor of Beta, BTW), they both had similar
prerecorded movies out, and the prices of the machines were almost
identical, what made VHS "win the war" was the fact that Sony thought it
owned the market; and saw no reason to upgrade or add features to their
machines. The competition did. [...]
Pure idiocy (in my opinion anyway) on the part of Sony management.

It's not clear that they HAD similar prerecorded movies: one "fact" (no idea
if it's correct, or an urban legend) about beta which is often repeated:
Early on, Sony didn't permit porn being distributed on Beta (although it
clearly either did later, or couldn't stop it - I've a porn tapes on Beta
that I picked up on clearance well into the 90s, and saw a few on the
shelves in Mexico in the late 80s where rental tapes in general - not
specific to porn - were easier to find on Beta.)

Well, in all the local video-stores around here at the time, if you
could rent a movie in one format, it was almost certain you could rent
it in the other ... and it wasn't just a blank-tape-copy either.

I think, with the court decisions against them, Sony gave up on trying
to limit which movies were copied and went with the flow of competition.
They made more money that way.

 

[krw]

Oh, pray tell, why? Do you believe you know the PC
business better than Intel, AMD, Dell, HP, ... who have
decided to manufacture chipsets and computers without ECC?

Do you believe ~50 US$/box is better spent on ECC than on
improved capacitors, mobo layers, cabling, cooling or shielding?

There are always many improvements possible. The key is to
choose the best ones. Not fixate like a kid in a candy store.

No, it's more like; there are always many cost reductions possible.
The key is to choose the ones least likely to bite one in the ass,
not fixate on quality like an adult running a business.

 

[krw]

No, unless used for yield improvement, std 64/72 ECC must cost
at least 12.5% more for components. Due to lower market
volume, it actually costs 30-50% more

Kinda a tail-chasing argument there. It costs 30-50% more because...
It certainly wouldn't cost the manufacturers 30-50% more.

Why do you doubt their design choice to omit ECC?

It's not a design choice. It's a marketing choice.

Wrong. From the 8088 thru 486, almost all PCs -- IBM and
clones had parity memory. Macs did not. Only with the
Pentium SIMMs did parity morphed to ECC and begin to drop.

ECC came about because the word size got big enough that ECC cost
nothing over simple parity. Both were dropped because memory was a
significant fraction of the cost of the system.

Reference please! BSoD can have many causes.
I suspect software and software patches mostly.
I keep Linux machines up for ~1 yr w/o ECC.

I agree with you here. ECC isn't going to fix M$'s crap.

No, because spikes often hit the busses in parallel.

So does Windows.

Again, you presume you know better than Intel, AMD,
Dell, HP, etc.

Than their marketeers? About technical issues? Well, yes...

Life critical computing and control machinery does
not run on PCs or with MS software.

Here you are wrong, unless you consider medical equipment
(defibrillators, life-signs monitors, and such) "non-life critical".
I couldn't count high enough to track all the PC-based hardware and
Windows crap when I was in the hospital, earlier this year. No,
being there, seeing that, didn't give me the warm fuzzies.

Except I've run several just fine without anything
resembling BSoDs with uptimes around a year.

Anecdotes <> data

Reference please!


Perhaps this is true for most, but I've run intense software
memory testers like memtest-86+ for days and weeks yet never
seen an inexplicable error.

"There are no known bugs, to which I am aware", or perhaps, "nothing
can go wrong, go wrong, go wrong, go wr..."

 

[Robert Redelmeier]

Kinda a tail-chasing argument there. It costs 30-50% more because...
It certainly wouldn't cost the manufacturers 30-50% more.

Yes, the argument is somewhat circular. But matches data.
No, in full quantity, it'll cost 12.5% of DRAM.

It's not a design choice. It's a marketing choice.

Semantics. They choose the design they will market.
I cannot distinguish between a design choice and a
marketing choice unless marketing is pushing vaporware.

ECC came about because the word size got big enough that ECC
cost nothing over simple parity. Both were dropped because
memory was a significant fraction of the cost of the system.

.... and was not perceived as offerring commensurate benefits.
A hard disk is a bigger fraction of system cost. A floppy perhaps
somewhat smaller. Both were retained for perceived benefits.

I agree with you here. ECC isn't going to fix M$'s crap.

I doubt anything will

Than their marketeers? About technical issues? Well, yes...

But ECC isn't a technical issue. Including it or leaving it off does
not fundamentally change the system. It is an option to improve
reliability. The real question is: How much? If it is a lot, it
will be included. If it is not, it will be omitted for cost savings.

Again I ask: What BER data do you have to justify ECC?

Here you are wrong, unless you consider medical equipment
(defibrillators, life-signs monitors, and such) "non-life critical".
I couldn't count high enough to track all the PC-based hardware and
Windows crap when I was in the hospital, earlier this year. No,
being there, seeing that, didn't give me the warm fuzzies.

I deal with industrial controllers. I haven't seen these
in hospital, but they'd scare me too!


-- Robert

 

[Robert Redelmeier]

If you can identify the motherboard used on your system, perhaps we can save
you the trouble of returning those ECC dimms before you buy them.

If the chipset - or the motherboard implementation thereof - doesn't
explicitly support ECC dimms, you're gonna be SOL...

Not to mention that few mobos support both 168 pin PC100 and
184 pin DDR PC2100.

-- Robert

 

[Noozer]

Yes, the argument is somewhat circular. But matches data.
No, in full quantity, it'll cost 12.5% of DRAM.

....and add a 12.5% increase in the chance that the memory won't overclock as
well, since you're adding an extra chip to the timing scheme.

 

[daytripper]

In alt.comp.hardware.pc-homebuilt daytripper

In alt.comp.hardware.pc-homebuilt daytripper

[blah blah ...]
Now most professionals, people who build server-boxes, probably wouldn't
be caught dead putting non-ECC memory into a system. But THEY know what
it is and is-for, what it does, and how it improves reliability.

Well, no. The *customer* has a check list, and ECC is on it. Otherwise the
"people who build server boxes" would put non-ECC memory in it and pocket the
cost differential.

However geeks like that don't talk to home-PC customers.
I'll bet most put ECC memory in their own systems at home though.

So, how many desktop chipsets actually even support ECC these days?

ALL of them do.
It's on the DIMM, not the motherboard.
Completely transparent. The mobo doesn't even know it's there.
The DIMM is just a bit bigger; and has extra chips on it.
Fits in the same slot.
You just buy PC2100 ECC memory instead of PC2100 non-ECC memory.
They meet the exact same specifications.

The only difference is the ECC memory doesn't have errors.

Ummm....No. And might I add, you really stepped in the dog poo this time...

Well ... I plan on buying some ECC memory shortly, and putting it in my
computer. PC2100 ECC memory, instead of PC100 non-ECC memory.
We'll see if it fits.

If you can identify the motherboard used on your system, perhaps we can save
you the trouble of returning those ECC dimms before you buy them.

If the chipset - or the motherboard implementation thereof - doesn't
explicitly support ECC dimms, you're gonna be SOL...

/daytripper

 

[Frank McCoy]

In alt.comp.hardware.pc-homebuilt Robert Redelmeier

Not to mention that few mobos support both 168 pin PC100 and
184 pin DDR PC2100.

Um ... Typo.
That was *supposed* to be PC2100 in both cases.

 

[Frank McCoy]

...and add a 12.5% increase in the chance that the memory won't overclock as
well, since you're adding an extra chip to the timing scheme.

Shouldn't be overclocking anyway.
Not for a reliable system; which is the point of buying ECC memory.

 

[Robert Redelmeier]

...and add a 12.5% increase in the chance that the memory won't
overclock as well, since you're adding an extra chip to the
timing scheme.

Not really that bad -- the extra chip is in parallel with
the others. And the ECC circuitry is pipelined with the
rest of the memory controller. When it detects an error, it
raises and SMI (System Mgmt Interrupt) and the BIOS routines
step in to fixup the error and log it.

-- Robert

 

[daytripper]

Not to mention that few mobos support both 168 pin PC100 and
184 pin DDR PC2100.

-- Robert

Now, now....Don't eff up the education by picking on his typo...

/daytripper

 

[~misfit~]

Somewhere on the interweb "Frank McCoy" typed:

In alt.comp.hardware.pc-homebuilt Robert Redelmeier

It shouldn't add more than 10% to the price of memory; which would be
about 2% or less of the price of the computer itself.

The problem isn't Intel or anybody else with the possible exception of
IBM; but there only slightly.

The problem is custom and history.
They didn't do it in the past, for fairly good and decent reasons.
They don't do it *now* because they didn't do it in the past.
That is NOT a good reason.

The problem is:
WITH ECC built in, probably over half the cases of "Blue Screen of
Death" or computer crashes and foulups *could* be things of the past!

Even in cases where things like poor capacitors cause spikes, having
ECC memory in the machine would obviate a large portion of those
problems.

The original reasons of the extra logic and extra expense just ARE NOT
that relevant any more. They shouldn't even SELL non-ECC memory, for
the relatively tiny price-differential versus the HUGE difference in
reliability. It's like selling retread tires as new ones for almost
the same price. Sure they're CHEAPER ... marginally.

The worst part is, people could actually be KILLED by such mistakes
made by a computer that might have been corrected with ECC ... Yet
nobody will trace it back to that; just: "Sorry, the computer
crashed!"

That's unlike a bad tire, which eventually *will* get noticed after
enough people die.

Worse-yet, people aren't even being educated as to what the difference
is. Essentially they're told and even believe that non-ECC memory is
just as good, only cheaper.

"I've ran my computer for years without ECC; and it ran just FINE!"
Only that ignores the freezups, crashes, blue-screens, and other crap
that got attributed to software instead of memory failures. ;-{

These days people seem to *expect* such failures, when 99.99% of the
ones caused by bad memory (probably well over half) could be fixed.

Most people ass-u-me that their memory is good; never EVER running a
memory-test other than the completely useless crap on boot. Hell,
most people, if a computer is crapping out, just replace the whole
thing.

In fact, many computer-repair places *encourage* their customers to do
just that ... It makes more money for the company; while running a
good memory-test takes up very valuable technician time and space in
the repair-shop.

For a mere pittance in extra cost these days, especially if ECC memory
was the *standard* instead of the rarely-used, the "extra cost" would
be a huge monetary *gain* instead of a loss. Most especially so in
customer satisfaction.

Still, they don't count "customer satisfaction" as worth a dime these
days, not in comparison to saving ten or twenty cents on a part, do
they?

Ahh, looks like you're lost. Utopia is two doors down, to the right.