RDRam

[Diane]

I have a PC with an Asus P4-TE mobo. Right now I have 2 Rimms, each 256. My
manual shows that the board can handle 2 GB of memory.

The manual does not say that the memory must be added in pairs but I seem to
have read that and it sticks in the back of my mind. The manual also does
not tell me if the memory I have is ECC or not. Would some kind soul tell me
if I can add two 512's to the two 256's I have or a 512 & 256 or must I add
2 more two 256's.

Computer builder has been slow in responding so I thought someone here could
help

Thank you.

Diane

 

[RusH]

Would some kind soul tell me if I can add two 512's to the two
256's I have or a 512 & 256 or must I add 2 more two 256's.

just dump this board with rimms and buy something normal


Pozdrawiam.

 

[George Macdonald]

I have a PC with an Asus P4-TE mobo. Right now I have 2 Rimms, each 256. My
manual shows that the board can handle 2 GB of memory.

The manual does not say that the memory must be added in pairs but I seem to
have read that and it sticks in the back of my mind. The manual also does
not tell me if the memory I have is ECC or not. Would some kind soul tell me
if I can add two 512's to the two 256's I have or a 512 & 256 or must I add
2 more two 256's.

I don't recall the rules on DRDRAM but the chipset is certainly dual
channel and according to NewEgg's product listings "all 16-bit RDRAM must
be used in pairs". Also check the manual here:
http://usa.asus.com/mb/socket478/p4t-e/overview.htm

Have a look at the modules you have to see if they're labeled ECC.

Computer builder has been slow in responding so I thought someone here could
help

He's probably slow because the stuff is expensive (about 2-3 times price of
similar sized DDR-DRAM) and scarce to get what you might need: the only
512MBs NewEgg has are non-ECC; for 265MB they seem to have both. For a bit
more than the price of two 256MB RIMMs and a lot less than two 512MB, you
could get a decent i865 mbrd
(http://www.newegg.com/app/ViewProductDesc.asp?description=13-131-484&depa=0
or
http://www.newegg.com/app/ViewProductDesc.asp?description=13-130-450&depa=0)
and two sticks of 512MB DDR400-DRAM:
http://www.newegg.com/app/ViewProductDesc.asp?description=20-146-541&depa=0

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Bob Knowlden]

I don't own a system that uses Rambus memory, but...

P4 Rambus machines require that the memory be installed in pairs. (There was
a form of Rambus memory that was essentially a pair in a single package, but
you have the regular 16 bit version.) I believe that the pairs can be of
different capacity, so it'd be OK to add a pair of 512 MB RIMMs. I normally
seek memory guidance at www.crucial.com, but they don't sell Rambus memory.

As the P4T-E can use ECC or non-ECC memory, I suggest that you use a utility
to tell which sort is installed. (CPU-Z is a small one:
http://www.cpuid.com/cpuz.php .) You may be able to mix ECC and non-ECC, if
you turn off ECC in the system's BIOS.

Out of curiosity, I checked prices at www.newegg.com. It appears that 512 MB
PC800 RIMMs are over $200 (US) each. You could pick up an Asus P4P800SE
mainboard for about $100, and a pair of 512 MB PC3200 DIMMs (1 GB total) for
less than $150. A 2.4 GHz "Northwood" CPU (800 MHz FSB) costs less than
$150, in the full retail package. For about the cost of 1 GB of RIMMs, you
could upgrade to a faster DDR system, with 1 GB or RAM. (This is a little
disingenuous, as I neglect the value of the labor to do the upgrade.
Swapping mainboards is a lot more work than inserting new memory.) I mention
this to show that another poster's response wasn't quite as impertinent as
it might have seemed.

(Socket 478 and DDR are on the way out. Intel is now pushing Socket T, also
known as LGA 775, and DDR2 memory. I believe that the Socket T CPUs aren't
much more expensive than their Socket 478 equivalents, but DDR2 is still
pricey.)

I hope that you are happy with your choice.

Address scrambled. Replace nkbob with bobkn.

 

[Diane]

I don't own a system that uses Rambus memory, but...

(Socket 478 and DDR are on the way out. Intel is now pushing Socket T,
also known as LGA 775, and DDR2 memory. I believe that the Socket T CPUs
aren't much more expensive than their Socket 478 equivalents, but DDR2 is
still pricey.)

I hope that you are happy with your choice.

Address scrambled. Replace nkbob with bobkn.

Thanks for your comprehensive answer. I did notice as I was searching around
for memory, that the chips were costly and hard to find. My PC was custom
made and I have been very pleased with its performance. 3 yr warranty runs
out in March and your suggestion about upgrading makes economical sense.

I didn't have a clue about Rambus memory when I bought this PC, but one
learns something new everyday...<<sigh>>. Thanks again.

Diane

 

[keith]

Thanks for your comprehensive answer. I did notice as I was searching around
for memory, that the chips were costly and hard to find. My PC was custom
made and I have been very pleased with its performance. 3 yr warranty runs
out in March and your suggestion about upgrading makes economical sense.

If it was custom built, you *should* (doesn't always work) be able to
simply replace the processor/motherboard, and memory and keep the rest of
your installation (hardware and software).

I didn't have a clue about Rambus memory when I bought this PC, but one
learns something new everyday...<<sigh>>. Thanks again.

You should have dropped in here three years ago. You would have learned
more about the law firm of RamBus, ShamBus, and ScamBus than you ever
wanted to know. ;-)

 

[nobody]

just dump this board with rimms and buy something normal


Pozdrawiam.

And while you are at it, dump the P4 altogether. A64 provides much
better value AND performance. Thar old P4 board definitely is not
worth the memory upgrade since it still will be an old P4 system.

 

[jack]

: On Sat, 11 Dec 2004 05:55:20 -0500, Diane wrote:
:
::
:: ::: I don't own a system that uses Rambus memory, but...
:::
::
:: <<SNIP>>
::
::: (Socket 478 and DDR are on the way out. Intel is now pushing
::: Socket T, also known as LGA 775, and DDR2 memory. I believe that
::: the Socket T CPUs aren't much more expensive than their Socket
::: 478 equivalents, but DDR2 is still pricey.)
:::
::: I hope that you are happy with your choice.
:::
::: Address scrambled. Replace nkbob with bobkn.
:::
::
:: Thanks for your comprehensive answer. I did notice as I was
:: searching around for memory, that the chips were costly and hard
:: to find. My PC was custom made and I have been very pleased with
:: its performance. 3 yr warranty runs out in March and your
:: suggestion about upgrading makes economical sense.
:
: If it was custom built, you *should* (doesn't always work) be able
: to simply replace the processor/motherboard, and memory and keep
: the rest of your installation (hardware and software).
:
:: I didn't have a clue about Rambus memory when I bought this PC,
:: but one learns something new everyday...<<sigh>>. Thanks again.
:
: You should have dropped in here three years ago. You would have
: learned more about the law firm of RamBus, ShamBus, and ScamBus
: than you ever wanted to know. ;-)

Really, must we open then door once again to the likes of "Tim Sullivan" /
"John Corse" etc.? Humph!

j.

 

[Bob Knowlden]

Rambus wasn't necessarily a mistake at the time. However, it turned into a
dead end technology.

Take what you hear about Rambus memory with a little skepticism. The company
is widely hated, as they are an intellectual property house (they make
little or no hardware). They launched several lawsuits, which appeared to
many to be extortionate. Also, there was a PIII chipset for RIMMs that gave
worse performance than PC133 SDRAM, although Rambus memory cost a lot more
than SDRAM. The PIII version was single channel, so the dual-channel P4 was
much better. The damage to the reputation of Rambus memory was already done,
though.

Count yourself lucky. Some of my co-workers have Dell Optiplex GX240
machines. These are P4s that use PC133 SDRAM. With a 1.7 GHz P4, one of
those is slower (at least for floating point number crunching) than the
Optiplex GX150 on my desk, which uses a 1 GHz PIII (and PC133 SDRAM ). I
don't know whether to blame Dell or Intel, as the machines use a genuine
Intel chipset. About the only good thing I can say about that combination is
that it wasn't on the market long.

Bob Kn.

 

[keith]

: On Sat, 11 Dec 2004 05:55:20 -0500, Diane wrote:
:
::
:: ::: I don't own a system that uses Rambus memory, but...
:::
::
:: <<SNIP>>
::
::: (Socket 478 and DDR are on the way out. Intel is now pushing
::: Socket T, also known as LGA 775, and DDR2 memory. I believe that
::: the Socket T CPUs aren't much more expensive than their Socket
::: 478 equivalents, but DDR2 is still pricey.)
:::
::: I hope that you are happy with your choice.
:::
::: Address scrambled. Replace nkbob with bobkn.
:::
::
:: Thanks for your comprehensive answer. I did notice as I was
:: searching around for memory, that the chips were costly and hard
:: to find. My PC was custom made and I have been very pleased with
:: its performance. 3 yr warranty runs out in March and your
:: suggestion about upgrading makes economical sense.
:
: If it was custom built, you *should* (doesn't always work) be able
: to simply replace the processor/motherboard, and memory and keep
: the rest of your installation (hardware and software).
:
:: I didn't have a clue about Rambus memory when I bought this PC,
:: but one learns something new everyday...<<sigh>>. Thanks again.
:
: You should have dropped in here three years ago. You would have
: learned more about the law firm of RamBus, ShamBus, and ScamBus
: than you ever wanted to know. ;-)

Really, must we open then door once again to the likes of "Tim Sullivan" /
"John Corse" etc.? Humph!

Sure. L'il Timmy has to make a living somehow! How else than pump-n-dump? ;-)

 

[keith]

Rambus wasn't necessarily a mistake at the time. However, it turned into a
dead end technology.

Oh, good grief! The techies here even said it was dead meat before it was
shipped. It was annother "good idea" that wasn't.

Take what you hear about Rambus memory with a little skepticism. The
company is widely hated, as they are an intellectual property house
(they make little or no hardware).

....not to mention that it did nothing for the end-user. It has been
forever a solution looking for a problem. Its "benefits" (few as they
are) have never come close to balancing its deficits.

They launched several lawsuits, which
appeared to many to be extortionate. Also, there was a PIII chipset for
RIMMs that gave worse performance than PC133 SDRAM, although Rambus
memory cost a lot more than SDRAM. The PIII version was single channel,
so the dual-channel P4 was much better. The damage to the reputation of
Rambus memory was already done, though.

Not to mention that DDR SDRAM was right around the corner. Dual=channel
DDR shortly after. The issue was never bandwidth, which is easily
handled. DRDRAM never delivered on *latency*. The fact is that it was
*another* technology backed by Intel to try to segment the market. Once
again, they fell flat on their face.

Count yourself lucky. Some of my co-workers have Dell Optiplex GX240
machines. These are P4s that use PC133 SDRAM. With a 1.7 GHz P4, one of
those is slower (at least for floating point number crunching) than the
Optiplex GX150 on my desk, which uses a 1 GHz PIII (and PC133 SDRAM ). I
don't know whether to blame Dell or Intel, as the machines use a genuine
Intel chipset. About the only good thing I can say about that
combination is that it wasn't on the market long.

At least one can buy PC133 DRAM at a reasonable price. This was never
even the intention behind ScamBus memory.

 

[nobody]

....snip...

Count yourself lucky. Some of my co-workers have Dell Optiplex GX240
machines. These are P4s that use PC133 SDRAM. With a 1.7 GHz P4, one of
those is slower (at least for floating point number crunching) than the
Optiplex GX150 on my desk, which uses a 1 GHz PIII (and PC133 SDRAM ). I
don't know whether to blame Dell or Intel, as the machines use a genuine
Intel chipset. About the only good thing I can say about that combination is
that it wasn't on the market long.

Bob Kn. ....snip...

Count yourself - and your co-workers - lucky. I recently finished a
project for one company where us developers (both employees and
consultants) had to put up with some 4 year old P3-866. It was quite
an achievement when we persuaded the management that these clunkers
needed at least memory upgrade to HUGE (by company standard) 512 MB
(and this for .NET development!) from then-current 128. And nothing
better than old fading 17" monitors. Of course secretary's desk
sported brand new P4 with HT (presumably at least 2.4) and 19" display
- probably to increase her speed of typing in MS Word.

 

[Ishkabibble]

Count yourself - and your co-workers - lucky. I recently finished a
project for one company where us developers (both employees and
consultants) had to put up with some 4 year old P3-866. It was quite
an achievement when we persuaded the management that these clunkers
needed at least memory upgrade to HUGE (by company standard) 512 MB
(and this for .NET development!) from then-current 128. And nothing
better than old fading 17" monitors. Of course secretary's desk
sported brand new P4 with HT (presumably at least 2.4) and 19" display
- probably to increase her speed of typing in MS Word.

Sigh.

You've just described my work PC. It's got 384Mb only because I chucked a
256Mb DIMM
in there myself. Everyone knows that Win2K works just peachy with 128Mb of
RAM. You
forgot to mention those ultra-zippy 5400RPM, 6Gb drives in those old
boxes...

Don't pick on the secretary too much - at least s/he uses the machine to do
something. Our
PC techs sitting there with schpankity new P4's with all the goodies just to
run MS Outlook
are what get to me ...

 

[Mike Tomlinson]

Some of my co-workers have Dell Optiplex GX240
machines. These are P4s that use PC133 SDRAM. With a 1.7 GHz P4, one of
those is slower (at least for floating point number crunching) than the
Optiplex GX150 on my desk, which uses a 1 GHz PIII (and PC133 SDRAM ). I
don't know whether to blame Dell or Intel, as the machines use a genuine
Intel chipset.

These will presumably be the ones with the buggy Intel MTH (Memory
Translator Hub) which allowed the i820 chipset to use SDRAM at the
expense of performance.

 

[Tony Hill]

These will presumably be the ones with the buggy Intel MTH (Memory
Translator Hub) which allowed the i820 chipset to use SDRAM at the
expense of performance.

The Optiplex GX150 uses the i815E chipset and the Optiplex GX240 uses
the i845 chipset. Both straight SDRAM controllers, no RDRAM or MTH's
involved. There were actually very few PCs ever built using that MTH
chip; I don't know if any of Dell's Optiplex line ever used them.
Compaq definitely never used such a setup in their Deskpro or Evo line
and I don't think HP did in their Vectra or Kayak lines either.

Mostly those buggy chips only found their way into the much more
price-conscious (and less reliable) consumer grade systems. Even
there they were pretty rare since most people opted for VIA chipsets
for such systems instead (at that time VIA had a larger share of the
chipset market than Intel did).


The real problem with that Dell GX240 system was that the P4 and SDRAM
just didn't mix well and the i845 wasn't all that great of a chipset.
It was reliable enough, which is why it was fairly widely used, but
until the second or third revision the performance was pretty weak.
Not only did this chip just not have enough bandwidth to keep the P4
well fed, the latency on the memory wasn't all that hot either. When
it came out, DDR was already available for the P4 from VIA and others,
making the first i845 chips rather uninteresting. On the upside, it
didn't last that long before being replaced by the DDR-capable i845
which improved performance significantly. Variations of that updated
chipset are still being sold today in bargain-basement systems.

 

[George Macdonald]

Rambus wasn't necessarily a mistake at the time. However, it turned into a
dead end technology.

Take what you hear about Rambus memory with a little skepticism. The company
is widely hated, as they are an intellectual property house (they make
little or no hardware).

They make no hardware at all... other than a few prototype -- proof of
concept if you like -- chips which are obviously foundried. The hate is
more due to the high ratio of lawyers they employ and contract... plus the
hint that one of the principals appeared to jump ship before a university
could get its hands on the patents. One *does* wonder though: who was the
sponsor of the original research??

They launched several lawsuits, which appeared to
many to be extortionate.

It *was* extortionate - the courts have reduced the infingement count to be
tried at the next trial from 54 to 4(I think those numbers are right?).
All this angst over 4 dubious patents is absurd.

Make no mistake here - the entire industry narrowly escaped being held up
for ransom by a nest of shyster lawyers. When they were at the muscle
flexing stage, they had declared that they would go after every DDR
interface on the planet: FSBs, IDE-UDMA, SATA, AGP etc. etc.... all were to
be subjected to licensing. You couldn't use a countdown register or a DLL
without submitting to their inspection.

In case you hadn't noticed, they currently have their attention on a
hi-jack of the PCI-Express connection interface. IOW it ain't over yet and
the industry may yet find itself subjected to legal wrangling... if
Infineon loses the next round and then Micron capitulates, prepare to bend
over at the waist.

Also, there was a PIII chipset for RIMMs that gave
worse performance than PC133 SDRAM, although Rambus memory cost a lot more
than SDRAM. The PIII version was single channel, so the dual-channel P4 was
much better. The damage to the reputation of Rambus memory was already done,
though.

The i820 was a dog, the i840 showed a slight advantage on AGP "long
transfers", which hardly any software ever used and which was withdrawn
from the specs for AGP 3.0 anyway.

FWIW, and counter to the apportioning of blame to MTH, here's one of the
more interesting articles related to DRDRAM, Rambus and Intel:
http://www.reed-electronics.com/electronicnews/index.asp?layout=article&articleId=CA50587.
Note the comments on (negative) timing margins. A disgruntled
ex-employee?... perhaps but how many err, gruntled ex-employees are there
and recall the reduction of the RIMM count and the aborted rollout of
(non-MTH) DRDRAM mbrds in Sep '99.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[chrisv]

: You should have dropped in here three years ago. You would have
: learned more about the law firm of RamBus, ShamBus, and ScamBus
: than you ever wanted to know. ;-)

Really, must we open then door once again to the likes of "Tim Sullivan" /
"John Corse" etc.? Humph!

Don't forget RayO and Chris Pitzel, who also disappeared right about
the same time DRDRAM did. 8)

 

[cquirke (MVP Win9x)]

On Sun, 12 Dec 2004 17:51:45 -0500, Tony Hill

Mostly those buggy chips only found their way into the much more
price-conscious (and less reliable) consumer grade systems.

Not sure about that - here, the i820 never did fall into affordability
before the scandal blew up. I stayed with the venerable i440BX in
those days, even if it required a slotket card; then when 815 and 815e
came out, I jumped over to that as native S370.

I haven't seen or sold a single i820, tho I built one i850 for a big
multimedia dev system.

The real problem with that Dell GX240 system was that the P4 and SDRAM
just didn't mix well and the i845 wasn't all that great of a chipset.
It was reliable enough, which is why it was fairly widely used, but
until the second or third revision the performance was pretty weak.

It wasn't long after i845 that the i845G came out, which was
DDR-based, and I switched to that. I think I built about 3 pre-DDR
SDRAM-based P4 systems; I don't remember them being slow.

didn't last that long before being replaced by the DDR-capable i845
which improved performance significantly. Variations of that updated
chipset are still being sold today in bargain-basement systems.

Yep; first the i845D, then the G with SVGA + AGP.

------------ ----- ---- --- -- - - - -

The most accurate diagnostic instrument
in medicine is the Retrospectoscope

 

[Tony Hill]

On Sun, 12 Dec 2004 17:51:45 -0500, Tony Hill


Not sure about that - here, the i820 never did fall into affordability
before the scandal blew up.

Perhaps I should have said "price-conscious and buzzword happy"
consume grade systems. Dell sold a few of those as their "high-end"
systems because they had the latest and greatest chipset from Intel.
for a time it was also the ONLY Intel chipset that support PC133
memory, and since Dell was locked in to their standard
nothing-but-Intel deal, they didn't have the option of using the much
faster VIA chipsets like everyone else did.

I stayed with the venerable i440BX in
those days, even if it required a slotket card; then when 815 and 815e
came out, I jumped over to that as native S370.

That's what a lot of people did, though many others started using VIA
chipsets instead as they offered better performance and a MUCH better
price than anything Intel had.

I haven't seen or sold a single i820, tho I built one i850 for a big
multimedia dev system.

I remember one friend had an i820 system with the MTH chips and SDRAM.
Piece of junk. I never ran into any troubles with the memory, but the
first few revisions of the i8xx drivers were terrible and caused all
kinds of troubles. It wasn't until around the time of the i815
chipset that Intel finally got their drivers in order, and the result
was a rather respectable chipset.

It wasn't long after i845 that the i845G came out, which was
DDR-based, and I switched to that. I think I built about 3 pre-DDR
SDRAM-based P4 systems; I don't remember them being slow.

"Slow" is, of course, a relative term. 10% slower than other P4
systems doesn't seem like much if you're comparing it to a PII 300MHz
box. However, when compared to the top-end PIII systems being sold at
that time, running at 1.0GHz with the i815E chipset, the performance
was nothing to get excited about. The 1.0GHz PIII systems were often
as fast or faster than the 1.5 and 1.6GHz P4s, let alone the slower
1.3 and 1.4GHz boxes.

Yep; first the i845D, then the G with SVGA + AGP.

And now the i845GV, which is basically the same as the G but without
AGP. This is the one that you will find on many low-end Celeron
systems, though it's being replaced now with either the i865GL or the
new i910GL. You can still find it in, for example, HPaq's Pavilion
a700y or their Presario SR1010V.

 

[keith]

Don't forget RayO and Chris Pitzel, who also disappeared right about
the same time DRDRAM did. 8)

But to be fair to (e-mail address removed) (hmm), Timmy keeps comming back to
pump every time RMBS gets a little good news.