Intel may have known DDR2 was a short-term technology

Toggle sidebar Toggle sidebar
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

"Yousuf Khan" <yjkhan@gmail.com> wrote in message
news:1114098062.272509.298500@o13g2000cwo.googlegroups.com...
> It may have known DDR2 was only a short-term technology all along, but
> decided to foist it on the world anyways.
>
> CoolTechZone::Industry Insider Report: Intel Knew DDR2 was Poor Design
> http://www.cooltechzone.com/index.php?option=content&task=view&id=1197
>
> It must've been trying to get AMD to adopt it and then watch as it was
> forced to adopt DDR3 in short order.

Another silly anti-Intel post. This one the silliest of all. DDR2 isn't
going anywhere anytime soon. Prices are only 5%-10% greater than that of
DDR uno right now. It's already scaling to 800 MHz and could scale to 1
GHz. I seriously doubt it's going anywhere anytime soon. But hey, I'm sure
gundeep is one deep dud....dooo
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

Judd wrote:

> Yousuf Khan wrote:
>
>> It may have known DDR2 was only a short-term technology all along,
>> but decided to foist it on the world anyways.
>
> DDR2 isn't going anywhere anytime soon. Prices are only 5%-10%
> greater than that of DDR uno right now.

Hold on just a minute! Is your calculator equipped with a Pentium CPU?
Because it looks like you've been bitten by the FDIV bug all over again.

Unbuffered, unregistered, SDRAM DIMMs
512-MB DDR-400 = US$35
512-MB DDR2-533 = US$60

That's +71% in my book, not +10%...
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

Judd wrote:
> "Yousuf Khan" <yjkhan@gmail.com> wrote in message
> news:1114098062.272509.298500@o13g2000cwo.googlegroups.com...
> > It may have known DDR2 was only a short-term technology all along,
but
> > decided to foist it on the world anyways.
>
> Another silly anti-Intel post. This one the silliest of all. DDR2
isn't
> going anywhere anytime soon. Prices are only 5%-10% greater than
that of
> DDR uno right now. It's already scaling to 800 MHz and could scale
to 1
> GHz. I seriously doubt it's going anywhere anytime soon. But hey,
I'm sure
> gundeep is one deep dud....dooo

So you're saying that you don't mind paying extra for a technology that
may not even get cheaper because it may not have much production life
left in it; and it may get phased out long before you're ready to
switch your processor out, not allowing you to salvage any of your
investment in it? If that's the case, then obviously you're a man of
great wealth. For the rest of us, who are trying to decide on whether
to purchase a new processor/mobo or make the existing one last a little
longer until the technology stabilizes, well this would be useful to
know. Even people considering buying AMD systems need to know this,
because if Intel itself doesn't believe in DDR2, then there's not much
point in waiting for an AMD system that supports DDR2 as there likely
won't be one.

Also the article lists that both manufacturers are looking at DDR3 and
Rambus XDR as possible future technologies. Depending on how successful
DDR2 is, those future technologies could be pretty far off, or pretty
near.

Yousuf Khan
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

David Wang wrote:
> Was the 90nm process technology only a short term technology, and
should
> we have gone from 130nm straight to 65nm? Better yet, should we not
> skip 65nm altogether and go straight to 45nm?
>
> If DDR2 is only a short term technology, perhaps we should have
skipped
> from SDRAM to DDR4. DDR3 may not last very long either.

Great rhetoric, but you know very well that there is a difference
between transparent technology like process technology, and technology
which you have to buy new supporting hardware for like DRAM standards.

A processor at 90nm works much like a processor at 130nm, and it may
even fit into the same socket as the previous generation processor --
thus making it transparent. If they had made DDR2 compatible with DDR,
where you could simply exchange the old RAM for the new RAM and plug it
into the same slots, then no problem -- that's transparent technology,
and no one would care if was really DDR1, DDR2, DDR3, or whatever
underneath.

> Going from DDR to DDR2 is a natural progression. "Not having much
> production life left in it" isn't a conspiracy by Intel to drive
> the screws to AMD. It's just that DRAM datarates are currently
> doubling every 3 years or so, and DDR2's range of 400 to 800 is
> only good for 3 years. The DRAM manufacturers can keep DDR2 around
> for a while and drive it up past 1 Gbps, but the timing pressure
> on the DRAM core becomes greater.

It's not the progression from DDR1 to DDR2 that's being questioned
here. It's the possible progression from DDR1 straight to DDR3 that
might be spoiling the party here.

> AMD will support DDR2, then transition to DDR3. Otherwise, sometime
> down the road, Intel will be shipping products with DDR2-667 while
AMD
> will be shipping DDR-400, and that scenario will last a year while
> everyone sits around waiting for DDR3.

Which is exactly the point. Are they going to sit around waiting for
DDR3?

Yousuf Khan
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

"Grumble" <devnull@kma.eu.org> wrote in message
news:426819cf$0$31040$626a14ce@news.free.fr...

> Hold on just a minute! Is your calculator equipped with a Pentium CPU?
> Because it looks like you've been bitten by the FDIV bug all over again.
>
> Unbuffered, unregistered, SDRAM DIMMs
> 512-MB DDR-400 = US$35
> 512-MB DDR2-533 = US$60
>
> That's +71% in my book, not +10%...

You can run the numbers any way you want. I think it's best to use the
best price/performance point for each technology and then compare MB/$.

For DDR: PC3200 512Mb -> $35
For DDR2: PC2-4200 512Mb -> $45
http://www.buyaib.com/index.asp?PageAction=VIEWPROD&ProdID=2342
http://www.18004memory.com/cart/addcart.asp?Try=Yes&itemID=502568&itemTYPE=cat

That gives me about 30%.

DS
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

In comp.sys.ibm.pc.hardware.chips Yousuf Khan <yjkhan@gmail.com> wrote:
> Judd wrote:
> > "Yousuf Khan" <yjkhan@gmail.com> wrote in message
> > news:1114098062.272509.298500@o13g2000cwo.googlegroups.com...
> > > It may have known DDR2 was only a short-term technology all along, but
> > > decided to foist it on the world anyways.

Was the 90nm process technology only a short term technology, and should
we have gone from 130nm straight to 65nm? Better yet, should we not
skip 65nm altogether and go straight to 45nm?

If DDR2 is only a short term technology, perhaps we should have skipped
from SDRAM to DDR4. DDR3 may not last very long either.

> > Another silly anti-Intel post. This one the silliest of all. DDR2 isn't
> > going anywhere anytime soon. Prices are only 5%-10% greater than that of
> > DDR uno right now. It's already scaling to 800 MHz and could scale to 1
> > GHz. I seriously doubt it's going anywhere anytime soon. But hey, I'm sure
> > gundeep is one deep dud....dooo

> So you're saying that you don't mind paying extra for a technology that
> may not even get cheaper because it may not have much production life
> left in it; and it may get phased out long before you're ready to
> switch your processor out, not allowing you to salvage any of your
> investment in it?

DDR2 will be around for a bit. DDR simply cannot scale to the same
datarate (at comparable yield levels) as DDR2. PC3200 DDR SDRAM was
a stop gap of sorts, with bumped voltages. In negative spin parlance,
PC3200 DDR SDRAM is "overclocked memory" that took over what was
rightfully a planned DDR2 speed bin, PC2-3200.

The sweet spot for DDR2 will hit around DDR2-667, and DDR2-800 will
be pushed out if DDR3 gets pushed out.

Going from DDR to DDR2 is a natural progression. "Not having much
production life left in it" isn't a conspiracy by Intel to drive
the screws to AMD. It's just that DRAM datarates are currently
doubling every 3 years or so, and DDR2's range of 400 to 800 is
only good for 3 years. The DRAM manufacturers can keep DDR2 around
for a while and drive it up past 1 Gbps, but the timing pressure
on the DRAM core becomes greater.

> If that's the case, then obviously you're a man of
> great wealth. For the rest of us, who are trying to decide on whether
> to purchase a new processor/mobo or make the existing one last a little
> longer until the technology stabilizes, well this would be useful to
> know. Even people considering buying AMD systems need to know this,
> because if Intel itself doesn't believe in DDR2, then there's not much
> point in waiting for an AMD system that supports DDR2 as there likely
> won't be one.

AMD will support DDR2, then transition to DDR3. Otherwise, sometime
down the road, Intel will be shipping products with DDR2-667 while AMD
will be shipping DDR-400, and that scenario will last a year while
everyone sits around waiting for DDR3.

> Also the article lists that both manufacturers are looking at DDR3 and
> Rambus XDR as possible future technologies. Depending on how successful
> DDR2 is, those future technologies could be pretty far off, or pretty
> near.

Just a bit of FUD then.

--
davewang202(at)yahoo(dot)com
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

In comp.sys.ibm.pc.hardware.chips Yousuf Khan <yjkhan@gmail.com> wrote:
> David Wang wrote:
> > Was the 90nm process technology only a short term technology, and
> should
> > we have gone from 130nm straight to 65nm? Better yet, should we not
> > skip 65nm altogether and go straight to 45nm?
> >
> > If DDR2 is only a short term technology, perhaps we should have
> skipped
> > from SDRAM to DDR4. DDR3 may not last very long either.

> Great rhetoric, but you know very well that there is a difference
> between transparent technology like process technology, and technology
> which you have to buy new supporting hardware for like DRAM standards.

If you want a 90nm processor with faster bus/interconnect, you
will have to move to a new board. You can make the 90nm processor
with the same bus/interconnect (datarate) as the previous generation,
Then again, if the point is to get a higher datarate bus/interconnect,
there is no point in staying with the same system board.

> A processor at 90nm works much like a processor at 130nm, and it may
> even fit into the same socket as the previous generation processor --
> thus making it transparent. If they had made DDR2 compatible with DDR,
> where you could simply exchange the old RAM for the new RAM and plug it
> into the same slots, then no problem -- that's transparent technology,
> and no one would care if was really DDR1, DDR2, DDR3, or whatever
> underneath.

Take a look again. DDR2 device command set is a superset of DDR
devices. DDR2 devices can be made to pretend that they're plain
jane DDR devices. The mode registers are supersets, the extra
per nibble differential strobes can be turned off and used just
as per byte single ended reference strobe. The voltages are a bit
different. Then again, 90nm processors typically have slightly
different voltages from 130nm processors. DDR2 devices are compatible
with DDR devices. You just have to ask DRAM module manufacturers to
create these oddball DDR2 device in DDR format modules for you to
fit into your criteria of "compatible", and plug the bastard DDR2
modules into DDR slots. (with no gains in datarate)

> > Going from DDR to DDR2 is a natural progression. "Not having much
> > production life left in it" isn't a conspiracy by Intel to drive
> > the screws to AMD. It's just that DRAM datarates are currently
> > doubling every 3 years or so, and DDR2's range of 400 to 800 is
> > only good for 3 years. The DRAM manufacturers can keep DDR2 around
> > for a while and drive it up past 1 Gbps, but the timing pressure
> > on the DRAM core becomes greater.

> It's not the progression from DDR1 to DDR2 that's being questioned
> here. It's the possible progression from DDR1 straight to DDR3 that
> might be spoiling the party here.

There is no such possibility. Anyone suggesting such possibility is
not familiar with the progressions of the DRAM market. He/she can
just as well suggest that 65nm technology is a short term technology,
and we shouuld jump from 90nm technology to 45nm.

> > AMD will support DDR2, then transition to DDR3. Otherwise, sometime
> > down the road, Intel will be shipping products with DDR2-667 while AMD
> > will be shipping DDR-400, and that scenario will last a year while
> > everyone sits around waiting for DDR3.

> Which is exactly the point. Are they going to sit around waiting for
> DDR3?

No.


--
davewang202(at)yahoo(dot)com
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

On 4/22/2005 07:56, Yousuf Khan wrote:

> great wealth. For the rest of us, who are trying to decide on whether
> to purchase a new processor/mobo or make the existing one last a little
> longer until the technology stabilizes, well this would be useful to
> know. Even people considering buying AMD systems need to know this,

I find that it's never a good idea to try and guess what technology to buy
based on what "the market" *will* do sometime down the line. It's an
endless treadmill and you can never keep up.

When I buy a new mobo I assume I will be buying new memory and processor
at the same time. I would be foolish to think any other way.

It's a fact; the tech. never stabilizes and the next great thing is always
just around the corner.

~Jason

--

Posted Via Usenet.com Premium Usenet Newsgroup Services
----------------------------------------------------------
** SPEED ** RETENTION ** COMPLETION ** ANONYMITY **
----------------------------------------------------------
http://www.usenet.com
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

On Fri, 22 Apr 2005 15:15:17 -0700, "David Schwartz" <davids@webmaster.com>
wrote:

>
>"Grumble" <devnull@kma.eu.org> wrote in message
>news:426819cf$0$31040$626a14ce@news.free.fr...
>
>> Hold on just a minute! Is your calculator equipped with a Pentium CPU?
>> Because it looks like you've been bitten by the FDIV bug all over again.
>>
>> Unbuffered, unregistered, SDRAM DIMMs
>> 512-MB DDR-400 = US$35
>> 512-MB DDR2-533 = US$60
>>
>> That's +71% in my book, not +10%...
>
> You can run the numbers any way you want. I think it's best to use the
>best price/performance point for each technology and then compare MB/$.
>
> For DDR: PC3200 512Mb -> $35
> For DDR2: PC2-4200 512Mb -> $45
>http://www.buyaib.com/index.asp?PageAction=VIEWPROD&ProdID=2342
>http://www.18004memory.com/cart/addcart.asp?Try=Yes&itemID=502568&itemTYPE=cat
>
> That gives me about 30%.
>
> DS
>

So, did anyone mention power yet - or has the conspiracy theory not yet run
its course?
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

Jason Gurtz wrote:
> I find that it's never a good idea to try and guess what technology to buy
> based on what "the market" *will* do sometime down the line. It's an
> endless treadmill and you can never keep up.
>
> When I buy a new mobo I assume I will be buying new memory and processor
> at the same time. I would be foolish to think any other way.
>
> It's a fact; the tech. never stabilizes and the next great thing is always
> just around the corner.

Actually I've been fortunate enough to find solutions that maintain
backward compatibility for me. I carried a set of SDRAM from a K6 to an
Athlon system. The Athlon system that I had, had the ability to accept
both SDRAM and DDR memory, so once I was ready to upgrade the RAM, I was
already prepared. And I fully expect to carry my DDR over to a new
Athlon 64 system, whenever i'm ready to buy one of those things. The DDR
is not top of the line anymore (PC2100), but it will still work in the
A64. Then if I feel like upgrading the RAM some more, then I can go for
more uptodate DDR.

Yousuf Khan
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

David Wang wrote:
> Take a look again. DDR2 device command set is a superset of DDR
> devices. DDR2 devices can be made to pretend that they're plain
> jane DDR devices. The mode registers are supersets, the extra
> per nibble differential strobes can be turned off and used just
> as per byte single ended reference strobe. The voltages are a bit
> different. Then again, 90nm processors typically have slightly
> different voltages from 130nm processors. DDR2 devices are compatible
> with DDR devices. You just have to ask DRAM module manufacturers to
> create these oddball DDR2 device in DDR format modules for you to
> fit into your criteria of "compatible", and plug the bastard DDR2
> modules into DDR slots. (with no gains in datarate)

Well, if they could've made DDR2 physically and electrically compatible
with DDR1, then they should have.

Yousuf Khan
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

"Yousuf Khan" <yjkhan@gmail.com> wrote in message
news:1114098062.272509.298500@o13g2000cwo.googlegroups.com...
> It may have known DDR2 was only a short-term technology all along, but
> decided to foist it on the world anyways.

And then again, they may not have "known" this.

> CoolTechZone::Industry Insider Report: Intel Knew DDR2 was Poor Design
> http://www.cooltechzone.com/index.php?option=content&task=view&id=1197
>
> It must've been trying to get AMD to adopt it and then watch as it was
> forced to adopt DDR3 in short order.

Oh, the horror!

--

... Hank

http://home.earthlink.net/~horedson
http://home.earthlink.net/~w0rli
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

In comp.sys.ibm.pc.hardware.chips Yousuf Khan <bbbl67@ezrs.com> wrote:
> David Wang wrote:
> > Take a look again. DDR2 device command set is a superset of DDR
> > devices. DDR2 devices can be made to pretend that they're plain
> > jane DDR devices. The mode registers are supersets, the extra
> > per nibble differential strobes can be turned off and used just
> > as per byte single ended reference strobe. The voltages are a bit
> > different. Then again, 90nm processors typically have slightly
> > different voltages from 130nm processors. DDR2 devices are compatible
> > with DDR devices. You just have to ask DRAM module manufacturers to
> > create these oddball DDR2 device in DDR format modules for you to
> > fit into your criteria of "compatible", and plug the bastard DDR2
> > modules into DDR slots. (with no gains in datarate)

> Well, if they could've made DDR2 physically and electrically compatible
> with DDR1, then they should have.

The point of making a new standard with DDR2 is that you CAN make these
new DRAM parts with lower voltages, and as daytripper points out, uses
less power.

You CAN make DDR2 devices pretend that they're DDR devices, you can even
make sure that the DDR2 parts are tolerant of DDR voltages. Then again,
there would be no point in it. You'd win nothing in terms of power
consumption or increased datarate.



--
davewang202(at)yahoo(dot)com
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

Yousuf Khan wrote:
> It may have known DDR2 was only a short-term technology all along, but
> decided to foist it on the world anyways.
>
> CoolTechZone::Industry Insider Report: Intel Knew DDR2 was Poor Design
> http://www.cooltechzone.com/index.php?option=content&task=view&id=1197
>
> It must've been trying to get AMD to adopt it and then watch as it was
> forced to adopt DDR3 in short order.
>
> Yousuf Khan
>

Some high end graphics cards are already running DDR3.
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

On Sat, 23 Apr 2005 06:00:01 GMT, No one <bibbe@yahoo.com> wrote:

>Yousuf Khan wrote:
>> It may have known DDR2 was only a short-term technology all along, but
>> decided to foist it on the world anyways.
>>
>> CoolTechZone::Industry Insider Report: Intel Knew DDR2 was Poor Design
>> http://www.cooltechzone.com/index.php?option=content&task=view&id=1197
>>
>> It must've been trying to get AMD to adopt it and then watch as it was
>> forced to adopt DDR3 in short order.
>>
>> Yousuf Khan
>>
>
>Some high end graphics cards are already running DDR3.

Yes but aren't those ram chips soldered right to the cards PCB?
I would think making sticks for motherboards isn't as easy.

Ed
 
Archived from groups: comp.sys.intel,comp.sys.ibm.pc.hardware.chips (More info?)

In comp.sys.ibm.pc.hardware.chips No one <bibbe@yahoo.com> wrote:
> Yousuf Khan wrote:
> > It may have known DDR2 was only a short-term technology all along, but
> > decided to foist it on the world anyways.

> > CoolTechZone::Industry Insider Report: Intel Knew DDR2 was Poor Design
> > http://www.cooltechzone.com/index.php?option=content&task=view&id=1197

> > It must've been trying to get AMD to adopt it and then watch as it was
> > forced to adopt DDR3 in short order.

> Some high end graphics cards are already running DDR3.

GDDR3 is not the same as DDR3.



--
davewang202(at)yahoo(dot)com
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom