NEC buys into Transmeta and Longrun technology

[Black Jack]

http://www.internetnews.com/infra/article.php/3331601

NEC seems to think it's going to need the LongRun technology to be
able to run their chips once they enter the 90, 65, and 45 nm process
nodes. There was also a rumour that Intel was looking at this
technology sometime back too. Intel might need it if it doesn't want
to adopt SOI.

Yousuf Khan

 

[Tony Hill]

http://www.internetnews.com/infra/article.php/3331601

NEC seems to think it's going to need the LongRun technology to be
able to run their chips once they enter the 90, 65, and 45 nm process
nodes. There was also a rumour that Intel was looking at this
technology sometime back too. Intel might need it if it doesn't want
to adopt SOI.

My take on this is that the main thing NEC was after was the dynamic
threshold voltage transistors that Transmeta announced a little while
back. Interesting technology, though I wouldn't be surprised if
Intel, AMD and IBM are able to get the same sort of design without
licensing Transmeta IP (of course, there will probably be some kind of
lawsuit on it's way with regards to this).

This doesn't really have that much of anything to do with SOI, except
for the idea of reducing leakage current. Either way, Intel is
planning on using SOI in the future, just not quite the same type of
SOI that AMD and IBM are currently using.


Ugg.. I gotta grip about one line in the linked article though:

<quoting>
Enderle says it was Transmeta that woke Intel up, and the result was
their ARM-based chip architecture initiatives: Pentium M and Centrino.
Both chips are purpose-built for mobile devices and laptop computers.
<end quote>

WTF?!?! And they pay this guy money for this totally incorrect trash?
First off they're talking about Centrino as if it's a chip rather than
the marketing campaign that it is, but what in the hell are they doing
mentioning ARM in there?! ARM has absolutely ZERO to do with either
the Pentium-M processor or the Centrino marketing effort! X-Scale,
yes; Centrino, no.

Actually I think maybe the anal-yst may have had it right here and
it's just the author of the article that completely misunderstood
things and mixed everything up. Still, such blatant errors that could
have been fixed with about 10 seconds of fact-checking really don't
lead to much confidence.

 

[Robert Myers]

On Fri, 26 Mar 2004 15:38:25 -0500, Tony Hill

Ugg.. I gotta grip about one line in the linked article though:

<quoting>
Enderle says it was Transmeta that woke Intel up, and the result was
their ARM-based chip architecture initiatives: Pentium M and Centrino.
Both chips are purpose-built for mobile devices and laptop computers.
<end quote>

WTF?!?! And they pay this guy money for this totally incorrect trash?
First off they're talking about Centrino as if it's a chip rather than
the marketing campaign that it is, but what in the hell are they doing
mentioning ARM in there?! ARM has absolutely ZERO to do with either
the Pentium-M processor or the Centrino marketing effort! X-Scale,
yes; Centrino, no.

Before you wig out completely and/or I google myself blind trying to
figure out things I only halfway understand, maybe you or someone else
would want to take a look at

http://www.newswireless.net/articles/031015-transmeta.html

and offer any comments that come to mind. The only relationship that
ARM has with Centrino that is relevant to the article may be a common
threat from Efficeon. The future is x86. The future is mobile. The
future is low power.

"Nobody is going to launch a pocket PC to take seriously, based on
Windows and X86," the article says. Okay, how about Linux and x86?
And just because the display in your pocket PC is less than dazzling
doesn't mean it will stay that way when you get to your desk. Walk
around with your *real* PC in your pocket, plug it into a real display
and keyboard when you're able to, and live with a munchkin display
when you can't? Sounds more attractive than laplink, briefcase, and
file synchonization and whatever other half-assed solutions there are
that allow you to split your time and your attention between a pocket
device and/or a laptop device and your "real" PC.

RM

RM

 

[Tony Hill]

On Fri, 26 Mar 2004 15:38:25 -0500, Tony Hill



Before you wig out completely and/or I google myself blind trying to
figure out things I only halfway understand, maybe you or someone else
would want to take a look at

http://www.newswireless.net/articles/031015-transmeta.html

and offer any comments that come to mind. The only relationship that
ARM has with Centrino that is relevant to the article may be a common
threat from Efficeon. The future is x86. The future is mobile. The
future is low power.

.... and the future is not here yet. :>

"Nobody is going to launch a pocket PC to take seriously, based on
Windows and X86," the article says. Okay, how about Linux and x86?

Not any time too soon. The chips used in palm-top systems (almost all
ARM these days) consume a maximum of 500mW of power. That's an order
of magnitude less than the lowest powered chips that Transmeta makes.
Even VIA, who has some chips that consume a max of only 3W of power is
not at all in the running, let alone Intel's who's lowest powered x86
chip consumes 7W.

And just because the display in your pocket PC is less than dazzling
doesn't mean it will stay that way when you get to your desk. Walk
around with your *real* PC in your pocket, plug it into a real display
and keyboard when you're able to, and live with a munchkin display
when you can't? Sounds more attractive than laplink, briefcase, and
file synchonization and whatever other half-assed solutions there are
that allow you to split your time and your attention between a pocket
device and/or a laptop device and your "real" PC.

Maybe someday, but that day is several years away at the very least.
The ultraportable and tablet PC market where Intel competes with
Transmeta, and the palm-top computing market where the X-Scale and
other ARM chips live are two very different markets at this time.
Many years into the future (probably at least 5, which is an eternity
in computerland) they may merge, but not today.

 

[Robert Myers]

... and the future is not here yet. :>


Not any time too soon. The chips used in palm-top systems (almost all
ARM these days) consume a maximum of 500mW of power. That's an order
of magnitude less than the lowest powered chips that Transmeta makes.
Even VIA, who has some chips that consume a max of only 3W of power is
not at all in the running, let alone Intel's who's lowest powered x86
chip consumes 7W.

The future may not be _quite_ as far off as you estimate:

http://www.intel.com/design/intarch/designgd/27380201.pdf

400MHz Ultra Low Voltage Mobile Celeron on 0.13micron process, typical
power 3.4W, maximum power 4.23W. As to performance, if you can live
with a 1Ghz Via Nehemia, you can probably live with the 400MHz ULV
Mobile Celeron:

http://www.stealthcomputer.com/pdfs/TollyTS203124IntelCeleronSept2003.pdf

The power budget is close enough to an order of magnitude more than
the 500mW you quote for a palm-top system, but the September, 2003
study commissioned by Intel would seem to indicate that Intel means
business.

I don't own or use a PDA, but the step that would get me interested
would be more a matter of software than hardware. I have too many
different things to explain to too many different people and systems
to be constantly explaining to my PDA what it's relationship is to my
"real" computer, a problem I already have with my laptop.

Your five year timeline may be right (I think it is pessimistic), but
a pocket x86 PC is where the world is headed.

RM

 

[Tony Hill]

The future may not be _quite_ as far off as you estimate:

http://www.intel.com/design/intarch/designgd/27380201.pdf

400MHz Ultra Low Voltage Mobile Celeron on 0.13micron process, typical
power 3.4W, maximum power 4.23W. As to performance, if you can live
with a 1Ghz Via Nehemia, you can probably live with the 400MHz ULV
Mobile Celeron:

For a palm-top system I'd rather live with a 520MHz ARM chip consuming
only 500mW.

http://www.stealthcomputer.com/pdfs/TollyTS203124IntelCeleronSept2003.pdf

The power budget is close enough to an order of magnitude more than
the 500mW you quote for a palm-top system, but the September, 2003
study commissioned by Intel would seem to indicate that Intel means
business.

They still aren't even close. These power consumption figures are
fine and dandy for a ultra-portable or tablet PC, but they just aren't
going to cut it on a palm-top. They really need to get power down to
under 1W before they can start competing.

I don't own or use a PDA, but the step that would get me interested
would be more a matter of software than hardware. I have too many
different things to explain to too many different people and systems
to be constantly explaining to my PDA what it's relationship is to my
"real" computer, a problem I already have with my laptop.

Your five year timeline may be right (I think it is pessimistic), but
a pocket x86 PC is where the world is headed.

Could be, but I still stick with my 5 year time-line. They aren't
close now and a 50% drop in power isn't going to get them there. Even
the lowest powered x86 chips still need at least a 75% drop in power
consumption before they can really start to compete. Of course, when
they get there, you then have to ask "is it worth it?". ARM chips
perform reasonably well, they are manufactured by everyone+dog and
they are dirt-cheap. Are the advantages of x86 really that great?
Given the very small screen used for palm-top systems, as well as the
lack of a keyboard, often times the GUI needs to be redesigned for
them anyway. If you need to support two different versions of the
software anyway, just how much of an advantage is it if they both use
the same processor?

 

[Robert Myers]

Could be, but I still stick with my 5 year time-line. They aren't
close now and a 50% drop in power isn't going to get them there. Even
the lowest powered x86 chips still need at least a 75% drop in power
consumption before they can really start to compete. Of course, when
they get there, you then have to ask "is it worth it?". ARM chips
perform reasonably well, they are manufactured by everyone+dog and
they are dirt-cheap. Are the advantages of x86 really that great?
Given the very small screen used for palm-top systems, as well as the
lack of a keyboard, often times the GUI needs to be redesigned for
them anyway. If you need to support two different versions of the
software anyway, just how much of an advantage is it if they both use
the same processor?

I think it's a big deal. I don't want another little gadget. My life
is littered with them. Something that integrated naturally and
automatically with my principal desktop (easy to accomplish all kinds
of ways), and that shared files, including binaries, would really get
my interest, both as a user and as an applications developer. While
it _may_ take us five years to get into a power profile that works for
a pocket x86 PC, we will

-> never <-

see truly platform-independenct software. If, say, Java had worked
out the way it was supposed to, we wouldn't need ISA-compatibility.
Java hasn't worked out the way it was supposed to, nothing else is
going to, and ISA-compatibility will win.

RM

 

[Jan Panteltje]

The future may not be _quite_ as far off as you estimate:

http://www.intel.com/design/intarch/designgd/27380201.pdf

400MHz Ultra Low Voltage Mobile Celeron on 0.13micron process, typical
power 3.4W, maximum power 4.23W. As to performance, if you can live
with a 1Ghz Via Nehemia, you can probably live with the 400MHz ULV
Mobile Celeron:

http://www.stealthcomputer.com/pdfs/TollyTS203124IntelCeleronSept2003.pdf

The power budget is close enough to an order of magnitude more than
the 500mW you quote for a palm-top system, but the September, 2003
study commissioned by Intel would seem to indicate that Intel means
business.

I don't own or use a PDA, but the step that would get me interested
would be more a matter of software than hardware. I have too many
different things to explain to too many different people and systems
to be constantly explaining to my PDA what it's relationship is to my
"real" computer, a problem I already have with my laptop.

Your five year timeline may be right (I think it is pessimistic), but
a pocket x86 PC is where the world is headed.

RM

I think we also see a trend to make cellphones into PDAs, and with UTMS here
that may become very interesting.
I do not have a PDA either, but my cellphone can do most a PDA does, plus
has a camera, the new ones you can use as memo recorder, and can even record
avi video.
Mine communicates with the PC via IRDA, the newer ones do Blue tooth.
So maybe PDA is dead, long live the cellphone?
Of cause mine has WAP, email, and I can do my banking via it..
Ad the battery lasts a week or so..
Anyone here been at the Cebit in Germany?
Just curious what the next thing is
JP

 

[Tony Hill]

I think it's a big deal. I don't want another little gadget. My life
is littered with them. Something that integrated naturally and
automatically with my principal desktop (easy to accomplish all kinds
of ways), and that shared files, including binaries, would really get
my interest, both as a user and as an applications developer. While

As an applications developer though, you are in a bit of the minority.
Shared files is critical, but it's fairly well supported on PDAs
already. Shared binaries is probably a non-issue for most users. And
even for developers, is it enough if you can share the source code?
Many PDAs will run either a Win32 OS or a Linux OS, so often a
recompile for a new architecture is all that would be required.

it _may_ take us five years to get into a power profile that works for
a pocket x86 PC, we will

-> never <-

see truly platform-independenct software. If, say, Java had worked
out the way it was supposed to, we wouldn't need ISA-compatibility.
Java hasn't worked out the way it was supposed to, nothing else is
going to, and ISA-compatibility will win.

What was the saying about saying "never"? :>

Actually I tend to agree with you on this point, but we do already
have some kinda-sorta platform -independent software. First, as you
mentioned, there is Java. There is also Microsoft's .Net thing,
though it's still a bit up in the air as to how compatible that will
be. However, beyond that, often you can get source code that will
compile on multiple platforms. Open source projects have really
opened the door in the regard, particularly when running on Linux.
While almost all Linux applications are written for x86, a fair chunk
of the more useful ones have active ports to other architectures
(including ARM), so it's just a matter of recompiling them. Since the
source code is freely available, this is often pretty easy.

 

[Robert Myers]

As an applications developer though, you are in a bit of the minority.

No news there. Permanently out of step is my status in life.

Shared files is critical, but it's fairly well supported on PDAs
already. Shared binaries is probably a non-issue for most users. And
even for developers, is it enough if you can share the source code?
Many PDAs will run either a Win32 OS or a Linux OS, so often a
recompile for a new architecture is all that would be required.

I run a mixed Windows/Linux cluster with really good network
connectivity and all kinds of tools to make things interoperable.
It's impressive, if I may say so, but it's not good enough, and it's
my model for why I don't want yet another environment, even if it
knows how to talk to one or more gateways on my network.

What was the saying about saying "never"? :>

You notice I didn't whisper it. In this particular case, I would be
so happy to be wrong that I tempted the gods to the maximum extent
possible.

Actually I tend to agree with you on this point, but we do already
have some kinda-sorta platform -independent software. First, as you
mentioned, there is Java. There is also Microsoft's .Net thing,
though it's still a bit up in the air as to how compatible that will
be.

You're saying all this with a straight face?

However, beyond that, often you can get source code that will
compile on multiple platforms. Open source projects have really
opened the door in the regard, particularly when running on Linux.
While almost all Linux applications are written for x86, a fair chunk
of the more useful ones have active ports to other architectures
(including ARM), so it's just a matter of recompiling them. Since the
source code is freely available, this is often pretty easy.

If you'd like to have a peek into such a future, do what I just did
and google "arm cygwin".

RM