Intel drops HyperThreading

Tony Hill wrote:

On Thu, 25 Aug 2005 17:36:17 GMT, CJT wrote:


George Macdonald wrote:

Its not even close - you can get a benchmark comparison .pdf here
http://www.tollygroup.com/DocDetail....cNumber=205107


Now show a study _not_ sponsored by Intel. And that addresses the watts
of power used by each processor.


Ok.. how's this?

http://www.tomshardware.com/cpu/20020605/


Quoting from the "Conclusions" page of that article:

"Looking at performance and power output in terms of a ratio, the C3
blows away its competitors."


I'm no big fan of Tom's, but the fact of the matter is that the VIA C3
at 1.0GHz really struggles to match the performance of an ancient
Celeron 667MHz processor. Now figure that the Celeron-M at 900MHz
offers a greatly improved core, 4 times as much cache, 6 times the bus
bandwidth and 3 times the memory bandwidth. It all adds up to the C3
just not being at all competitive.

Even at 2.0GHz I suspect that the yet-to-ship C7 processor will have
difficulty competing with a 900MHz Celeron-M, and it will do so with 4
times the power consumption (20W for the VIA chip vs. 5W for Intel).

Ohh, here's another set of slightly dated numbers comparing the C3 at
800MHz to a PIII at 500MHz:

http://www.dansdata.com/c3.htm


-------------
Tony Hill
hilla underscore 20 at yahoo dot ca


--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form .

CJT wrote:
Tony Hill wrote:

On Thu, 25 Aug 2005 17:36:17 GMT, CJT wrote:


George Macdonald wrote:

Its not even close - you can get a benchmark comparison .pdf here
http://www.tollygroup.com/DocDetail....cNumber=205107


Now show a study _not_ sponsored by Intel. And that addresses the watts
of power used by each processor.


Ok.. how's this?

http://www.tomshardware.com/cpu/20020605/


Quoting from the "Conclusions" page of that article:

"Looking at performance and power output in terms of a ratio, the C3
blows away its competitors."


quote

As expected, the C3 processor is not able to compete with other
processors at similar clock speeds. Depending on the particular
benchmark, an old Celeron 667 is either considerably faster or
considerably slower, making it difficult to specify a recommendation
for VIA's C3.

The C3 has definitely won the power and temperature race: no other
desktop processor consumes as little energy and wastes as little power
as the C3. Looking at performance and power output in terms of a ratio,
the C3 blows away its competitors.

/quote

The "Old Celeron 667" that is used to support this comparison being one
that was never designed for low power operation. At 17W, it's not
particularly power-hungry and isn't exactly "blown away" by the 12W
Via. Celeron power consumption from

http://users.erols.com/chare/elec.htm

As long as Tom's was going to dig into old proecessors to compare for
conclusions about power, he should have been using the PIII-667, but
that's a nit. Any of the ULV processors that Intel has brought out,
whethere branded Celeron or Pentium, would blow away Via. A 1GHz
Pentium-M ULV draws 5 Watts.

As to comparison of "ratio" with desktop chips, the only exercise that
makes any sense is to compare power consumption at equal performance or
performance at equal power consumption.

RM

Robert Myers wrote:

CJT wrote:

Tony Hill wrote:


On Thu, 25 Aug 2005 17:36:17 GMT, CJT wrote:



George Macdonald wrote:


Its not even close - you can get a benchmark comparison .pdf here
http://www.tollygroup.com/DocDetail....cNumber=205107


Now show a study _not_ sponsored by Intel. And that addresses the watts
of power used by each processor.


Ok.. how's this?

http://www.tomshardware.com/cpu/20020605/


Quoting from the "Conclusions" page of that article:

"Looking at performance and power output in terms of a ratio, the C3
blows away its competitors."



quote

As expected, the C3 processor is not able to compete with other
processors at similar clock speeds. Depending on the particular
benchmark, an old Celeron 667 is either considerably faster or
considerably slower, making it difficult to specify a recommendation
for VIA's C3.

The C3 has definitely won the power and temperature race: no other
desktop processor consumes as little energy and wastes as little power
as the C3. Looking at performance and power output in terms of a ratio,
the C3 blows away its competitors.

/quote

The "Old Celeron 667" that is used to support this comparison being one
that was never designed for low power operation. At 17W, it's not
particularly power-hungry and isn't exactly "blown away" by the 12W
Via. Celeron power consumption from

http://users.erols.com/chare/elec.htm

As long as Tom's was going to dig into old proecessors to compare for
conclusions about power, he should have been using the PIII-667, but
that's a nit. Any of the ULV processors that Intel has brought out,
whethere branded Celeron or Pentium, would blow away Via. A 1GHz
Pentium-M ULV draws 5 Watts.

As to comparison of "ratio" with desktop chips, the only exercise that
makes any sense is to compare power consumption at equal performance or
performance at equal power consumption.

RM

Here's something current and on point:

http://www.computerworld.com/hardwar..._PM&nid=104017


--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form .

CJT wrote:
wrote:

On Wed, 24 Aug 2005 17:24:57 GMT, CJT wrote:

...snip...

A focus on watts could drive Itanium even deeper in the hole.



Focus on watts in laptops and SFF boxes? Paramount. In blade servers
- ditto. Even at the expence of raw speed. But in desktops there is
always a place for a fan or two, and there is always need for speed,
be you a gamer, software developer, or heavy graphics user. Much more
so in 4U+ servers. The top $ are paid for top notch performance of
"mission-critical" databases and like. The heat produced by a
high-performing chip is a problem that can be and usually is
reasonably solved. That is, unless you deal with Prescott core that
doubles as a space heater.

If people calculated how much per month it's costing to power their
"gaming" machines, it might quickly become an issue. "Power user"
is closer to true than one might imagine.


Well, if the processor is an extra 100 watts (large number) and
electricity is 15 cents/kwH (on the high side) and the processor draws
the extra 100 watts even when nobody is using it, it comes to 36
cents/day, or about 10 dollars/month. So the extra electricity to
assure optimim game play over 6 months equals the cost of a game. On
the other hand the high speed internet connection costs 40 dollars per
month. QED Power consumption in gaming PC is not a significant economic
factor.

--
Del Cecchi
"This post is my own and doesn’t necessarily represent IBM’s positions,
strategies or opinions.”

CJT wrote:


Here's something current and on point:

http://www.computerworld.com/hardwar..._PM&nid=104017


What's this got to do with Via? Green Destiny is very old news: LANL
scientists playing in the sandbox. With any luck, we won't be seeing
more of that kind of nonsense. LANL p'd away a ton of taxpayer money
wiring together everything they could lay their hands on, even paid
Cray to custom-build them their sandbox dreams, and the bottom line is
that we're going to be using piles of generic server boxes for the
forseeable future. The real breakthrough in performance/watt came from
IBM's Blue Gene.

As to Transmeta, I haven't seen head-to-heads, but I'll be really
surprised if *they* do better on a performance per watt than Pentium-M
ULV.

RM

Del Cecchi wrote:


Well, if the processor is an extra 100 watts (large number) and
electricity is 15 cents/kwH (on the high side) and the processor draws
the extra 100 watts even when nobody is using it, it comes to 36
cents/day, or about 10 dollars/month. So the extra electricity to
assure optimim game play over 6 months equals the cost of a game. On
the other hand the high speed internet connection costs 40 dollars per
month. QED Power consumption in gaming PC is not a significant economic
factor.

Nobody would argue about the importance of power consumption for
servers, for HPC, or for mobile applications. The question is: why
worry for stationary machines, such as those used for gaming? As far
as I can tell, because there is no other way to get more performance
into an acceptable power and cooling envelope, and I'm assuming that
machines used for gaming will continue to have an insatiable demand for
greater performance. The only way to get it, as I currently understand
the situation, is more cores operating at a point that is less than
optimal from the POV of single-thread performance. It will be
interesting to see how long power-no-consideration single-thread
workhorses will survive. I expect them to become an endangered species
for all but the the most specialized applications (very high-end HPC,
for example).

RM

Robert Myers wrote:
Del Cecchi wrote:


Well, if the processor is an extra 100 watts (large number) and
electricity is 15 cents/kwH (on the high side) and the processor draws
the extra 100 watts even when nobody is using it, it comes to 36
cents/day, or about 10 dollars/month. So the extra electricity to
assure optimim game play over 6 months equals the cost of a game. On
the other hand the high speed internet connection costs 40 dollars per
month. QED Power consumption in gaming PC is not a significant economic
factor.


Nobody would argue about the importance of power consumption for
servers, for HPC, or for mobile applications. The question is: why
worry for stationary machines, such as those used for gaming?

Because while the additional power consumption might be
insignificant or irrelevant on a personal basis, it is *very*
significant on a national or global basis.

If 100 million home computers in North America are replaced with
machines that need an additional 100 Watts each, then an
additional 10 GW of generating capacity is needed. More likely
double that when you consider the fact that in most parts of
North America still more power is going to be wasted by air
conditioners working just a little harder to remove that
additional 100 Watts from the house/apartment/office.

I often wonder if power rationing is the only way to keep in
check the rising power demands caused by obsessions with
increasingly power-hungry toys.

Increasingly power-hungry TVs and home computers are not fully to
blame for North America's growing energy crisis, but they are a
significant and highly symbolic factor.

Rob Stow wrote:
Robert Myers wrote:

Del Cecchi wrote:


Well, if the processor is an extra 100 watts (large number) and
electricity is 15 cents/kwH (on the high side) and the processor draws
the extra 100 watts even when nobody is using it, it comes to 36
cents/day, or about 10 dollars/month. So the extra electricity to
assure optimim game play over 6 months equals the cost of a game. On
the other hand the high speed internet connection costs 40 dollars per
month. QED Power consumption in gaming PC is not a significant economic
factor.


Nobody would argue about the importance of power consumption for
servers, for HPC, or for mobile applications. The question is: why
worry for stationary machines, such as those used for gaming?


Because while the additional power consumption might be insignificant or
irrelevant on a personal basis, it is *very* significant on a national
or global basis.

If 100 million home computers in North America are replaced with
machines that need an additional 100 Watts each, then an additional 10
GW of generating capacity is needed. More likely double that when you
consider the fact that in most parts of North America still more power
is going to be wasted by air conditioners working just a little harder
to remove that additional 100 Watts from the house/apartment/office.

I often wonder if power rationing is the only way to keep in check the
rising power demands caused by obsessions with increasingly power-hungry
toys.

Increasingly power-hungry TVs and home computers are not fully to blame
for North America's growing energy crisis, but they are a significant
and highly symbolic factor.

TV's use less power than they used to. You do remember vacuum tubes,
right? And I was responding to the assertion that gamers should take
the cost of electricity into account when selecting a system. There
aren't 100 million gamers, either.

--
Del Cecchi
"This post is my own and doesn’t necessarily represent IBM’s positions,
strategies or opinions.”

On Fri, 26 Aug 2005 00:11:23 GMT, CJT wrote:

George Macdonald wrote:

On Thu, 25 Aug 2005 17:36:17 GMT, CJT wrote:


George Macdonald wrote:

Its not even close - you can get a benchmark comparison .pdf here
http://www.tollygroup.com/DocDetail....cNumber=205107


Now show a study _not_ sponsored by Intel. And that addresses the watts
of power used by each processor.


C'mon this was not even a competition - the Celeron was in a different
class and did not fail any tests.

Who's to say whether, if VIA were the sponsor, a test could be found
that the Celeron failed.

I'm no Intel fan but I know which way I'd bet. There are plenty of
"consultant" companies ready to offer to do the job for VIA.

--
Rgds, George Macdonald

On Fri, 26 Aug 2005 02:54:59 GMT, CJT wrote:

wrote:

On Wed, 24 Aug 2005 17:24:57 GMT, CJT wrote:

...snip...

A focus on watts could drive Itanium even deeper in the hole.


Focus on watts in laptops and SFF boxes? Paramount. In blade servers
- ditto. Even at the expence of raw speed. But in desktops there is
always a place for a fan or two, and there is always need for speed,
be you a gamer, software developer, or heavy graphics user. Much more
so in 4U+ servers. The top $ are paid for top notch performance of
"mission-critical" databases and like. The heat produced by a
high-performing chip is a problem that can be and usually is
reasonably solved. That is, unless you deal with Prescott core that
doubles as a space heater.

If people calculated how much per month it's costing to power their
"gaming" machines, it might quickly become an issue. "Power user"
is closer to true than one might imagine.

When you pay someone hourly, or, alternatively, when you are paid
for end result, the energy cost is negligible comparing to the cost of
time it takes to (compile the code, render the frame, etc. - pick what
suits you best). When your life in virtual universe is more important
for you than real life (that's the case for many gamers) who cares
about the energy bill?
But wait, with the rates of energy cost increases everyone will
start taking it into account quite soon ;-(