PDA

View Full Version : Difference between x86 and PPC


Sieb
07-10-2003, 05:53 PM
I would add this to another thread, but figured it was a heavy enough read to be mentioned alone. Differences between PC processors and Mac processors (in layman terms).

http://www.osnews.com/story.php?news_id=3997

Edjumakate yoself!

"Dude, RISC is coming back!" -Hackers

mark_wilkins
07-10-2003, 06:11 PM
Worth checking these out too if you're interested in this question:

http://arstechnica.com/cpu/02q2/ppc970/ppc970-1.html

http://arstechnica.com/cpu/03q1/ppc970/ppc970-0.html

-- Mark

Sieb
07-10-2003, 06:54 PM
Good good, OSnews also brings out some good points to look into dealing with the future of procs. All good reads.

Add them if ya got them. :beer:

NO PERSONAL OPINIONS! No flame wars, this is an all technical thread to put to rest any mac/pc wars. I want to try to get more specific technical info on this board for readers.

Thalaxis
07-10-2003, 07:20 PM
That was a pretty good article, I thought.

Lockstar
07-11-2003, 09:54 AM
Good article.

The following information is from an unconfirmed and anonymous source. As such, authenticity is always uncertain, but due to the content of the piece, was felt to be of sufficient interest for publishing. Of interest, MacBidouille has posted similar (but fewer) details in their unconfirmed rumors of the PPC's timeline. This information below may represent corroboration -- or simply a common source. Take, as with all rumors, with an appropriate amount of skepticism.

Apple and IBM have been working on parallel development of the Power5 and PPC 980. The PPC 980 is a single core version of the Power 5. While prototype forms of this chip exist, it is almost a year away from shipping in Macs.

Improvements in the PPC 980 include Hyperthreading, eLiza error correction, and more massive parallelism. IBM's implementation of hyperthreading provides a 30% gain over Intel's. eLiza technology will reduce the bottlenecks when the branch prediction unit fails. Altivec will split into 3 pipelines (vs 2 in the 970), 4 Integer and 4 Floating point units. 980 will have to be built on a 90nm processor due to heat dissipation requirements.

Steve's comment of 3GHz in 1 year will not be accomplished with the G5 (970) - which will top out at 2.6-2.8GHz. The PPC 980 will start at speeds of 2.6-3GHz and top out around 4.5-5GHz. The G5 will make its way into PowerBook lines in Jan/Feb, Xserve's later this year, and iMacs in approximately one year.

Marklar's project size has decreased, but remains ongoing. There are four generations of the PowerPC including and beyond the 970 that are in development and planning. Besides the 980 chips (targetted at end of 2004), there are plans for 990 chips on a 65nm process in 2005/2006 @ 6GHz and scaling up to near 10GHz. Beyond this, the PPC 9900 starting on a 45nm process is targetted in 2007/2008 starting at 9-10GHz and reaching up to 20-25GHz by 2010-2011.

Obviously just rumours, but it does illustrate that the PPC has a very healthy future. Other vendors and platforms will have answers (i've got my eye on Opteron and Itanium Thalaxis) but it's good to see the PPC in the running again.

Apple and OS X will once again gain kudos in the creative arena, but in my opinion, the only way to gain a larger acceptance of the PPC in the market place is through Linux. If the PPC can grow with linux adoption over the next couple of years, then Intel/ AMD will have to sit up and take notice. In my opinion the PPC will not only evolve into a formidable chip in itself, but it will also become the catalyst (perhaps not solely) for Intel/ AMD etc producing not just faster chips, but more efficient (heat dissipation, power etc) chips.

This can only be good news for everyone, regardless of platform, and to allow for a wider choice of tools.

:beer:

Thalaxis
07-11-2003, 01:33 PM
Originally posted by Lockstar
If the PPC can grow with linux adoption over the next couple of years, then Intel/ AMD will have to sit up and take notice. In my opinion the PPC will not only evolve into a formidable chip in itself, but it will also become the catalyst (perhaps not solely) for Intel/ AMD etc producing not just faster chips, but more efficient (heat dissipation, power etc) chips.

Check out the Pentium-M. :)

beaker
07-11-2003, 06:42 PM
Originally posted by Lockstar
(i've got my eye on Opteron and Itanium Thalaxis) but it's good to see the PPC in the running again.

Right now I still think the Itanium is out of the picture for most people if they take TCO into the picture. The power those things suck up, heat dissapation and space they take up is frickin huge. Itanium power(120-130w) is 2x of an Opteron(60-70w) and 3x of a G5(30-40w). Then the extra AC on top of that to keep them cool. Also, what is the smallest machine they have itaniums in(2U or 3U rack space boxes, I could be wrong on this)? The Opteron and G5 were both made with blades and 1u racks in mind(IBM is making Opteron and G5/970 blades).

The Opteron and G5 look like much better solutions(especially for rendering) for the cgi world even though the Itanium 2 give higher numbers.

Thalaxis
07-11-2003, 07:32 PM
Originally posted by beaker
Right now I still think the Itanium is out of the picture for most people if they take TCO into the picture. The power those things suck up, heat dissapation and space they take up is frickin huge. Itanium power(120-130w) is 2x of an Opteron(60-70w) and 3x of a G5(30-40w). Then the extra AC on top of that to keep them cool. Also, what is the smallest machine they have itaniums in(2U or 3U rack space boxes, I could be wrong on this)? The Opteron and G5 were both made with blades and 1u racks in mind(IBM is making Opteron and G5/970 blades).


That's not entirely accurate. I've heard power figures for the G5 that imply that it's not far from 90watts, and IBM themselves say that when it's undervoltaged and driven at 1.1-1.2 GHz, it's around 20 watts.

The Madison power consumption is around 90 watts, so even if
the G5 power figures are on the high side, it's not as far ahead of
Itanium2 as one might expect.

The Deerfield version of Itanium2 is cooler still.


The Opteron and G5 look like much better solutions(especially for rendering) for the cgi world even though the Itanium 2 give higher numbers.

We're not talking about just "faster" here... in floating point
the Itanium2 that was just launched is close to doubling the
performance of the fastest P4. The performance difference is
potentially huge. The catch is that the software isn't there yet, sot
that potential isn't accessible yet.

For high-density computing solutions,the winner looks like it will
be none of the above. I think Dothan will take that cake, since it
has a power consumption of around 10 watts at 1.8 GHz, and its
predecessor (Banias, aka Pentium-M) has shown itself to be quite
powerful already.

Sieb
07-11-2003, 07:33 PM
More PROOF! Less opinions! :)

I looks as if its pretty clear that Itanium will in up in a very nich market, but its inevitable that AMD's Hammers and IBMs Power 4 and 5 will start taking up the brunt of the market. Intel will always be speed king, but at a price. But its also clear at what cost that speed is coming at (efficiency).

Thalaxis
07-11-2003, 07:45 PM
Originally posted by Sieb
More PROOF! Less opinions! :)

I looks as if its pretty clear that Itanium will in up in a very nich market, but its inevitable that AMD's Hammers and IBMs Power 4 and 5 will start taking up the brunt of the market. Intel will always be speed king, but at a price. But its also clear at what cost that speed is coming at (efficiency).

Intel announced Deerfield pricing at around $800. HP's Madison-
based workstation costs less than IBM's Power4 workstation, and
is considerably faster.

The price issue is skewed in Intel's favor, not the other way
around. What I'm wondering is how aggressively Intel will push
the lower end of the Itanium2 family, since they would start to
compete directly with the Xeons at some point.

beaker
07-11-2003, 09:54 PM
>>More PROOF! Less opinions! :)

Ok, here is proof with URL's to back it up.

Originally posted by Thalaxis
That's not entirely accurate. I've heard power figures for the G5 that imply that it's not far from 90watts, and IBM themselves say that when it's undervoltaged and driven at 1.1-1.2 GHz, it's around 20 watts.

Don't know where you heard 90 watts, it's 42 watts to be exact :)
http://www.arstechnica.com/cpu/02q2/ppc970/ppc970-1.html


The Madison power consumption is around 90 watts, so even if
the G5 power figures are on the high side, it's not as far ahead of
Itanium2 as one might expect.

HP's own web page says it's 130 watts, plus another 40 watts for the power regulator(170 total) http://www.hp.com/products1/itanium/infolibrary/analyst_reports/mx2_white_paper.pdf
(page 3, column 2, 1st paragraph)

The Deerfield version of Itanium2 is cooler still.

Yea, I searched for this one and they said it would be in the 60-70W range. But by the time it ships the power 5 based PPC 980 will be shipping soon after and it is supposed to have 3/4 to half the power consumption of the 970.

Also we have the biggest problem of all, 99% of the apps all of us use still don't run on itanium. When Itanium first came out the economy was still doing good and everyone was like "sure we will port to itanium". Then the economy went in the shitter and very little studios actually showed interest in using itanium. I wouldn't doubt that much of the plans/money for porting went out the window.

DaveW
07-11-2003, 10:01 PM
Sieb that article you posted is marketing crap. He even says so in the beginning of the article. He leaves out stuff, makes opinion and speculation sound like facts, and skews everything in favor of PPC. The point of that article was to sell PPC systems from his company, not to provide an unbiased comparison of x86 and PPC processors.

- PPC systems are often sold in dual-cpu configurations so they can perform closer to single cpu x86 systems. This makes power consumption and heat output about the same as the x86 systems, and the "inefficient" x86 systems still perform faster at most tasks and are much cheaper than PPC systems.


- PPC laptops are slower than Centrino laptops and have similar power consumption and heat dissipation. PPC used to be the choice when battery life was more important than speed, but that isn't the case anymore. And you can still get a P4-M laptop if performance is more important than power consumption. And PPC laptops are more expensive than x86 laptops.

When the 970 systems finally hit the market they will help to balance the scales, but they will have tougher competition by then as well.

Thalaxis
07-11-2003, 10:32 PM
Originally posted by beaker


That's typical, not max, and at 1.8 GHz, not 2 GHz. Until IBM releases an updated spec sheet, we won't know for sure.

[b]
HP's own web page says it's 130 watts, plus another 40 watts for the power regulator(170 total) http://www.hp.com/products1/itanium/infolibrary/analyst_reports/mx2_white_paper.pdf
(page 3, column 2, 1st paragraph)


That information is out of date.
http://www.intel.com/design/itanium2/datashts/250945.htm

According to this, the TDP for the 1.5 GHz Itanium2 with 6 MB 3rd level cache is a 107 watt maximum.

It's clear that the power consumption advantage of the PPC970 is
not as large as expected.


Yea, I searched for this one and they said it would be in the 60-70W range. But by the time it ships the power 5 based PPC 980 will be shipping soon after and it is supposed to have 3/4 to half the power consumption of the 970.


Also: http://news.com.com/2100-1006_3-1024439.html?tag=fd_top

Deerfield is due later this year. PPC980 isn't due until next.


Also we have the biggest problem of all, 99% of the apps all of us use still don't run on itanium. When Itanium first came out the economy was still doing good and everyone was like "sure we will port to itanium". Then the economy went in the shitter and very little studios actually showed interest in using itanium. I wouldn't doubt that much of the plans/money for porting went out the window.

There are quite a few content creation software vendors who
have announced ports to Itanium (my guess is that Intel is
subsidizing them).

Sieb
07-12-2003, 05:15 PM
Sieb that article you posted is marketing crap.

I didn't really see that since most of what he said was backed up in the ArsTechnica writeups. Granted he works for a G4 mobo company, the article wasn't bleeding with personal opinon. He even has the Arstechnica article listed for further reference. Opinion or not, it was still a good read for those that didn't want to pour through the pages of tech specs of the ArsTechnica article.

CGTalk Moderation
01-15-2006, 02:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.