Slashdot Mirror
Mobile Processor Showdown
AnInkle writes "The Tech Report has a head-to-head comparison between the Pentium M760 and the Turion ML-44. From the article: 'AMD has done well with Opteron in servers and the Athlon 64 in desktops, but surely AMD's K8-derived mobile competitor doesn't match up with the Pentium M. Does it?' Conventional wisdom (or marketing genius) says Pentium M's power-saving features and performance-per-watt leave AMD's Turion 64 gasping for batteries. Even though the next-gens are just around the corner, countless mobile systems will sell with these chips over the next year; find out which to choose, whether for performance, battery life or a combination of both."
Am I the only one running into relatively new laptops that overheat? I shouldn't have to keep the bottom elevated on a hard surface to keep from crashes.
I'm waiting for vast improvements in battery life before I decide. I've waited 10 years. I can wait 10 more.
They mention in the beginning that MTs are lower power than MLs (they are 25W vs. 35W T.D.P. in fact), yet they didn't throw one into the comparison.
That's lower power, and faster, than even the infamous Core Solo (T1300 1.66GHz 27W TDP).
There is a 1.666GHz Core Duo LV which is lower power. But, if you don't have much use for dual-core, AMD seems the way to go.
With all the talk about AMD not yet on 65nm it would seem AMD is still, not just competitive, but ahead of Intel in low-power CPUs, and performance. (It seems like nobody is talking about the benefits of SOI, for some reason)
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
personally, i would settle for a 700MHz PC (running linux of coarse) that lasted until the cow came home (it's a small farm :P ). it would probably be best to use the ARM archatecture though. im not interested in running Quake 4 when all im doing is writing a college paper, browsing and chatting... isn't that what a desktop computer is for? well, that and adding fancy lights, a see-through panel, oh... and not to mention the harddrive with a window. :)
Anons need not reply. Questions end with a question mark.
"Turion 64 test system consumed a third again more power than the Pentium M system at 100% CPU load".
I guess there is always a price to be paid for more performance.
I have to say that I'm impressed with AMD's offering here. For a while I've been under the impression that they were being trounced by Intel in the mobile market (which I'm sure they are in term of sales). However, this review shows that they have improved to a point where they are pretty much on a par with Intel.
They still have a bit of work to do with the maximum power consumption, but they've managed to get the idle consumption down to where the Pentium M is with similar overall performance. Good work AMD.
it's somewhat pointless as they don't compare the best of each company's current offerings.
This test pits a Pentium M against a Turion 64. Granted, this was the comparison for 6 months ago. But Intel now has the Duo Core (Yonah) processor which has a slightly different architecture than Pentium M.
Not to mention that while Conroe and Merom will be based on the same design principles it is a fresh design.
I believe the key to Intel's new design will not be its close approximation to the Athlon in performance. The secret is in performance per watt, as they say. High performance computing with as little engergy consumption and heat dissipation as possible. The Athlon 64 architecture looked cool compared to the toaster oven called Netburst, but even against the old Pentium III it is quite hot and hungry. Lifestyle PCs, laptops, and blade servers will all favor the much cooler design from Intel.
While Athlon 64 will continue to compete on performance and price, without a major architectural change they will be stuck in the hot seat for the next couple of years. And it will only get worse before it gets better because Intel's chip design is truly superior, only held back by a dated bus architecture slated for replacement in 2007.
While Intel will "win" technologically, they will burn a lot of capital to remain competitive until they do. Lots of Pentium M chips have been stockpiled. By the time Yonah reaches mass production it will be replaced by Merom. Lots of stockpiled Pentium D chips that will be replaced by Conroe. Intel will need to slash prices for processors nobody wants anymore only to flood the market with brand new PC's that don't need to be replaced by the superior technology they so desperately need to release.
Maybe Intel will sooner push the P4's into a landfill than cut their own throats? Or maybe 2006 will be a good year to start up your own server farm in the basement.
There has got to be something wrong with the new Intel chips Apple is using. No one from Apple wants to talk about battery life. After all the talk about 'performance per watt' something has to be seriously wrong with power consumption.
It is too bad Jobs refused to pay for a mobile 970 chip to go along with the killer quad-970 workstations they are shipping right now, Apple would be rocking right now instead of plummeting in the stock market.
The chip to compare to the Turion is the Intel Core Duo. It has superior floating point performance to the PentiumM, SSE3 support, and two cores instead of one.
In other words, it crushes the Turion.
hmmm, an intel performing more work per clock cycle than an equivilant amd chip, while using only 3/4 the watts under load! what a reversal from the norm. AND the intel chip is faster in gaming!?
i disable sigs
It has never been advisable to run a bottom vented laptop on a plush/cloth surface that could restrict the flow of air to the vents. What happens if you cover all the holes in your PC case? Eventually it will overheat.
My Pentium M laptop rarely has heat issues, but I do have several customers with Pentium 4 based laptops that will char your thighs. Ouch.
But it didn't really come out ahead in overall performance. Plus it costs almost $70 more than the Pentium M it was compared to.
There's a lot of AMD pole smoking going on in the comments and it's starting to make me nauseous. I love AMD's desktop processors, but I'm in no way a brand loyalist. I can't stand the thought of buying an inferior product based on brand. The Pentium M still comes away with a lead in this test when you factor in the cost difference and power consumption.
The AMD processors embed the DDR controller into the processor. There is no FSB to speak of. The AMD Northbridge takes the hypertransport from the processor and splits it to PCIe, SATA, and your other peripherals. The DDR core that is embedded with the AMD processor is a heat hog, but at least there is no FSB, which is worse. The dual core Opteron has two memory channels on the processor.
The Intel processor does not embed the DDR controller. The DDR controller is part of the northbridge for both single and dual core designs. There is an 500-800MHz front side bus connecting the proc to the NB in Intel arch.
To properly compare to AMD power consumption with Intel, you have to compare the both processor and the chipset. These fundamental differences make direct processor power comparisons meaningless.
The reason they chose these two chips was that they wanted to test the AMD chip against an Intel chip which cost about the same.
It seems to me that AMD will price their chips so they will sell. If they have to price their chips at a tenth of what Intel charges for its best chips, they will. If they can't quite match Intel's performance, they will compete on price alone. Mind you, they won't give away the chips just for the joy of it. They will price them where they produce the best profit.
In any event, most people have all the computer performance they need. These days, most people won't pay double for a faster machine (gamers excepted). The bottom line is that there will still be a market for AMD chips.
Here is an interesting battery life benchmark at tomshardware:
d uo_notebooks_trade_battery_life_for_quicker_respon se/page21.html
http://www.tomshardware.com/2006/01/16/will_core_
I haven't heard much about the VIA C7 Chips, which were supposed to perform a bit below the Pentium M's, but were also supposed to have extremly low power consumtion and heat output. They are supposed to have come out by now, but I have yet to hear anything other than the general specs articles.
If you've never heard of the C7 this article give a pretty good outline of it:
http://www.pcstats.com/artvnl.cfm?articleID=1833
I'm not looking forward to it mostly because the socket has changed, so i can't upgrade my turion based laptop :(
SearchIRC - Now with live chat directory!
It's pretty annoying that they put the NVidia card in for some benchmarks, and then didn't just LEAVE IT IN. They explicitly mention in one test that they think ATI graphics are having trouble with OpenGL (no shock, ATI drivers have sucked rat fur in this area for many years). Instead of twigging to 'hey, let's set this to be as fair as we can', they just accept the screwed up results! That's really dumb... they're not thinking it through. They claim to be testing the CHIPS, not the LAPTOPS.
I get so frustrated with benchmarks in general... they so often miss really obvious stuff like this. If you're trying to test a CPU, then you do your best to remove as many other variables as possible. Use the same damn video card. Test what you SAY you are testing. Sheesh.
I think it would have been interesting to see power consumption scores both with and without the NVidia card, too. It'd be nice to try to separate the video power requirements from the CPU/chipset requirements.
Amd can have a nice offering, the laptop with AMD cpu are a little less easy to find and don't have the others things that made them shine !
Ceci n'est pas une Signature !
I must say that these recent advances in chip performance vs battery life have definitely made laptops more viable. There's still plenty to be desired of course certainly we are ages away from seeing a laptop that can seriously handle a video card and chip set worth playing up to date games on. Over all for business users I believe the laptop is truly coming of age. 7+ hour charge times on normal use is nothing to sneeze at and it's what I typically see on mine.
I have ALOT of experience with Laptops, I have 3 personal laptops and 1 provided by my company. I thought it might be helpful to some if I posted brief notes about two of them.
HP ZD8110 - 3Ghz P4 HT, Radeon X600 PCI-E 128MB, 2GB Ram, 17inch Widescreen, Ubuntu Breezy - This is a workhorse. It does overheat periodically if it does not get full venting from the three bottom mounted fans. It has even overheating during an overnight compile session once or twice. During heavy use acpi -V shows me CPU temps upwards of 67C. Battery life is rarely more than 30 minutes. 200watt power supply can't even run off of most dc-ac car power inverters. I do love this laptop, It is my main computer. But these powerful laptops are what they are.
IBM Thinkpad t42p - ~1.83ghz Pentium M, ATI FireGL, 2GB Ram, Windows Server 2k3 - I do all my Windows development on this laptop. During a normal days use I typically have 3 or 4 copies of Visual Studio running, a few instances of Visio, SQL Server 2005, SQL Server 2000, mySQL, IIS, and Apache2. This laptop stands up to a massive workload and never gives up, it always runs perfectly cool, and in power saving mode I can get as much as 7 hours of battery life using all the extra-size battery options available from IBM. I think the huge cache on the Pentium M helps this machine deal with so much multitasking, and I often feel it is more responsive than my P4 laptop, although obviously slower for floating point intensive work. This thing is light and easy to carry around. It is great for travel.
I guess the point I am trying to make here is that there are pro's and con's to power intensive and power saving styles, hopefully the few details I mentioned help others make a decision.
Big ones, small ones, some as big as yer 'ead!
Give 'em a twist, a flick o' the wrist...
Any self-respecting geek that's flying to Japan has an extra battery or an inline APC battery, etc. The power consumption life only counts in SoCal and Houston where they pay $1/kw. Heat dispersion is another thing. Got a nice fan? Good. You need to blow that stack with some cool breeze.
Where does it matter: bang for the buck. Both seem to do that. AMD has better math, but this is no surprise as their FPU has eaten Intel's for years now. Bad memory moves? Ah yes, that damn FS bus. Sigh.
The bucks? The same, roughly. And now that dual cores are here, as mentioned above, it's all irrelevant for those of us that must fly with nitro-injected rendering machines or compilers from hell. Multi-core or die. Mobetta cores, please. I got juice, and I got code to burn. Get out of my way, slow stuff.
---- Teach Peace. It's Cheaper Than War.
The two processors represent two quite different approaches to getting the most performance. The Pentium M has an enormous cache and good memory bandwidth. The Turion, on the other hand, has much better memory latency as well as AMD's traditionally strong scalar arithmetic.
The benchmarks come down to:
If the code and data fits in Pentium M's cache, Pentium M wins hands down.
For tasks like media encoding, where the problem doesn't fit into PM's cache, Turion wins hands down.
If you are spending much time at 100% CPU usage, Pentium M will give you better battery life.
Oh, and games? Both suck about equally well. If you want to play games, get a desktop.
Centrino is not a processor, but a group of Intel technologies bundled under one brandname.
Not only that, but isn't comparing the Intel Core Duo to the single core Turion like apples to oranges? Single core vs. single core makes for an even comparison.
Sure the new laptops are thinner, lighter, and use less power but there is a drawback. PERFORMANCE SUCKETH! The powersavers are especially slow and underperforming. The only decent laptops are the battery draining monsters with the full size heatsinks, real video cards, and faster harddrives.
I am not entirely sure why people even keep buying laptops with hotels now offering Internet kiosks. Why lug a laptop, have to show it to homeland security at the airport, then worry about it getting broken, damaged or destroyed just so you can run e-mail, excel, and word?
Incidentally, I think a laptop is one of the few purchases in which the value of the item depreciates faster than of a new car. That's impressive.
I suppose it's simply an inherent fact of capitalism that even if (and I have no doubts WHEN) AMD produces a more efficient cpu for laptops it won't help them nearly as much as it should. Intel simply has all the hardware contracts with OEM manufacturers and the enthusiast base for do-it-yourself laptops isn't remotely comparable to that of desktops. Besides brownie points among those in the know (of which there are few) what does AMD have to gain from putting so much effort into mobility processors? They're still trapped in their much smaller market. The average user has no clue what Centrino means. They see Ghz and pass judgement. The average consumer (and thus, the almighty dollar) tends not to understand that not all Hertz in the CPU world are created equally.
The author assumed that a notebook CPU runs with 100% load. I have two applications for a notebook: office stuff like writing a message or reading a document, or playing games. Even the latter hardly requires 100% CPU load all the time. For these applications I find a cheap notebook with a software solution like CPU Idle quite adequate. Why spend more money on "mobile" processors? CPU Idle also works fine for desktop PCs.
Suriously i have never heard of these before
TO START
PRESS ANY KEY
Where's the 'ANY' key? I see Esk, Kitarl, and Pig-Up...
I thought it was great when the smarter part of the PC community refused to play along when Intel offered a poor solution for the customer by trying to make everyone use RDRAM. AMD took the lead on price/performance, and Athlon was the big thing.
When Intel finally freed themselves from the RDRAM shackles, they debuted their 800MHZ HT chips and showed everyone that there it was possible to get higher performance with only moderately higher power levels.
But then when Intel went to their 22-stage pipeline power-hog disasters, the community did the right thing and moved to Athlon 64 and X2. AMD was providing higher performance at much better costs and using less power.
The community's move to AMD's superior solution spurred Intel to make a huge change in their strategy, abandoning NetBurst (P4) and moving to a much better solution.
When the community follows the best solution, the industry has responded.
Which is why I find it baffling that people let the wool be pulled over their eyes on AMD's mobile offerings. They back AMD unconditionally against Intel and make excuses about it too.
Ever since the Pentium M LVs and ULVs, AMD has not been able to keep up on performance/Watt. And if you compare the most recent offerings from both companies it is abundantly clear.
So I say please, make the wise move. Continue to back the company that is making the right moves. And that seems to mean Intel for low-power solutions and AMD for high-performance solutions. It could change at any time, so keeping informed is essential.
http://lkml.org/lkml/2005/8/20/95
Intel beat out the AMD in nearly every performance test. They try and make an argument for the AMD on power consumption. I'll paraphrase: The AMD chip uses less energy in an idle state and since most usage is idle, the AMD chip uses less energy.
AMD is clearly the overall winner if you don't use your computer.
http://www.askthevoid.com
Seems like they are all labeled a mobile workstation or a notebook, precisely from this reason.
There's a big difference between a dual-stream processor (like SUN's new multi-stream offering) and a true dual-core processor. But a dual-core processor works almost exactly like two separate chips, just in a single package. In fact, for AMD, they are exactly the same, for Intel, the dual-core is a bit better off than the two separate chips since they share cache better than two separate Intel chips (but less well than any AMD offering).
As to the "more under the pedal" stuff of the GP, I can see why you say that, but it's really because the dual-core machine cannot hand all its horsepower to a single process even if it wants to. A single core chip can do so, and will in the case of a single CPU consumptive task. An OS could be designed to never hand over all the CPU to a single task and then a single core would have "more under the pedal" too. But it turns out to generally reduce performance overall.
I have had several single processor machines and several dual processor ones. I have never felt like I would never want to go back to single processor. Dual processor is nice (my current machine is dual core) but until recently, dual processor (core) just didn't make financial sense. A single core has almost always been much more cost effective than two slower processors because the two processor setup not only requires two chips, but also requires specialized motherboards (and recently big power supplies too).
But with affordable dual-core single-chip solutions that fit on run-of-the-mill motherboards it seems pretty likely that I'll have more dual-core machines in the future.
http://lkml.org/lkml/2005/8/20/95
On the other hand, reducing the power consumption can be done in good ways and bad ways. One of the laptops here is nice, although it can't be used on a lap because it gets so hot. The battery life is decent on paper, but in reality you can't use it much on battery because the system slows down to a crawl to conserve battery life. I'm sorry, but if I wanted a 1999 class laptop, I would have bought one, and not this $1500 behemoth that pretends to be a desktop when it's plugged in, and otherwise is slower than a PIII.
Why do I need that power? Because software up until now has depended on Moore's law, and require twice the processing power every six months to do the equivalent job at equivalent speeds. Sure, you can keep on running six years old software too, but then you could just as well buy an old laptop on eBay too.
The first few Pentium Ms were good. That's an extended P3 that doesn't draw much power when running at full speed. The problem with the newer M's and Turions are that they doesn't require much power when running like molasses. Sorry, I don't buy that as efficiency.
Dhrystones per watt, now that would be a more useful benchmark.
--
*Art
It is too bad Jobs refused to pay for a mobile 970 chip
Who was selling this mobile 970, dare I ask?
One thing the Turion's have that the Intel Mobiles don't is 64bit support. Is that an issue? For the average user, probably not for a while. Although, I imagine that a few corporations will try to start standardizing on 64bit hardware in (a long) preparation for 64bit software.
I agree, it's too bad they didn't test the Turion MT. These benchmarks conclude that (1) the Turion ML-44 "only" offer performances similar to the Pentium M 760, and that (2) the Turion system consume as a whole a third more power than the Pentium system at full load (but, good news, a bit less at idle). From my point of view, in order to improve (1) and (2), I would:
But isn't the above obvious ? Why neither the article conclusion, nor slashdot readers have pointed this out ?
I better get some sleep. :)
"Enjoy what you're doing! If it becomes drudgery, you're doing it wrong!" - Jim Butterfield
They were comparing a 2.4 turion to a 2 ghz pentium M. Not only did the pentium lose by small margins (as i think it should being the slower chip), and had considerable power gains. I am hear sitting in anticipation waintg for my new dell inspiron 6000. I was a little worried about getting the pentium M, but after these results, the turion will be a better choice in the future. AMD is going to kill Intel, just not today.
By 800MHz HT chips I meant 800FSB HT chips. The standard at the time was 133MHz for a memory bandwidth of 1.0GB/sec. (AMD was stuck at even less!) With the 800FSB (200MHz quad-pumped) chips and dual-channel DDR RAM, Intel opened up the bandwidth floodgates, providing 6.4GB/sec.
This made a huge performance difference in applications that needed bandwidth, like gaming.
Hyperthreading was a help, it wasn't the boost Intel made it out to be, but it was helpful in getting the most out of Intel's long-pipeline chips.
http://lkml.org/lkml/2005/8/20/95
or which laptop runs linux better? i'm not kidding. what i have noticed that centrino laptops have a much better support of linux than amd laptops. and that would definetly put amd in a bad position, for me at least.
:-(
when you get a centrino laptop you get the intel wifi card, which is pretty much supported. you also get an intel video card, which AFAIK runs 3d with open source drivers (can someone confirm this?).
on the other hand you've got an amd laptop with different brands of wifi cards, which range between having totally open source drivers or shitty proprietory like broadcom (you gotta use ndiswrapper). and you're stuck with either nvidia/ati/sis video card which aren't open source friendly.
i am currently in the market for a laptop. i do believe that as an architecture/performance/price amd64 is better, however i'd be stuck with a pain in the ass laptop to configure to run linux. i'd rather pay more money for a fully functioning laptop than having to rely on ndiswrapper and buggy closed sourced video drivers.
running knoppix on a centrino laptop of a friend successfully configured everything, even chromium ran without a hitch (thats why i assumed 3d worked with open source drivers). which was cool showing off to the guy that everything worked without the hassle of installing drivers
i'll be getting an acer centrino laptop, BECAUSE IT COMES WITH LINUX NOT WINDOWS XP. fyi acer in the middle east have been bundling linux with pcs, a distro called linpus (kinky name), for a while now. although i'm pretty sure that 99.9999999% of the population reinstall a pirated xp
Okay, we've seen the benchmarks for 32 bit operation. Let's see what happens when we move to 64-bit stuff.
Oh, wait... the P4M doesn't do that. Guess we know the winner there.
what is killing batteries faster is the screens. Throw in CD and DVD drives which some people seem to think run on magic and you have the recipe for short battery life. One industry that needs to wake up is the game industry which loves to have CD/DVDs that must be loaded to play and are accessed during play. Hello!
The processors already have technology to slow down when not being used to their fullest extent. Now what is needed are methods to reduce reliance on spinning up the harddrive; or make it more efficient; and new screen technologies that are not as energy intensive.
Would you put up with more ghosting of the display for longer battery life? Most probably would not as they would associate that problem with inferior technology.
* Winners compare their achievements to their goals, losers compare theirs to that of others.
> battery life should be equivalent to the Powerbook G4.
That's unacceptable! The 70 minutes my 17" PowerBook lasts is ridiculous. When flying, by the time I get the thing out and can start working, the battery is almost spent. I was hoping Apple would try for once to make a usable high-end laptop. My old 12" iBook would last over 4 hours. It was a much better laptop and about 1/3 the price. So instead, Apple has decided to make a portable desktop again. Thanks for nothing. I bought my first Apple in 1981, and I try to continue to support them but they make it very difficult.
There are a lot of comments about "the Pentium M came out ahead in most benchmarks". That's true of the graphics benchmarks using an AGP video card. (Which is curious given the speed difference of the FSB in each setup.) I do have to wonder how this is relevant in a test that is supposed to be comparing laptop performance, where there isn't generally an option to upgrade the graphics. Maybe this suggests that to get a relevant comparison between laptops, you need to benchmark the actual systems you're comparing - duh!
The "cpu only" benchmark results are mixed. The 2M cache on the Pentium looks like a real win, but the AMD still 'wins' a number of tests. As always, the conclusion seems to be 'you need to test with your own applications to see which will work better for you'...
Don't push your fancy thermodynamics on us!
Am I the only one that noticed that ALL of these tests were really for a 32 bit system, with the AMD only running in 32 bit. I have an Athalon notebook (not sure which processor). HP admits on the tech site that they don't have ANY support for 64 bit. On an unsuccessful Linux install, it screamed, right up until it hung. The 64 bit performance was about 4 times as fast for those tasks. To really be useful, it would be nice if the comparisons were to take into account a few 64 bit apps. Too bad Windows only exists in 32 bit, and the manufacturers don't have any drivers for 64 bit.
As a 32 bit system, you are wasting over half of the processing power, and it was still a very close comparison.
Pentium M system at 100% CPU load, but unless you're using your
laptop to
For typical use, it seems likely the Turion 64 would be competitive
with the Pentium M on the battery life front.
It would be nice to test, rather than speculate, on "typical use" battery life in a laptop review. They missed:
1. The companies buying $1500+ laptops often deploy them as desktop replacment units.
2. These companies don't want their employees waiting for machines to update screens.
3. Why bother with wireless mouse or WPA if you are mostly connected to an AC adapter?
You replace 800MHz P3 desktops in remote locations with 2400MHz machines since full load performance counts.
Old winbond southbridge... no USB2.0 or firewire. :-(
Kinda necessary for a portable IMHO... USB1.1 is dismally slow.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
Because they are measuring power consumption at the wall. At idle, the CPU is only a small part of the total power consumption, you're mostly measuring the chipset power. And on desktop chipsets. Laptop chipsets not only use less power, but use even less power when they aren't plugged into the wall (by reducing performance somewhat). There is no way to do an accurate on-the-go total system idle power measurement on a laptop by plugging it into a Kill-A-Watt (as much as I may like them) and measuring the draw from the wall socket.
The point of measuring idle power would be to get an idea of how long your batteries would last if you weren't plugged in. These numbers cannot be used to determine that because the chipset is in "plugged into the wall mode" and in fact, as a desktop chipset, it's unclear if it even has a power saving mode.
So I wouldn't place much creedence on the idle test. The full bore is test flawed a lot too, but due to the large size in CPU power consumption a little bit of info sneaks through the noise.
Review sites are going to have to adapt, to measure power consumption in a more useful fashion, especially if they want to evaluate laptops.
BTW, the Intel chip is $70 cheaper. Although I dunno how much that matters since you must buy each chip with its own specific chipset and we don't know how much those cost. If I had to decide, I'd pick whichever total system was cheaper I guess.
http://lkml.org/lkml/2005/8/20/95
What about the sleaze factor? Intel is winning that one hands down.
Yes, the most important difference between the Turion and the Sonoma platform is the optimized chipset. The most important optimization Intel made was dropping DDR for DDR2 on these mobile platforms.
Why is it such an improvement? Because DDR2 offers two obvious power improvements over DDR1:
1. The voltage is lower (1.8v versus 2.5v). Since power consumption is proportional to the voltage squared in CMOS devices, this should reduce power consumption by almost 50% (ignoring leakage).
2. The internal design of DDR2 adds more buffers (4 versus 2 for DDR1), so the memory inside runs at half the speed of an equivilantly-clocked DDR1 chip. This means that, with the exception of external control logic, DDR2 667 chips are running as fast as DDR1 333 chips. Thus, you can get higher throughput (at the expense of increased latency) with very little operating frequency, which means little increase in power consumption due to frequency. The Pentium M hides that high latency well with that huge low-latency L2 cache.
Here is a link discussing the two technologies in detail.
Personally, I'm looking forward to this year, because AMD knows they need a low-power dual-core solution for the mobile platform, and they know they need to reduce power consumption even further on the desktop to stay one step ahead of Conroe. DDR2 will help them deliver that.
Man is the animal that laughs.
And occasionally whores for Karma.
My understanding is also that the dual core P4s are just two separate cores in the same chip (on the same die)?
The Intels that are better off together instead of separate are the Yonahs (Core Duos). They can signal to each other a little better, and if you turn off one core (for power), the other can use its cache as additional cache.
But no Intel does as good as job as AMD does right now in terms of interprocessor communication and cache utilization.
http://lkml.org/lkml/2005/8/20/95
It seems to me that all the HTPC systems posted to Slashdot recently are suped-up hot-rod systems that are completely inappropriate for home theater purposes. I built my own desktop Turion HTPC recently, and it worked out very well. If anyone is considering a desktop Turion, I highly recommend it.