Slashdot Mirror
Centrino Mobile Equals Desktop Pentium 4 in Speed
Spy Handler writes "On Wednesday during the launch of its new Sonoma Centrino Mobile, Intel put on a demonstration running a video game on a laptop. It matched the performance of a high-end Pentium 4 desktop running the same game, declared Intel. The contenders were a laptop sporting a 2.13 GHz Pentium M processor, 1GB RAM, and the Alviso chipset versus a desktop with a 3.6 GHz Pentium 4 with hyperthreading, 1GB RAM, and the Grantsdale chipset. Is this a testament to how far the Pentium Mobile architecture has come, or a sad comment on the clockspeed-pushing design of the Pentium 4?"
Intel's finally learning the lesson everyone else knew about 5 years ago. Too little, too late? Or can Centrino save them?
I'll put a Pentium M in my next desktop. Hey, how about dual P-M?
Is this a testament to how far the Pentium Mobile architecture has come, or a sad comment on the clockspeed-pushing design of the Pentium 4?
I think it's a testament to the fact that whatever game they were running doesn't bottleneck at the CPU. Most video games are not CPU-limited beyond a GHz or two.
hmm I would like to know which video game it ran to get equal performance. Also, was the game software rendered or was there a graphics chipset involved?
They are different chipsets, cores and technology, not to mention the 1.5ghz+ difference in frequency.
The wireless device is not the issue here.
> Is this a testament to how far the Pentium Mobile architecture has come, or a sad comment on the clockspeed-pushing design of the Pentium 4?
No, it's probably only a testament on not showing enough info about the benchmark system/conditions to provide any useful technical data, but only marketing data.
Who know? Maybe the game was simply framerate limited by the similar integrated graphics chipset.
I'm not saying that the Pentium M isn't fast, or as fast as a desktop P4; only that probably that demo don't prove that.
Just my 2c.
Bye!
SeqBox
Billy
Centrino is the Pentium-M with a wi-fi chipset. I think the article was meant to highlight the pentium-M's performance versus the pentium *4* performance
I am sure that they got together with NVidia and came up with some crazy optimized drivers (read: cheating).
Sounds like Intel is ready to write off the P4 as done for, and is putting all of thier eggs in the Centrino basket until the launch of their dual-core chips...
...Intel confirms that its desktop chips and chipsets suck?!
If someone says he and his monkey have nothing to hide, they almost certainly do.
that your mature, fast chip that people have been designing for for the better part of a decade combined with spanking new memory and hardware works well! my god, it boggles the mind! carl not like that's stopping me from representin with a wack AMD64 chip and gear that makes the inside of my tower a good substitute for a microwave oven(tea, anyone?))
...with the fan attached or without?
I had the "pleasure" of performing a heavy number crunching on a P4 laptop. Luckily it was winter and one of the rooms in my house is unheated. Leaving the laptop there (temp. about +3C) with bottom lifted off the floor by some books to allow free access to the built-in fan prevented it from entering thermal throttling mode and allowed it to run at full speed...
45 5F E1 04 22 CA 29 C4 93 3F 95 05 2B 79 2A B2
I 'upgraded' from a P3-1 ghz to a P4-2.26 ghz and noticed hardly any difference. I upgraded from my P4 to an Athlon64 3400+ and it not only smokes it, it also has a variable clock speed which only ramps up when needed.
I've been a loyal intel user since the Pentium 90 came out, but after building several cheap and stable AMD systems for friends and family I took the plunge myself, and I'm more than happy.
Hal Spacejock: Science Fiction with Nuts
As an earlier poster mentioned, most newer games depend more on the GPU than the CPU; anything over 2Ghz is almost overkill.
Intel and AMD are in the awkward position of needing to create a market for new processors in a world where a 1Ghz processor will do most office tasks brilliantly. They pushed speed, speed and more speed for so long that the average consumer doesn't give a whit about HyperThreading or anything else. Tech heads and researchers and universities are different, but is that enough to support to very large chip manufacturers forever?
Don't be a looter...and yes, I know that it's spelled with an "A" instead of an "E".
Intel must have not been using a laptop with "Intel Integrated Extreme Graphics", because otherwise it would have been far outclassed by any PC with a decent Radeon 9200 or equivalent. A nice clockspeed means nothing if your video card isn't up to snuff.
If you think it's only using 50% of the CPU (and it's not just windows reporting it oddly), then try running two instances of the same game and seeing how the performance does. Obviously there are other factors here (graphics card, enough RAM), so choose your game carefully. It'd probably be easier just to use some simpleish single-threaded app. Also you can generally turn off HT in the bios, if you're really interested in seeing what effect it has. -J
Slashdot - Mutual Assured Discussion
I guess Centrino beats P4 in speed of falling off the desk as well.
45 5F E1 04 22 CA 29 C4 93 3F 95 05 2B 79 2A B2
Is this a testament to how far the Pentium Mobile architecture has come, or a sad comment on the clockspeed-pushing design of the Pentium 4?
Or perhaps a testament to how fill-rate limited the game was? Honestly, what was the game? Doom 3? Or Monkey Isnald 3? It makes a difference.
Refuse to make a statement in your sig!
They don't really mention any of these factors about the laptop. What good is having good performance if it weighs 10 pounds and has a battery life of an hour?
Monstar L
Its time to do what we used to do back in 1990 before the Pentium arrived, run benchmarks to determine how fast the machine is.
The only interesting thing about using a game as a benchmark is if the thing will run. Its not unusual to find that a game simply does not run on a laptop.
Looking for an Information Security student project suggestion?
Try http://dotcrimeManifesto.com/
I took the plunge and bought an Athlon 64 3700+ laptop (a Compaq Presario). Not only is it cheap, at well under US$2000, but my kids say it plays HalfLife 2 a lot faster than their P4 3.4ghz system with whatever was the fanciest graphics card a few months ago.
I don't see any reason to roll the dice on Intel again based on the price/performance of the Athlon 64 these days.
Cheers,
Here
http://www.heise.de/newsticker/meldung/55276
support Intel pretty well.
If you don't get the German, don't worry, it's all Geek to me.
Now please give me fast disks on a laptop. One gig on mine plays away as a RAMdrive (heavy P'shop load) and it runs way too hot for comfort.
Is this a testament to how far the Pentium Mobile architecture has come, or a sad comment on the clockspeed-pushing design of the Pentium 4?
Yes.
That's part of the beauty of the Pentium M: most of the notebooks based on it are in the 6lbs or less range, and I've yet to see one with a battery life of less than four hours.
As I had mentioned in a previous comment, the front side bus speed is the biggest limiting factor on Pentium M processors. The day we see an 800MHz FSB Pentium M is the day the direct MHz comparisons will apply (i.e. 1.8GHz P-M vs. 1.8GHz A64). Even the Tom's Hardware Guide review of the new Sonoma chipset for P-M shows fairly marginal gains and proves the FSB is the limitation, PLUS they do the stupid thing here and put in DDR-2 which does little for performance but increases system costs.
The poser of the question (that started this thread) signals his ignorance of microprocessor design and underscores what AMD has said all along, and everyone else who hasfollowed the industry since when there was much more competition in the microprocessor, namely you can't juxtpose microprocessors on clock frequency alone. Anyone remember the Intel i860? Or when MIPS was a stand alone company competing against Intel (seemingly). If you say no to these things, that probably explains why you're even pondering this stuff. Nothing to ponder, some of us have known this all along.
Intel has been pushing a "Get the most GHz for you über-boxen !!! Our competitors don't have that much of those !!!" marketing strategy for years.
:
:
So much, that
- It hurts their sale of Pentium M processor, because customer will prefere Pentium 4 "Because there's more GHz inside", as Intel has taught them,
- It will hurt them more when the next generation of desktop processors won't have a pipeline as deep as the current one, and will have less GHz for the same effective speed. (Intel said they wanted to use some Pentium M technology in next generation's desktop chip. And actually, some of P-M's cache hit and branch prediction optimisation are ported into current P-4)
Someone in Intel's department said : "oops ! wrong move we made them start starve for Ghz and now we won't give them those Hz".
What you see now is just one piece off their back-off strategy
- first they changed the name of their processor using model numbers (ala AMD) instead of GHz.
"See ? GHz doesn't matters that much. What matters is to have the latest model !"
- then they make tehcnical demo : "See ? The CPU speed doesn't matter that much as long as you 0ur n3w k001 über chipset. Please buy it ! Please !!! Don't look only on the GHz !!!! You don't need them anymore !!!! Please quit buying AMD !!! Their GHz are lie !!!!"
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
It's a shame all the benchmarks disagree. Have a look at Benchmark
This benchmark also shows that a Pentium M 2.3 (Yes it is overclocked) is as fast as a AMD Athlon64 FX-53 (2.4 GHz)
One area where Pentium M is fantastic is in scientific and engineering simulation software. My company produces such a piece of software called OrcaFlex (www.orcina.com). The code is mainly old fashioned 8087 FPU instructions doing 3 dimensional vector operations.
In the past few years clock speed has become much less important than memory architecture in determining how fast the simulations run. Of current architectures P4 stinks and is comprehensively stuffed by Opteron. However, PM even beats Opteron. Our fastest machine for OrcaFlex is a DELL Centrino notebook! This just edges out our top of the range Opteron workstation.
Has anyone else out there seen anything similar with other applications?
It's a valid point. the question: "Is this a testament to how far the Pentium Mobile architecture has come, or a sad comment on the clockspeed-pushing design of the Pentium 4?"... I'd say the latter. The pentium mobile architecture hasn't come a long way, it's been dragged along by AMD's use of similar technology (hell, and even IBM with its PPC970) to run better at lower clock speeds.
It's a sad comment on how damned long the clockspeed-pushing went on for.
So what.. if Intel is going to compete with AMD, they need to make a 64 bit version of the P-M chip. AMD already has a mobile Athlon 64.
used AMD instead of P4 as comparison, that way in their already unfair marketing, they would put down a high end AMD(with a slower FSB) as being equal to a lower end Intel chip.
Make your computer faster: rm -rf
Here is a link to a benchmark that show that intel might be right.
This benchmark shows that a Pentium M 2.3 (Yes it is overclocked) is as fast as a AMD Athlon64 FX-53 (2.4 GHz) in many games
I call douchebag of the day on you.
/. about the Pentium-M platform and read the arcticles linked therein? You will find many benchmarks that will demonstrate that, yes Virginia, there's plenty of FPS to be gained by a faster CPU. Hell, feel free to read any CPU article that compares game performance.
Why don't you look up the various articles posted on
You do have a point here: for most games all you need is a decent graphics card; I mean, in some cases you don't even need that: fable on the XBox look very good and is fun to play. However I must say that my IBM T42p (dothan) laptop just feels much more responsive than my P4 2.6 desktop, especially for scientific applications having a 2MB cache makes all the difference!
Btw. and testing one overclocked product against several stock clocked products just doesn't seem right.
I now personally look for other stuff, mainly performance per $ and how cool these things operate.
Would someone explain to me the point of having a wifi device inside the processor? Does it vastly improve performance or reduce power consumption? I personally would think a typical addon card is preferrable. If your network changes usb standards, you don't have to replace your processor.
In the good old days, when the P-III was being replaced by the P-IV, you could roughly multiply the P-IV speed by 0.6 to get the comparable P-III speed. (Specifically, a 2Ghz P-IV was about the same as a 1.2Ghz P-III.)
So a P-IV at 3.6 Ghz, is roughly equal to a 2.16 Ghz P-III.
And guess what? The Pentium-M is a repackaged version of the Pentium-III core.
Maybe Intel and AMD and the rest of the world will start using something useful, like SPEC results to market their processors, instead of Hz ratings?
Centrion isn't just the processor, it's a Pentium M, an Intel chipset on the motherboard, and an Intel wireless chipset.
I've read some reviews on the Pentium M, and its power requirements are impressively low. I wonder what kind of performance you could get out of it if you overclocked it to the point where its power requirements were similar to the P4. There may be other reasons why that's impossible, but I'm sure you could still overclock it significantly, and that seems like a resonable thing to do for desktops which can afford the power requirements.
You...are...a...tit.
Centrino is a (collection of) chipsets (like NForce is, like the VIA ones are, etc, etc) A northbridge + southbridge + other chips.
The WiFi is part of the chipset (like the ethernet controllers on most motherboards are these days). Moving the WiFi there is the next step in the process that began with IDE controllers, continued through sound, graphics, USB, SATA, Ethernet and RAID controllers. Now WiFi is added to the list.
Just like ALL those controllers, having itin the chipset doesn't prevent you adding another controller and disabling the on-board one. Lots of people do this with Sound and Graphics.
The wifi isn't in the processor - that would be as stupid as your comment on changing the processor if "your network changes usb standards" - networks don't use USB; you wouldn't want to change USB standard (what to?!?) and neither means (or has ever meant) changing PROCESSOR.
TIT.
I had a professor who still swears by the P3-based Xeon for his work and that it will always smoke anything that the P4 has to offer. Why? Strong integer performance.
The professor I speak of is Bob Hyatt, and his research is on computer chess (specifically Crafty, the chess engine we all know and love). The reason the P3-family of chips does such a good job with it is because of the strength of integer calculations. Dr. Hyatt has repeatedly stated that there is not a single floating point instruction to be had in Crafty.
However, the FP unit in the P3 sucks big time. Intel made a processor with a much longer pipeline in order to improve floating point performance--FP is now world class, but the integer stuff won't be as good as it was with the P3 family. (The shorter pipeline is what makes the Athlon a superior performer in some aspects to this day.) This is why we slashdotters are always screaming that raw clock speed will never indicate the supreme chip.
So Intel did the selection of boards and processors. We have no way of knowing what was done to assign core processes. Wasn't there a graphics board company that wrote their drivers to boost perfomance under benchmarking because of the specific set of conditions? We learn from history, I wonder if Intel did? Me, I will stick with my Athlon64.
All Intel need to do is get off their arse and produce a Pentium-M with an 800MHz FSB and 64-bit support, and AMD finally have some competition in terms of performance again.
So what would I get if I paid for the new P4 over that Centrino? Why would I?
--
make install -not war
Bah. I have a 10 pound laptop, plus accessories, an external drive and a leather case, the whole package comes out to around 18 pounds; HT'd 3.0 P4m or whatever runs games okay; it's just when you need to do anything that involves shuffling a lot of memory around that you notice how goddamn slow laptops are. I proudly walk a mile to work each day with that thing slung over my shoulder. You'll never hear me complain about laptop weight.
...cos I'm too busy complaining about perpetually sore shoulders. I sleep terribly at night, and just yesterday I pulled a calf muscle crossing the street.
But it's all worth it when I ask some sucker to "hold my bag for a sec". Nerddom has its price.
Intel recently dumped a major portion of their roadmap citing heat and power issues. What most expect to see in the future are new chips built on the Pentium M architecture. I think this demo is nothing more than to get "us" all talking about how amazing it is how the M can stand up to the standard P4. Then when Intel makes the transition, the GHz shock is not so dramatic. This corresponds to Intel's move to start code numbering their chips - why in the world would Intel dump their marketing lead of using GHz ratings for marketing and start using obscure codes like AMD? Of course AMD did it because they did not want customers focusing on GHz. When Intel starts using the M as the core, their new systems will take a step backwards in GHz, and that's much easier to hide if all your chips are known only by their codes and everyone already thinks the M core is similar to P4 in performance.
This is way better than the real article!
Centrino is an Intel markenting campaign and not the Intel CPU called Pentium M. A laptop that's called Centrino not only has a Pentium M CPU but also an Intel wireless card and an Intel chipset on the mainboard. It's clever marketing from Intel that makes people think Centrino = Pentium M. Since Intel started this campaign, the sales of Intel wireless cards and mainboard chipssets went up and not because they're especially good but because they are cleverly bundled with the CPU...
PM is not bad, but I just can't use a 64 bit desktop CPU anymore! Don't ask me why I need 64 bit computing, you fool! For using 'int' instead of 'long long', at least :-)
3.6 GHz is more speed than 2.13 GHz!
It's like saying a G5 2.5 GHz is faster than a P4 3.6 GHz, psttt!
Intel said it, not me.
That's part of the beauty of the Pentium M: most of the notebooks based on it are in the 6lbs or less range, and I've yet to see one with a battery life of less than four hours.
My Dell Inspiron 500m (Centrino, first generation) has a battery life of about 1h 40m... Windows or Linux both, and yes, cpu scaling does work.
Pentium-M has 2MB of on-die cache.
Athlon 64 has a cache size of 512kb or 1MB; interestingly 3800+ and 4000+ both run at 2.4Ghz but the latter gains in speed by having a double sized cache... so what if we add more cache?
To effectively compare Pentium-M architecture (by architecture NOT price!) to Athlon 64 we would need a 2MB cache version of it...
Intel was extremely dishonest in its design of the P4. They pushed the pipeline from around 10 stages to around 23 stages. This goes against everything you will ever learn in any engineering class about pipeline design -- as soon as there is a branch instruction (of which there are a ton in x86 code -- it is much less efficient about this than say the ARM), the whole pipeline has to be flushed, so you get a huge latency hit for every condition. The pipeline flush isn't so bad with speculative evaluation / branch prediction. The whole reason Intel lengthened the pipeline is that it allowed them to push clockspeeds up further (because less work had to be done per clock cycle), and people bought CPUs based on clock speed. This gave them temporary headway against AMD, although clock cycle for clock cycle, the P4 was the least efficient processor I think anyone has ever released. More efficient branch prediction, as well as advances like hyperthreading, helped a little, but AMD chips were actually still in the lead performancewise, even though the clock speeds were much lower. The Pentium-M was apparently based on an earlier (Pentium-Pro?) architecture, but wasn't even designed by Intel directly -- they contracted out to a company in Israel (anyone know the details?). That team focussed on actual processor efficiency rather than devious marketing tactics, and hence the clock speeds of the Pentium-M are comparable to AMD offerings, and Intel finally bit the bullet and started using model numbers like AMD had to to originally compete with Intel's FUD. Intel finally acknowledges that it isn't just CPU speed that counts. The Pentium-M is the first Intel processor I have liked since Intel started being evil several years ago.
Full speed that's about right, try clock limiting it to 1ghz or 600 mhz (whatever the lowest clock rate is). That's how you get the impressive battery time. I can get 3 hours straight up under full utilization with a 14.1" screen and 1.5 ghz P-M underclocked. If you run it full speed it's no better than any other cpu for battery utlization. My now dead iBook G4 was the same, about 2 hours at 800 mhz with a 12.1 screen, and about 3.5 underclocked. (obviously a bit better power usage at lower speed than the P-M).
It does run at 600 MHz most of the time.
That really doesn't sound like it, and if so, then you might want to look into if your battery is failing.
The last generation Centrino processor, the Dothan-bsaed Pentium-M, was just as good as games as this new generation. Nothing has really changed. In fact, the Pentium M has always been very good at games, though it really came into it's own with the Dothan core upgrade.
The Pentium M is a very impressive chip. It's not all fun and games though (pardon the pun), it's very good at some things (games) and very bad at others (media encoding).
Well, dell laptops are __CRAP__...
Cheap plastic construction, corners cut at every opportunity, comparable quality but higher price than the cheapest of brandless laptops available from asia..
Theyre unreliable, the battery life is poor, the build quality is terrible, in the cheap nasty ones you cant even remove/replace the cdrom etc, virtually none of them have onboard serial ports, and linux compatibility with dell laptops has been getting worse and worse... One bios upgrade for the latitude 2500 breaks X11 completely by initializing the displaycard in a nonstandard way, doesnt provide any benefits, it was just done to break non-windows os's..
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
For those of us in the laptop market, which one do we buy? I have been looking at the Centrino, P4, the AMD Athlon XP-M and the AMD 64. But the problem is, I can't figure out which one I want. I _think_ that I want AMD, but I am unsure about whether I want a XP-M or and AMD64. Can some explain what an IS major who wants to use the thing for programming and a little mobility would want? Does anyone know of a good site that compares the four in performance?
The views expressed are mine own and do not express the views of my employer.
We need parallel processing chips.
l
http://radio.weblogs.com/0105910/2003/12/01.htm
Something like Merrimac .
The reason a processor with half the clockspeed shows the same performance is that both processors just wait for the magnitude slower clocked memory. Why waste 1000$ for the current top Intel Chip? A 500MHz CPU does the same job when working with the same RAM/Chipset. Try this: throttle down the CPU clock while keeping memory at full speed (if that is possible with your system ;) - you won't notice a performance drop for quite a while...
If you remember your history, the Pentium-M is based off of the Pentium-III. You know, the same P3 that kicked the crap out of the P4 when it came out, despite the fact that the P3 was topping out around 1.0Ghz and the P4 started at 1.4Ghz.
Unless you mean that the the Pentium M has no hyperthreading, which I suppose is a valid point, as hyperthreading is about all the P4 has going for it right now.
I really doubt that the difference is as large as a minute or two and an hour. Even when comparing my Athlon XP 2Ghz and my P4 1.5Ghz, the difference in speed is only about 15-20% in favor of the Athlon. Noticable, but not a big deal to me.
I have a PIIIm laptop - which time and time again is faster than the new P4 machines I work on at the computer repair shop.
Google cannot tell me which tools are good. An experienced user, can.
Is this a testament to how far the Pentium Mobile architecture has come, or a sad comment on the clockspeed-pushing design of the Pentium 4?
The Pentium III has veen embarassing the Pentium 4 as long as the Pentium 4 has been shipping. This is merely another act in the continuing Greek Tragedy that is the Pentium 4.
Well duh. I already have read reviews where the much cheaper, much cooler, slower clock speed Pentium M has kept up with or beaten P4/P4EE/A64/AXP systems. Its just funny that Intel has finally admitted it.
I wonder if theyre going to give a public apology for the P4. What a crappy design. Since the Pentium M was an offshoot of the P3, they really should be sorry that they pushed their inferior designs when they knew they had better.
On a side note, I love AMD, but my next desktop is going to be a Pentium M. Maybe well see dual boards by then...wow.
When the revised DFI 479 board comes out, it will SMOKE AMDs and Pentiums alike.
Slashdot "libertarians": Small government for me, big government for those I disagree with. -1, I disagree with you
No, if you look at: http://www.anandtech.com/linux/showdoc.aspx?i=2308 &p=9
The compilation speed (a task which generally involves a lot of context switching) on the Pentium-M is simply horrible. Both the K7 and the P4 are much more capable (with the P4 more so due to Hyperthreading). Comparing a P3 to a P-M is a very very bad call as about the only thing the two share is a similar execution core. Banias has a lot of things the P6 core didn't and I'm guessing one of those things are causing the poor context switching performance.
Lol nice link in your sig. :)
"We invented personal computing." - Bill Gates
...does it run Lin--ah, just about everything does.
But, does it run Racer --ugh, runs that too...
BUT, does it use x86-64? YES, something it can't do! But then you all knew that.
You can hold down the "B" button for continuous firing.
Well, it all depends on what you're using it for, right? From the same benchmark, the floating point performance still isn't all that good - in ScienceMark an Athlon64 at 2.2GHz still beats a Dothan OC-ed to 2.3GHz. OK, you're going to say Dothan is not about FP, but you have:
1. a (marginally) higher clock on the Dothan
2. 4 times more cache on the Dothan (2M vs. 512k)
3. A64 running in 32bit mode, which means 8 SSE2 registers instead of 16 (which would have given its top performance if optimized for that)
4. OC possibilities (apparently the 90nm A64 is a hell of an OC-er too, as people have had it jacked up to 2.4-2.6GHz on stock cooling)
Still, nobody is going to deny that Dothan has a heck of a strong performance on average. For many things it could well be the best tool for the job, if only Intel would drop the price and maybe someone else (nVidia? ATi?) would make a cheaper desktop chipset for it.
The point is that they are only able to finish the workload in the time wich is given by the effective memory bandwidth.
When the CPU waits 200 cycles on average to go to memory and back it performs exactly the same amount of payload effective operations than a CPU with half the clockspeed wich still waits 100 cycles to go to memory and back.
DMA is limited by the RAM speed as well.
The @home applications may be more memory bandwidth agnostic than everything but they weren't my intention when I bought my gear.
In addition to the other replies defending Intel's choice of DDR2, note that the JEDEC DDR1 spec's top speed is DDR400 while DDR2 is now up to DDR2 533 (and will go higher). Single/dual channel DDR2 533 matches/sychs up perfectly with the 533MHz bus of the new Pentium-M CPUs.
Also, the chipsets that use Intel's integrated GMA 900 graphics "shares" (steals) memory with the CPU. Therefore, the CPU and the GPU can use all the extra memory bandwidth they can get. Dual-channel DDR2 533 gives 266MHz more memory speed than dual-channel DDR400.
TO START
PRESS ANY KEY
Where's the 'ANY' key? I see Esk, Kitarl, and Pig-Up...
Imagine how smooth Doom 3 could be if Intel would finally address the effective memory bandwith issue. But go ahead and waste your bucks (or your daddies) on pooly designed hardware.
It may surprise a lot of people that CPU speed is still a huge determining factor for 3D gaming performance.
Some people might look at a game like World of Warcraft and jump to the conclusion that since it doesn't appear to have all the latest 3D eye candy, it must not be CPU-bound. But the reality is that the WoW client is heavily CPU-bound, like most of the 3D MMOG clients out there. When you have to manage several dozens of players on-screen, plus the hundreds of static meshes that comprise the scene, it's extremely difficult to keep even a mid-range GPU well-fed with data. You can also take a look at Doom 3 or FarCry for good examples where even a top-end CPU simply cannot keep up.
Game performance isn't a black-and-white deal where it's either the CPU or the GPU that's the bottleneck. It's almost always a combination of both at various stages of rendering. I do this stuff for a living, and I'll take anything extra CPU performance I can get. Faster CPUs mean I can get more characters on the screen, more complex scenes, better NPC behavior, etc.