cache memory in AMD vs Intel

[kpit]

Hi,

I have a confusion.
While reading many of the threads here; many people suggested AMD quad core processros as best when compared to Intel processors.
But while reading the details of the processors, I see that AMD processors has less chache memory than Intel processors.
I feel that cache memory will help lot for performance specially in case of gaming and grafics.

Does any one can give more insight into this....?

 

[corElement]

AMD phenom II x4 965 be = best vfm gaming cpu combined with a 890gx chipset mobo running on a 5770 gpu. advantages is usb 3 sata 3 but you dont even get hardware which uses that but for 7k those mobos are a blast

i5's on the other hand are better than x4 965 be but their mobos are not feature rich
i5 cpu = 10kish
p55 mobo = 12kish

i7's own everything atm but to own one properly it costs 30k for mobo and cpu.


Also, what do you mean by gaming and graphics, graphics is a very broad term, what are you going to do. video editing, animation, multi thread applications? what? depending on your use there are cpu's which are practical for you, however dont bother about 1 single thing like cache in understanding the cpu.

 

[gobble]

Hi,

I have a confusion.
While reading many of the threads here; many people suggested AMD quad core processros as best when compared to Intel processors.
But while reading the details of the processors, I see that AMD processors has less chache memory than Intel processors.
I feel that cache memory will help lot for performance specially in case of gaming and grafics.

Does any one can give more insight into this....?

you get amd cpus with both 512kb per core and 1mb per core.

AMD has a reputedly better cache optimization algorithm that reduces the negative impact of a cache miss for the instruction set than Intel. 1mb per core is good.

Cheers

 

[spirovious]

Intel markets on L2 cash & Amd has both L1 & L2 cash.
Theoretically L1 is faster than L2 & practically AMD can be faster than intel.

eg Intel C2D has 32kb L1 & Amd some CPU has 512KB L1, here AMD is faster.

 

[asliarun]

Intel markets on L2 cash & Amd has both L1 & L2 cash.
Theoretically L1 is faster than L2 & practically AMD can be faster than intel.

eg Intel C2D has 32kb L1 & Amd some CPU has 512KB L1, here AMD is faster.

Actually, that is not the whole picture. Historically, and as it stands today, Intel has always given much better CPU cache than AMD. The reason for this was that quite a few years ago, AMD migrated from the front-side-bus architecture to a point to point interconnect architecture that was both higher bandwidth and lower latency.

Remember, cache exists only as a buffer, like a bucket to store the data flowing into a CPU, so that slower interconnects and I/O will not result in CPU having to twiddle its thumbs while it waits for the data to arrive.

Intel, on the other hand, has traditionally had a much better manufacturing process than AMD, which allowed them to throw huge amounts of cache in their CPU so that they could retain the slow front-side-bus interconnects, while still remaining competitive with AMD in terms of performance.

In the "i" series of CPUs, Intel has upgraded to a very good point to point interconnect architecture itself (which is actually very similar to AMD's interconnect). However, in addition to this, Intel still retains its superior manufacturing process, and have also managed to do a very very good job of designing the CPU correctly. This is the reason why across the board, Intel is completely dominating AMD in all counts, performance, latency, scalability, efficiency and performance-per-watt.

If at all, Intel has priced its CPUs at a bit of a premium as it always does, and this has allowed AMD to survive, mainly by undercutting Intel on price and selling their old architecture CPUs at essentially throw-away prices to retain market-share.

Anyway, back to cache, the cache that Intel fitted in its Core2 series of CPUs was actually one of the finest caches made by either Intel or AMD, and was and is much superior to anything AMD produced. This is actually the reason why Core2 whipped Athlon in performance even though Core2 was using an FSB. The "i" series cache is actually larger, but has a higher latency.

As far as OP's question goes, cache is a minor aspect to consider when it comes to CPUs - if performance is really a concern, then you should look at building a "balanced" system that contains no major performance bottlenecks. To use a car analogy, fitting a 500hp engine on an Ambassador might have its wow factor, but it will not win any races in a circuit.

As things stand today, AMD is a good option mainly because it has very reasonably priced motherboards, and remains competitive with Intel because of its low prices. An alternative to consider is Intel's i3 or i5. In any case, if you are going to do graphics intensive work, what you really need is a powerful GPU or graphics card, and a really fast hard drive (or multiple hard-drives in RAID configuration, or better still, an SSD). An average CPU will do.

 

[spirovious]

I have even read in past that intel was good at multimedia purpose(ending to Divx etc)
AMD was better for games.

 

[gobble]

Actually, that is not the whole picture. Historically, and as it stands today, Intel has always given much better CPU cache than AMD. The reason for this was that quite a few years ago, AMD migrated from the front-side-bus architecture to a point to point interconnect architecture that was both higher bandwidth and lower latency.

Remember, cache exists only as a buffer, like a bucket to store the data flowing into a CPU, so that slower interconnects and I/O will not result in CPU having to twiddle its thumbs while it waits for the data to arrive.

Intel, on the other hand, has traditionally had a much better manufacturing process than AMD, which allowed them to throw huge amounts of cache in their CPU so that they could retain the slow front-side-bus interconnects, while still remaining competitive with AMD in terms of performance.

In the "i" series of CPUs, Intel has upgraded to a very good point to point interconnect architecture itself (which is actually very similar to AMD's interconnect). However, in addition to this, Intel still retains its superior manufacturing process, and have also managed to do a very very good job of designing the CPU correctly. This is the reason why across the board, Intel is completely dominating AMD in all counts, performance, latency, scalability, efficiency and performance-per-watt.

If at all, Intel has priced its CPUs at a bit of a premium as it always does, and this has allowed AMD to survive, mainly by undercutting Intel on price and selling their old architecture CPUs at essentially throw-away prices to retain market-share.

Anyway, back to cache, the cache that Intel fitted in its Core2 series of CPUs was actually one of the finest caches made by either Intel or AMD, and was and is much superior to anything AMD produced. This is actually the reason why Core2 whipped Athlon in performance even though Core2 was using an FSB. The "i" series cache is actually larger, but has a higher latency.

As far as OP's question goes, cache is a minor aspect to consider when it comes to CPUs - if performance is really a concern, then you should look at building a "balanced" system that contains no major performance bottlenecks. To use a car analogy, fitting a 500hp engine on an Ambassador might have its wow factor, but it will not win any races in a circuit.

As things stand today, AMD is a good option mainly because it has very reasonably priced motherboards, and remains competitive with Intel because of its low prices. An alternative to consider is Intel's i3 or i5. In any case, if you are going to do graphics intensive work, what you really need is a powerful GPU or graphics card, and a really fast hard drive (or multiple hard-drives in RAID configuration, or better still, an SSD). An average CPU will do.

Frankly it sounds more like marketing bull to me. Sorry I cant buy it. I vaguely remember when AMD had the advantage for a while. You make believe that Intel has trounced AMD in all departments because you are interested in this version of things. Having an advantage NOW (if they do) is not the same as having had the advantage at all times.

Are you a marketing or product manager for Intel brand?

-G

 

[sam9s]

Frankly it sounds more like marketing bull to me. Sorry I cant buy it. I vaguely remember when AMD had the advantage for a while. You make believe that Intel has trounced AMD in all departments because you are interested in this version of things. Having an advantage NOW (if they do) is not the same as having had the advantage at all times.

Are you a marketing or product manager for Intel brand?

-G

Gobble I did not read the entire post but, its true currently Intel Processors are outperforming AMD (now out performing does not mean better), but benchmarks shows an upperhand for Intel. AMD was the king when Intel had those Pentium 3sss and 4ss around...........after Core2Duo, the story has been much in favour of Intel.....

Similarly there was a time when Nvidia was the king with GPUs, but gradually ATi has taken over and have out performed Nvidia GTX in every benchmark with their HD line or series......

To the OP cache is not the "only" thing that decides the performance, like for example the Core i7 series architecture is a complete rework from Intel much like core2duo was from Pentium 4...........a small example would be, an integrated PCIe controller. An integrate PCI Express controller, would allow the chip to connect directly to a graphics card, just as it would connect directly to the main memory (thanks to the memory controller integrated as well). This could mean faster transfer and elimination of same on the supporting platform. This is just one of the many changes i7 has brought.

Support for 3 Channel DDR3 memory is another thing that is there with the i7 series.....

 

[thevortex]

AMD offers more muscle for the same money. Both the 965 BE and the newer 890GX chipsets offer far more than, say, the Intel i5 and a compatible motherboard with USB3.

That should make an easy decision.

 

[marsilians]

While reading many of the threads here; many people suggested AMD quad core processros as best when compared to Intel processors.

Unfortunately we live in times where "best" usually is associate only for certain criteria such as speed, performance on a specific type of application, speed, complexity. Without any specification of such criteria the above statement makes no sense and whoever quoted as such has not provided the necessary facts or referenced to back such a claim.

So if you are confused then there is good reason for that.

To counter argue above statement, Intel quad cores are more efficient, use lower power and are better at over clocking.

But while reading the details of the processors, I see that AMD processors has less chache memory than Intel processors.
I feel that cache memory will help lot for performance specially in case of gaming and grafics.

Does any one can give more insight into this....?

It depends on which level cache you are referring to. YOu may want to review Tom's Hardware or AnandTech for detailed reviews to help make a decision.

 

[spirovious]

Again for faster PC we shld see Ram speed,MB FSB & other factors too.
Its like system matching in HiFi.

 

[asliarun]

Frankly it sounds more like marketing bull to me. Sorry I cant buy it. I vaguely remember when AMD had the advantage for a while. You make believe that Intel has trounced AMD in all departments because you are interested in this version of things. Having an advantage NOW (if they do) is not the same as having had the advantage at all times.

Are you a marketing or product manager for Intel brand?

-G

There is not need for you to be rude by making insinuations. Care to tell me why you call my post "marketing bull" when you yourself provide no backing proof except for having a vague memory.

Even though you are clearly baiting me, I have absolutely no interest in indulging in a flamewar, especially on a subject like Intel/AMD, where people have been at it hammer and tongs as if either AMD or Intel were religions or something!

I will just say this.. AMD's Athlon used to rightfully beat Intel's inefficient Pentium4 CPUs in all counts. However, since 2006, Intel has been comprehensively beating AMD in both performance and efficiency when Core2 or Merom-Conroe-Woodcrest was launched.

If you care to read just about any online review, you would see the facts for yourself. Or, let me give you some snippets, all of them taken from highly respected neutral websites and reviewers.

Here is a review from 2007:
Intel vs. AMD: Today's generation compared - The Tech Report - Page 1

To quote directly from the article:
"Intel has had a lock on the overall performance lead since the Core 2 Duo first hit the scene, but AMD has made clear its intention to maintain a competitive price-performance ratio."

Please note that in 2007, Intel was providing, on an average, 2MB cache per core, wheras AMD was providing 1MB cache per core. This is both L2 cache, by the way.

Please also note that architecturally speaking, Core2 never had any L3 cache.

You also quote:

You make believe that Intel has trounced AMD in all departments because you are interested in this version of things. Having an advantage NOW (if they do) is not the same as having had the advantage at all times.

I have never said that Intel has had the advantage over AMD at all times. You are misquoting me or more likely, misunderstanding me. Once again, I urge you to put your emotions aside. You may be cheerleading AMD's cause because they are the underdogs and fair enough, I buy that stance, but not when it starts influencing facts. I'm only interested in having a technical discussion and ensuring accuracy.

In fact, I consider AMD's chip design to be superior and much more visionary than Intel's mainly because they are forced to counter Intel's manufacturing superiority time and again with better design! Their 64bit implementation is also superb, which is why it became the de facto standard, and Intel was forced to copy it in the later versions of P4 onwards. Intel is also quite often hindered by a lot of internal bureaucracy, which is why you often see Intel slipping up on the design front, until they get a kick in the pants by AMD which forces them to cut the bureaucracy and actually deliver good products.

On the other hand, the one thing you simple cannot touch Intel on is process technology. AMD is small fry - actually relies on IBM to do its basic research. Intel has, time and again, over the last couple of decades, led everyone, and that includes all semiconductor manufacturers, when it comes to manufacturing (fab) process. At any given time, Intel is usually 6-12 months ahead. This means that 50% of the time, when AMD is using say 45nm process, Intel is already manufacturing its chips at 32nm.

The only reason why I am posting my comments in this thread is because I noticed some mistakes if what people are saying and recommending when it comes to CPUs. In the CPU and GPU world, the tables turn so fast and so often, that perceptions held by most people are quite often outdated. Unfortunately, these perceptions also end up resulting in sub-optimal choices and recommendations.

For example, in 2001, AMD was supposed to be merely a clone manufacturer like Cyrix, that made slightly faster chips than Intel but were also cheaper and comparitively unreliable. This perception was not helped by some cases where AMD chips famously burnt out because they didn't have thermal sensors. Even though this problem was fixed almost immediately (which company wouldn't!), the perception lasted years and years, and even today, hardware assemblers will tell you that AMD is "unreliable"!

Similarly, Intel has had its own share of perception issues, starting with the infamous Pentium rounding off bug, and later on, the huge amount of heat that P4 would generate leading to the "P4 space heater" jokes.

Let me put it another way. If this was an audio recommendation, and if someone asked for the "best" speaker or the speaker with the "biggest woofers" or "big bass", would you recommend Wharfedale?? Even if you did recommend it, you would carefully word your recommendation by saying this for the target price range, Wharfedale 9.2 or 10.2 is probably the best value-for-money product.

However, I see only blanket statements being made in gereral about AMD and Intel chips, that are mostly based on outdated perceptions and in some cases, not strictly correct.

You yourself made a statement in an earlier post that AMD has better cache optimization algorithms and that 1MB per core is good is not correct.

Please read this Anandtech article that was written in 2006 when Core was launched:

Intel Core versus AMD's K8 architecture - AnandTech :: Your Source for Hardware Analysis and News

To quote the article verbatim:
"The Core CPUs have much bigger caches and much smarter prefetching than the competition. The Core architecture's L1 cache delivers about twice as much bandwidth (Measured by ScienceMark), while it's L2-cache is about 2.5 times faster than the Athlon 64/Opteron one."

If you want to read about Nehalem and its cache design, please read the excellent but very detailed article written by David Kanter:
Real World Technologies - Inside Nehalem: Intel's Future Processor and System

Or, the article by Jon Stokes of ArsTechnica:
What you need to know about Intel's Nehalem CPU

If you want to see the trend of cache over time and over multiple CPU architectures, please read:
Nehalem Part 3: The Cache Debate, LGA-1156 and the 32nm Future - AnandTech :: Your Source for Hardware Analysis and News

When you made your 1MB cache recommendation, that is actually only relevant if the cache is low latency L2 cache.

For L3, the number should be 2MB minimum. This is based on what Intel's chief CPU architect of Nehalem, Ronak Singhal says about this subject:

(quoting from the Anandtech article on caches)

"Ronak wanted the L3 to be bigger on Nehalem; at 8MB thats only 2MB per core and merely sufficient in his eyes. There are two 32nm products due out in the next 2 years, I suspect that at least one of them will have an even larger L3 to continue the exponential trend I showed in the second chart above.

Could the L2 be larger? Sure. But Ronak and his team ultimately felt that the tradeoff between size/latency was necessary for what Intels targets were with Nehalem. And given its 0 - 60% performance increase, clock for clock, over Penryn - I cant really argue."

Edit: You can find benchmarks for most CPUs here: http://www.tomshardware.com/charts/2009-desktop-cpu-charts-update-1/Far-Cry-2-1.0.1,1401.html

The page I have linked to gives you the real world gaming benchmark for FarCry.
You can see that even the Core i3 530 does an excellent job in gaming.

 

[gobble]

I have never said that Intel has had the advantage over AMD at all times. You are misquoting me or more likely, misunderstanding me. Once again, I urge you to put your emotions aside. You may be cheerleading AMD's cause because they are the underdogs and fair enough, I buy that stance, but not when it starts influencing facts. I'm only interested in having a technical discussion and ensuring accuracy.

There ... now re-read your earlier post and try and understand how the tone of your post was of a triumphalist nature with one side emerging gloriously victorious forever in a battle of odds. Maybe you can start with Wren and Martin grammer book to understand your own sentence construction better?

I carefully put in the word "vaguely" because I do not care about proving anything on this front. Sites like Anandtech and Toms hardware, I outgrew years back. Before you tit for tat, re-read your OP and consider with a cool head.

Cheers

Edit: You made an Intel Win All vs AMD Loses All type of post, which reeks of fanboyism. My post was because such mindless fanboyism misleads people into making a more expensive purchase than they would. Ok there are easy to copy-paste "facts" out there on the Internet like how Intel is currently the reigning champion. Frankly I'm not interested ... except to counter your triumphalist tone.

 

[asliarun]

There ... now re-read your earlier post and try and understand how the tone of your post was of a triumphalist nature with one side emerging gloriously victorious forever in a battle of odds. Maybe you can start with Wren and Martin grammer book to understand your own sentence construction better?

I carefully put in the word "vaguely" because I do not care about proving anything on this front. Sites like Anandtech and Toms hardware, I outgrew years back. Before you tit for tat, re-read your OP and consider with a cool head.

Cheers

Edit: You made an Intel Win All vs AMD Loses All type of post, which reeks of fanboyism. My post was because such mindless fanboyism misleads people into making a more expensive purchase than they would. Ok there are easy to copy-paste "facts" out there on the Internet like how Intel is currently the reigning champion. Frankly I'm not interested ... except to counter your triumphalist tone.

Do you really think that my first post, or my subsequent post on this topic was biased or "reeks of fanboyism"?
Seriously??

Just because you have probably encountered a hundred AMD or Intel fanboy does not mean that everyone you will encounter henceforth in your life will also be a fanboy.

As I mentioned earlier, I am trying to be as factual as I can, and have made no personal references or attacks, unlike you.

If you say that you outgrew sites like Anandtech, it is actually quite a pity. People like Anand Lal Shimpi, Jon Stokes, David Kanter, etc. - the people who gave these sites its reputation - are people from whom we can learn a lot. I've actually been reading Anand Lal Shimpi's articles since roughly 2002/2003 (and he was actually quite young then), and he has consistently been unbiased, objective, and deeply insightful. Jon Stokes is another very good example - he has actually been soundly berating Intel in Arstechnica for a long long time now.

Finally, please remember, one can sound slightly dramatic (or triumphalist as you call it) and still be fair and objective at the same time. After all, the nature of our passion, viz. computers, audio, video etc. is something that is purely technical and dry, which prompts us to use adjectives like "warm", "mellow", "colored", "blazing", "amazing" etc. However, just because these adjectives are used does not necessarily make the statement false. If you really read what I have written about, not just in this thread but in other similar threads as well, I don't think you will find any bias or "fanboyism".

I've been in this game for a very long time as well (long by internet standards at least - about 10 years or so), and frankly, I don't have the energy to do this "mera Intel tera AMD ko boxing match mein beat karega" kind of nonsense. Nonetheless, I don't like skewed statistics or incorrect facts that end up in people making wrong choices.

If you notice my first post, I recommended AMD to the OP, stating that it offers better value for money, and a better ecosystem (i.e. better motherboards). Yet you call me an Intel fanboy? Oh well.

Edit: Deleted some stuff I posted earlier. You DO have a really dry sense of humor though, and on first reading, sounds a bit provocative! I'm just trying to have a constructive discussion here

 

[corElement]

The term best is so baseless. Static term for a highly dynamic subject.

 

[doors666]

though not related to the topic, both of them are garbage CPUs which are surviving because of microsoft's lack of will power to move its OS to a better CPU. Both of these are CISC CPUS, while RISC CPUs are far superior. CISC just exists because of compatibilty reasons. These were invented in the first place because of lack of memory in earlier systems, while memory has now grown to astronomical levels, the CPUs are still using the old rotten technology.

OK, the CISC fans can flame me now

 

[reignofchaos]

though not related to the topic, both of them are garbage CPUs which are surviving because of microsoft's lack of will power to move its OS to a better CPU. Both of these are CISC CPUS, while RISC CPUs are far superior. CISC just exists because of compatibilty reasons. These were invented in the first place because of lack of memory in earlier systems, while memory has now grown to astronomical levels, the CPUs are still using the old rotten technology.

OK, the CISC fans can flame me now

What you say is true to an extent or rather was true not so long ago. The legacy x86 baggage is pretty crappy and honestly a complete waste of CPU die space.

However calling a modern CPU garbage is not entirely correct because no other chip exists which is faster irrespective of the fact that x86 is garbage or not. Every other architecture is mostly dead because of lack of software support. All modern CPUs (both AMD and Intel) have moved towards a RISC like architecture internally since the Pentium days and x86 instructions are simply a layer of translation into the CPU's microcode. No self respecting developer these days who had optimizes code will dare touch these legacy instructions anyway.

Coming back to the topic, AMD and Intel architectures are like the north and south pole. You can always invent a benchmark which will favor either depending on whether it exploits the strengths of the architecture. Comparing X megs of cache on an intel chip with Y megs on an AMD chip has absolutely no relevance. Both are fast enough for any job you throw at them and getting the fastest chip is simply a way of increasing the size of your e-penis :lol:.

 

[sam9s]

though not related to the topic, both of them are garbage CPUs which are surviving because of microsoft's lack of will power to move its OS to a better CPU. Both of these are CISC CPUS, while RISC CPUs are far superior. CISC just exists because of compatibilty reasons. These were invented in the first place because of lack of memory in earlier systems, while memory has now grown to astronomical levels, the CPUs are still using the old rotten technology.
OK, the CISC fans can flame me now

I am not a CISC fan but, just saying RISC is better or vice versa is half the truth. Its like comparing Multi Core Single Processor and Single core
Multi Processor and claiming that one is better.

CISC and RISC are simply two different architecture with their own sets of Pros and Cones. Simply saying RISC are fast and superior is not justified, if RISC is fast that is bacause their hardware is simpler but that puts a greater burden on the software.......programmers require more lines of code to produce the same results and are increasingly complex. This increases the size of the application and the amount of overhead required..........hence and so forth......
Anyhow today's CISC chips use many techniques formerly associated with RISC chips, so ......you see its not that simple or a one sided affair.........

 

[marsilians]

though not related to the topic, both of them are garbage CPUs which are surviving because of microsoft's lack of will power to move its OS to a better CPU. Both of these are CISC CPUS, while RISC CPUs are far superior. CISC just exists because of compatibilty reasons. These were invented in the first place because of lack of memory in earlier systems, while memory has now grown to astronomical levels, the CPUs are still using the old rotten technology.

OK, the CISC fans can flame me now

There are as many business reasons for not moving as technological ones. The cost of changing or moving to new architectures is very expensive (as we see how the 64 bit is being embraced). Simply looking at the history of last 15 or so years, it is pretty clear that M$ was very smart to stick with Intel due to its market share.

If you look at the some of the heavy weights behind RISC technologies such as Motorola, IBM and HP, they have all embraced Intel architecture very tightly into their product development plans.

Also like ROC has noted, Intel has been using RISC techniques from 386 version onwards and has increasing been using it through the later family of pentium processors.

What you say was true during the early years of the technologies but Intel's market share is a clear indication that it has got its fundamentals right in its approach to building processors.

Another example is Apple's embrace of Intel CPUs for macs. Its not so bad after all!

 

[asliarun]

There are as many business reasons for not moving as technological ones. The cost of changing or moving to new architectures is very expensive (as we see how the 64 bit is being embraced). Simply looking at the history of last 15 or so years, it is pretty clear that M$ was very smart to stick with Intel due to its market share.

If you look at the some of the heavy weights behind RISC technologies such as Motorola, IBM and HP, they have all embraced Intel architecture very tightly into their product development plans.

Also like ROC has noted, Intel has been using RISC techniques from 386 version onwards and has increasing been using it through the later family of pentium processors.

What you say was true during the early years of the technologies but Intel's market share is a clear indication that it has got its fundamentals right in its approach to building processors.

Another example is Apple's embrace of Intel CPUs for macs. Its not so bad after all!

To add, the only CPU architecture that can (and will) rival x86 (let me be very clear, i mean both Intel and AMD ), will be ARM, in my humble opinion.

If we look at things from the perspective of form-factor, we see x86 trying to push its way from large to small (workstation -> mini desktop -> HTPC -> laptop -> netbook. Decreasing the form factor beyond netbooks is seeming to be extremely difficult, mainly because energy consumption (and heat dissipation) becomes really crucial in these form factors.

On the other hand, we also see ARM push its way from small to large form factors (embedded -> handhelds -> smart phones -> home entertainment digital/internet devices -> tablets. Here, the challenges in increasing the form factor are exactly the opposite - lack of performance, lack of robust software platforms etc.

Intel and possibly AMD will try to attack in the first front by improving their power consumption by means of constantly evolving process shrinks - Atom was the first crude effort, Moorestown will be releasing over the next 6 months or so, and it will be interesting to see how this pans out, as Moorestown is efficient enough for tablets and MIDs (probably not smart phones yet).

ARM on the other hand, has an equally promising future, and will try to push towards bigger form factors by making their designs much more powerful and multi-threaded, such as the Cortex CPUs. A more powerful thrust is being provided by rapidly evolving software ecosystems such as Apple's iPhone, Google's Android, and possibly Nokia's Maemo software platforms. Soon enough, at least in the consumer space, people won't really care for legacy support for x86 applications if alternative software platforms are robust enough that it provides them with lots of viable software alternatives.

(The business world will be a much harder nut to crack for ARM, but that's a different battle to fight, and quite likely, ARM will never bother fighting this battle.)