If desktop CPUs are already fast enough, why doesn't AMD CPUs sell?

[Mfusick]

AMD doesn't sell because for normal folks they appreciated the extra power and lower heat/noise/energy of the 1155 Intel for about the same cost.

AMD does not hold a very significant price advantage and they are not as effificient or powerful as INTEL.

Also- If you really wanted power you go with a LGA2011 making the high end for AMD suck.

AMD is good for integrated GPU only. Otherwise it's crap today. I might buy it for an integrated GPU build that is affordable... but that is it's only benefit.

with a graphic card- Intel is a better choice. IMO.

 

[Idontcare]

It is easy to blame the foundry but is the foundry the only problem?

Bulldozer was a 45nm product. The 45nm SOI node was already very mature and did yield descent clocks and still they could not compete with Intel designs with the time, and it is safe to say that they could not compete with AMD designs of the time.

Have a look at the HEDT and server market. AMD is in the same node as Intel but still they cannot be competitive at all with Intel offers. Mobile, same thing. They trail Intel despite devoting more CPU area than Intel to their chips.

In the three cases here we had situations where AMD designs, the ones you say are fine, could not compete with Intel designs at the same node or even with AMD own designs, and in the case of the 45nm Bulldozer there wasn't a foundry partner to blame. Designs *are* a problem for AMD.

I think it is fair to say that there is a subpar design being manufactured in a subpar foundry.



How do you explain a 75% slump in purchase commitments and inventory ballooning with the market falling just 1.2%? Care to show the math here?

There is no question the microarchitecture design itself could have been better, the existence of piledriver on the same exact process node is proof that bulldozer itself could have been improved and released as piledriver in the first place.

That isn't to say that piledriver itself addresses all the shortcomings, just saying it is proof that AMD's design engineers obviously had been directed to design bulldozer knowing in advance that there were things that bulldozer needed and could use if the design engineers had more time/money/resources to do it. (and when given more time/money/resources they did, they created piledriver)

What does this tell us? It tells us that the project management triangle is alive and well, even at AMD, and at GloFo.

In project management the saying is "you can pick only two of the three".

You can be like Intel, move along at a nice tick-tock model that coincides with a wonderful 2-yr node cadence all thanks to a well-funded and well-resourced project model.

You pay the piper which then enables the project managers to focus on schedule and scope. schedule = tick/tock and nodes every 2yrs, scope = IPC capabilities, power efficiency, HKMG, Finfet.

AMD (and GloFo) were not so flush with cash as to pay the piper to the same degree as Intel could/did. That required the project managers to factor in cost in a way that had to impact either the schedule or the scope of the project itself.

So you don't get HKMG at 45nm, nor finfets at 20nm. And you don't get an amazing thuban-trouncing microarchitecture at 45nm, you have to delay it to 32nm only to then strip it down even further in scope and leave out things you know it needs but you delay those features even further to the piledriver core (delay the schedule to delivering on the same scope, because of the cost-constraint).

I've seen how these decisions, both process node and IC design, are made. I was a project manager myself on a number of critical decisions and had to factor in all kinds of concerns (risk to timeline delivery is a challenging one to adequately capture).

From my perspective, AMD's situation is one of inevitability. Given the competition, the resourcing gap, the only way AMD stood a chance to deliver competitive products was if the competition's R&D efficiency went down the drain...and for a while it briefly did at 130nm and again at 90nm with Willamette and Prescott.

And unfortunately AMD did not capitalize on those brief opportunities enough such that they cemented their lead over Intel in terms of annual revenue. Without that lead, there is virtually no way AMD's project managers were going to be able to deliver superior in scope projects on a competitive timeline while operating with pennies to the dollar of their equivalent project leaders over at Intel.

To be sure AMD had its missteps, intentionally delaying the 65nm node and products, and the parasitic situation with GloFo is going to be to both their detriments in the long run. But in the end, I do believe that even if AMD and GloFo had fired on all cylinders and hit home-run after home-run across the board on every project they still would not have been able to persist in competing against a competitor whose R&D investments absolutely dwarfs AMD and GloFo combined.

That is how the USA beat the USSR in the cold war, we outspent them in every way thanks in part to have a near 4:1 GDP advantage. That is how AMD has managed to beat Via in every way, with a near 10:1 revenue advantage. And it is of no surprise that Intel is/has done the same to AMD.

 

[riva2model64]

Very insightful stuff, IDC:thumbsup:

So if everyone wants AMD to perform better, but most people are buying/recommending Intel products, how will AMD get the money it needs to fund R&D efforts to make a better product?

In other words, AMD won't have enough R&D funds to compete with Intel, but most people expect AMD to be a lot more competitive with Intel.

 

[ShadowVVL]

The first Bulldozer was a 45nm 6 core product, it was canned by Dirk Meyer in 2009, and now the reasons are clear. If a 32nm 8 core could barely keep up with Thuban, what would a 6 core at 45nm do? Probably wouldn't even keep up with Athlon X4.

Ah ok my bad I wasn't aware.

 

[Idontcare]

Very insightful stuff, IDC:thumbsup:

So if everyone wants AMD to perform better, but most people are buying/recommending Intel products, how will AMD get the money it needs to fund R&D efforts to make a better product?

In other words, AMD won't have enough R&D funds to compete with Intel, but most people expect AMD to be a lot more competitive with Intel.

Its a vicious feedback loop to be sure. Which is why it is all the more critical that management make the right decisions at the right time.

Sander's made one fatefully wrong, and fatal, decision when he decided Ruiz would succeed him.

It all goes back to that. Had Ruiz not taken the helm then there was a chance the ship would not have gone down.

Look at Apple and what happened when Jobs was ousted. Then their second-chance when Jobs came back

Look at TSMC and what happened when Morris Chang stepped down (45nm debacle), and their second-chance when he came back (28nm is the biggest success in TSMC's history).

Intel when Grove stepped down, Barret was a near-disaster (that's also when the unethical stuff inside Intel started happening), then Otellini came in and turned Intel around - both the products and the business ethics.

 

[Fjodor2001]

@Idontcare:

You have a point. But still quality and performance does not necessarily scale exactly with R&D budget. There are lots of companies that have spent tons of R&D money but only managed to deliver a crappy product anyway. In the same way, some other companies come up with brilliant ideas and manage to create fantastically clever products with a very limited budget.

Also, remember that during the Athlon 64 & Athlon 64 X2 period, Intel still had much more resources than AMD. Yet AMD managed to deliver a superior product.

So R&D budget is no guarantee for success, although it often helps.

 

[Fjodor2001]

I think I perhaps should clarify my original post to highlight the conflict I'm interested in:

Question: Are current desktop CPUs "fast enough" for most people?

No => Then how come Intel is focusing on lowering TDP etc instead of increasing CPU performance?

Yes => Then how come AMD desktop CPUs aren't selling so well, despite that they are not that much slower than Intel CPUs, and they also have better IGPU and lower price?

 

[peonyu]

AMD chips are fast enough today, but currently a intel chip will remain "fast enough" for a longer period of time compared to AMD. Given the option of one or the other, why go for the slower chip that will become noticably slow in games/apps sooner than the intel chip will ?


Whats fast enough today will become slow 2 years from now etc..A intel may be fast enough for 3 years.

 

[frozentundra123456]

I think I perhaps should clarify my original post to highlight the conflict I'm interested in:

Question: Are current desktop CPUs "fast enough" for most people?

No => Then how come Intel is focusing on lowering TDP etc instead of increasing CPU performance?

Yes => Then how come AMD desktop CPUs aren't selling so well, despite that they are not that much slower than Intel CPUs, and they also have better IGPU and lower price?

I think most people did understand your question. For most people, yes, any cpu is fast enough. The problem on the desktop is that AMD offers no compelling reason to accept lower cpu performance and higher power usage. If you look in big box stores where most people buy their computers, the AMD price advantage is minimal, partially because one, almost 2 generation old Intel chips are more than competitive and the price has come down to very near and in some cases lower than AMD levels. I do not consider the igpu a compelling advantage in the desktop. Just too easy to add a discrete card and if you are the normal light user, Intel igps are adequate as well. Unfortunately for AMD some makers are even using Brazos in desktops to reach a low price point, which IMO does a disservice to both AMD and the consumer, as it is really not a desktop chip.

 

[ShintaiDK]

Question: Are current desktop CPUs "fast enough" for most people?

No => Then how come Intel is focusing on lowering TDP etc instead of increasing CPU performance?

You aint happy enough with Haswells >2x performance boost for AVX/AVX2?

 

[Idontcare]

@Idontcare:

You have a point. But still quality and performance does not necessarily scale exactly with R&D budget. There are lots of companies that have spent tons of R&D money but only managed to deliver a crappy product anyway. In the same way, some other companies come up with brilliant ideas and manage to create fantastically clever products with a very limited budget.

Also, remember that during the Athlon 64 & Athlon 64 X2 period, Intel still had much more resources than AMD. Yet AMD managed to deliver a superior product.

So R&D budget is no guarantee for success, although it often helps.

That is why I mentioned R&D efficiency and the necessity of good management.

This is not unique to AMD/Intel or even semiconductors. In every industry, in every R&D budget, there is a metric of assessment that captures R&D efficiency. Too low of an efficiency and you have a Bell Labs (or NASA) type situation where the budget is HUGE but the returns are minimal in comparison.

But your efficiency can not scale to silly heights either. No matter how efficient Chinese national R&D may have been in the 1960s, they were never in contention for getting a man to the moon before the USA.

Anyone can waste money, that is the trivial outcome, but you cannot get from here to the moon on a budget that can barely acquire a shovel and a wheelbarrow, let alone create the necessary technology.

I think I perhaps should clarify my original post to highlight the conflict I'm interested in:

Question: Are current desktop CPUs "fast enough" for most people?

No => Then how come Intel is focusing on lowering TDP etc instead of increasing CPU performance?

Yes => Then how come AMD desktop CPUs aren't selling so well, despite that they are not that much slower than Intel CPUs, and they also have better IGPU and lower price?

Brand awareness and risk management.

80% of consumers have experience with an Intel product, and unless that experience was negative they are likely to buy their next rig with the thinking of "why fix what isn't broken, just get the Intel rig because you know it'll work, I can't say for certain the AMD one will work fine because I have no experience with it."

And for the 20% who had AMD rigs, a certain percentage of those will not be recaptured in the next upgrade cycle because of other factors - performance/watt, sales, peer-pressure, absolute performance with their apps of interest, etc.

At the business level it is all about risk management. No one wants to have their ass hauled in on a Sunday to deal with a critical-down situation with their hardware (a problem that likely goes to software issues) and then on Monday they are expected to explain to their bosses why they didn't just go with Intel hardware...because they can never prove the problem would have existed even with an Intel server, and the non-IT guys simply don't care to understand the IT challenge beyond the superficial.

So it is an uphill battle for the IT guy to convince management to buy AMD gear, and if it doesn't go smoothly then it is their own career path that is imperiled. Who wants that headache or personal career risk? Very few, less than 5% by last IDC market share numbers.

 

[coffeejunkee]

This Newegg user review probably sums it up:

Cons: None, it's an INTEL

Taken from the Pentium G530 reviews, which costs exactly the same as an A4-3400 ($48,99). Like MFusick said, there is no significant price difference anymore, back in the Pentium/K6 days you could save $100, now you can save (barely) $10 by getting an A4-3300. So why would you buy the unknown AMD when you can get the well-known Intel?

For the informed buyer AMD is currently only really viable in a few niche situations, such as cash-strapped gamers who have no money for a cpu and graphics card. Llano works for them but as soon as you add a discrete card you would have been better off with a G530 in the majority of games. Or HTPC afficionados who can't stand the 23.97fps bug. And they would be better off with a G530 and a discrete card most of the time as well.

Brazos is very nice for netbooks though but as you probably know, they have been pretty much killed off by tablets .

 

[Leyawiin]

Same story its always been - momentum. Its always been hard to compete against a behemoth with deep pockets and a tight marketing leash on PC manufacturers (glaringly apparent in the Athlon 64 days when AMD was significantly better in design...but still couldn't make market share headway because Dell et al wouldn't use them). The only thing that's changed now is AMD has slipped in design which only exacerbates the problem. Inept management hasn't helped, but stellar management probably wouldn't have saved them. AMD has always been a one-legged man in an ass kicking contest.

 

[ShintaiDK]

Same story its always been - momentum. Its always been hard to compete against a behemoth with deep pockets and a tight marketing leash on PC manufacturers (glaringly apparent in the Athlon 64 days when AMD was significantly better in design...but still couldn't make market share headway because Dell et al wouldn't use them). The only thing that's changed now is AMD has slipped in design which only exacerbates the problem. Inept management hasn't helped, but stellar management probably wouldn't have saved them. AMD has always been a one-legged man in an ass kicking contest.

AMD back then was capacity constrained. And they got so fond of themselves they delayed 65nm as well as developments. And then on top bought ATI instead of investing in fabs. They got exactly what they deserved.

 

[AtenRa]

I think I perhaps should clarify my original post to highlight the conflict I'm interested in:

Question: Are current desktop CPUs "fast enough" for most people?

No => Then how come Intel is focusing on lowering TDP etc instead of increasing CPU performance?

Yes => Then how come AMD desktop CPUs aren't selling so well, despite that they are not that much slower than Intel CPUs, and they also have better IGPU and lower price?

It has nothing to do with performance,

Its all about availability, execution on time, lower prices, better deals than your opponent etc.
For the last 2-3 years AMD havent been able to maintain any of the above.

Llano at the hands of Intel would sell billions of chips but at AMD it had manufacturing problems, it was late with low availability. It took 2-3 quarters for it to reach acceptable numbers in retail and when that happened Trinity was on the way.
Then Trinity was released and there were no Motheboards anywhere to be found.

Llano laptops where like Uranium, hard to find and expensive, same with Trinity. Capacity constrained or not if you cant sell a better product then you seriously have a problem.

OEMs care about execution on time, they dont care if Trinity consume 10W more than CPU B. They care about having the chips on time at the volume they ordered, with a better deal than your opponent. When you cant give them that then they will not buy your product even if it is better than the competitor.

Retail,
Trinity is perfect for a micro-iTX PC/HTPC etc BUT, there are no micto-iTX motherboards to be found anywhere two months after the original launch of desktop products. They buried there own product with low availability and luck of innovation for new products (Intel NUC).

Intel is selling at higher prices not because they always have the better product, but because THEY HAVE A PRODUCT TO SELL.

 

[magomago]

I don't accept that modern processors are fast enough. Sure if all you do is web browse and watch video then you really don't need anything high performance at all. But just today I managed to make Libre Office calc lockup for a minute calculating a graph with 28,000 points in it as a line chart. That isn't a lot of points when we are talking about frame time captures from Fraps.

I have compilations that take 10's of minutes to complete where realistically I want them to take less than a second. I have games that are still CPU limited and delivering less than 60 fps. Coding might be specialist but playing an MMO or plotting a graph is not, its basic stuff and my monster rig can't cope with these simple tasks. I haven't even got to the point of video encoding, or the oddness in playing back videos in VLC with blocks appearing to try and optimise seeking so that its quick enough to be useful rather than producing a decent image. Even some web pages take multiple seconds for the CPU to render in chrome.

I simply don't accept the premise that CPUs are fast enough, I have lots of programs I want to write but which I can't validate they are correct because you can't buy a CPU that can run the program.

AMD doesn't sell as well because its not as good a CPU. What they are currently doing is selling CPUs with a cupholder (the GPU) which is fine if you want a bundled GPU but it doesn't make the CPU any more competitive in its main function, it actually makes it worse.

Just a point that needs to be emphasized because you barely gave it lip service- if you are trying to plot 28k points in an excel spreadsheet you are doing it wrong. You need the right tool for the right job.

Libreoffice & Excel is the WRONG tool for plotting 28,000 points and trying to do any meaningful analysis. I could do the exact same in Matlab (and I'm assuming Octave works just as well for the OpenSource Folks) and be done instantly.

 

[riva2model64]

Just to clarify my earlier post, of course product performance is not only dependent on R&D, but also having a talented team, and the right market vision, taking the right risks, owning the right patents blah blah blah. . There is also a little bit of a randomness factor involved with innovation (accidental discoveries for example). I just wanted to point out the vicious cycle I saw when reading IDC's post.

To address OP's question, I believe it has already been addressed, but I will try to answer it from my own observations:

Because Intel has more control over manufacturing, they can meet volume needs much better than AMD currently can, and consequently have more flexibility with pricing.

In the enthusiast market, where AMD is well known, it is clear to see that Intel's offerings generally outperform AMD offerings at all price points for gaming. For those looking for silence (HTPC or audiophile) Intel chips use much less power and are easier to cool. Those on a budget will buy Intel for power efficiency without losing performance. So even for those buyers who know about AMD and think all new processors are "fast enough," there are still reasons to switch to an Intel processor.

Regarding the consumer market, your average Joe will most likely buy what he/she is familiar with and know works. Since Intel has most of the market, that answer will likely be Intel. And OEMs will have mostly Intel-powered computers because of sales momentum, ability to meet volume requirements, and more modest power requirements.

For those who enjoy wasting money, aka Apple users (just joking don't flame me), Apple computers for consumers ONLY use Intel processors.

Regarding Server or Research or other professional environment, I am not a reliable source of information. But if Xeons/Opterons are anything like their desktop counterparts, then its a clear choice to pick Xeon because you will save $$$ in the power bill. Opterons are very muscular and really fast, but only in a select few programs. Intel LGA2011 provides higher performance all-around (compared to Opteron 8/6 core).

 

[mrmt]

There is no question the microarchitecture design itself could have been better, the existence of piledriver on the same exact process node is proof that bulldozer itself could have been improved and released as piledriver in the first place.

Once we analyse Bulldozer today, they are not a conclusive step ahead from Thuban. Even some die hard AMD fanboys didn't find compelling an upgrade from the first BD and only now with Vishera are thinking about upgrade. Still, Vishera improvements pale in comparison with Intel improvements with Nehalen, and Nehalen was already far ahead of whatever AMD threw at them.

That said, and given the more technical discussions about the concept here, is Bulldozer really a budget problem, or a problem akin to the Netburst architecture, a fundamentally flawed concept that no matter how much money you sink on this pit, the chip won't work?

AMD (and GloFo) were not so flush with cash as to pay the piper to the same degree as Intel could/did. That required the project managers to factor in cost in a way that had to impact either the schedule or the scope of the project itself.

At the engineering level, that's what you do. But there are things you can do at the management level and AMD management isn't deserving to earn the millions they assign to themselves every year.

What AMD has always tried to do is to become a "me too" Intel, so once Intel fields something new you can expect two or three years to have AMD "me too" product on the market. Servers, netbooks, ultra-thin notebooks, you name it. This does not work for the reasons you explained above.

But look at, for example, Nvidia. It does compete with Qualcomm? Yes and no. Nvidia does not touch a lot of segments, but in a given segment, premium phones and tablets, they do. Instead of using the dumb AMD strategy of trying to become a "me too" Qualcomm, they focused their meager resources in a given market segment, and in fact they did bring a competitive solution to face Qualcomm. You don't limit the scope of your projects, you limit the scope of your company by choosing where to fight.

AMD always refused to do that, and instead react to each Intel move in a Pavlovian mode. Intel focusing on servers? Then let's build an architecture from the scratch tailored for servers. Ops, but now Intel is focusing on ULP chips? Let's make a chip for here too. Intel focusing on netbooks? Gotcha, let's have our netbook chip too.

 

[frozentundra123456]

Once we analyse Bulldozer today, they are not a conclusive step ahead from Thuban. Even some die hard AMD fanboys didn't find compelling an upgrade from the first BD and only now with Vishera are thinking about upgrade. Still, Vishera improvements pale in comparison with Intel improvements with Nehalen, and Nehalen was already far ahead of whatever AMD threw at them.

That said, and given the more technical discussions about the concept here, is Bulldozer really a budget problem, or a problem akin to the Netburst architecture, a fundamentally flawed concept that no matter how much money you sink on this pit, the chip won't work?



At the engineering level, that's what you do. But there are things you can do at the management level and AMD management isn't deserving to earn the millions they assign to themselves every year.

What AMD has always tried to do is to become a "me too" Intel, so once Intel fields something new you can expect two or three years to have AMD "me too" product on the market. Servers, netbooks, ultra-thin notebooks, you name it. This does not work for the reasons you explained above.

But look at, for example, Nvidia. It does compete with Qualcomm? Yes and no. Nvidia does not touch a lot of segments, but in a given segment, premium phones and tablets, they do. Instead of using the dumb AMD strategy of trying to become a "me too" Qualcomm, they focused their meager resources in a given market segment, and in fact they did bring a competitive solution to face Qualcomm. You don't limit the scope of your projects, you limit the scope of your company by choosing where to fight.

AMD always refused to do that, and instead react to each Intel move in a Pavlovian mode. Intel focusing on servers? Then let's build an architecture from the scratch tailored for servers. Ops, but now Intel is focusing on ULP chips? Let's make a chip for here too. Intel focusing on netbooks? Gotcha, let's have our netbook chip too.

At one time, such as in the A64 days, AMD did innovate. I remember they were one of the driving forces for 64 bit processors and were first with integrated memory controllers and "true" quad core chips. They are still "innovating" with fusion, the Bulldozer module concept, and gpu computing. The problem to me seems more to be one of poor decision making and execution. They seem to always be looking for some pie in the sky concept that will put them magically ahead of intel instead of producing a superior product for the current environment. On the execution side, they have been late on product after product, and are at a severe disadvantage because of their lack of fabs and intel's process advantage.

 

[Torn Mind]

There are a multitude of reasons.

From an actual performance per dollar standpoint, the only AMD chips with decent value for the dollar are the Trinity and the FX-6300.

Hidden costs. The FX-line requires more wattage from PSUs. Trinities require high speed RAM to get the best out of its GPU. And of course, the hidden variable cost of electricity.

General consumer behavior. AMD is not well known or viewed with suspicion among the average, typical consumers, who are hardware-illiterate and use various heuristics to buy a computer.
These average consumers also are trending towards mobile products such as laptops, tablets, etc, reducing sales of both AMD and Intel desktops.

Those with Phenoms IIs and Bulldozers probably feel that the performance gains from Vishera aren't enough over Bulldozer and they are waiting for Steamroller. Or they bolted to Intel i5s after Bulldozer. In other words, previous gen processors are still "good enough" for some people and they are sticking with what they have. This latest "good enough" is not "better enough" for them.




Pretty much, Intel has corneedr the low end with the Celerons while the the i5s own the $200 and above end. So, AMD is left trying to find niche consumers in the above 50 and under 200 dollar range. Namely, the brand loyalists, the ones who need GPU power on a budget, and guys who like tweaking with their processor to get more performance regardless of stability or added hidden costs.

 

[pc999]

I would blame 32nm vs 22nm, their CPU as they are would be really good on 22nm (~half the die size and lower price) and its expected power usage (~30-40% less)!

 

[pablo87]

Very good question. For me, trust.

How can you trust a company with an unhealthy relationship with its main supplier/banker/shareholder, complete mgmt turnover, constant layoffs, no CEO for 9 months, over promising all the time, losing money hand over fist, one time charges the norm rather than the exception, new CPUs launched with old chipset, etc...

the processor is good enough but the company ain't.

 

[lakedude]

At one time, such as in the A64 days, AMD did innovate. I remember they were one of the driving forces for 64 bit processors and were first with integrated memory controllers and "true" quad core chips..

Yes and I remember when they finally came out with their "true" quad core chip and it wasn't as good as the Q6600 which had been out for some time. That was the beginning of the end.

For what it is worth there is a good chance that I'll buy another AMD laptop soon. I'd rather have an Intel Ivy Bridge but the A-10 is matched up with the awesome HD 7970m for a lot less money than Intel based systems. The great GPU is more important to me than the so-so A10 CPU.


http://www.pro-star.com/index.cfm?mainpage=productdetail&model=GX60-1AC-021US

 

[OBLAMA2009]

Hi,

A lot of people claim that the current desktop CPUs are already "fast enough" for the average user. This is also said to be the reason that focus instead lately has been on e.g. improving the iGPU, integrating VRM and Memory Controller, and lowering power consumption instead.

But if that is true, then how come AMD desktop CPUs are not selling so well? After all they should be fast enough for the average user too, since they are not that much slower than the corresponding Intel CPUs, right? And since the AMD CPUs are cheaper too, it should be a better option for a lot of consumers. But for some reason most people still buy Intel CPUs. How come?

because when dudes say "the current desktop CPUs are already 'fast enough' for the average user" they are referring to intel cpus since no one uses amd chips. amd chips arent fast enough.

 

[frozentundra123456]

Yes and I remember when they finally came out with their "true" quad core chip and it wasn't as good as the Q6600 which had been out for some time. That was the beginning of the end.

For what it is worth there is a good chance that I'll buy another AMD laptop soon. I'd rather have an Intel Ivy Bridge but the A-10 is matched up with the awesome HD 7970m for a lot less money than Intel based systems. The great GPU is more important to me than the so-so A10 CPU.


http://www.pro-star.com/index.cfm?mainpage=productdetail&model=GX60-1AC-021US

Yes that looks very attractive. I am sure you saw Jared Walton's preview on that, and that he will come out with a full test. I would wait for that to see if you are severely CPU limited in more demanding games.
If you are gpu limited you can turn down some settings, but if you are CPU limited there is not much you can do.