As much as I wanted a Haswell rig, now I'm thinking twice about it.

[poohbear]

U hafta ask urself what u wanna do with ur system. If u dont game, why would u neednto upgrade in the first place? Ur rig seems like its more than enough for ur needs.

 

[Homeles]

As usual, it seems like AMD is out-innovating Intel, what with them giving 8-core CPUs to the consumer.

If your definition of innovation = core count, sure. That is a very narrow view, in my opinion.

Once AMD was hailed as the great innovator for integration. They were the first to put their memory controller on die. They've stagnated in that regard, while Intel has all systems go in the direction of having SoCs. AMD's too worried about socket compatibility -- yeah, I can upgrade 4 years from now. Great. But progress is slowed as a result.

That's not to say AMD isn't innovating -- they certainly are. They took the Bulldozer plunge, they've been increasing their developer relations in a push for OpenCL support, there's HSA, and the passable graphics on a budget chip. I'm not going to continue listing their progress, but they're certainly not slacking.

However, Intel is innovating as well. We've seen their push for Larrabee/MIC. Haswell will have on-package voltage regulation, which should prove to be a huge boon for overclockers and average consumers alike. Intel's stopped fooling around when's it comes to graphics. They'll be the first to have a system on a package with Haswell GT3. They'll be the first to have embedded DRAM on package. They've had a huge hand in ensuring Android's compatibility with x86.

So while you may see this as their CPU development stagnating, is it not a bit selfish to think that if Intel isn't developing in the area you want them to, that they're not developing at all? I'm not saying that this is your position, but it is a common sentiment in enthusiast communities.

 

[Ajay]

8 Core? You are going to be waiting awhile. Even 6 core is likely Skymont at the earliest. And that's not even a given, especially if Intel makes the mainstream "default" laptop processor 17W. I still think Intel is going to use any die savings on integrating more functions (eg: chipset, wifi, ram, disk) onto the die/package, maybe more GPU EUs.

IIRC, Broadwell will have a new, and more powerful, iGPU. So, yes, Intel will be using some of those extra xtors for non-CPU functionality; Intel will also keep beefing up the cores to stay well ahead of ARM.

 

[phillyman36]

Can someone give a sort of brief know differences between Haswell and Ivory Bridge(other than graphics) with an easy to understand definition? Trying to understand the whole system on package, embedded dram, etc etc. Thanks

 

[Charles Kozierok]

The only "innovation" associated with AMD's "8-core" CPUs is in their marketing department.

 

[SunRe]

Hi,

Somewhere in a leaked table with the Haswell SKU I noticed that most of the parts, with the exception of the unlocked ones, if I recall correctly, were missing VT-x/VT-d instructions for virtualization.

Any news on that?

 

[Idontcare]

Not really. I agree with him - Intel has slowed down their innovation, by not giving us more than 4 cores on a consumer CPU. Instead, as he has shown, they are just fattening their profit margins, because AMD is doing so poorly in competing with Intel.

Edit: Not that I would think that any powerful, monopolistic corporation would do any different, really. But I think that the claims that Intel wouldn't stagnate innovation, because they have to compete with themselves, is false. I do think that if AMD fails to compete, that Intel will stop competing with itself, and we will be stuck with the modern equivalent of the P4 for years to come. (P4 meaning the current status quo of CPUs at the time, rather than what could have been.)

Edit: Especially if the desktop market shinks, in response to mobile and ARM market. Intel will be a lot less willing to pour R&D funds into desktop CPUs, and their performance will level out, and then Intel will just simply milk what they have, for as long as they can, and as long as the market exists.

In fact, some, like the above poster, may seem to think that this is already happening, and the lack of anything more than a quad-core (which was introduced in 2007-2008) in the consumer space, is prima facie evidence of such.

As usual, it seems like AMD is out-innovating Intel, what with them giving 8-core CPUs to the consumer.

If you are a consumer who is otherwise not a professional in the field or industry from which the consumed product is derived then all you know of "innovation" in that product is what you have been told to think you know from various marketing arms of the company trying to sell the product.

There is a lot of innovation that goes on behind the scenes that we laypeople simply aren't brought up to speed on by marketing folks.

Take for example the fact that you probably don't really have a clue as to what makes Intel's CPU's have higher IPC than AMD's at the circuit level...but the difference is called innovation and it came from talented folks who were resourced and directed to make it happen.

What you get spoon fed to view as innovation is not to be confused with the actual innovation that is going on in engineering, accounting, marketing, and general management in a modern-day semiconductor business. You are too far removed from that reality to even begin to comprehend all that you don't even know exists behind the scenes.

(just another way to say that an illiterate man has little choice but to judge a book by its cover, a cover that was engineered by the publisher to convince the passerby that the contents of the book are worthy of the pricetag listed on the back of the book)

I know what I know of semiconductor market which only prepares me to accept that I know virtually nothing about those industries for which I have no direct personal experience.

All I know of the innovation that went into my Prius is what Toyota's marketing team decided I needed to know. But I am smart enough to know that the list of innovations doesn't end with the marketing blurb, there's a lot more that goes into the hybrid power train than what can be put into a single-page advert in popular science.

 

[frozentundra123456]

Oh just what I wanted, a snot-nose to repeat something every Joe in the world knows about tech by reading People magazine.

My response was to the question of why Intel doesn't throw more cores at these processors, for one. My answer is that they have gotten lazy and greedy, and that will eventually bite them in the ass. This seems to bother you...

So, lets see how well Intel's strategy is working for them, shall we? Let's just keep it to facts :

Intel's Profit Falls 27% as PC Sales Drop - The New York Times
https://www.google.com/url?sa=t&rct...La4IWMPffB7P9WZMw0m8vKg&bvm=bv.43828540,d.b2U

Wow, nice job! Of course, still very profitable, for now...


And as far as the teenage banter of "they can't do anything about it blame physics" statement you made, please spend a few moments educating yourself :

http://techreport.com/review/23663/as-amd-struggles-intel-chip-prices-stagnate


"...almost two years after the release of the Core i5-2500K, Intel still doesn't offer an unlocked quad-core processor for less than $200. Meanwhile, Intel's gross margin has climbed to an eye-popping 63.4%, nearly 20 points higher than AMD's."



http://factions.in/article/Hardware...Intels_Lead_Over_AMD_Has_Help_Worsen_It/74347

"...Ivy Bridge duals have a die size of 94mm2. Do you realize how many i3s fit on a wafer? FX 8350 is 319mm2. i5s IBs are either 133mm2 or 160mm2. Do you have any idea what kinds of margins Intel must have on those chips? "


Basically, intel does not enhance the performance or add cores because they don't have to right now. They add to their profit margin with the smaller die, and charge a premium if you want a die 2/3s the size of an fx-8350 (ie, the i7 39xx 6 core chips).

That, just in case your great intellect missed it, has nothing to do with physics. It has to do with finance.

And of course, I don't suppose you ever considered why mobile is so big, now? I mean, we've had mobile a long, long time. Early 90s, we started getting smart mobile devices. Late 90s, we had touch screens. Why now?

Oh, that's right, because the devices got powerful enough to do cool things just like a desktop, right? Just like a desktop...

Is that the same desktop that's stagnated for the last 5 years maybe?

So maybe Intel ate its own lunch, you think, maybe? I mean, if they had actually had competition the last 5 years maybe the desktop would be something different, and maybe mobile would still be seen as a poor shadow of the desktop.

But it isn't, not anymore. My iPad browses amost as well as my macbook, and my 'apps' have replaced 'web sites'.

Never mind that a dual core ARM CPU from 2011 is only 1/5 the power of a dual core Atom, it is good enough for tasks we were doing 6 years ago on the desktop - and those are the same tasks we're doing today.

So yeah, you are right, Intel and MS are going to get killed by mobile - but it's not because it's mobile, it's because Intel and MS stupidly wasted half a decade.

Well, you are basically saying why Intel is doing everything right from a business standpoint. If they can make a big profit on a quad core that basically matches an 8 core from AMD in all but a few areas, and is faster in some, why shouldnt they? It sounds harsh, but they are a business and in it to make money, not cater to a very small portion of the market that demands absolute highest performance. I do think they are taking a risk relying on hyperthreading and high IPC instead of throwing more cores at performance, but I am sure they could bring out a hex core at a lower price if they need to.

 

[frozentundra123456]

Adding cores is not innovation. And just because the industry doesnt go your way doesnt mean its all sour grapes as you write. You can already get the cores you wish, but as usual you do not want to pay for it.

Performance/watt have been increasing dramaticly. As well as integration. And this is what 95%+ of all customers ask for.

Haswell also contains the ability massively outperform SB/IB when AVX2 takes off.

And AMD dont give you 8 core. They give you a quadcore with CMT. Intel gives you a quadcore with SMT if we exclude LGA2011. Its simply 2 different paths, but both are quadcores.

Not to mention, I doubt you see those "8 cores" on FM2 or FM3. And AM3+ is dying fast. Sofar no SR core update in view.

When that is said, R&D budgets havent been bigger. Intel for example is throwing more money than ever before after x86. So they are certainly not slacking. And AMD is certainly not out-innovating Intel.

I agree with you in principle, but I will give AMD credit for one thing at least--They are pushing their CPUs to the limit of their performance, while Intel seems to be leaving a lot of performance on the table. Granted for most things, AMDs architecture is not as efficient as Intel's, but every new revision pushes TDP and clockspeed to the max, while Intel stubbornly seems to resist even increasing clockspeed, much less adding more cores.

 

[Idontcare]

I agree with you in principle, but I will give AMD credit for one thing at least--They are pushing their CPUs to the limit of their performance, while Intel seems to be leaving a lot of performance on the table. Granted for most things, AMDs architecture is not as efficient as Intel's, but every new revision pushes TDP and clockspeed to the max, while Intel stubbornly seems to resist even increasing clockspeed, much less adding more cores.

Depending on how you want to look at it, they are doing exactly what the market told them it wanted when they struggled to sell Prescott against AMD X2's.

If the market really wanted 125W TFP 8-core chips then Rory would be having a very different conversation with shareholders and analysts.

Intel learned from the market 7yrs ago that the market wants sub-100W TDP chips and it doesn't really want/value anything that comes with more than 4 cores.

It was the magic formula for AMD and no one faulted them for it. Intel figured it out, improved on it (drove TDP to 77W now) and reaped the same rewards AMD thought they were going to receive back when they were counting their chickens before they hatched.

The market doesn't want or value chips pushed to their absolute max. AMD is proof of that, proof 7 yrs ago and proof today. The problem for AMD is they seem to have forgot what made their products desirable back then, and Intel hasn't forgot what made their products undesirable.

 

[Fjodor2001]

With the PS4 and XBOX720 we'll be seeing 8 core x86 consoles this year.

Therefore we'll also see games making use of 8 cores, and hence likely also a demand for 8 core x86 gaming computers.

So if Intel does not want to provide 8 core mainstream CPUs in the next few years to cover this market segment, perhaps AMD will.

 

[Ajay]

Hi,

Somewhere in a leaked table with the Haswell SKU I noticed that most of the parts, with the exception of the unlocked ones, if I recall correctly, were missing VT-x/VT-d instructions for virtualization.

Any news on that?

It usually the other way around, with unlocked CPUs leaving out VT-d.

 

[tweakboy]

I think this i5-3570k will be my last quad core. Its either going to be IVB-E, HSW-E or Steamroller... which ever comes out first.

Ivy Bridge E 6 or 8 core monster

All I gotta do is buy it and slap it on my current rig.. and wow 16 threads. gl

 

[ShintaiDK]

With the PS4 and XBOX720 we'll be seeing 8 core x86 consoles this year.

Therefore we'll also see games making use of 8 cores, and hence likely also a demand for 8 core x86 gaming computers.

So if Intel does not want to provide 8 core mainstream CPUs in the next few years to cover this market segment, perhaps AMD will.

Its far from certain consoles will use those cores for gaming. They didnt on the previous consoles.

 

[Ajay]

Not me. I'm holding out for 14nm goodness. 22nm is already old school

Wish I could wait, maybe I can if I find a i7 980 on the cheap.

Edit: NOT! 980X CPUs are going for >= $500, waste of money spending that much on old tech.

 

[shady28]

Well, you are basically saying why Intel is doing everything right from a business standpoint. If they can make a big profit on a quad core that basically matches an 8 core from AMD in all but a few areas, and is faster in some, why shouldnt they? It sounds harsh, but they are a business and in it to make money, not cater to a very small portion of the market that demands absolute highest performance. I do think they are taking a risk relying on hyperthreading and high IPC instead of throwing more cores at performance, but I am sure they could bring out a hex core at a lower price if they need to.

No, that isn't exactly what I was saying (referring to my original post).

What I'm saying is they are doing everything *typical* of a big business, from a financial standpoint.

That does not necessarily mean it is good for their long-term viability. You probably don't recall the massive collapse of DEC, a company that was crushing IBM at one time. Nor do you likely recall that IBM was thought to be about to die. One of these survived and thrived, one did not. One of these branched out into many different products, one consolidated and focused.

This is not something to be lauded or praised, it simply is.

Perhaps the right question needs to be asked :

"Why don't you need a more powerful CPU?"

I would guess that the answer is - you have nothing to use it on. And that, folks, is due to - lack of innovation.

 

[Piroko]

If you want to know where a company is investing the bulk of their R&D you need to look no farther than to the product they are releasing first on any given brand new node.

I don't say that as a matter of opinion, that is a matter of fact and accounting when it comes to semiconductor design and manufacturing.

The day Intel pushes out Atom or Xeon Phi or Itanium on a new node before they push out "big" core mainstream x86 products is the day you will have confirmation that their R&D monies have stagnated or declined in the big core x86 area.

Until that happens you have proof of where they are spending the bulk of their R&D, look at what was released first on 32nm, first on 22nm, and planned to be first for 14nm.

I'm not arguing that the big cores don't get the largest piece of the pie, but in the last two years they've massively accelerated Atom development and in the last months they've pushed back IVB-E. Those are decent indicators that priorities aren't as clear cut as they used to be and additional R&D budget isn't focused on big cores imho.

 

[Homeles]

I would guess that the answer is - you have nothing to use it on. And that, folks, is due to - lack of innovation.

Well good. Then it's AMD's fault as well. Or software devs. So really, this has nothing to do with Intel at all. Might as well start blaming the clouds for your problems.

 

[Denithor]

Not me. I'm holding out for 14nm goodness. 22nm is already old school

Actually in this case waiting for the next die shrink instead of grabbing the newest micro-architecture might be an excellent move. A large part of the expected improvement in Haswell seems to come from AVX2 and TSX. Both of which are going to have to be implemented to some degree before being truly useful and providing a significant boost.

Perhaps by the time the next 'tick' comes around those will have some level of support and you'll get the full advantage of both those and the new smaller node.

 

[Fjodor2001]

Its far from certain consoles will use those cores for gaming. They didnt on the previous consoles.

Perhaps because it was only PS3 that had 8 cores? I.e. the game developers do not want to create a game that only can make use of 8 cores on one of the consoles.

Now both the XBOX720 and the PS4 (and possibly future AMD desktop CPUs) will have 8 cores, and they will also both be on x86. I.e. designing games for 8 cores will result in greater returns than before.

 

[ShintaiDK]

Perhaps because it was only PS3 that had 8 cores? I.e. the game developers do not want to create a game that only can make use of 8 cores on one of the consoles.

Now both the XBOX720 and the PS4 (and possibly future AMD desktop CPUs) will have 8 cores, and they will also both be on x86. I.e. designing games for 8 cores will result in greater returns than before.

The Xbox360 got 3 cores. I dont even think a single game used all. They like to assign cores to different functional tasks on consoles. 1 alone goes to the OS.

I dont think near future AMD desktops will have 8 cores. The next SR FM2 core is still "quadcore" and on 28nm. And AM3+ seems dead, meaning the "octocore" is actually dead on the desktop for now.

 

[Fjodor2001]

The Xbox360 got 3 cores. I dont even think a single game used all.

So the PS4 and XBOX720 will have 8 cores, but max 2 of them will ever be used by any game. Is that's what you're saying?

Then how come Sony and MS didn't order a 2 core CPU for their next gen consoles?

 

[lau808]

So the PS4 and XBOX720 will have 8 cores, but max 2 of them will ever be used by any game. Is that's what you're saying?

Then how come Sony and MS didn't order a 2 core CPU for their next gen consoles?

he is saying all 8 won't be used by the game. at least 1 will be strictly for the os and maybe others for streaming or downloading etc etc. at most those games will be coded for 7 cores. weird number tho so maybe 6?
just my opinion on how i took his post and it makes sense to me

 

[Homeles]

So the PS4 and XBOX720 will have 8 cores, but max 2 of them will ever be used by any game. Is that's what you're saying?

Then how come Sony and MS didn't order a 2 core CPU for their next gen consoles?

Games are notoriously difficult to make use of multiple threads. They're lightly threaded apps by nature. You can't make a processor run faster by making a simple 1 + 1 add operation run on multiple cores. More cores aren't the answer to everything.

 

[Revolution 11]

he is saying all 8 won't be used by the game. at least 1 will be strictly for the os and maybe others for streaming or downloading etc etc. at most those games will be coded for 7 cores. weird number tho so maybe 6?
just my opinion on how i took his post and it makes sense to me

More likely that the 7th core will be kept as a spare and 6 will be available for games. Though it is just plain hard to make game software multithreaded enough to use more than 2 or 3 cores.