Intel's Haswell Architecture Analyzed: Building a New PC and a New Intel

http://www.anandtech.com/show/6355/intels-haswell-architecture

by reitzensteinm • 14 years ago

60 17 14 years ago

mtgx• 14 years ago

Anand is too optimistic about Apple adopting Intel chips in iPads or iPhones in the future. Personally, I don't think that's ever going to happen, and I don't think it makes any sense either. Apple is trying to become an ever more integrated company, and apparently making their own chip is becoming very important to them. And recent rumors start pointing to the same conclusion, although we probably won't see this happening until 2014:

http://www.techradar.com/news/computing/apple/apple-reported...

From what I hear Haswell will be significantly more expensive than Ivy Bridge, too, so again, to me it looks Intel is becoming less competitive with ARM, not more. But I think having full control over their chips matters more than anything else to Apple, and I think they will move there at least gradually. Apple's A6 is a significant step in that direction, and I don't see them moving back from it.

reitzensteinm• 14 years ago

Where does he say that? He seems to be saying the opposite here, about Apple dropping Intel for laptops:

"There's even more reason to worry if the post-Steve Apple/Intel relationship has fallen on tough times. While I don't share Charlie's view of Apple dropping Intel as being a done deal, I know there's truth behind his words."

bryanlarsen2• 14 years ago

Anand is probably talking more about Windows 8 tablets. We're already getting Ivy Bridge based tablets, so it's obvious that we're going to get Haswell tablets. The article just shows us how compelling those tablets are going to be, as long as you don't paying a few extra dollars.

I suspect that this also means that we're going to get Haswell based Android tablets, which along with initiatives like Ubuntu for Android might be really compelling to a certain subset of HN readers.

bornhuetter2• 14 years ago

I really hope that the threat of Apple switching to ARM causes Intel to push Linux as an alternative to OSX.

I'm not holding my breath, but I can only hope.

chucknelson• 14 years ago

While some Linux distributions have come a long way, I still don't think this will happen any time soon. The ease of use is still far behind OS X and Windows in my opinion.

Granted, I've only used Ubuntu (12.04 LTS mostly), but that's one of the most user-friendly distributions I think.

bornhuetter2• 14 years ago

For the average user, I think Windows and OS X are easier to use than current Linux distributions, but for a developer Ubuntu is getting pretty good.

I disliked Unity at first, but it's really starting to grow on me - especially after using Ubuntu Tweak to change a few things I didn't like. There are some great shortcut keys, like Super-W, Super-S and Crtl-Alt-Arrow to switch between workspaces and windows, that I think are as good as OS X gestures. And the package manager is superior in many ways to the App Store and the Windows 8 Store.

Intel have done a good job so far of kicking OEMs into shape over ultrabooks - and manufacturers are finally starting to get it. If they were to make a similar push for linux based developer laptops, where all the drivers work properly, I could see it potentially being very good for the industry.

tedunangst2• 14 years ago

Intel has supported Linux for a long time. There's an intel compiler for Linux but not for osx...

octopus• 14 years ago

Sure it is, Intel has a C++ and a Fortran compilers for OSX ...

mdasen• 14 years ago

Linux is unlikely to be an alternative to OS X (for most users). The key is in looking at why most users decide to buy an OS X machine. Many people on here probably bought one because it provides a decent nix style OS with normal command line interface and the ability to run the things they use for development in much the same way as their Linux deployments. However, that isn't the majority of OS X users.

The majority of OS X users probably bought it because it felt more like a consumer electronic device than their Windows computer did. Apple's integration between the hardware and software means that things like trackpad drivers, sleep mode, ambient light sensors, etc. all just work nicely and smoothly. Many PC manufacturers offer features thinking that a feature is just a boolean: it has it or it doesn't have it. For example, a friend of mine has a Sony laptop with a backlit keyboard - just like my MacBook. Except that, in practice, it isn't like my MacBook. Either the light sensor is faulty or it doesn't use one, but it doesn't change the backlight brightness automatically. When you manually change the brightness, sometimes a box on the screen comes up showing you raising and lowering it and sometimes it doesn't. The same can be said for the WiFi helpers that companies seem to install to manage the WiFi: they feel cheap and poorly executed. So, for many users, a MacBook is more like buying a consumer electronic where you feel like it's been manufactured as a cohesive product a little more than the Windows environment where it feels like an OS on top of hardware with the pieces supporting hardware specific functionality created hastily and poorly.

If Intel pushes Linux as an alternative, it could help Linux along for those wishing they had a Windows with a nix development environment (which is what I think you mean when you say "alternative to OS X"). It's unlikely to create something that most OS X users would want since mainstream Linux would mean the same hastily made pieces to do hardware specific things. Even look at Android. It's great, but what does everyone complain about: the ways in which manufacturers customize it to make themselves seem different from everyone else.

In terms of the likelihood that Intel will do this, I think it's highly unlikely. You talk about Apple (potentially) switching to ARM and Intel losing that. However, if Intel pours money into Linux to make it a better OS than it is today, that isn't helping Intel's case. Linux can transition to ARM a lot easier than OS X can. Most Linux software is open-source meaning that even if the original maintainer is gone, someone else can recompile the software. For OS X and Windows, it would be a lot harder for them to transition processor architectures since they would need developers to do that work or provide a translation layer like Apple did moving to Intel.

Making Linux the best OS out there would only make it easier for people to leave Intel processors. With OS X and Windows, Intel has a certain amount of lock-in. It's growing smaller, but there's a lot more lock-in than with Linux.

Ultimately, Intel has to compete to make a processor that people want.

bryanlarsen2• 14 years ago

I think to most people, Android feels more like a consumer electronics device than even OS X. Yes, Samsung Android is different than HTC Android, but only geeks care about that.

Intel is putting a lot of effort into making sure that Android runs well on Intel. As a side effect, stock Linux will run well, too, since the vanilla and Android Linux kernels are converging.

dmorgan• 14 years ago

>I really hope that the threat of Apple switching to ARM causes Intel to push Linux as an alternative to OSX.

And what exactly would that achieve? It's not as if consumers listen to Intel about what OS to buy.

There is a way for Linux to matter as an alternative to OS X (and Windows) and that is for it to get more like OS X (or Windows): Intel, Google or some player(s), could hire an experienced project manager guy, and a 100-200 person team, plus QA testing and documentation writers. Preferably most of the core team would be the top volunteers at their specific projects. The then should spends some in the range of a billion a year for a couple of years, and fix EVERY pain point Linux has for consumers, especially multimedia use which is crap (e.g, a Joe guy trying to make a 5 minute video of his holidays). They should also use all that money to influence design decisions in a "benevolent dictatorship" manner, so that we finally have things like long term binary AND driver stability and third party apps for Linux.

Someone• 14 years ago

I would think the threat of Windows moving to ARM is much bigger. Microsoft already announce an ARM version of Windows 8, and there is more revenue to lose there for Intel.

However, I doubt Intel will start pushing Linux in response. The strategy of building more power efficient x86 CPUs seems more logical.

batgaijin• 14 years ago

I still can't help but think that the name is a little jab at the added hardware level STM support...

Symmetry• 14 years ago

Good article. I've always wondered how the bypass network on Intel chips works, though. Is it between the ports or directly between the execution units? I also can't imagine it being fully bypassed in either case, so I wonder what it looks like.

theevocater• 14 years ago

Intel is making the necessary steps to stay relevant as we see smaller lower power devices become the norm. Ultrabooks (and their silly netbook cousins of the past) are merely stepping stones on the path to more embedded, power sipping devices everywhere in our lives.

I'm curious to know what this means for Nvidia. Nvidia is now essentially competing on both big fronts. On one end, nvidia has to stay relevant with its video cards. Things like Tesla are going to be big business, but how do you compete with intel's on chip offering and keep power demands low enough? On the other side of the wall you have Tegra. Intel is a much easier target here as the Tegra is arm based and can already be found in android phones and tablets. And as the article indicates Haswell is still not equipped to truly compete with Tegra at the low end of the power spectrum.

Also, and it hurts to mention it, what about AMD? Bulldozer was at best a lukewarm release and piledriver still isn't out. On the other hand, AMD's graphics offerings still compete well with Nvidia. Big box retailers don't offer much in the way of AMD but ATI still sees significant usage. Dell, for example, only offers opterons in its servers, but offers plenty of AMD graphics cards in consumer offerings. But as AMD still has the same problem as Nvidia: this side of the business is being disrupted by things like Ultrabooks and tablets though so its hard to see how this will last. For AMD to compete with intel, it would seem they need to focus on SoC designs.

Lastly, one thing I found very interesting was the various winks and nods to the fact that intel creates a lot of interesting features that are more expensive, but only 'one' manufacturer ever uses them with the need for the stick. That manufacturer is clearly Apple.

People often like to parrot the meme that Macs are simply more expensive PCs but it seem clear here that other manufacturers require the stick to get them to make improvements to their boards.

> Intel gave one example where an embedded controller on a motherboard was using 30 - 50mW of power. Through some simple firmware changes Intel was able to drop this particular controller's power consumption down to 5mW. It's not rocket science, but this is Intel's way of doing some of the work that its OEM partners should have been doing for the past decade. Apple has done some of this on its own (which is why OS X based notebooks still enjoy tangibly longer idle battery life than their Windows counterparts), but Intel will be offering this to many of its key OEM partners and in a significant way.

You don't drop power consumption by an order of magnitude that easily unless the original chip was seriously poorly designed.

Intel's Haswell Architecture Analyzed: Building a New PC and a New Intel

Related Stories