Slashdot Mirror

For those that are interested in a closer look behind the scenes of the Frostbite 2 engine DICE recently held a 1 hour talk about the inner workings of the graphics in BF3. It's pretty amazing what can be done with DX 11 these days.

It's possible that I've misunderstood something, but it was my impression that one thread was dominant so that if they both want the same resource at the same time it'd always come first.

But how should it choose which thread is dominant? Generally speaking, neither the OS nor the HW has any idea which thread is most important.

I believe the only policy for HT resource allocation is more or less a fairness policy -- because if anything they want to ensure that no starvation takes place. It's also known that some resources are just allocated in a fixed 50/50 split (e.g. reorder buffers in Nehalem) to simplify the implementation.

That way a heavy single threaded application would get nearly the same performance as before unless it was blocked by calculations already in progress, while simultaneously letting it do light work on the side. Early hyperthreading performance was the same for most workloads, indicating to me that one thread run practically all the time and the other didn't get any work done at all. That is at least how I interpreted the numbers.

I'd interpret it as much of the list being single-threaded benchmarks, especially the games (at that time it was uncommon for games to do computation on >1 thread). Anand didn't have a lot of multi-threaded tests in his toolkit back then!

What makes you think Intel has one fast thread and one slow thread filling the gaps? As far as I know, both threads share the core's resources equally and the CPU doesn't favor one over the other.

It's possible that I've misunderstood something, but it was my impression that one thread was dominant so that if they both want the same resource at the same time it'd always come first. That way a heavy single threaded application would get nearly the same performance as before unless it was blocked by calculations already in progress, while simultaneously letting it do light work on the side. Early hyperthreading performance was the same for most workloads, indicating to me that one thread run practically all the time and the other didn't get any work done at all. That is at least how I interpreted the numbers.

I suspect part of the issue is its easier to justify power budgets and not as easy to justify 8000 more per server to upper admins.

Well, it's not that much alone but it's another thing added to the total cost that lowers the price AMD can charge. A Bulldozer uses about 70W more than a 2600K, that's about 600kWh/year. Say 5 years lifetime and 10 cent/kWh and that is $300 more per CPU to operate it, not counting scaling up the power supply or the air conditioning. Compared to $8000 that's little, but as part of the profit margin AMD could have had it's probably a lot. Of course a 2600K wouldn't be in direct competition on the server anyway, but I'm to lazy to look up the appropriate Xeon.

iPhone: 640 x 960 Galazy[sic]: 480 x 800

Galaxy Nexus: 720 x 1280 (4.65" screen 315 ppi)

Plus, while it may be a personal preference, SAMOLED >> LCD.

Galaxy Nexus: 720 x 1280 PenTile. http://www.anandtech.com/show/5000/galaxy-nexus-pentile-discussion-confirmed

The earlier article i read must have been way off then.

Here's a set of graphics displaying the actual architecture.

http://www.anandtech.com/Gallery/Album/754#7

You might consider waiting for a zenbook from ASUS. It is essentially a macbook Air without the apple logo. ASUS usually have pretty decent linux support too. http://www.anandtech.com/tag/zenbook

Supposedly the fixed one BSOD bug a few days ago. That wouldn't be with this controller anyway, but their record isn't spotless. Then again, Intel managed a SSD blemish too so... you're seeing an industry moving at breakneck speed, just make sure yours isn't on the line.

Using OCZ 8 months now. I tend not to get BSOD's on Linux anyhow.

10/17/2011 : After months of end user complaints, SandForce has finally duplicated, verified and provided a fix for the infamous BSOD/disconnect issue that affected SF-2200 based SSDs.

wow, that's not something anyone wants to see, a bug in their hard drive. CPU I can replace, ram I can replace... pretty much everything I can swap out, but my hard drive is where everything is stored, I can't risk losing data because of a bug.

Intel just had problems too that cause loss of data:
"JULY 13TH, 2011 : Intel has recently acknowledged issues with the new SSD 320 series, where by repetitively power cycling the drives, some may become unresponive or report an 8MB drive capacity."

I was waiting on an SSD until they worked out the bugs and there were no articles about problems for awhile but with stories like these I'll keep waiting, it's just not worth the risk.

Supposedly the fixed one BSOD bug a few days ago. That wouldn't be with this controller anyway, but their record isn't spotless. Then again, Intel managed a SSD blemish too so... you're seeing an industry moving at breakneck speed, just make sure yours isn't on the line.

From what I've read, the galaxy s II is actually a step above the new nexus prime. Based on the soc they used and the graphics process in that soc vs the galaxy s II

How so? The S2 was less than 10% faster than the previous generation OMAP 4430 in the Optimus: http://www.anandtech.com/show/4686/samsung-galaxy-s-2-international-review-the-best-redefined/14

With the clock stepping jump alone from the Optimus to the Prime (20% on the core, ~25% in the graphics engine), the Prime should edge out the S2. If there are hardware acceleration advantages to Ice Cream Sandwich, the difference could be dramatic until the S2 gets updated.

Actually looking over the two...I am unsure why anyone would pick the prime / galaxy nexus over S2

1) Galaxy Nexus is on Verizon.
2) The OMAP 4430 at 1GHZ (LG Optimus HD) was already neck and neck with the Exynos in the S2. The Galaxy Nexus has a 1.2 GHZ OMAP 4460, which should be enough to cap out all these charts: http://www.anandtech.com/show/4686/samsung-galaxy-s-2-international-review-the-best-redefined/14
3) The Galaxy Nexus has more pixels. For folks with good vision, that's going to be visible.

They had a top notch cool phone, but now it's par, if not lagging behind many android devises.

Citation needed.

And, let's make it clear, you won't find a citation to support your comment because it is false.

Here - unlike you, I'll back up my claim:

http://www.anandtech.com/show/4951/iphone-4s-preliminary-benchmarks-800mhz-a5-slightly-slower-gpu-than-ipad-2

The iPhone 4S is faster than virtually every mobile device currently on the market with the sole exception of a two top-tier tablets (the iPad 2 and, in one test, the Samsung Galaxy Tab, which have more space enabling more power). Against other phones, the 4S is currently ahead of everything else.

So, how is it that you can claim that the iPhone 4S is only on par, if not lagging behind Android devices when the facts don't support that?

Actually the T2 is one of the poorest performers and last for about half the iPhone's battery life. The iPhone scores the highest (note this benchmark is from a year ago so doesn't include the latest and greatest from Android).

http://www.anandtech.com/show/3967/the-tmobile-g2-preview/4

Battery benchmarks are about the middle of the page for a variety of smartphones.

This is the comparison of a8 as a standalone cpu versus 1100T. In this test, the onboard die is not used. this is not how this platform should be run. It should be run as its onboard gpu in hybrid crossfire with a 6670 external gpu :

http://www.anandtech.com/show/4476/amd-a83850-review/6
http://techreport.com/articles.x/21730/8

- First, there is the huge delay intel caused by engaging in fraud by paying pc makers to not use amd chips, right at the time amd was at an advantage.

Yes, in the distant past.

- Then there is the fact that these synthetic benchmarks use intel's proprietary libraries, which were proven to work ineffectively when 'non genuine intel' architecture was detected.

Yes, in the past. Is this still the case? Do INTC compilers still fuck over non-'GenuineIntel' silicon? I don't think so.

- Then there is the fact that this is a new platform, and its just out, and the main deal with this is being easily increasable in cores. so amd will just add more cores without any research being needed. expect 32 core cpus in a year or so. 16 cores already out.

with kW powerdraws at the wall? I'm exaggerating, but you get my point. For a 32nm process, the physical size and power-consumption of this chip is excessively high at idle AND load.

- As you can understand these cpus are geared more for server environment, and will take that environment over.

We need to see some benches (realworld AND synthetic) before we can state that for a fact.

- Amd is moving to trinity in one year or so. Trinity is the APU format that all amd cpus will take from then on. Llano apus have been quite successful in gaming fro example 50-80 fps in starcraft 2 (crossfired and not) -> you dont need to buy an external card anymore, and if you do you can crossfire it with the cpu contained one. http://www.anandtech.com/show/4476/amd-a83850-review/6 http://techreport.com/articles.x/21730/8 intel is worlds behind in this one.

Can't argue there... although INTC is closing the gap.

and then there is the ultimate question of what the fuck i am going to do if i grab a powerful processor. really. i bought an overclockable board, and an unlocked cpu. and when i played games, i found out that it was mostly the video card i added that did most of the thing. the cpu i had was way, way over any potential requirements and needs of these games. i didnt need to buy a powerful one at all.

i went about hardware/software forums asking what i could do with a powerful computer. answers have been 'video encoding', 'benchmark', 'seti'. as it seems, any daily usage for cpus are WAY behind the power of modern cpus. to utilize your cpu power at all, you need to do unorthodox, unnecessary shit, or be in a profession that works on these.

Good points.

so i think all this performance talk is bullshit. there is no way in hell you will use that performance, even in hardcore gaming with an eyefinity 3 monitor setup in 5000x resolution, with 2x antialiasing and full detial. (and i just have 2x 5670 cards).

wat

1. The better the performance, the faster you can do shit (when you are doing it).
2. The better the performance, the longer the life of your box (you can upgrade at a later point in time).

future is in the heterogeneous chips i think. llano already has been a success, and its possible to save 30% on the cost of cpu + mobo + graphics card if you go the llano way over anything intel, and gaming performance is incomparable. when trinity comes, i think there will be a big change in computing. especially when amd puts out a computing platform like cuda.

AMD needs to push OpenCL more, but maybe they'll do that between now and when Trinity comes around. Right now OpenCL sucks ass compared to CUDA.

Thank you for your interesting post.

Quick summary timeline...

Dude sees an old MacBook Pro on Craigslist listed by Seller as broken.
Dude buys it thinking maybe he can fix it.
Dude does indeed fix it, requiring reflowing of parts, adding parts (ram, HDD, etc.), chronicles it at Anandech, noting that it seems to be no ordinary MBP.
http://forums.anandtech.com/showthread.php?t=2165252&highlight=macbook+antenna - Prototype Macbook Pro with 3G: In my shop now!

Dude then sells it on Craigslist himself as he has no use for it and doesn't yet realize its uniqueness.
Buyer takes it to an Apple Store for some service, Apple Store Genius bar says "This is not an Apple product." on account of weird things in there.
http://www.macrumors.com/2011/08/30/apple-genius-bar-didnt-recognize-macbook-pro-3g-prototype-apple-now-wants-it-back/ - Apple Genius Bar Didn't Recognize MacBook Pro 3G Prototype

Buyer sues Dude, wins (in part based on Apple Store findings), Dude is out moneys.
Dude thinks 'wtf', though, and takes a closer look at the MacBook Pro, asking around on forums.
Dude learns that the red motherboard implies it's a prototype.
Weeks pass and Dude does what anybody who isn't a fanboy would do - puts it up for sale on e-bay.
http://news.cnet.com/8301-27076_3-20092180-248/3g-equipped-macbook-prototype-pops-up-on-ebay/ - 3G-equipped MacBook prototype pops up on eBay

e-bay bids go up to $70k, listing is pulled due to request from Apple.
Dude then hears nothing, sits around waiting for some manner of official explanation for days on end.
CNet, however, now wants to know what happened, so arrange an interview, in which they of course also call Apple.

Apple suddenly takes very keen notice.
http://m.cnet.com/Article.rbml?nid=20099494&cid=null&bcid=&bid=-248 - Apple wants its 3G MacBook prototype back

Dude gets call - Apple wants their hardware back and they can have somebody stop by Dude's private residence that evening.
Dude says 'I think not, my lawyer will be in touch'.

Lawyer says Apple have no case.
Lawyer and Apple chit chat.
Lawyer says having no case matters shit all when you're Apple, so give up or incur huge costs.

Apple thus sends over a PI to pick up Prototype MBP.
Dude hands over the MBP.
Dude then sits around again wondering wtf just happened while waiting to see if he gets compensated in any way at all.
Apple does nothing.
Dude then petitions to Apple to get his shit back.
Apple says nothing, but does send an unmarked FedEx box with parts back.
http://news.cnet.com/8301-27076_3-20117512-248/prototype-3g-macbook-buyer-gets-parts-back - Prototype 3G MacBook buyer gets parts back

Dude now left with little option but either go "oh well", or sue the original Seller for incurred costs. Seller however says he received the MBP in earnest.

It would have been nice of Apple if they had arranged an exchange for a shiny new MBP and cover Dude's costs, as there's no reason to believe that this prototype was stolen and - as of the latest reports - Apple never filed it as such either.

The 'best' part? Being on IRC, watching a guy go from not being a fanboy but certainly an admirer of Apple, to being completely disenchanted.

Anandtech passed 300W trying to overclock this beast.

No they didn't.

http://www.techpowerup.com/reviews/AMD/A8-3850_Fusion_GPU_Performance/5.html

http://techreport.com/articles.x/21730/8

http://www.anandtech.com/show/4476/amd-a83850-review/6

- First, there is the huge delay intel caused by engaging in fraud by paying pc makers to not use amd chips, right at the time amd was at an advantage.

- Then there is the fact that these synthetic benchmarks use intel's proprietary libraries, which were proven to work ineffectively when 'non genuine intel' architecture was detected.

- Then there is the fact that this is a new platform, and its just out, and the main deal with this is being easily increasable in cores. so amd will just add more cores without any research being needed. expect 32 core cpus in a year or so. 16 cores already out.

- As you can understand these cpus are geared more for server environment, and will take that environment over.

- Amd is moving to trinity in one year or so. Trinity is the APU format that all amd cpus will take from then on. Llano apus have been quite successful in gaming fro example 50-80 fps in starcraft 2 (crossfired and not) -> you dont need to buy an external card anymore, and if you do you can crossfire it with the cpu contained one. http://www.anandtech.com/show/4476/amd-a83850-review/6 http://techreport.com/articles.x/21730/8 intel is worlds behind in this one.

and then there is the ultimate question of what the fuck i am going to do if i grab a powerful processor. really. i bought an overclockable board, and an unlocked cpu. and when i played games, i found out that it was mostly the video card i added that did most of the thing. the cpu i had was way, way over any potential requirements and needs of these games. i didnt need to buy a powerful one at all.

i went about hardware/software forums asking what i could do with a powerful computer. answers have been 'video encoding', 'benchmark', 'seti'. as it seems, any daily usage for cpus are WAY behind the power of modern cpus. to utilize your cpu power at all, you need to do unorthodox, unnecessary shit, or be in a profession that works on these.

so i think all this performance talk is bullshit. there is no way in hell you will use that performance, even in hardcore gaming with an eyefinity 3 monitor setup in 5000x resolution, with 2x antialiasing and full detial. (and i just have 2x 5670 cards).

future is in the heterogeneous chips i think. llano already has been a success, and its possible to save 30% on the cost of cpu + mobo + graphics card if you go the llano way over anything intel, and gaming performance is incomparable. when trinity comes, i think there will be a big change in computing. especially when amd puts out a computing platform like cuda.

Here you go http://www.anandtech.com/bench/Product/434?vs=363.

Doesn't mention features at all. Doesn't even filter by socket type much less manufacturer. I am trying to burn thru the marketing and figure out the fastest CPU to buy that supports virtualization for $X on my motherboard of choice.

I'm thinking an expert system where I tell it "Intel, Socket B, virtualization mandatory, gimme the performance vs price table, and if you must narrow it down, narrow it down to sub-$300 please".

I can probably do this by hand in at most a couple hours of work, which is worth a couple bucks for a website to do it for me.

Specifically I Really Wanna Run KVM or Xen, so if some obscure model simply won't work with the linux/KVM then I'm totally uninterested. For example, I know only certain MBs support USB passthru, so I'm exclusively interested in the CPUs supported by those MBs, furthermore I have to make sure that the virtualization features have not been lasered off.

Don't much care if there exists a chunk of silicon that is faster yet completely unusable for me. Just don't care. Much like my first criteria for a mythtv FE card is "does it work with vdpau" and I have zero interest in a "better" card that does not support vdpau. Marketing wants to "help" me by having the name of the card include the word "sniper", or by placing a picture of a girl in a chainmail bikini on the box, so they're useless (marketing, not the chainmail girl)

Its almost as much of a pain in the ass as buying a car...

Just select stuff from the left hand pair of menus:

http://www.anandtech.com/bench/CPU/342

As usual, I feel somewhat obligated to post up AnandTech's review, which always seems much more in-depth and polished than almost all the sites out there:

http://www.anandtech.com/show/4955/the-bulldozer-review-amd-fx8150-tested

Here you go http://www.anandtech.com/bench/Product/434?vs=363.

See the impact of current generation LTE chipsets on battery life. Combined with the very low deployment of LTE and the fact that actual LTE speeds are only marginally better than HSPA+, including LTE would have been a huge mistake. The phone would have to be larger to make room for the LTE chipset, larger and heavier to allow room for a larger battery, more expensive for both of those reasons, and being larger, would have required new case designs, dock designs, etc. All for little or no difference in download speed for a very small percentage of customers. That would be bad engineering and bad marketing. Next year, the situation may be different, but LTE is not ready for mass market mobile devices this year.

http://www.anandtech.com/show/4760/arms-mali400-mp4-is-the-fastest-smartphone-gpufor-now

... Since Thunderbolt is Intel tech, Microsoft will support it as well. ...

They already do. Thunderbolt is nothing more or less than 4-lane PCIe Gen. 2 over a wire instead of a board. Don't believe me? Ask Anand!

(Okay, it's DisplayPort too. But they're muxed/de-muxed at the chip, so the OS doesn't know or care.)

-Ster

The only glaringly obvious omission seems to be sticking with 3G instead of adding LTE or HSPA+ support.

Current LTE chipsets are too bulky and use too much power. It also has HSPA+.

One company can not compete with 10+ in terms of hardware innovation. Apple can release a phone maybe once a year, there's a new Android super-phone out every 3 months, and lesser new Android phones even more often.

Which just leads to people waiting for the next superphone, then waiting for its price to drop a third within two months and then still not buying it, because until then the next superphones got pre-announced. Or finally get one and then see it drop and drop in price and newer and newer models being annonced while their own phone doesn't get updated anymore and feels like four years old half a year later...

Only geeks can think model avalanches are what people want. Most people want, if they spend lots of money on such a thing or are locked into a contract for two years, some peace of mind and not the nagging feeling that what they got will be forgotten history after half a year that nobody cares for anymore. What Apple does is just sane model management that you need to do if you want people to part with their money. And if the iPhone 4S indeed has got the same CPU/GPU combination as the iPad2 it will blow the competition out of the water anyway, very much as the iPad (which is still at least as twice as fast as all other tablets even by naked benchmarks -- look at anandtech.com for details).

Sorry, are you from the past?... The Dual Graphics option for Llano has been in the news since, well, basically since the existence of Llano is known. It also has been featured in basically all the Llano reviews (like this one from June) so I am not sure what do you mean by "not making much words about this"

Your phenom II (I'm guessing a 965 at $125) isn't much lower performing than some of the newer $300 (wait, no $190 for an i5 2400) i5s (it's more like 30% compared to an i5 2400 generally http://www.anandtech.com/bench/Product/102?vs=363), but neither is an i3 2100. In fact, so much so that the i3 2100 will beat your CPU silly most places (http://www.anandtech.com/bench/Product/102?vs=289), and costs $125 too ;)

Okay then...
Pentium G840 –$88.99
ASRock H61M-GE LGA1155 Intel H61 1PCI-E16 1PCI-E1 2PCI SATA2 VGA GBLAN Motherboard - $62.99
Team Elite 8GB 2x4GB PC3-10666 DDR3-1333 9-9-9-24 Dual Channel Memory Kit - $39.99

That puts me $6 cheaper, and as you can see, even if you succeed in unlocking your cores, it beats your system 9 times out of 10. If you don't, well, it demolishes your system:
http://www.anandtech.com/bench/Product/188?vs=405
http://www.anandtech.com/bench/Product/121?vs=405

Okay then...
Pentium G840 –$88.99
ASRock H61M-GE LGA1155 Intel H61 1PCI-E16 1PCI-E1 2PCI SATA2 VGA GBLAN Motherboard - $62.99
Team Elite 8GB 2x4GB PC3-10666 DDR3-1333 9-9-9-24 Dual Channel Memory Kit - $39.99

That puts me $6 cheaper, and as you can see, even if you succeed in unlocking your cores, it beats your system 9 times out of 10. If you don't, well, it demolishes your system:
http://www.anandtech.com/bench/Product/188?vs=405
http://www.anandtech.com/bench/Product/121?vs=405

Well here is a page doing some comparisons but sadly no hard numbers. From what I gather as long as the CPU doesn't hit 100% it doesn't matter because the video will play smooth. This page here was written by a guy doing 720p with a Geforce 7200 and AMD XP2600 CPU so if that combo will play 720p I have no doubt the 520 PCI with a Pentium D will play it no problem.

And from what I understand once its passed to the GPU that's it, there is no further interaction with the CPU. I know I've set up AGP cards with hardware decode and AGP certainly isn't feeding data back to the CPU and it worked great. So I still say the 520 PCI plus a Pentium D would make a fine box for surfing and watching videos. Both of my nephews are running Pentium Ds and they even play MMOs with it and never drop below 30FPS, so for something less strenuous like video I think it would work just fine.

This card couldn't touch the new Intel Graphics Family stuff, which is pretty decent ( http://www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/11 ). Still, that would require a whole new system build at about $300 minimum, even using the built-in graphics. The card is probably less than that.

Here's the story I was thinking of:

http://www.anandtech.com/show/4569/village-instruments-ceo-promises-graphics-card-enclosure-for-thunderbolt

Basically it's feasible because on the Mac at least Thunderbolt gives you direct access to the bus, not sure if all PC implementations will do that yet.

If Newegg's shipping department can't learn that shipping hard drives and other components in bubble mailers is a bad idea, then no, they likely won't survive. After receiving 3 hard drives in a two week period earlier this year from Newegg shipped in bubble mailers, I've shifted what were previously regular Newegg purchases to TigerDirect and Amazon. Newegg has a long history of recurring problems with improper or inadequate packaging and it makes me wonder just how many of the negative reviews of "This drive/[insert part here] died after two days" were cases where parts had been improperly packaged for shipping.

If Newegg's shipping department can't learn that shipping hard drives and other components in bubble mailers is a bad idea, then no, they likely won't survive. After receiving 3 hard drives in a two week period earlier this year from Newegg shipped in bubble mailers, I've shifted what were previously regular Newegg purchases to TigerDirect and Amazon. Newegg has a long history of recurring problems with improper or inadequate packaging and it makes me wonder just how many of the negative reviews of "This drive/[insert part here] died after two days" were cases where parts had been improperly packaged for shipping.

That's because quick shipping time is one of Newegg's top goals. It reminded of this old article about how Newegg's warehouses work.

http://www.anandtech.com/show/1945

The relevant quote is, "Newegg is trying to target 2 day turnaround for all shipments, regardless of shipping method, thanks to strategic placement of their warehouses all over the country."

The article itself is worth a look if for nothing else but the pictures.

From Anandtech http://www.anandtech.com/show/4863/the-samsung-ssd-830-review:

"The Samsung SSD 830 will be available to consumers starting in mid October. Although Samsung isn't announcing pricing at this time, I've been told to expect the drive to be priced around where the SSD 470 is today."

AnandTech has a review up. For your questions, the drive has TRIM support, Samsung has been the SSD manufacturer of choice for Apple so I'd say OSX support is a given, costs will be in line with the SSD 470 which is within the range of the OCZ Vertex 3, Crucial m4 and Intel SSD 510 and it goes up to 512gb, which is the sample that has been reviewed.

Notably, they say that this is the first really exciting release by Samsung. Apparently, garbage collection is delayed to moments with low IO activity, making torture tests dip down to as low as 50mb/s, but on the flip side this boosts peak speeds. In normal operation such issues are not quite as problematic. Another thing to note is that this SSD is less dependent on compression. SandForce based drives like the Vertex 3 suffer tremendously with incompressible data, whereas Samsung's offering doesn't dip that much.

Did you miss the if you were slamming it part of the sentence? I've actually worked with these chips and like I said liked them enough I have a EEE netbook with one on the way and i found there was only really two ways to slam it hard enough to get it that high. 1.-Load up a game far more advanced than the machine should be running, like say Starcraft II or Bioshock II with settings maxed, or try to do multi-transforms with Virtualdub on the chip.

Now be honest: How many people are actually gonna try either of those? how many who would seriously think about ARM on the desktop would try either of those? To use a /. car analogy it would be like saying the Prius is a bad car because it lugs when you are trying to drag a boat. Duh don't drag a boat with a Prius!

But if you are talking everyday surfing and basic office tasks, you know the kind of thing one would reasonably use something like that for? We are talking between 6w and 9w, less than the cable modem. It doesn't even really jump when doing HD because the Radeon 6310 takes the hand off and does the decoding in graphics hardware instead of slamming the general purpose CPU.

Anyway here is a nice article about Brazos and gives all the details and some cool benches. TLDR? More powerful than Atom+ION while using less power. For a low power desktop it is just about as nice as one could ask for.

Obviously it's too soon to know how well this will work but according to the article on anandtech, the OS is not even aware of the existence of this fifth core.

Android isn't aware of the fifth core, it only sees up to 4 at any given time. NVIDIA accomplishes this by hotplugging the cores into the scheduler. The core OS doesn't have to be modified or aware of NVIDIA's 4+1 arrangement (which it calls vSMP). NVIDIA's CPU governor code defines the specific conditions that trigger activating cores. For example, under a certain level of CPU demand the scheduler will be told there's only a single core available (the companion core). As the workload increases, the governor will sleep the companion core and enable the first GP core. If the workload continues to increase, subsequent cores will be made available to the scheduler. Similarly if the workload decreases, the cores will be removed from the scheduling pool one by one.

Obligatory:
http://www.theonion.com/articles/fuck-everything-were-doing-five-blades,11056/

Seriously, a low-performance core doing administrivia type work sounds great, but won't this require OS support? I can't imagine this detail is completely abstracted from the kernel.

Anandtech also has an article up on this. From the sound of it this isn't really different from other multi-core processors that are able to power down or turn off individual cores. At low system demand, the CPU switches to the companion core and reports a single core available for task scheduling; if system demand is too high for the companion cube, er, core to handle the CPU switches to the main core(s). Sounds like a slight delay going from the companion to main (Anandtech quotes it at 2 ms), but as far as the OS is concerned it is no different than the situation we have now where one or more cores can be turned off independently.

Apparently they're talking both. I suspect many .NET apps would run on the ARM version just fine, if it wasn't for the fact that MS are, apparently, limiting said version to apps developed against the new Metro API.

Source: http://www.anandtech.com/show/4771/microsoft-build-windows-8-pre-beta-preview/5

(Yes, I know I pasted something very similar elsewhere on this article, but I really think the API incompatibility is much more egregious than x86 binary incompatibility.)

The stupid part is that it should never have gotten to this. Microsoft should, at the very least produced an architecture analogous to LLVM where devs could build and compile their apps just the once and it didn't matter if it was running on ARM or x86 or x86-64 or MIPS or anything else. When the app installed the system could compile native code from the LLVM bitcode and run that. It might not help with x86 emulation but it would offer a migration path, one far more compelling than what's on the table at the moment.

What, like .NET you mean? Nobody has officially come out and said that .NET binaries (which don't interface with native code) will work, but I suspect they will. The real problem here is not the lack of x86 emulation, or the lack of a CPU-agnostic runtime. The real problem is that MS aren't even providing an ARM version of the full Win32 API, just the API needed for Metro apps. I wouldn't be surprised if the desktop "app" just isn't there in the ARM version - only the Metro UI.

Source: http://www.anandtech.com/show/4771/microsoft-build-windows-8-pre-beta-preview/5

AnandTech has some more details.

"For those of us power users with big desktops and multiple screens with 10+ windows open... guess what... that's not going away. You just launch Explorer, and have a full desktop window manager."

except if you dl the actual dev preview you will see that no, you can not just dismiss metro like the xp/vista/7 theme. explorer IS metro now. your 'desktop' app is just a task bar and wallpaper. the start menu is not the normal metro ui start screen. also you will have a metro ui panel docked on the right hand side of the screen.
http://images.anandtech.com/doci/4771/Charm.png thats what the new 'desktop' looks like. but if you press the start button or the windows key on the keyboard THIS returns.
http://images.anandtech.com/doci/4771/StartWin82.png
no way to remove it or disable it. THIS is the new windows ui, no if's ands or buts. Microsoft has jumped off the deep end and seems to have declared the tried, true desktop method that everyone is used to 'dead'

"For those of us power users with big desktops and multiple screens with 10+ windows open... guess what... that's not going away. You just launch Explorer, and have a full desktop window manager."

except if you dl the actual dev preview you will see that no, you can not just dismiss metro like the xp/vista/7 theme. explorer IS metro now. your 'desktop' app is just a task bar and wallpaper. the start menu is not the normal metro ui start screen. also you will have a metro ui panel docked on the right hand side of the screen.
http://images.anandtech.com/doci/4771/Charm.png thats what the new 'desktop' looks like. but if you press the start button or the windows key on the keyboard THIS returns.
http://images.anandtech.com/doci/4771/StartWin82.png
no way to remove it or disable it. THIS is the new windows ui, no if's ands or buts. Microsoft has jumped off the deep end and seems to have declared the tried, true desktop method that everyone is used to 'dead'

incorrect.
http://www.anandtech.com/show/4771/microsoft-build-windows-8-pre-beta-preview/1
the window's 7 like desktop is only there for compatibility with programs not made for the metro ui. it provides only a task bar and desktop. the old start menu is gone. launching it will result in being kicked back to the metro ui. also on the right side of the desktop will be a metro ui panel.