Click

Saturday, September 12, 2009

ATI and NVIDIA: Quick Look at HDTV Overscan Compensation

ATI and NVIDIA: Quick Look at HDTV Overscan CompensationIt has been a while since ATI released their HDTV dongles, which provided HDTV output support for most of their Radeon line. In fact, we probably first experimented with their HDTV dongle back in July or August of 2002. Back then, HDTV output support was plagued by the overscan issue.

And for those of you unfamiliar with "overscan", it is simply the part of the picture that is cropped. Depending on whom you ask, others also describe it as the space that bleeds or "scans" beyond the edges of the visible area of the screen. Typical televisions can have a loss of up to 20% of the image due to cropping. This portion of lost image is what is commonly known as overscan. Technically speaking, the information of the "lost picture" is not actually lost, but it is outside the range of the visible area of your TV screen. A similar situation on the computer end is when you view a picture in 100% scaling on a monitor with a lower set resolution than the picture, i.e. a 1600 x 1200 picture in a 1280 x 1024 desktop environment. The difference is that on a computer, you can move the picture around to see portions cut off by the visible area of the monitor.

We should clarify that overscan is a not necessarily a bad thing. It is implemented deliberately on TV sets because of the different video input formats: composite, s-video, etc., all of which the TV needs for which to provide support. If overscan was not implemented as a factor of these different formats, there would likely be underscanning of different degrees on different TV sets. This is due to the different protocols and inherently different signals that the TV needs to handle. (Underscanning would be the opposite of overscanning, where the image is smaller than the area on which it is being displayed.) It would be tantamount to zooming out when you look at a picture; though in the case of TV sets, the space that doesn't reach the edges of the display would be black. The deliberate use of overscanning allows the screen to be completely filled, as opposed to underscanning, which could have varying degrees of underscanned margins.

The reason why we notice overscan more so on a computer is that we already know what the signal is suppose to look like: where the start menu button should be, where the clock should be, where the corners of the screen should be in relation to our desktop shortcuts. A DVD to a DTV or even a regular TV usually encounters some measure of overscan, though we hardly notice it because we aren't use to its native signal. One way that DVD player manufacturers' or TV manufacturers' compensate for this is to provide a zoom out function, where you tell the system essentially to underscan. This is why we go crazy when we notice overscan from a X-Box rather than from DVD signal; you know where the game menu is supposed to look like.

In theory, if a HDTV was designed for only HDTV, there would be no overscan from component computer video output. The main issue with overscan is that DTV are programmed to do more than just DTV signals. They accept many legacy signals: camcorders, s-video, composite, etc All of this means that there must be a cross platform support for all formats, and the only way for that to occur is to either overscan or underscan. Underscanning would be more frustrating to the consumer, since the signal would be smaller than the displayed area with the black bars surrounding the image. Overscan ensures that video signal always fills the screen, though this gets to be increasingly frustrating when you get an X-Box or output video from your computer and the signal is overscanned.

And as Keith Rochford (Chief Engineer of eVGA) explained, when you switch to a DTV, you are now talking about a high resolution display, and backwards engineering a pixel technology to a line scan technology isn't a simple task. This backwards engineering or transfer is what leads to the large 10% to 15% margins of overscan that we have been accustomed to when we output from a computer to a DTV. For those who own something like a plasma display that can do direct computer video output via VGA or DVI, this obviously isn't an issue, since there is no backwards conversion needed. It is essentially like a really big computer monitor, since it keeps the video card's native output.

In the most practical sense, overscan is something you don't want to have or at least want to minimize. Using your HDTV set as a substitute for your monitor can be awesome, but the limitation of having part of the picture cropped gets to be a major deterrent, especially when you want to plan games, surf the web, or watch videos on that nice big screen.

There are more than just ATI and NVIDIA cards on the market, but most of us are still going to be stuck with one or the other. In which case, you are most likely going to get some degree of overscan. Keep in mind that we can't track down every or even most DTV sets and check the degree of overscan, and even if we could, overscan varies between TV sets because of the manufacturer's design, which isn't a bearing on the video card. For these practical reasons, we are going to focus primarily on how ATI and NVIDIA approach HDTV overscan compensation.

Apple's 2009 MacBook Pro: Battery Life to Die For

I was so focused on the iPhone 3GS and Snow Leopard announcements from this year’s WWDC that Ialmost missed the gravity of the MacBook Pro announcements.

Apple announced price drops on nearly all of its laptops. The new lineup looks like this:

MacBook

MacBook Pro 13-inch

MacBook Pro 15-inchMacBook Pro 17-inch
CPUCore 2 Duo 2.13GHzCore 2 Duo 2.26GHzCore 2 Duo 2.53GHzCore 2 Duo 2.8GHz
GPUNVIDIA GeForce 9400MNVIDIA GeForce 9400MNVIDIA GeForce 9400MNVIDIA GeForce 9400M + 9600M
Memory2GB DDR22GB DDR34GB DDR34GB DDR3
HDD160GB160GB250GB500GB
Battery LifeUp to 5 hoursUp to 7 hoursUp to 7 hoursUp to 8 hours
Price$999$1199$1699$2499

If you want an all aluminum body, you have to buy a MacBook Pro. There’s only a single MacBook model and it’s the white chassis that’s been around for a while now.

Apple added a 13” MacBook Pro to the lineup to fill in the gap, although it’s not clear to me whether this 13” MacBook Pro uses the same LCD panel as the old 13” aluminum MacBook or a derivative of the 15” MacBook Pro’s panel, which is superior.

Of course there are different models within each one of these categories that you can purchase, but they are irrelevant to the discussion we’re about to have. Look at the battery life row in the table above; Apple is claiming up to 7 hours of battery on the new MacBook Pros. The old specs used to be up to 5 hours.

Apple did some clever work on its own here. Standard lithium ion batteries are made up of cylindrical cells, similar to AA batteries. The problem with these batteries is that they waste a lot of space within a notebook (try cramming a lot of cylinders into a box, you end up with wasted space). This wasted space translates into larger batteries than are necessary, which makes for larger notebooks.

In order to continue to drive laptop thinness down, Apple started experimenting with using custom lithium polymer batteries instead of the industry standard lithium ion parts. Lithium polymer cells aren’t made of cylindrical cells (they’re rectangular), so there’s no wasted space. Not only does this make the batteries more compact, but it also gives you greater capacity since you’re using all available chassis volume for the battery.


Makes sense. Courtesy, Apple.

Apple also found that it was wasting space in the removable enclosure for the batteries as well, so its lithium polymer offerings are no longer user removable. I suspect this part of the equation has more to do with cutting costs than saving space though.

Apple first used this lithium polymer battery technology in its MacBook Air. It gave Apple a very thin battery that allowed it to create the MacBook Air’s sweet form factor. Then came the new 17” MacBook Pro, without a removable battery. Apple claimed that this battery would last for five years before it needed replacing and resulted in up to an 8 hour battery life.

The extended life is supposedly due to an on-battery sensor that communicates with the system's management controller that can dynamically sense the needs of each lithium polymer cell and feed that info back to the charging circuitry. The result is slight variations in charging current designed to optimally charge each and every cell; apparently reducing wasted charge cycles significantly. Apple claims that most cells will hit 80% of their life after 200 - 300 charge cycles, but its special lithium polymer batteries will hit the 80% mark after as many as 1000 charge cycles. Apple claims its unique battery chemistry and microprocessor managed charging (Adaptive Charging) is responsible for these gains but it’s a difficult statement to prove; we’ll have to wait and see what happens after a few years of use.