Last Active
  • Skilled labor shortages could be the next big chip supply problem

    This is where chip makers need to step in and work with universities or institutions. Many big corporations donate equipments to schools for training, research and hiring after graduation. Establishing mini or virtual chip fabrication lab in University maybe tough but something is needed to be done.
    They already do. A college near me has had a semiconductor manufacturing degree plan for over 20 years. Analog, Intel, Texas Instruments, and a few smaller manufacturers all hire the graduates. Now that TSMC is expanding into the US, I assume they will as well.

    What's really cool is the program also offers early enrollment and dual credit for people still in high school. Spend half your day at the high school, half your day at the college, and you have around 20 hours of college EE credit when you graduate high school.
  • HDMI cable purchasing is about to get a whole lot more complicated

    Beats said:
    Apple should create a new cable (USB-C?) that makes HDMI obsolete. HDMI has so many issues nowadays it needs to be discontinued.

    A new high bandwidth cable that isn’t dependent on itself, so it doesn’t matter if you bought it at a thrift store or a high end audio website it acts the same. The bandwidth is future proof to 20 years+ so it can handle anything that’s thrown at it. New features will be dictated by the hardware/software and the cable is just a tube for features. For example PlayStation 6 wants to add 3D images to a TV: done. Apple TV 6 wants to add an exclusive audio standard like Spatial Audio to receivers: done. The cable you bought is irrelevant because the capability is so high from the start.

    Can a new “HDMI 3.1a gen 3x Series 2” standard add these features? Sure. But the fact it has the same form-factor adds more chaos to sort through. HDMI needs to end ASAP.
    That would be nice, but the problem is we're at the limits of what simple copper cables can do. This is why Thunderbolt, HDMI, and several other standards are moving to active cables, which is what leads to all of these problems where a cable with USB-C on both ends could have any of a dozen performance levels.

    We're also at the current limit of what affordable active cables can do, which is why these standards aren't moving forward even faster. Timing correction and error correction chips which can operate at higher speeds are enormously more expensive. You want something faster than HDMI 2.1's 48 gigabits? Get ready to pay $350 for a single 1 meter cable.
  • Apple promotes ways of staying connected with new SharePlay feature

    Of note: SharePlay isn't sharing a video stream. It is synchronizing playback in a separate application instance on each device. That means each device must be able to use the application on its own. In most cases, that means every person watching must have their own subscription to the service involved.
  • 9-year-old unlocks unconscious father's iPhone with his face to call 911

    If the generator is off, it isn't generating any CO.
    Not creating any once it’s off, sure, but it creates a lot more than you might expect until the engine and exhaust path are up to operating temperature. Run a cold engine for just a few minutes, and it can easily make enough CO to kill you several times over. It also accumulates in the blood for quite a while, so even once ambient concentration drops to normally-tolerable levels, that can still be enough to push somebody already exposed past the point of unconsciousness.

    The risks involved in running a generator (any internal combustion engine, really) are pretty manageable, but they’re really, really dangerous if you don’t know to manage them.
  • Google launches Pixel 6, Pixel 6 Pro with Tensor processor

    avon b7 said:
    zimmie said:
    HBCan said:
    Bayer... as in Bayer filter.  An RGB pattern filter over the camera's sensor.  One color filter per pixel... Red, Green, or Blue.  The image colour data is captured and interpolated for the neighbouring pixels to produce a full colour image.  Virtually all commercial colour sensors employ a Bayer filter solution otherwise you would require three sensors to be used... one per colour.  Not easily implemented in such compact environments as a beam splitting prism would be required too.  Creator of the Bayer filter.... Bryce Bayer... who worked with Eastman Kodak.  Died in 2012 I believe. 
    Sure, but the Bayer pattern is naturally a tiled series of squares with two green, one red, one blue photosite per four pixels. Lines up nicely with Pentile display subpixel arrangements. So what in the world is "Quad Bayer"?

    Did a little research, and it turns out it's a Sony variant of the normal Bayer pattern. They turn each photosite into four separate, smaller photosites, then average their values as a way of reducing amplification noise. Thus, the "50 megapixels" is a lie. It has 50 million photosites, but they operate in clusters of four, producing one output pixel value which is still only one channel. It's the equivalent of a 12.5 megapixel sensor.

    Still no idea what Octa PD is supposed to be.
    Can't the sensor output at full 'resolution'?

    Pixel binning to lower megapixel counts is common nowadays but I thought the option was still there to output without pixel binning. 
    Yes, you are right. The sensor can output at full resolution as well.
    Ish. You can get a file out of it with 50 million pixels, but the resolving power is physically limited to about 13 megapixels even in that mode. Setting aside diffraction for the moment, due to how the same-color photosites are clustered (dense, then a gap, then dense, then a gap), they could never capture contrast data in the different channels to provide 50 megapixel level sharpness.

    But in reality, we can't set diffraction aside. With an f/1.85 lens at smartphone lens distances, the Airy disk diameter is 2.37µm or larger. The pixels on that sensor are 1.2µm, so any given Airy disk covers four of them. They physically can't resolve separate detail.

    Which is actually a good thing, as if they could resolve separate data, you would need a more serious antialiasing filter to make sure they couldn't. The main advantage of Quad Bayer sensors seems to be that they don't need as much optical antialiasing to avoid horrible Moiré.