Thu Mar 12, 2020 7:12 am
Actually, I stand corrected. There is a technology that could do low cost energy low cost uncompressed recording. I was thinking it only needs a thin sliver of ASIC to do the recording, but then realised that a certain technology can be custom programmed to do that at great efficiencies, but there maybe no chips setup to do that, as the tech hasn't gone wide market in a general purpose design over the last 15 or so years (but is used in a various custom applications). But that made me realise there are a variety of chips that could do the basics, but you loose out on the energy requirements of the rest of the chip, size and cost, but if you can dynamically power down the rest of the chip you can keep the energy envelope down, but the codec side is still going to take too much processing energy without the right processing solution, but I know somebody that can do it but it would cost millions to design.
Back to reality of the now, not reality of the could/should. BM could save a fortune over FPGA and be able to do action cam sized compressed visually lossless and lossless recording. By commisoning the advanced version of the greenarrays chip. There programmanle chip is one of the ones meant to outdo FPGA in performance energy and costs. A single processor might be 10,000-20,000 transistors plus memory. So, so, that's hundreds to tens of thousands of cores with memory per reasonable FPGA transistor counts. Once you slice up chips off a wafer, you get so many chips really cheap, because they are so small. One wafer could do the entire Blackmagic production run for years. Every product could be done in a year on one wafer (it really depends on how many cores and configuration is required, buy you could put the top camera chip into a very small product, and simply turn most of the cores off, as these chips are by one of the members of the asynchronous computing patent, and can be powered down and powered up dynamically in the execution stream, really simple neat stuff except only the light weight simple model is available to the public). I'm assuming the advanced model is a modest improvement on this with a low multiple of transistor counts here).
They probably would welcome a deal at the moment, as they have to change foundaries. So, I would order on the basis of commission on sales of the advanced chip to other people, as the investment would allow the to springboard into other markets, and on the basis of general sales, as there is a community of developers using their existing chips that would love to use the advanced chip. Plus, they are one of the few companies that I would point to, to do processing behind pixel technology. They run at natural silicon clocking speed, do low energy heat and interference. You could compress and write to storage and run the whole camera system off of the sensor chip (but realistically, I would stick a lot of the processing to the side of the pixel array, or drill through technology, which is still expensive and unlikely to be at the foundry). On top end you stick more to the side, on lier end you stick more underneath, just incase it gives you 12db+ better performance. So, the foundry costs to do on sensor at the sensor foundary will probably be lowish, as it was a rolled into the cost of the sensor development. Interesting. What do you guys think?
aIf you are not truthfully progressive, maybe you shouldn't say anything
bTruthful side topics in-line with or related to, the discussion accepted
cOften people deceive themselves so much they do not understand, even when the truth is explained to them