TD-Linux

38 ms·

361.
by TD-Linux 10y ago
There's less flexibility than you might think - you get only one choice of quantizer and quantization matrix for the entire frame. So pretty much your only option is to twiddle the values themselves. This is usually done with trellis q
362.
by TD-Linux 10y ago
Step 3), choosing more zeros, can be generalized with trellis quantization, which does a search for the best values to encode for each block for the best distortion-per-rate score, where distortion can be any metric (edit: apparently guetzl
363.
by TD-Linux 10y ago
mozjpeg also generates progressive JPEGs by default, for the same reason.
364.
by TD-Linux 10y ago
Well, I suppose it is one of the fastest CP/M computers around... especially because it's the PCPI Applicard (or rather, Franklin's licensed version) instead of the Microsoft Softcard, which means it runs at 6MHz and has its
365.
by TD-Linux 10y ago
You do need part of that - a lot of it is capacitors wired directly to the ribbon cable, for the COG's charge pump. There's also a temperature sensing chip and a couple of external mosfets, one of which is to fix a bug on older CO
366.
by TD-Linux 10y ago
It's way simpler than that, you just write the displayed image more times based on temperature. If you don't do it your display will just have less contrast in the code. The chip-on-glass has tons of registers to set drive voltage
367.
by TD-Linux 10y ago
I have a copy of WordStar running on an Franklin Ace 1200 (Apple II+ clone from 1983)... it runs WordStar via a Z80 peripheral card that runs PCPI CP/M while talking to the Apple 6502 to use the rest of the peripherals. So even then pe
368.
by TD-Linux 10y ago
BU doesn't implement a hard fork but rather potentially infinite hard forks, depending on two user controlled parameters. If the AD parameter is low (the default), then it will just be a long orphan chain, with effectively miner select
369.
by TD-Linux 10y ago
You can, I've used Pervasive Displays (I bought them from Digikey, they also have breakout boards): http://www.pervasivedisplays.com/ But yeah they are kind of complicated to drive. I've only used the small ones,
370.
by TD-Linux 10y ago
You don't even have to invent a new cryptocurrency. Bitcoin has had a non-turing-complete script from day 1. And if you want to run something more complicated, you can use zk-SNARKs combined with a tiny HTLC script on-chain.
371.
by TD-Linux 10y ago
Huh? Even BIP-101, the original Bitcoin XT BIP to have forever increasing block size, mentioned this as a risk [1]. It argues that the effect is insignificant, but I would hardly call it watertight. [1] https://github.com/bi
372.
by TD-Linux 10y ago
As someone who will never run BU, I'd rather have segwit activate, which will bump the blocksize to ~2.1MB and also enable a lot of better payment channel implementations, like lightning, that can lessen the demand. It seems very clear
373.
by TD-Linux 10y ago
Yes and yes, though that would require your users to pay $100 per year, install Xcode, and compile your app.
374.
by TD-Linux 10y ago
Hangouts doesn't use WebRTC, it uses a NPAPI plugin on Firefox and nonstandard APIs on Chrome.
375.
by TD-Linux 10y ago
Note that this is unrelated to Zcash, the zk-SNARKS based cryptocurrency perhaps more known to HN.
376.
by TD-Linux 10y ago
Being unable to represent 0.1 exactly is unrelated to any precision limitation. It's because the quantization of binary and decimal is different.
377.
by TD-Linux 10y ago
Most of the nonstandard features that Hangouts uses are built into Chromium and the JS APIs it exposes to Hangouts. In particular, SDES, google-ICE, RTP data channels, and plan B are all nonstandard features exposed that way. [1] [1] https
378.
by TD-Linux 10y ago
Hangouts uses a bunch of non-standard Hangouts-specific APIs in Chrome. Maybe it's be best to call it "Google-RTC". https://webrtchacks.com/hangout-analysis-philipp-hancke/ In summary, they use weaker en
379.
by TD-Linux 10y ago
... and making it open-source?
380.
by TD-Linux 10y ago
Yup, the format is called SDP. It's an old, crufty, and mostly effective way to specify not only the port and IP, but also whether a TCP fallback is being used, video and audio formats, and other things you can mostly ignore when just
381.
by TD-Linux 10y ago
That's because you don't need to implement any of those things for a basic client-server architecture. Unfortunately it looks like the author saw the massive and complex webrtc.org implementation and gave up, rather than try writi
382.
by TD-Linux 10y ago
Yup, it's not an exact tradeoff, and near-0 it breaks down entirely. At zero it's worse than that, because of the diminishing block subsidy schedule, having zero fee pressure will reduce the incentive for mining, and therefore sec
383.
by TD-Linux 10y ago
There is no global mempool, it's a per-node thing. I think the default for Bitcoin Core is to hold 300MB of transactions, which means for most nodes the number of transactions in the mempool is constant. A better metric is transaction
384.
by TD-Linux 10y ago
Larger block sizes allow more transactions, so although per-tx fees will be lower, there will be more transactions. The computational costs of larger block sizes are borne on the entire network, however the effects are most severe for miner
385.
by TD-Linux 10y ago
Being in the EU is not a magical bullet against software patents. https://arstechnica.com/tech-policy/2017/02/blackberry-sues-...
386.
by TD-Linux 10y ago
LEDs don't produce polarized light. You might be seeing the matrix scanning.
387.
by TD-Linux 10y ago
IIRC many of them use SPI flash or EEPROM, which you should just be able to pop off and replace with a similar part.
388.
by TD-Linux 10y ago
You can actually evaluate work you've delivered, but only for a very small subset of "work" - basically, things that can be represented and evaluated entirely with software. Here's an example: https://bitcoinc
389.
by TD-Linux 10y ago
Maybe someone can enlighten me, but it doesn't actually look any more low level. The only lower level thing I see is the texture attachment load/store ops. Can you commit a command buffer more than once? (edit: I realized this API
390.
by TD-Linux 10y ago
Well, sysadmins have a way simpler solution: just tell their users to use a different browser. This would have to be coordinated between browsers to have any effect at all.

More ›