HN user

mumumu

83 karma
Posts1
Comments44
View on HN

Amdahl's law is common sense, but also the one of the reasons premature optimization is the root of all evil.

A small speedup on the longest (by time) part of the program is usually better than infinite speedup on a short part of the program.

The bottleneck is bandwidth and storage not computer power.

Data from sensor and to display are usually (but not always) raw and uncompressed.

To go beyond 8k 120fps we will need fiber, or better compression.

We have good codecs to move data online. And specialized ASIC to do live encoding with good quality.

The current compression defined by HDMI is very rudimentary.

*avoid branches

The blog post I mentioned:

Paper: Parsing Gigabytes of JSON per Second https://branchfree.org/2019/02/25/paper-parsing-gigabytes-of...

Another related post from Lemire:

Ridiculously fast unicode (UTF-8) validation https://lemire.me/blog/2020/10/20/ridiculously-fast-unicode-...

Those algorithms are fast. But to put them in perspective. A single x86 CPU can write 64B per cycle. At 5GHz, the theorical maximum bandwidth is 320 GBps. IIRC, the read bandwidth is twice that.

There are others botlenecks, and is very hard to write code that writes at every cycle.

A interesting consequence, is that the theorical maximum bandwidth is logarithmical to the number of cycles. Again, talking about branchless streaming application.

This is true on a few streaming application (such as parsing).

And most of the speedup is because of tricks to avoid doing beaches. There is a great blog post from one of the authors of JSON SIMD discussing this.

I'm on mobile, there is a link for the blog post on the simd JSON github repository.

It is hard to find yerba mate that doesn't taste stale outside of south of South America.

Friends that drink it ask for people traveling to SA to bring them some. They are also very particular about how to choose it.

Something I realized is that the dark patterns change from country for country.

Brazilian websites uses patterns that aren't that common on American sites. All news organizations and all large e-commerce uses the browser's notification and localization API (some ask before trying).

To view the content on those sites, I need to deal with about 5 interruptions. The previously two, plus newsletter, "accept cookies" modal, "special offer" pop-up, "free shipping" sticky window.

And those are >1 bi usd companies.

This is not new. It is news because it's from Intel.

I looked into that a year or two ago and there were papers on the this.

Anyone who is familiar with Euler Video Magnification and with neural network likely though of that.

Does this work in encoded videos? I doubt. Intel probably can add a a feature the video encoder and sell it as an authentication service for webcam communication on Intel Plataform.

Every once and a while I have a look at the changelog of the FF dev. It is good way to find the flags to disable features and to have an idea how the web is changing overall.

The downside of cherry picking features is that your browser fingerprint becames pretty much unique.

In about:config you can enable or disable features individually.

Recently they added ":modal" that will make paywalls harder to bypass. Most sites haven't implement yet, but I've already disabled this "feature" on config.

Me too. But nowadays you can define them with with something like a ascii table.

Also, check "massonary".

There are so many cool features in css being released today that soon that banner "Best viewed on Netscape 3 on 800*600 resolution" may be common in some niche design sites.

But more likely as "Best viewed on Google Canary xxx, on flip screen devices with 10 bit depth colorspace".

By the way, did you know there is a css spec for rounded screen devices?