An Update on Spline Licensing

An Update on Spline Licensing

Today I have pushed changes to spiro and the spline research repositories, changing the license from GNU General Public License to Apache 2 + MIT. After 37 years in the technology licensing business (dating from the filing of my first revenue-generating patent), I have come to understand that ideas are worth very little, almost all the value is in the execution. I’m finding much more receptiveness to opportunities to adapt spline research to specific applications, than licensing of the existing research code.

Source: raphlinus.github.io

Facebook: Breaking Up Facebook Is Not the Answer

Facebook: Breaking Up Facebook Is Not the Answer

In messaging, we’re not even the leader in the top three markets — China, Japan and, by our estimate, the United States — where we compete with Apple’s iMessage, WeChat, Line and Microsoft’s Skype. Almost all of our revenue comes from digital advertising, and most estimates say Facebook’s share is about 20 percent of the United States online ad market, which means 80 percent of all digital ads happen off our platforms. In Facebook’s case, our size has not only brought innovation, it has also allowed us to make a huge investment in protecting the safety and security of our services.

Source: www.nytimes.com

Using Spark to Prove 255-Bit Integer Arithmetic from Curve25519

Using Spark to Prove 255-Bit Integer Arithmetic from Curve25519

One of the most important functions used in the proof is a function that converts the array of 10 integers to the big integer it represents, so when X is an Integer_255 then +X is its corresponding big integer. Another case of inductive reasoning in my project is a lemma, whose proof present a very useful technique when proving algorithms using recursive ghost functions in contracts. I had a lot of recursive functions in my contracts, but also quantifiers… I did not hesitate to split proofs into Ghost procedures in order to reduce context size, but also wrap some expressions into functions.

Source: blog.adacore.com

Microamp: Asymmetric Multi-Processing on Microcontrollers

Microamp: Asymmetric Multi-Processing on Microcontrollers

This can be seen in the output of the program where each core reports a different address of the variable . For example, if we place the and sections of both images (core #0’s contains the variable) at address (that is in BTCM0) it would become possible to break Rust aliasing rule with no code. Regardless of the performance, we must default to a memory safe program layout so we’ll pick the following layout:

Rationale: References to stack variables can not be sent to a different core because they have non-static lifetimes and to place a value in a static variable it must only contain static lifetimes ( there’s an implicit bound on all static variables).

Source: blog.japaric.io

Japanese railway company starts testing 249mph bullet train speeds

Japanese railway company starts testing 249mph bullet train speeds

This week, Japanese railway company JR East showed off its new Alfa-X, a high-speed bullet train that is designed to achieve a top speed of 400kph, or 249mph, which would make it the fastest commercial train in the world. That long lead time suggests that there might be an opening for another high-speed bullet train option to overtake the Alfa-X Shinkansen train in speed for commercial railway service. Virgin Hyperloop One, arguably the best-funded Hyperloop startup, may be the closest: it has a test track outside of Las Vegas where it has been able to log a speed of 240mph (386kph) in less than a quarter-mile (300 meters).

Source: arstechnica.com

Google Creates ‘Dedicated Placement’ in Search Results for AMP Stories

Google Creates ‘Dedicated Placement’ in Search Results for AMP Stories

Google says it will soon introduce a “dedicated placement” in Google Search for AMP Stories in specific categories, like travel, along with components that let AMP Story creators embed interactive content. AMP Story creators will soon be able to embed new types of content — specifically Twitter posts, Google Maps, and YouTube videos — through components that will debut individually this year. Today’s announcements follow months after Google said it would leverage AI to “construct” AMP Stories and surface them in search results, starting first with stories about celebrities, athletes, and other “notable people.”

Source: venturebeat.com

Uber Is the Worst Performing IPO in History

Uber Is the Worst Performing IPO in History

According to University of Florida professor Jay Ritter, Uber’s 7.62 percent decline since hitting the NYSE makes it “bigger than first day dollar losses of any prior IPO in the U.S.”
In terms of percentage losses, Uber’s dip doesn’t even scratch the surface of the worst IPOs. But the staggering valuation of the company makes it, in raw scale, “among the top 10 IPOs ever” including companies outside the U.S., Ritter told Gizmodo in a phone interview. Consider also that Uber’s debut valuation of $76.5 billion was a considerable drop from the between $90 billion and $120 billion the company had been worth in some analysts estimation just a month earlier—one meant to stanch the forthcoming bleeding that had begun with competitor Lyft’s bellyflop IPO.

Source: gizmodo.com

To reinvent the processor

To reinvent the processor

The fundamental work of an instruction
A typical instruction in a typical instruction set performs something like the following:
Read the operands from the global register file. Out-of-order execution also introduces costs for instruction issue and wakeup: In a fully out-of-order processor, any write to the LRF could make any locally-queued instruction ready to issue, and any write the the GRF could make any instruction ready to issue from any BW. You can get most of the motivation needed to invent the Mill’s belt by noticing how much more headroom operand forwarding has than large global register files, primarily because of the comparatively miniscule number of sources.

Source: medium.com

Windows Implementation Libraries: type-safe C++ interfaces for Windows patterns

Windows Implementation Libraries: type-safe C++ interfaces for Windows patterns

From here, you can either invoke CMake directly:

Or through one of the scripts in the scripts directory:

If you initialized using Ninja as the generator, you can build the tests like so:

Or, if you want to only build a single test (e.g. for improved compile times):

If you initialized using MSBuild as the generator, there will be a file in the root of the build directory. If you used the initialization script(s) mentioned above, or if you followed the same directory naming convention of those scripts, you can use the runtests.cmd script, which will execute any test executables that have been built, erroring out – and preserving the exit code – if any test fails. You will want to repeat this process for the other architecture (e.g. by using the “x86 Native Tools Command Prompt for VS 2019”)

This project welcomes contributions and suggestions.

Source: github.com