Good morning! Secret face scanning in campus vending machines has prompted an investigation revealing no disclosure to students about this data collection and analysis as well as potential privacy law violations. Additionally, a hardware issue has been discovered affecting some high-end 13th & 14th gen Intel processors that can cause system instability and crashes under heavy multithreaded workloads. Lastly, a new general-purpose compression library called LZAV was recently released that is optimized for fast in-memory data compression and delivers excellent compression ratios compared to popular alternatives.

— Forrest Knight & Meghanadh Vasireddy

Secret Face Scanning in Campus Vending Machines

A software error exposed facial recognition capabilities in vending machines at the University of Waterloo. The machines were running a Windows application called “Invenda.Vending.FacialRecognitionApp.exe” which failed to properly launch.

This prompted an investigation which found:

• Invenda brochures stating the machines utilize facial recognition and machine learning to classify users by age and gender in order to boost sales

• No disclosure to students about this data collection and analysis

• Potential violations of Canadian privacy laws

The vending machine company, Adaria, claimed no identifiable user data is actually stored. However, facial images constitute highly sensitive personal information. Consent was not obtained for any collection or use of student biometric data.

While the machine learning models may not retain user images, they could still reveal identifiable markers encoded from facial scans that students were unaware of.

In response to privacy concerns, the university is now removing the vending machines, prompting backlash from outraged students.

Read More Here

Intel Processor Instability Causing Oodle Decompression Failures

A hardware issue has been discovered affecting some high-end 13th & 14th gen Intel processors. When doing heavyweight multithreaded workloads, a small percentage of the CPUs exhibit system instability and crashes. This includes intensive tasks like decompressing Oodle-compressed game data, running benchmark apps, and other demanding programs using many threads.

The root cause seems to be overly aggressive default BIOS settings that push the processors outside Intel's recommended safe operating range for factors like clock speed, power draw, and voltage. Even without explicit overclocking enabled, common "auto-overclocking" features may be pushing the limits too far.

Potential workarounds include:

• Using Intel XTU to reduce the P-Core multiplier

• Disabling enhanced turbo, multi-core enhancement, and performance boost features in the BIOS

• Reducing voltage and power limit settings to Intel spec

Additionally, turning off XMP memory overclocking is recommended, as instability in this area can have similar symptoms.

If crashes still occur frequently in stress/benchmark programs after tuning, RMA of the processor may be required. A hardware defect causes the stability issue in only a small percentage of chips.

Read More Here

LZAV - Fast Data Compression Algorithm (in C/C++)

A new general-purpose compression library called LZAV was recently released that is optimized for in-memory data compression. LZAV is written in C/C++ and uses the LZ77 algorithm. Key features include:

• Fast compression and decompression speeds - averages 530 MB/s and 3500 MB/s respectively on modern 64-bit CPUs

• Strong compression ratios - up to 14% better than LZ4 on some data sets

• Portable C code works across platforms and compilers

• Safe to use on untrusted/corrupted data

• Adaptive performance optimized for different data types

Benchmark results published by the author show LZAV achieving excellent combination of compression ratio and speed compared to popular alternatives like LZ4 and Snappy. It compresses 14% smaller than LZ4 on the Silesia corpus at only 35% slower speeds. Decompression through LZAV is still very fast.

For applications like databases and file systems where compression ratio directly reduces storage costs, LZAV presents a compelling drop-in replacement for existing libs. Its portable C code can enable compression across diverse systems.

Read More Here

🔥 More Notes

• AT&T is handing out a bill credit to make things right after massive outage

• The latest billionaire trend? Doomsday bunkers with a flammable moat

• VSCode + WSL makes Windows awesome for web development

• Falsehoods Junior Developers believe about becoming Senior

• Large Language Models Are Drunk at the Wheel

🎥 Youtube Spotlight

AT&T blames software update for nationwide cell phone outage

AT&T experienced a nationwide cell phone outage due to a software update, disrupting services for many customers, businesses, and first responders. The outage caused inconvenience, with some users resorting to Wi-Fi for communication, and even led to unnecessary 911 calls. The interruption also affected government operations, but service was eventually restored in the afternoon.

Was this forwarded to you? Sign Up Here