OLIVER ILIFFE

I describe a lock that is in many regards much faster than an XCHG-based lock for x86 machines. I made this as part of my Master's thesis. I also release my frustrations about computer architecture.

Cache coherence protocols that exist on modern hardware are often proprietary, but this doesn't stop us analysing side channels to figure out how these protocols might be implemented.

A swift and easy guide to Ruby, an old relational HDL for proving the correctness of circuits.

I describe my experience making a machine learning classifier for acute kidney injury (AKI) in pure C. It's interesting because I use a cool technique to compress the data.

How CXL 1.0 and related concepts work.

I work through various known hashtable designs, and also a new one, hoping to see some performance increase.

I work through a derivation for the average number of probes in a hashtable, then try and verify it through real data.

I make a hashtable with SIMD probing for a drag race.

Some of the work I did for a summer research project with one of my very favourite professors. It's quite rare that you get to do 'algorithms research'. I was looking at how we can quickly associate an integer with a specific tree in a nice, ordered way. WARNING: this is one of my first articles I wrote on this website and hence, it's a pretty tough read. Proceed with caution.

I discuss a new idea for safety that I call "item-scoped unsafe code". The idea is to try and get as few things as possible be 'unsafe' and hopefully have everything that is be verifiably safe at runtime.