Performance

The Most Useful Statistical Test You Didn't Learn in School

In performance work, you will often find many distributions that are weirdly shaped: fat-tailed distributions, distributions with a hard lower bound at a non-zero number, and distributions that are just plain odd. Particularly when you look at latency distributions, it is extremely common for the 99th percentile to be a lot further from the mean than the 1st percentile. These sorts of asymmetric fat-tailed distributions come with the business. Often times, when performance engineers need to be scientific about their work, they will take samples of these distributions, and put them into into a $t$-test to get a $p$-value for the significance of their improvements.

Fixed Point Arithmetic

When we think of how to represent fractional numbers in code, we reach for double and float, and almost never reach for anything else. There are several alternatives, including constructive real numbers that are used in calculators, and rational numbers. One alternative predates all of these, including floating point, and actually allows you to compute faster than when you use floating point numbers. That alternative is fixed point: a primitive form of decimal that does not offer any of the conveniences of float, but allows you to do decimal computations more quickly and efficiently.

You (Probably) Shouldn't use a Lookup Table

I have been working on another post recently, also related to division, but I wanted to address a comment I got from several people on the previous division article. This comment invariably follows a lot of articles on using math to do things with chars and shorts. It is: “why are you doing all of this when you can just use a lookup table?” Even worse, a stubborn and clever commenter may show you a benchmark where your carefully-crafted algorithm performs worse than their hamfisted lookup table.

Racing the Hardware: 8-bit Division

Occasionally, I like to peruse uops.info. It is a great resource for micro-optimization: benchmark every x86 instruction on every architecture, and compile the results. Every time I look at this table, there is one thing that sticks out to me: the DIV instruction. On a Coffee Lake CPU, an 8-bit DIV takes a long time: 25 cycles. Cannon Lake and Ice Lake do a lot better, and so does AMD. We know that divider architecture is different between architectures, and aggregating all of the performance numbers for an 8-bit DIV, we see:

The Meaning of Speed

A lot of the time, when engineers think of performance work, we think about looking at benchmarks and making the numbers smaller. We anticipate that we are benchmarking the right pieces of code, and we take it for granted that reducing some of those numbers is a benefit, but also “the root of all evil” if done prematurely. If you are a performance-focused software engineer, or you are working with performance engineers, it can help to understand the value proposition of performance and when to work on it.

Performance Numbers Worth Knowing

When you design software to achieve a particular level of performance, it can be a good idea to be familiar with the general speed regimes you are working with: fundamental limitations like storage devices and networks can drive software architecture. Here are a set of common benchmark numbers that can help you anchor performance conversations and think about the components that your software will interact with. As with all guidelines, these numbers are all slightly wrong, but still useful.