The Tyranny of Terminology

Posted: 05.28.10
Source: Willi Heidelbach 


One of the running jokes about the difference between the English, French and German languages are how rich and efficient English is and how complicated and wordy French and German can be.

We struggle with terminology here all the time, trying to come up with meaningful, efficient, and entertaining nomenclature to describe the incredibly complex thingies we deal with every day.

We are currently struggling with two developmental products that as of yet have not been totally defined or turned into a comfortable acronym. Our attempts this week concerned the new class of processors, which have graphics embedded in them, and the new class of PCs that will deliver stereovision.

HPU. The new processors from AMD (Fusion—Llano and Ontario) and Intel (Clarksdale, Arrondale, and Sandy Bridge) have multicore x86 processors, and multicore SIMD GPUs. In one sense they are neither fish nor fowl, and in another sense they are the ultimate manifestation of Moore’s law and massive integration. They are the epitome of the long promised heterogeneous processor, and as such we are designating them the HPU—Heterogeneous Processor Unit. We experimented with various forms of integrated graphics processor (IGP), integrated processor graphics (IPG), and processor integrated graphics (PIG). None of these terms or their subsequent acronym was commutative or satisfying, and seemed to cause considerable confusion because of the similarity to the established chipset IGP acronym—something new was needed and we think HPU satisfies that need. AMD prefers the term APU (Accelerated Processor Unit), and Intel thinks no term is needed other than i5 or i7.

We’ll try HPU and let the consumers, press, and financial analysts tell us if we got it right.

Everyone is in a hurry and looking for shortcuts. Shortcuts in their work process, shortcuts in their commute to and from home, and shortcuts to marketing a product. One of the tools we all try to use are abbreviations. The medical industry is often criticized and made the brunt of jokes about their lexicon of strange coded words. The computer industry is no different. I once sat in a meeting at Intel and listened to Intel people speak to each other almost entirely in acronyms. I even commented at one point, do you realize your sentence didn’t have a single word in it? However, that was a very efficient exchange between the Intel people. And although I didn’t have the secret decoder ring to be able to follow what they were saying, my understanding was not critical to the meeting or the information transfer. So we do speak to each other in secret languages, and they are not designed to exclude people or talk behind their backs, but rather to get the maximum amount of information exchanged as quickly as possible—they were in a hurry.

The naming of acronyms, perhaps taken to new levels of absurdity by the military and various government agencies, is critical to their acceptance and often understanding. Sometimes, a brand gets established (like CUDA) and then later as it gains acceptance an acronym is developed for it. An acronym not only has to be an abbreviation of the terms it’s describing, but it has to sound cool, and be memorable. Sometimes acronyms form words (FAT, NIC, or FOG (Free Open-source Ghost), etc.) and sometimes the letters of the acronym are pronounced (ASAP, AIB, CPU, etc.). Someone commented that since Adam and Eve were the first people they got to name everything. As far as we know, they didn’t use acronyms but then they didn’t have to deal with PCMCIA or OLPC or NSIT. Snake, on the other hand is a fairly simple concept. Though, perhaps they should have told each other, WOFTS (Watch Out For That Snake).

So we will go on creating acronyms to try and efficiently communicate complicated ideas and we’ll try to come up with appropriate ones that improve communication rather than hinder it– because we are always in a hurry.