Wikidata:Property proposal/transistor density

transistor density

Originally proposed at Wikidata:Property proposal/Computing

   Done: parts per area (P13745) (Talk and documentation)

Description	density of transistors for classifying the complexity of discrete circuits
Represents	transistor density (Q135397613)
Data type	Quantity
Example 1	GeForce GT 1010 (Q132176771)→24,300,000 transistors per square millimeter
Example 2	GTX 980 (Q133796828)→13,100,000 transistors per square millimeter
Example 3	Intel Arc B570 (Q131460527)→72,100,000 transistors per square millimeter
Example 4	Radeon RX 7900 XTX (Q115118503)→109,100,000 transistors per square millimeter

Motivation

I was told this was outside the scope of density (P2054) so a new property is needed--Trade (talk) 21:59, 16 July 2025 (UTC)

• @Peter James, PantheraLeo1359531, Vicarage, Wd-Ryan, Arlo Barnes: --Trade (talk) 22:00, 16 July 2025 (UTC)

Discussion

There are only about 4 density properties, none of which fit the requirement for chips, but I'm reluctant to have a swathe of new properties for transistors, populations, sowing, etc. I'd prefer creating a new property linked to areal number density (Q108914965), which could have qualifiers for transistors, people, seeds etc. — Vicarage (talk) 16:14, 16 July 2025 (UTC)

You are free to try and make such a proposal. I'll withdraw this if it ever gets made — Trade (talk) 23:02, 16 July 2025 (UTC)

Speaking of seeds, "transistor density" is an established term of art in the literature, and individual sources (example) could be used to seed the database and its reference sections.  Support as-is. Arlo Barnes (talk) 00:53, 17 July 2025 (UTC)

 Oppose too narrow a scope. Vicarage (talk) 02:49, 17 July 2025 (UTC)

 Support I agree because it is a common aspect that is used when it comes to integrated circuits. It may be narrow, but because of the broad usability (basically for any microprocessor) and the common term, I agree --PantheraLeo1359531 (talk) 07:43, 17 July 2025 (UTC)

 Support Density is indeed already defined by natural sciences while this is a different property from the technical world. It is also an important vendor independent metric for chipset manufacturing. Matthias M. (talk) 07:24, 18 July 2025 (UTC)

I'm thinking about what @Vicarage: proposes - number per unit area is what we are talking about. If we had a statement "GeForce GT 1010" "areal number density" "24,300,000 per square millimeter" it would need a qualifier to indicate we were talking about transistors and not something else (other electronic components perhaps). Putting "transistors" in as part of the unit doesn't feel right for a generic property like that. Nevertheless I think a more general property for this would be better to start with. ArthurPSmith (talk) 14:22, 22 July 2025 (UTC)

Yes, the numerator is unitless or unity. I think we only need one qualifier for this. We can reuse class (P2308) for qualifier and call the property "instances per area". Datatype of quanitity so we are free to use units to our liking, or make new ones if needed. $\frac{1}{m^2}$ exists as reciprocal metre squared, but there isn't one for $\frac{1}{m{m}^2}$ yet. I haven't made a property that are involved in conversions yet, so I want to create this one for the experience. And if you're thinking "why don't we just call it 'instances per whatever'?". That's overgeneralized and you end up with unholy things like P642 (P642).

Overall I think I like Vicarage's proposal best. We don't need the bureaucracy another round of proposals, and proposals for general properties tend to take longer and there is more discussion any way. I think we agree it should be done, the only question is how. Vicarage's vote was clearly towards the how part. Infrastruktur (talk) 18:54, 22 July 2025 (UTC)

instances per area	24 300 000 reciprocal millimetres squared edit class (P2308) transistor (Q5339) 0 references add reference
		add value

Looks ok, or do you hate it? In the latter case be specific please. Infrastruktur (talk) 09:32, 25 July 2025 (UTC)

Works for me, though I'm not sure I like that as the main label in English (I do see if class (P2308) is the qualifier then that label makes sense). Will wait for others to weigh in. ArthurPSmith (talk) 21:00, 25 July 2025 (UTC)

I think "parts per area" is more descriptive and would avoid confusion. -wd-Ryan (Talk/Edits) 17:42, 3 August 2025 (UTC)

---

•  Done @Trade, Vicarage, ArthurPSmith, Wd-Ryan: Created as parts per area (P13745). Had to widen a constraint for the class qualifier. Let's hope I don't get an angry mob of expert ontologists after me with torches and pitchforks. I politely left them a note at least. Infrastruktur (talk) 08:02, 25 August 2025 (UTC)