2
$\begingroup$

Caches appears at many different levels or places in computer system. You can have on-chip caches, sytem caches and network caches. But the only tag about caches is apparently the tag , which seems very restrictive. Is that intentional, and why? Why not just ?

$\endgroup$
5
$\begingroup$

The tag is about caches in computer architecture and their relevance to programming. It so happens that we have many questions on that topic, mostly about CPU/memory architecture and about algorithm design.

The tag is definitely not too restrictive — if anything, it's overloaded, as the relevance of CPU caches to algorithm speed and the architecture of memory subsystems are very different topics. Other tags (e.g. or serve to distinguish these topics.

would be a useful tag for a question about the design of caches. I can't find any.

and are very different topics. The two tags could coexist, on a question about memory subsystem design (as opposed to what we have in the , which is about analysis of existing designs); there is not much correlation between their domains.

$\endgroup$
  • $\begingroup$ Can you give an example of a caching question that is computer science, but not about cpu-cache in the broader sense that babou proposes? $\endgroup$ – Raphael Jun 10 '14 at 21:19
  • 2
    $\begingroup$ @Raphael Something about caching in network protocols, is the obvious example that comes to mind. And in the other directions, most cpu-cache are not really about caching, or about a very specific aspect of it. $\endgroup$ – Gilles 'SO- stop being evil' Jun 10 '14 at 21:31
  • $\begingroup$ It seems that we could indeed potentially use both caching and cpu-cache with an overlap in questions that deal specifically with the impact (the design of) CPU caches on algorithms. $\endgroup$ – Raphael Jun 11 '14 at 5:49
0
$\begingroup$

I don't see any reason against that; as far as I know, the term caching is not overloaded (with similarly popular notions). Also, as far as computer science is concerned, we probably don't care where exactly the cache resides.

If we change the tag it shouldd probably be to go with our convention of using the plural, or alternatively . I'd associate the former more with "How does a cache work?" and the latter with "How do I analyse this algorithm considering caching?" but that may be my personal bias.

$\endgroup$
  • 1
    $\begingroup$ caches or caching do not make much difference since tags are intended to be used as approximations. I tend to like caching better since it emphasize the effect rather than a specific device to achieve it. But not worth a fight or even a disussion. How do we go about changing this? $\endgroup$ – babou Jun 10 '14 at 21:09
  • 2
    $\begingroup$ caching (or caches, I'm indifferent) is not at all the same topic(s) as cpu-cache. $\endgroup$ – Gilles 'SO- stop being evil' Jun 10 '14 at 21:16
  • $\begingroup$ @babou We wait a couple of days for more opinions to trickle in. If there is a consensus, we can mod-hammer the thing; in the meantime, you can suggest a synonym yourself; if it collects enough votes, it will be enacted independent of the discussion. A proper tag merge (which removes the old one), however, requires moderator action and thus, arguably, some result here. $\endgroup$ – Raphael Jun 10 '14 at 21:17
  • $\begingroup$ @Gilles: Is it a superset? That would be okay then, I guess. $\endgroup$ – Raphael Jun 10 '14 at 21:18
  • 1
    $\begingroup$ Calling cpu-cache a subset of caching sounds like calling regular-languages a subset of computation-models. $\endgroup$ – Gilles 'SO- stop being evil' Jun 12 '14 at 15:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .