Brice Goglin wrote: >> What hasn't been finished yet and to my opinion needs to be for v1.0, is >> the prefix/suffix/whatever to easily distinguish between physical and >> logical numbers in lstopo. >> > > I played with this today and arrived to these conclusions: > * Having 'l' or 'p' prefix without # or any other special character is > very confusing since we already have things like L1 for cache levels and > P for proc numbers > * Having nothing before the number is a bad idea, it would be confusing > with cache/misc levels > * I like '#' as a prefix since it's widely used as a symbol meaning > "number" and I found no other nice character > > So I see two solutions: > * p#2 and l#2 (prefixed with a space) > [...] > * #2p and #2l > [...]
I am asking people here, some are confused by all these --logical/--physical outputs. One idea that came is to always keep the logical index and print the physical index as an attribute. Something like: $ lstopo - Machine(1508MB) + Socket#0 (phys=0) + L2(phys=0,2048KB) L1(phys=0,32KB) + Core#0(phys=0) + P#0(phys=0) L1(phys=1,32KB) + Core#1(phys=1) + P#1(phys=1) I am interested by this idea. The physical index could actually be displayed as an attribute for Proc all the time since it's often interesting. The physical index of other objects is often useless and even confusing. So what about we: 1) keep --logical by default, add physical index in Proc attributes by default 2) if --verbose, we add the physical index in every object attribute 3) if --physical, we replace #0 with physical indexes as #0p (and don't print logical indexes anywhere) I would actually just drop (3). If we really need --physical to match other tools command-line options, we could just make --physical equivalent to --verbose. Brice
