What's in a name
|
Prev: D2007: Typecasting 32 bit integer/longword to 32 bit single floating point not possible ?!? (GPU floating point format questions)
Next: Ohh that's juicy: Inventors mysterisly killed in car-crash ?!
From: MitchAlsup on 7 Jan 2010 13:38 Let us imagine a black box that receives signals from various components and asserts signals to various components. Together the black box and the components are a processor. Now let us imaging that there is a well known implementation of this black box with a well defined name that receives and asserts exactly the same signals as this black box. However, the interrior logic of this black box is completely different than the well known implementation, and is capable of dealing with data-flow problems the well known implementation cannot. Does this black box deserve the same name as the well known implementation? Does it deserve a different name? For example: The well known implementation is a ScoreBoard, The black box can deal with simultaneous WAW hazards The black box can deal with conditioni codes The black box can deal with memory ordering issues The black box can deal with branch prediction, and recovery, and The black box is made with CAMs which are only used durring issue But the data-flow computations are performed as in the well known implementation Mitch
From: Anton Ertl on 7 Jan 2010 14:26 MitchAlsup <MitchAlsup(a)aol.com> writes: >Let us imagine a black box that receives signals from various >components and asserts signals to various components. Together the >black box and the components are a processor. > >Now let us imaging that there is a well known implementation of this >black box with a well defined name that receives and asserts exactly >the same signals as this black box. However, the interrior logic of >this black box is completely different than the well known >implementation, and is capable of dealing with data-flow problems the >well known implementation cannot. > >Does this black box deserve the same name as the well known >implementation? Does it deserve a different name? If the name refers to the interface, use the same name. If the name refers to the implementation, use a new name. If the name refers to both interface and implementation (because there was only one implementation until now), it's probably better to introduce a new name and make the interface and implementation relations explicit. But it depends on your target audience: If they don't care for the implementation difference, it's probably best to use the same name. >For example: >The well known implementation is a ScoreBoard, That's definitely an implementation, not an interface. The interface is out-of-order execution, and another implementation is Tomasulo's algorithm. - anton -- M. Anton Ertl Some things have to be seen to be believed anton(a)mips.complang.tuwien.ac.at Most things have to be believed to be seen http://www.complang.tuwien.ac.at/anton/home.html
From: Ken Hagan on 18 Jan 2010 05:39
On Thu, 07 Jan 2010 18:38:53 -0000, MitchAlsup <MitchAlsup(a)aol.com> wrote: > However, the interrior logic of > this black box is completely different than the well known > implementation, and is capable of dealing with data-flow problems the > well known implementation cannot. So turn your question around. The familiar implementation does not deserve the same name as your new box, because it can't handle all the same problems. |