Looks cool, but I think this is UnApple. Apple creates cool products which most people want by creating a simple single unified design, while a modular design is trying to satisfy absolutely everyone with infinite configurations but increases complexity and failure points. A modular design is not elegant because you create too many choices for the consumer and the connections between modules are complicated for the average user.
I would buy something like this. Your average person just wants to buy a workstation, login and go.
Your average consumer doesn’t buy the Mac Pro.
This (admittedly well-drawn) concept imagines potential configurations for consumers. It posits a nerd solution for a problem few consumers probably have.
Exactly. The people who still want a MacPro are looking for customization and expansion.
Its a good concept, but I agree unlikely.
Is it completely unfeasible (from a cost/processing standpoint) to connect two computers (not unlike the Mac Mini) and have them share processing?
I don’t think Thunderbolt is fast enough, but I’ll use it as an example…
CPU Unit A is connected by Thunderbolt to Drive A, Drive B, and optical Drive A. I decide I need a faster computer, so I buy CPU Unit B and connect it with CPU unit B… with a sufficiently fast interconnect, could both units efficiently share the workload?
Modular design is easy if it follows the Lego Rule… simple blocks, simple interconnects.
Your average Mac customer doesn’t buy a mac pro either. Moeskido is right, this solves problems for you and me, but Apple doesn’t do that much any more, which is my point.
Buzz-box, with those fans.
I had been hearing for a while now from an ex-Genius that Apple were playing with the idea of being able to daisy chain something like multiple Mac Minis together using Thunderbolt connectors (or, more accurately, the original Lightpeak) to add computing power via a building block process.
will look great with a samsung logo on it in the spring.
I don’t even see what problem this is solving.
People who … need an 4/6/8/12 core Xeon monster, but can’t abide the idea of a full tower?
Yeah,
all four of them will rush out and buy that – but everyone else in the
already constrained Mac Pro market will just get the full tower package
anyway.
More complexity, more cost, less structural strength…
for what? So a small portion of the already small market for them can
have a box six inches shorter?
It’s not unfeasible at all, if the task is easily parallelized.
Problem is, most common computing tasks tasks aren’t (though admittedly many Workstation tasks are) – and Apple’s had that for years. It’s called XGrid.
Looks cool, but I think this is UnApple. Apple creates cool products which most people want by creating a simple single unified design, while a modular design is trying to satisfy absolutely everyone with infinite configurations but increases complexity and failure points. A modular design is not elegant because you create too many choices for the consumer and the connections between modules are complicated for the average user.
I would buy something like this. Your average person just wants to buy a workstation, login and go.
Your average consumer doesn’t buy the Mac Pro.
This (admittedly well-drawn) concept imagines potential configurations for consumers. It posits a nerd solution for a problem few consumers probably have.
Exactly. The people who still want a MacPro are looking for customization and expansion.
Its a good concept, but I agree unlikely.
Is it completely unfeasible (from a cost/processing standpoint) to connect two computers (not unlike the Mac Mini) and have them share processing?
I don’t think Thunderbolt is fast enough, but I’ll use it as an example…
CPU Unit A is connected by Thunderbolt to Drive A, Drive B, and optical Drive A. I decide I need a faster computer, so I buy CPU Unit B and connect it with CPU unit B… with a sufficiently fast interconnect, could both units efficiently share the workload?
Modular design is easy if it follows the Lego Rule… simple blocks, simple interconnects.
Your average Mac customer doesn’t buy a mac pro either. Moeskido is right, this solves problems for you and me, but Apple doesn’t do that much any more, which is my point.
Buzz-box, with those fans.
I had been hearing for a while now from an ex-Genius that Apple were playing with the idea of being able to daisy chain something like multiple Mac Minis together using Thunderbolt connectors (or, more accurately, the original Lightpeak) to add computing power via a building block process.
will look great with a samsung logo on it in the spring.
I don’t even see what problem this is solving.
People who … need an 4/6/8/12 core Xeon monster, but can’t abide the idea of a full tower?
Yeah,
all four of them will rush out and buy that – but everyone else in the
already constrained Mac Pro market will just get the full tower package
anyway.
More complexity, more cost, less structural strength…
for what? So a small portion of the already small market for them can
have a box six inches shorter?
It’s not unfeasible at all, if the task is easily parallelized.
Problem is, most common computing tasks tasks aren’t (though admittedly many Workstation tasks are) – and Apple’s had that for years. It’s called XGrid.
That’s how all modern supercomputers work.