It is kind of funny. A couple of times articles on HN about complexity theory with assumed knowledge of terms like P-time or NP-hard. In these instances some commenters complained that they could not easily understand what it meant.
I really had to read closely to understand that this is about chip design.
The article mentions chips, stacking, packaging, memory and yield in the second paragraph, and foundries, dies, silicon, ICs and PCBs in the third (and that’s just 10% into the article), I thought it was obvious pretty quickly.
My understanding is that the cerebral cortex has several layers of active neurons "grey matter" in a folded but basically 2D structure and there are bunch of axons "white matter" that connect those layers to other parts of the cerebral cortex and also to other structures that are structurally different such as the brainstem, hippocampus, cerebrum, etc. that evolved at a different time.
I think maybe "Floats" and "Ints" in this case are not referring to numerical computation. Perhaps not the best terminology.
The point being, that typical components are organized on a 2D wafer's surface, aligned with the 2D's surfaces two axes. Transistors are not normally oriented with arbitrary rotations in the 0-360 range of possible rotations.
One reason for this alignment, is that a consistent angle between transistor layout and the silicon's preferred crystalline plane direction, allows for control of material stress and device electron mobility.
That would hold even stronger for 3D structures. Transisters are still likely to oriented in exactly the same way on the 2D surface of each wafer. Not oriented to any of three possible 2D grid orientations of a 3D grid (think: the three 2D surfaces that come together at a corner of a cube).
And certainly not oriented with arbitrary rotations across all three dimensions.
With 3D the cooler still goes on top, 2.5D vs 3D is more about how the signaling and interconnect is structured. When the interconnects are still mainly planar it will be 2.5D when most of the interconnects are throughsilcon it’s called 3D.
Cooling 3D is more challenging because you need to wick the heat through the stack. It’s usually done by embedding additional metal layers for heat wicking but there is also a lot of research these days into micro channels for in-die liquid cooling.
Just how tall would these be? Micrometers, it seems like. I guess that the materials themselves are not heat conductive enough to divert the heat evenly throughout the stack?
This sounds sooo crazy to me :), I've just got 0 intuition about this stuff!
Keep your electrons in the fridge when you're not using them, if your electrons are already at room temperature then you're at a heat disadvantage right off the bat
the classic example is "pandemonium!", a 3D platformer that only allows you to move on two axes. you move on the surface of an imaginary ribbon winding through the 3D space.
i've never heard 2.5D used to refer to that, i've only heard that used to refer to games that are rendered in 3D but are played in two dimensions, (where that wouldn't be assumed, so excluding, say, a tennis game.)
You would describe as an innate and obvious part to the environment. Interesting how something that took 1000s of years to articulate is just innate to the layperson in the modern world.
I'm not putting it on a pedestal. I'm afflicted with the need to know the source of beliefs. It could be with more study one might find the value is anything but arbitrary. I guess I should bow out of this thread since I'm never really sure what we are discussing. One certainly doesn't need to understand the foundations of understanding to get by in the modern world. I can't really cope without knowing the roots of why we do what we do.
I guess so... I lost track:) The original comment seemed to imply that we had a coping system for understanding 3-D. I had an undergrad drawing class that started with Brunelleschi and framing tool he used to frame the Palazzo Vecchio. The story goes that he spent about 5 years trying to accurately reproduce what he saw in the frame. This is how he developed perspective. If one looks at the Plazzo Vecchio it looks like an obvious place to work on perspective.
There is another comment regarding a theory that Brunelleschi was influence by the book of optics and I can see an influence. I'm not sure how flushed out the perspective system is. I have only done a cursory look and the book argues that light in our eye on is perceived at right angles. Off angle light is not detected. I can see Brunelleschi reading that while studying the Plazzo Vecchio
and developing perspective.
I wasn't asking for clarification because I thought I understood your statement.
As a side remark, it is probable [0] that Brunelleschi built his understanding of perspective from the Books of optics [1], written by Ibn al-Haytham around 1020 AD.
I read a little of the book of optics and I could see it influencing Brunelleschi. The other reference requires a login so I wasn't able to look into the claim.
I really had to read closely to understand that this is about chip design.