>1.1. by shortening average "wire" length No, the correct answer was "3D memory has nothing to do with increasing processor clock speed." I'm currently using the HMC (3D memory with a logic layer) for my research. We've talked about different sorts of applications, mostly revolving around memory-bound codes. The shortened wire lengths will indeed reduce latency and speed up such applications. However, this does not mean processors will run any faster. Memory is not part of the processor.
>1.2. shortening average wire length would lower "switching losses" of the whole clock tree thus further reducing power consumption Like I said, memory is not part of the processor. It will be great to reduce power draw from memory, considering it is such a large part of total system power draw. But its not going to change anything with the processor, except free up some power budget, when we start making hardware over-provisioned systems.
Grayson Morgan
Sometimes I think if you're not one of them, then you can think of their whole Standard Model as a hoax. I'll admit I do believe them, but I mean, are all these subatomic particles really "particles" as a someone ignorant like me would understand? The Higgs imparts mass huh? How does that work? Someone here must be somewhat familiar with this stuff. When you're taught what a Higgs is and how it functions, is it at all intuitive from what you know?
Carson Butler
Gravity Waves
Ethan Harris
> Gallium Arsenide
In what context?
Hunter Ramirez
IUT Theory
Nathaniel Reyes
Tetraethyl Lead is good for your car engine >it's not >the heavy metal residues left by TEL can cause catastrophic frictional damage to the cylinders and pistons >fouled catalytic converters
The social impact of TEL >Poisoned millions >Led to an increase in criminality >Government knew it was bad, was bought off by Standard Oil of New Jersey (AKA Exxon)
Brandon Kelly
>>it's not >> Government knew it was bad >>>>>>>>>implying
you one of those /pol/tards that got lost here by mistake? Or have you actually taken a course on applied thermodynamics?
Henry Turner
Here is a big one:
"physics"
Joshua Barnes
t. Indoctrinated amerifat.
>Pic very related
Wyatt Sanchez
>How the fuck will 3D memory help us increase processor clock speed? Clocks are as fast as they are going to get. That's why we went pipelined, to multi-core, and then massive parallel with GPUs. The future is different interconnect topologies, like a compute fabric made of arrays of cores that can only access their neighbors memory, not some global memory as in current machines.
>3D processors Heat dissipation will be the largest obstacle to 3D. Btw, modern CPUs are dozens of layers deep.
I think silicon optical buses for moving data across large distances on a die will be a thing in the near future (within 5 or 10yrs), but they will be useless for anything else. Electro-optical couplers and decouplers take up a lot of floorspace, so these buses can't be used for short distances. That being said, it will only very modestly boost performance.
I honestly think more interesting applications for silicon photonics will be a return to analog computation for things like even faster FFT hardware and signal processing.
