3 ms·

interesting, so essentially a giant memristor blob store which a bunch of compute nodes can simultaneously leverage? If the memristor promises are to be realiz

by spydum 12y ago

interesting, so essentially a giant memristor blob store which a bunch of compute nodes can simultaneously leverage? If the memristor promises are to be realized, TB if not PB levels of storage at >SSD speeds, accessed by many compute nodes as shown in diagram, which could also be many cores by themselves. Think of like a hadoop style cluster, self-contained on amphetamines? If any of this is true, why not start in phases where you have memristor replacing SSD/existing system memory.. why this monster design? unless they are trying to leverage their new fancy technology to disrupt the cisco/dell/supermicro enterprise server market?

techdragon 12y ago
Because that's exactly what HPA want to do, foolishly try and disrupt the market with unproven hardware no one can afford.
- sgt101 12y ago
  It might be the opposite, for corporates, this might be hardware that no one can afford not to have. People spend a lot on hardware, and a lot on electricity. If HP can do what others can do, but at 1/4 of the cost, and they charge 1/3 of the price, then money will flow to them like water.
- yitchelle 12y ago
  > foolishly try and disrupt the market with unproven hardware no one can afford. Sometimes, that is what research is all about. Some work out, some don't. I am actually quite excited about this for HP.
  - spydum 12y ago
    nobody objects to that.. the thing is, it sounds as if they have some very cool technology (memristors) which could make other existing products significantly better, but are explicitly withholding because they want to force some paradigm shift.. that seems like a pretty bold bet. What if nobody wants it?
phkahler 12y ago
>> If any of this is true, why not start in phases where you have memristor replacing SSD/existing system memory.. why this monster design? From what I've read, the memristors may not be working at scale. A single device works, but chips packed full of them have not AFAICT worked out yet.
- kahirsch 12y ago
  > From what I've read, the memristors may not be working at scale. Where have you read this? That sounds interesting.

by spydum 12y ago

techdragon 12y ago

Because that's exactly what HPA want to do, foolishly try and disrupt the market with unproven hardware no one can afford.

sgt101 12y ago
It might be the opposite, for corporates, this might be hardware that no one can afford not to have. People spend a lot on hardware, and a lot on electricity. If HP can do what others can do, but at 1/4 of the cost, and they charge 1/3 of the price, then money will flow to them like water.
yitchelle 12y ago
> foolishly try and disrupt the market with unproven hardware no one can afford. Sometimes, that is what research is all about. Some work out, some don't. I am actually quite excited about this for HP.
- spydum 12y ago
  nobody objects to that.. the thing is, it sounds as if they have some very cool technology (memristors) which could make other existing products significantly better, but are explicitly withholding because they want to force some paradigm shift.. that seems like a pretty bold bet. What if nobody wants it?

phkahler 12y ago

>> If any of this is true, why not start in phases where you have memristor replacing SSD/existing system memory.. why this monster design? From what I've read, the memristors may not be working at scale. A single device works, but chips packed full of them have not AFAICT worked out yet.

kahirsch 12y ago
> From what I've read, the memristors may not be working at scale. Where have you read this? That sounds interesting.