Slashdot Mirror

You're absolutely right, we can't.

But do not be surprised that some people want to talk about it. And certainly do not be surprised when foreign nations decide not to buy US hardware and software. I'm not saying Chinese hardware is any more innocent, that's not the point I'm making, the decision that the NSA made (was anyone consulted?) to subvert the security of every PC on the planet will have repercussions. And this will have significant impact on the IT economy of the US.

As an individual wanting to use a trusted computing platform, for my own computing needs and those of my small business, I now have to look far and wide for hardware I can trust. Personally I'm keeping a close eye on the Russian effort to shrug off x86 dependence.

s/x86/itanium

Fixed that for you

If it ain't broke, don't break it by trying to fix it. When HP decided that Itanium was the way forward, that was "HP-UX on Itanium" for their machines, so they didn't "[decide] HP-UX was yesterdays news" at that point.

Presumably by "decided HP-UX was yesterdays news and x86 was the way forward" the person whose statement you're "fixing" was referring to HP announcing Xeon-based Superdome servers not running HP-UX, and HP-UX not being ported to x86.

A diverless car without any kind of overiding system enabling a human to take control (for whatever reason) is like a fully autonomous airplane. You will never see them.

Seriously? I'll never see a fully autonomous airplane or driverless car? You see these all the time, especially in rural areas.

http://www.uavcropdusterspraye...
http://www.enterprisetech.com/...

So obviously these cases do fit into your "constrained set of specific roads".

Anyway, I do suspect there will be a period of time where there are fully driverless roads, fully human-driver-only roads, and mixed roads. I suspect the major highways will be among the first to be driverless only (you already aren't supposed to be a pedestrial or a cyclist on those roads).

And in that case you fucking want the ability for a human driver to override an AI if the need arises.

Only in the same sense that I want to remotely slam the brakes on asshole's car on the road if the need arises. Otherwise, at a certain point, I fucking *don't* want a human yanking the wheel from the machine. Yes, we aren't there yet, but you're ridiculous if you don't think that's happening.

AI can't take into account all possible modes of failure or unexpected events.

Literally nothing can, including humans. This is a meaningless goalpost. If you take it seriously, then cars should not be allowed, whether or not there is a human driver.

Googling the JEDEC document number JESD235 from the article found several references with NVIDIA talking about this for 2 years now for their Pascal series of chips after Maxwell.

future-nvidia-pascal-gpus-pack-3d-memory-homegrown-interconnect
http://en.wikipedia.org/wiki/High_Bandwidth_Memory
http://www.cs.utah.edu/thememoryforum

It looks like there are enough big names wanting to escape the grasp of Intel's grip to tip the scales but only time will tell. I found a painfully in depth article discussing the reasoning and driving forces.

Softlayer was only recently acquired by IBM and has been using CloudStack since before the acquisition. Here's an article mentioning Softlayer's IMS (Infrastructure Management System) and CloudStack:

SoftLayer Brings Hourly Billing To Bare Metal Cloud

My guess would be that the real perk is bandwidth and latency. Unless Intel really phones it in on integration, the FPGA should have about the fastest, lowest-latency, link to the CPU, possibly even some of the cache, especially if they throw in a big chunk of eDRAM, as they have for 'Iris Pro' parts, that money can buy.

As usual, the slashdot post has the absolute worst story link. compare http://www.enterprisetech.com/... which gives you links to where it gets its info, namely https://communities.intel.com/... and http://gigaom.com/2014/06/18/i... ... the latter is the interesting link because it tells us that the FPGA will have access to main memory. I personally would presume that means it's tied into the memory controller somehow.

Less of a "Hey, let's do this instead of GPU compute!" and more of a "It sucks that our weirdo application-specific operation is probably never going to be one of Intel or AMD's extensions to x86; but this is the closest we can get to having it added" thing.

What I began fantasizing about immediately upon reading the article was some sort of optimizer that would semi-automatically build functional units to perform whatever function the CPU was grinding on at the moment, with some sort of recognition engine and periodic updates garnered from participating customers to help special-yet-common cases. As well, seeing how customers actually use FPGA with their products will help Intel decide what functionality to add to their next (or next+1, etc) processor.

There are already options to add an FPGA to your Xeon system, with its own blob of RAM. Since they talk about this being fundamentally different, I'm not sure what makes sense except the idea of it being connected at the memory controller. Hopefully there will be a talk with some nice block diagrams released soon.