craigjb

They did not only risk their own lives. They put other people around them at an increase risk when it was unnecessary. That is the argument, that it was not necessary. This same demonstration could have been done on a track or other controlled environment where the public was not in danger.

There is also a lot of space between general purpose CPU and completely blank-slate FPGAs. GPUs are essential very wide data word processors (one instruction but a large data vector).

And, there are also configurable pipeline processors that consist of multiple ALUs (can be vector data) with reconfigurable connections between them. So rather than eat the overhead of generic bit LUTs in an FPGA, you reconfigure the interconnect in a fabric of ALUs. The fabric can contain specialized ALUs (or execution units), and varying densities depending on typical usage. This avoids a the technology mapping and place & route of FPGA design. Translating a description of hardware (HDL) into actual lookup table data is a massive compute problem for large modern FPGAs. However, if we collapse the routing to just data buses between compute units, the problem can be solved in real-time. This way, a compute pipeline could be reorganized by an application at run-time without using a full-fledged hardware description language with all kinds of very low-level constructs. Higher-level language compilers already exist for this kind of architecture in academia.

EDIT: The sea of ALUs can also contain memories, FIFOs, and other block elements. However, they would all operate on the same word size to reduce the routing problem and allow maximum density implementation.

In fact, this is the essence of how micro-coded instructions in a modern CPU work anyway--instruction 'scheduling' is basically figuring out how to route data transfers in a sea of execution units.

The article "Fundamental Underpinnings of Reconfigurable Computing Architectures" in the March 2015 Proceedings of the IEEE contains a wonderful introduction to all these concepts.

Xilinx invested heavily in TSMC's TSV interposer and likely their IN-FO fan-out packaging. These allow them to integrate more silicon area in the same package space. It can potentially offset the technology node gap, and maybe for lower cost. Yes, TSV is expensive, but a process node is even more expensive.

It's not a bubble! This time is different!

Deca Technologies - Phoenix, AZ - on-site - EDA Software Engineer

Deca Technologies is an electronic interconnect solutions provider that offers wafer level chip scale packaging (WLCSP) and fan-out wafer level packaging (FO-WLP) services to the semiconductor industry.

Deca's software engineers develop next-generation electronic design automation (EDA) tools and manufacturing software that enable Deca's unique approach to wafer level semiconductor packaging. EDA software requires cross-disciplinary skills and knowledge related to the semiconductor packaging technology and the complex CAD tools used during the design process. In addition, manufacturing systems require skills in high-reliability distributed systems that perform compute intensive tasks across large server clusters. As a software engineer, you will work with a team on developing and supporting EDA tools and distributed systems in a fast-paced environment. We need our engineers to have a passion for learning new technologies and the drive to tackle new problems as we transform the electronic interconnect industry.

Our technical stack consists of Python and C/C++ for high performance portions. These systems run in distributed configurations, both local high reliability clusters and cloud resources. Deca Technologies offers software engineers a chance to apply their skills to a new field and learn a vast new domain.

Contact: craig.bishop@decatechnologies.com

Solder ball size is definitely a factor in modern packaging cost! Silicon area is really expensive. That's why places like where I work are developing fan-out wafer level CSP. It has all the benefits of water level processing cost, but extra area for solder balls is made of cheap mold, not expensive silicon.

These WLCSP parts are definitely not solderable at home though. But, with the rise of low cost PCB and assembly services, I'm not sure that matters as much as it used to. For example, one project I'm working on with a 384 ball BGA cost me twenty dollars to have placed and reflowed locally. And, new projects involve way fewer chips because of the scale of integration. As an analog engineer, it makes me sad, but I appreciate what I can make at home now!

Nearly every chip in there is a WLCSP (wafer level chip scale package). Seems like the industry is trending toward no package is the package. Also, it will be interesting to see if there are any fan-out packages, like eWLB. Several IDMs have started integrating several ICs horizontally--usually things like audio codecs or USB transceivers. With multi-die fan out, you save a lot of PCB real estate. In small form-factor devices like the iWatch, packaging is a huge differentiator.

Here's an interesting paper about actually implementing this (sorry for paywall).

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumb...

The NSA is probably the worst-case of the general public not trusting government institutions. However, it seems that government has lost trust across all functions, despite only marginal increase in corruption. Is new, faster communication just making us more acutely away of the corruption that inevitably happens? I don't think it's bad to be aware, but we really need to learn to scale our response. A program can have a % of corruption and still achieve success. Also, I wonder whether mistrust of institutions stems from the record wealth disparity. In general, institutions seem much less trustworthy when a small percentage of people can disproportionately influence them.

I agree. How come as software engineers, we like to talk about larger emergent behavior from large groups of simple pieces, but when we talk about governments and nations (large groups of complex pieces) we use analogies to simple systems? For example, the national debt does not operate like a household debt. The cacophony of systems at work in a nation makes it exponentially more nuanced. Some rules are emergent only at the larger scale. Or, international relations. Nation to nation relations don't consist of just the heads of state, but whole organizations of people interacting. Again, a large group of complex things (people) interacting probably creates some emergent behavior not visible in small groups. It's frustrating to see these systems reduced without any thought.

I agree. I don't think interaction should be necessary in a good visualization for most questions. It should add usability on top of an already usable graphic. This graphic would not be usable if published in a paper.

They're also near impossible to buy outside of China or NDAs. So it may be open, but I couldn't build one.

Because they are not. Colonizing other worlds is.

Edison is the size of a postage stamp including DDR. The picture linked is of the module plugged into a board.

Notice the power specs are missing from the table...

Well I'm glad someone made the entire country pay for the Chancellor of the Exchequer's decisions.