Karl
5/29/2012 9:37:54 AM
User Rank Guru
Re: time for a new appellation
After sleeping on it, It does not matter so I am out of it. There is no wording that will enable the uneducated to understand. Every field has its jargon and one must learn the jargon to communicate. I do not understand all the computer science/programming jargon, so if they think that an FPGA is programnable using a programming language then a new acronym ain't a gonna matter.
Thankyou and good luck, Karl.
Karl
5/28/2012 3:53:02 PM
User Rank Guru
Re: time for a new appellation
Agreed. Just one point. Only the processors are programmable and there are many other peripherals to consider.
Let us wait.
Re: time for a new appellation
@Karl: I must admit that I like "Configurable" as compared to "Programmable".
I'm not sure about the "Virtual" -- I think you are using that to refer to the fact that the gates are actually implemented in LUTs and so aren't really there. This comes on to my next point which is...
I don't really like the "Gate" -- I know that everything boils down to gates and registers at the end (or transistors if we want to be pedantic), but when you think of a function like a dual hard core ARM Cortex-A9 MCU ... it's hard to think of the "Gate" part of the name as embracing this.
When I first heard the "All Programmable Device (APD)" term in a Xilinx press release it didn't really grab me, but I have to say that I'm starting to like it. I did just toy with the idea of "All Configurable Device (ACD)", but when we come to think of a component that contains conventional programmable (configurable) fabric with soft and hard processor cores, then I think we have to pick one or the other, because a name like "Configurable Device with Programmable Processors (CDPP)" is getting a bit too over-thought (and I can say this because it's me that's doing the overthinking :-)
Obviously we aren't going to settle this in one day ... I'm really looking forward to hearing what other folks think when they get back from their holiday weekend.
Karl
5/28/2012 2:21:08 PM
User Rank Guru
Re: time for a new appellation
Another view is the gates are virtual and the "fabric" is an array, net, or mesh.
Array and mesh imply a rectangular structure while net is more general.
Connectable and configurable are more appropriate than programmable and both imply ome intrconnect structure.
Configuable implies changable more than connectable.
Configurable Virtual Gate Device.
The device being defined is a chip made by a chip maker and that is not apt to change.
Configurable Virtual Gate Chip.
If the vendor happens to have put a hardcore or peripheral on the chip, that seems OK because that is still part of the configuration(system) and the virtual gates are the other part.
Maybe Configuarable Virtual Gate Component would focus on the "designable" part of the chip. Remember FPGAs being used as "glue" logic?
Virtual Gate Array Ccomponent may work because the chip is structured by row and column so it is rectangular and "Component" allows for the other components. If the general usage degrades to Virtual Gate Array Chip or Device that would not be too bad to live with. VGA is already taken by the PC world.
"Egad, the chip is a VGAD!"
Of course only those who understand the chip structure would see the need for virtual, but I hope that we are looking for technical accuracy as well as some cosmetics.
Re: time for a new appellation
@Karl: Re "Field Programmable Array/Ram Network" ... so "FPARN" (pronounced "pharn")? Hmmmm, It lacks a certain something while rolling off the tongue (grin)
Re: time for a new appellation
@Karl: The original reason for the "Gate Array" part of "Field Programmable Gate Array" is that the predominant ASIC technology of that time was the Gate Array (GA).
Of course the basic cells on the GA ASIC were just unconnected transistors and resistors and suchlike. It was only when they were connected together that they implemented primitive gates and registers.
I agree that the lookup-tables in an FPGA don't implement logic gates on a one-for-one basis ... but I still think that the original FPGA name made sense at the time.
The problem is that we are coming up on the FPGA's 30th birthday (2014 if we go from when the FPGA was first announced in 1984, or 2015 if we go from when the first real device was presented to the market in 1985). The capabilities of these devices have evolved so dramatically over the years that the term "FPGA" simply does not reflect the nature of today's devices.
Karl
5/28/2012 9:32:59 AM
User Rank Guru
Re: time for a new appellation
I just did the poll and Field Programmable Array/Ram Network came to mind. The memory blocks are ram, the LUTs are ram, the fabric is connected by a ram(shift register), MMIO makes peripherals look like ram to the cpu, etc.
(Here I go talking to myself reply to my post)
Karl
5/28/2012 9:12:06 AM
User Rank Guru
Re: time for a new appellation
Hi, William. I wonder if FPGA was ever correct -- there are no physical gates for the user. The designer can use primitives (schematic input, HDL) and get the function of a gate but that is the tool chain image.
The "programmable" part is a network of rams. Since a ram can be set up to generate an output equivalent to any Boolean combination of its address bits which are used as Boolean inputs, then the output emulates one or more logic gates.
The higher level blocks also on the chip are system components/building blocks.
Now we are getting into system design combined with logic design. It seems possible to separate the chip design into the two levels because the interface to the system components is probably MMIO or DMA and used by a processor and connected physically by a system builder tool. Very much like the MCU board that has a mixture of system components.
This gets cloudy because the MCU is an off the shelf, tested and verified item. The FPGA has to go through all the design steps rather than changing a PCB, which may also be necessary for the FPGA based system.
time for a new appellation
So true that the Devices have morphed beyond the name --
|
|
Can statistical or heuristic verification really work for FPGA designs?
One of the things I've been wondering is whether or not the "okWireOR" module is really just a giant OR, or if the order in which things are attached matters.
I am shocked and horrified. It appears that those little scamps at Planet Analog are writing blogs pertaining to field-programmable issues.
This week's live online chat takes place on Thursday, May 23, 2013, at 1:00 p.m. ET.
Duane has decided that the time is ripe to get his ZedBoard bolted onto his robot with a Linux distribution up and running. That was the ultimate plan anyway, so why wait?
|
Tweet This
