What makes a microprocessor different from a microchip? Is is the ability of a microprocessor to load information into transistors, use it/process it and then reset it`s state so that new information may be loaded and used again. Microchips are just an entity meant to direct information based on a hardcoded algorithm, rather than execute mathematical operations. Is this exact?
In exactly the same way as my house is multiple houses, right?
There were a number of processors thirty to forty years ago requiring multiple chips. Maye there still are some, but I haven't seen any multi-chip processors for quite some time.
In any case, you turned my analogy upside down. A processor (the functional part, the housing unit) resides in a physical unit, a chip, a building. Like an apartment can house multiple functional housing units, can a physical chip house several processors. In a multi-core CPU, some of the processors may be identical, but the chip may in addition house I/O-processors, graphic processors, debug processors and other specialized pones.
It can't, in economical terms, they a single piece.
Several of my friends have housing units which has a main house, a garage, a shed and even other houses. They are one housing unit, but spread on several physical units.
In the old days, you could have one main CPU, supported by a Floating Point Unit, possibly also a Memory Management Unit - the CPU, FPU and MMU being three different chips, making up one single processor.
If you go further back in history, even the CPU core was built on multiple chips: The iAPX432 processor had one chip to fetch and decode instructions, one to execute them, and a third chip controlling I/O.
For some years, many 16- or 32-bit CPUs were built from an array of "bit slice" chips - typically 4-bit AMD290x. The 290x were labeled 'bit slice processors', but they were not: They were hardcoded ALU logic that could be activated through control lines. The programming was external to the 290x. (That's exactly what you did when building a real processor from 4 or 8 290x chips.
Way back in time, CPUs were typically built from hundreds of 74 series chips, usually with support of a fair number of discrete components.
So in the old days, processors were built from several chips. Nowadays, several processors may be placed on the same chip.
You saying a chip can be a microprocessor and a microprocessor can be a chip?
"<x> can be a <y>" is the wrong way of saying it. To build the functionality of a processor, you might need to use several chips, although that is rarely the case today. And you can build the functionality of multiple processors onto the same chip. A chip is a physical electronic component. A processor is a functionality realized by (a) suitable chip(s).
My problem is understanding how this can be a problem to understand.
There's tons of other chips out there that are not processors, like arrays of AND gates, and whatnot.
The fraction, at least in transistor count, has been steadily falling, though. Years ago, you designed electronics from 74-series chips. Nowadays, that is for extremely simple cases.
The primary example of "dumb" chips today are memory chips of various technologies - but modern memory access protocols are so complex that the handling is getting close to a 'processing' task. I haven't yet heard of a memory chip flash update, but I won't be surprised the day it comes. (At least not as surprised as I was the first time a flash update was announced for my SLR lens. What?? Does a lens have a flash?? Nowadays, most SLR lenses do!)
What I might be saying probably isn`t science language but it seems that microprocessors have enough unique characteristics to make them stand out from the rest in the microchip family. Like it has become a breed of its own and deserves a distinct chapter.
The country I`m coming from doesn`t have a microprocessor printing plant so I`m not qualified to say what is the correct term for this or that in this field.
A chip (whether micro or not, I rarely hear people refer to microchips) is a collection of electronic basic components (transistors, resistors, capacitors, ...) in one physical package - a quite general term.
One chip, or possibly a well defined set of chips, may be designed as programmable: It may access a set of instructions from a more or less independent storage, that will determine how the chip(s) behave(s). Another set of instructions (i.e. another program) can make the chip behave differently. The programmability is what identifies the chip(s) as a microprocessor.
'Microprocessor' is functionality, not transistors etc. 'Chip' is any package of components like transistors etc.
You may of course press extreme definitions. E.g. How many changeable bits does it take to define it as a 'program'? If an I/O-controller reads a 4-bit set of flags, a 4-bit "program", behaving in one of 16 different ways, is it then a microprocessor? Depends on your understanding of what is a program.
Many advanced chips of today contain several microprocessors, sometimes arbitrarily programable, but some of them may be running a single program, read from flash memory (or even ROM). Yet, that set of electronic components is capable of running another program, if flashed in.
Are there still multi-chip microprocessors being made? There is this concept of a "chipset", doing lot of support tasks for the main CPU, especially related to I/O, but more and more of this is taken over by the main CPU chip. What is still left to the "chipset" are so advanced I/O-functions that the logic most definitely is programmable, and deserves to be called a microprocessor it its own right. Or some more specialized term, such as a GPU. These are distinct microprocessor; they are not a multi-chip single microprocessor. Maybe there still are multi-chip microprocessors around - thirty years ago, they were not uncommon.
In the old days, you saw a lot of chips that could do a single function, determined by how the components where hooked together, and no program store that can be updated or replaced to make the circuits do another job. Since the first super-simple 74-chips appeared, there has been a steady trend towards replacing dedicated circuitry with programmable, so we see more and more microprocessor crammed into a single chip, and more and more non-programmable chips being replaced by programmable ones. I tend to relate to non-programmable chips as in the same class as discrete resistors and capacitors, primarily used to adapt a more complex chip (with one or more microprocessors) to an outside world.
What was once single-function, non-programmable chips (say, a counter, shift register or clock) are so primitive by today's standards that there is no reason to make a separate chip for it. You integrate it on the microprocessor chip, treating it as a basic component, an 'advanced transistor' (or a sibling thereof).
Then, since I rarely hear anyone referring to 'microchip': Could there be a confusion with what is commonly referred to as a 'chiplet', i.e. part of the components making up a microprocessor? A number of new processors of today are made in smaller pieces, "silicon-wise", for greater flexibility. These silicon pieces are put together in a single package. If a processor is marketed in several variants, e.g. with or without a vector unit, there may be a single set of chiplets, but the vectorless chip omits the vector unit chiplet. Also, if one chiplet fails at testing (before integration on the chip), the loss is limited compared to a huge multi-billion-transistor chip failing.
Microchips are what's left after your partner gets to the chips first.
"Before entering on an understanding, I have meditated for a long time, and have foreseen what might happen. It is not genius which reveals to me suddenly, secretly, what I have to say or to do in a circumstance unexpected by other people; it is reflection, it is meditation." - Napoleon I
It's my first question in 2022, so "happy new year" to all!
I encounter a strange effect and really need an idea on how to find what is exactly happening and how to resolve the problem. Everything went perfectly fine in the exact same setup (as described in the following) before I moved house (coming from less than 1 mile away) a few months ago...
Whilst other devices use WiFi to connect to my private WLAN, one of the desktops is supposed to use maximum speed (72 Mbps download) and bandwidth via a CAT-5 LAN cable. The router is a standard model of the provider and has been exchanged already, the desktop is running Windows 10 and permanently updated. In addition, I updated all the device drivers using the "Avast Driver Updater" software.
What happens is that the Router (given it's connected via LAN) disconnects and reboots approximately every 1/2 hour. After approx. 1 minute it restarts (with the full procedure of lights blinking etc.), so that the interruption lasts around 2-3 minutes in total. For another 1/2 hour (this interval changes wildly, sometimes it is 1 hour, sometimes only 20 min or e.g. during the night even several hours) everything seems to be running normal with no indication of an error and at full speed.
I've tried using 2 different LAN devices of the desktop computer (1 PCI-Express and a built-in Gigabit-device) and also all 4 different LAN inputs of the router. No change that I would have been able to observe. Since I couldn't resolve the problem, I'm currently using a WiFi stick which works perfectly but only makes 60% the speed (approx. 47 Mbps).
Would someone please help me narrow down the error and/or give me hints on how/where I can find help on such a peculiar problem? Thank you very much in advance!
That would be possible – the (only) LAN cable is installed closely to an extension lead with 6 plugs... I'm gonna try to change that, test the function again over night and report. The idea would correlate with the fact that in the old house everything in that respect was absolutely fine.
I read from what you write that there's no chance the computer (through the LAN sockets) would send anything (probably a test signal that could be misinterpreted) causing the router to re-boot. If I can drop this suspicion, I feel at least a little bit more on a track.