much easier to design the processor with the security holes built in, then block them for secure customers.
That would of course require collusion between a processor manufacturer and a very major government.
Or use a small processor and some added front-end circuitry and encapsulate it to look like (and emulate) a bigger processor. It would run slower, but who actually subjects processors to speed tests?
Relief postman did similar to my mother last week as something had to signed for, M " I was in yesterday you Know" PM " I knocked ,maybe you didn't hear me" M" why didn't you ring the bell? PM "What bell" ? M "This one" at which point she clanged the ex fire engine bell which I gave her about 40 years ago* and was hanging on a bracket in the porch about 2ft from the posties ear.
*In those premobile days it was a solution to bring dad in from the fields for important phone calls, visitors etc. In the right atmospheric conditions a good clang can be heard about 2 miles away.
Well, of course, the above is all true, but it's also true that having a broader computing experience is useful. At SLAC I mostly did software in C, but I got brownie points because I was able to debug a small Z80 board that had been designed in the Operations Group, but where the engineer concerned hadn't checked the board layout that came back before sending it off to be built. So some small knowledge of how to use an HP logic state analyser with a Z80 probe came in useful.
Well the TNT bloke who delivered from Staples yesterday wasn't working on that basis. We were out for 30 mins when he first tried and left a card. Rang the CS number rescheduled delivery for next day. Then got a call from the driver are we in? He's just finishing his drops and will be passing, we were in and got our delivery less than 24hrs after placeing the order. Full marks...
Mind you he might have looked at the weather forecast and didn't want to be out here in the afternoon. Hartside and Killhope are closed due to snow, Whitfield Moor is probably not far behind, it'll certainly be interesting with blowing and drifting snow. That just leaves the "low" route to Brampton...
That wouldn't surprise me.
Our local Royal Mail Delivery Office is just 2.5 miles away and we can park for free. B-) I've not really looked where the various courier depots are but I suspect Carlisle, 30+ miles away. I'm not doing a 60 mile roud trip, they can damn well redeliver.
My point was that even when one is comfortable[1] working at a "low level" like that, you are not going to get he opportunity to re-engineer the microcode in your i3 processor or similar unless you happen to work for Intel in their CPU design department or similar...
[1] I find it quite quaint when people seem amazed that one can actually do useful things in machine language these days - its almost like you are speaking fluent Latin as far as many are concerned.
On Friday 25 January 2013 20:52 John Rumm wrote in uk.d-i-y:
That depends on the machine langauge too :)
I hated 6502 (though some liked it). I could stomach Z80 but it was not what I called fun.
VAX MACRO32 was indeed a lot of fun, as was 68000. I don't mind Atmel 8 bit AVRs either, though they program so well in C, there's rarely much need.
I think it is because I like highly orthogonal machines with loads of registers. Therefore I would hate RISC.
VAX IIRC has some string instruction whose cycle count was a multiple of the string length, and thus could take 100's of cycles to complete.
And yes - I often heard the system programmers talking of DEC sending out microcode patches.
I was deeply awed when the explained that the VAX 8600 was incapable of even doing it's first stage boot until the little MicroVAX (IIRC) next to it had loaded it's microcode into the CPU - which seemed to be on every reboot - or maybe ot was on a power cycle.
Not at all... its now quite common for "processors" to simply be dropped into FPGAs along with whatever other custom logic is required in the design. Most of the FPGA development tools have standard library modules available to simply program whatever processor core into the device you need. So it would not be difficult to create a device with a modified processor.
Having said that its also not uncommon for so called standard designs to not produce standard results in these fields. My colleague was doing some debug work on a FPGA based video switching / encoding system the other week, and concluded that it was basically not working at all. Alerted the hardware designer, who was convinced that it should be fine since it (the VHDL design) was proven in previous generations of kit. Then he realised that the actual FPGA itself was a different model in this equipment (something that should not make any difference). Closer inspection revealed that the compiler had not managed to generate quite the same functionality in the new device as it did in the previous device from the same code. End result was the VHDL was passed to the "guru of VHDL black magic fiddling", to "adjust" it so that it was the same but different, in the hope of convincing the compiler to actually generate what was asked for!
True, but in principle they are much of a muchness, at least at some level.
Agree on both.
If you like highly orthogonal machines with lots (well, 16) of registers, look at the DECSystem-10 (aka PDP-10). Very interesting - complete set of conditional jumps, for example, including one that always jumped and one that never jumped.
MOVSB could take a very large count and could use thoudands if not millions. At least it was interruptable and resumable.
I think the original VAX-11/780 was like that too - with a PDP-11/03 and an 8 inch floppy.
Think it was a power cycle (ours was an 8800, just the dual CPU version). I did like the command POWER OFF on the PDP-11 console.
Knowing what a woodscrew is, is not going to help me make my own! (Which won't matter unless society is totally destroyed.)
I once knew in detail how to make a simple multiplier using discrete components - separate transistors, diodes, resistors, etc. - though I don't think I ever used the knowledge except to track down problems.
By now I've probably forgotten more than I realise.
I've certainly forgotten how the sort-of-microcode used in a PDP9 functioned. It could be crudely described as being stored in a kind of read-only core memory, but that's about all I remember.
That sounds like a polynomial-calculating instruction. Which would be OK if it was capable of being interrupted by higher oriority events, then continued from where it left off.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.