AMD Introduces The Radeon Pro WX Series

[starshipeleven]

No two semiconductor foundries will produce identical dies. The fabrication process is a literal transformation of code into hardware and the process of which is determined by highly sophisticated software.

No it is not, quit this bs now. The complex stuff at a fab is needed to drive the system to print ICs at tiny scales, do quality-control and all that stuff that is needed but does not to change the design in any way.

Also printers aren't dumb, they use software to transform image files into ink on a paper and no two printers will print exactly the same because their software isn't exactly the same.

Also the hardware of the printers isn't exactly the same, nor the firmware running the controllers driving the printer's hardware is exactly the same. Software is needed to translate the design that into something the controllers driving the hardware can understand as they have to actually print it, but the goal is making a print that is as close as possible to the design, not making it better (how?).

Sure some (most) consumer printers are idiotic hell-spawned pieces of crap that flip your pages BY NON-CHANGEABLE DEFAULT and add 3cm of borders to everything (again not changeable), but that's because the firmware/drivers/softwares are a piece of shit that was coded by monkeys in the dark.

High end ones don't have that much issues, because those doing high-end printing cannot put up with that crap and pay for moar quality control.

I don't understand how you can think the transformation from design to physical hardware is 0 modifications. Obviously it's very highly sophisticated modification.

You said it (and repeated it) yourself, software is not hardware, so what applies to making software does not apply to making hardware (or only to a limited extent).

Making hardware is not like making software, there are no magical compilers that take a design that is perfectly fine and tweak it for lulz. A factory making physical objects (any physical object) makes as perfect as possible copies of the design at the lowest cost pssible, it's not the factory's work to do any kind of optimization, their task is mass-production.

A compiler is NOT a factory. You maybe think it is because you only stop at the surface because you are a goat. Yes it takes the source and spits an executable binary, but it does MUCH more and is crucial for software. Software that isn't optimized by a compiler but just compiled brutally runs like total crap. Compilers do all sorts of sanity checks and controls and whine like little bitches if they find things they don't like, and so on and so forth.

[duby229]

No it is not, quit this bs now. The complex stuff at a fab is needed to drive the system to print ICs at tiny scales, do quality-control and all that stuff that is needed but does not to change the design in any way.

Also the hardware of the printers isn't exactly the same, nor the firmware running the controllers driving the printer's hardware is exactly the same. Software is needed to translate the design that into something the controllers driving the hardware can understand as they have to actually print it, but the goal is making a print that is as close as possible to the design, not making it better (how?).

Sure some (most) consumer printers are idiotic hell-spawned pieces of crap that flip your pages BY NON-CHANGEABLE DEFAULT and add 3cm of borders to everything (again not changeable), but that's because the firmware/drivers/softwares are a piece of shit that was coded by monkeys in the dark.

High end ones don't have that much issues, because those doing high-end printing cannot put up with that crap and pay for moar quality control.

You said it (and repeated it) yourself, software is not hardware, so what applies to making software does not apply to making hardware (or only to a limited extent).

Making hardware is not like making software, there are no magical compilers that take a design that is perfectly fine and tweak it for lulz. A factory making physical objects (any physical object) makes as perfect as possible copies of the design at the lowest cost pssible, it's not the factory's work to do any kind of optimization, their task is mass-production.

A compiler is NOT a factory. You maybe think it is because you only stop at the surface because you are a goat. Yes it takes the source and spits an executable binary, but it does MUCH more and is crucial for software. Software that isn't optimized by a compiler but just compiled brutally runs like total crap. Compilers do all sorts of sanity checks and controls and whine like little bitches if they find things they don't like, and so on and so forth.

You're so wrong. I'm done trying to convince why and how you're wrong. Enjoy your bliss.

[starshipeleven]

You're so wrong. I'm done trying to convince why and how you're wrong. Enjoy your bliss.

More like you have no proof of the bs you say.

[drSeehas]

... the M.2 slots are populated by end users. I have started looking into whether there is anything we could do with AMD-branded SSDs though.

AFFAIK there are only 2.5" AMD-branded SSDs.

[drSeehas]

... I still have painful memories of entering boot loaders using front panel 1/0 switches, enough to read a cassette and bring in the code needed to interact with a teletype.

Lots of ones and zeros there

PDP-11?

[bridgman]

PDP-8 and IMSAI 8080, both on loan (PDP-8 was paper tape IIRC). Shortly after that we bought a SWTPC 6800 system and terminal then we could enter the cassette boot loader via keyboard. Did that for about 6 months before we could afford a floppy drive. I may have burned the cassete boot loader into EPROM somewhere in that 6 month period, typing in a boot loader doesn't seem like something I would do for 6 months

We did get a PDP-11 but that was maybe 5 years later. It was a PDP-11/03, so no front panel switches.

Most of the money we had was going into graphics/video processing logic - TI 9900 for geometry processing (it had 16-bit multiply and divide instructions !!) and an array of 6502s for pixel processing, plus a bunch of analog video processing feeding a 19" colour oscilloscope.

The analog bits could do interesting-for-the-70's things like quantize video from a camera, assign Z values to different slices, then rotate the results in 3D, display them on the colour oscilloscope, then capture the results into another TV camera, and the digital processing was for early VR efforts (yes, with little CRTs in the headsets).

[drSeehas]

PDP-8 ...

Oh, I assume you are older than me ...

... It was a PDP-11/03, ...

Was this a Unibus or a Q-Bus model?
AFAIR Q-Bus models allowed up to 22-bit physical addresses (= 4 MB of memory!).
Later our PDP-11 got an Ethernet card.
Although it was a 16 bit system, we worked in octal.

[drSeehas]

For those who are interested: I found a PDP-11 video:

https://www.csc.depauw.edu/~dharms/PDP11Video/Video/PDP11-640x480.mov

Beware of slooow download!

[bridgman]

Oh, I assume you are older than me.

I probably worked on computers that were older than you

I was actually really lucky, got to see a lot of different generations of technology when I was getting started. First full time job was at a life insurance company - primary system was an IBM 370/165 but the payroll software was still running on an emulated IBM 650, with the emulator running on an IBM 1401. Everyone was familiar with the big 1403 band printers, but until then I never knew there was a 1401 CPU and 1402 card reader/punch to go with the 1403 printer.

The vacuum tube 650 was still in the office but not running. (that is not me in the picture but I think I had that tie)

Was this a Unibus or a Q-Bus model?
AFAIR Q-Bus models allowed up to 22-bit physical addresses (= 4 MB of memory!).
Later our PDP-11 got an Ethernet card.
Although it was a 16 bit system, we worked in octal.

Pretty sure it was Q-bus. Starting to think it might have been an 11/23 rather than an 11/03 since I'm pretty sure it had more than 64KB of memory (128 or 256 IIRC). Probably semiconductor memory rather than core but I don't remember for sure, which is surprising because I usually took everything apart as soon as it arrived and core memory looks very different from semiconductor memory.

It was our workhorse Unix system used for software development including cross-compilation to various microprocessors.

I remember Ethernet being very exotic at the time, with big heavy cable and taps on the floor, but don't think we ever had it in house though... we were wired for RS-232 and damn proud of it. A lot of our work was high frequency analog (we designed a fair number of the consumer satellite receivers being built at the time) and I was the token computer guy.

I do remember some octal, not sure if it was on the PDP-11 though. The only thing I clearly remember is hating it... I could barely use it for some reason, maybe because I spent some of the time there building things out of AMD bit-slice logic where pretty much everything is 4-bit.