Live data from Hacker News

Home made GPU escalated quickly [video]

youtube.com

51–60 of 71 posts

Re: Home made GPU escalated quickly [video]

#51

Earlier quoted context omitted.

One could argue that it is 100% as every Watt that enters through the socket (not counting reactive energy) is going to become heat at some point in the very near future, including the air moved by fans, the photons emitted by the screen, etc etc.

So what happens to the 0.01%? Ah thats.. the light? The sound? Not sure.

My bet would be some lucky photons that end up heading to outer space from some window. Hahaha

Re: Home made GPU escalated quickly [video]

#52
post #14

This is great. It somewhat reminded me of Steve Ciarcia's build of a Mandelbrot-generating supercomputer from around 1990. That was also made from microcontrollers (Intel 8052 in that case).

The computer was animating a Julia set in one of the clips.

Re: Home made GPU escalated quickly [video]

#55

Earlier quoted context omitted.

My fav job was an internship with a EW company that made these massive rackmount radar systems. My boss, the lead engineer, loved blue LED status lights. So the massive boards we made had hundreds on them. There were thousands total in the rack. GPIOs, busses, fpga signal outputs, all got an individual LED and mosfet. It was crazy when it was running. You could visually debug a bus, if you wanted to.

My dad programmed minicomputers, which normally had LEDs on the front panel that showed the binary contents of the program counter, and maybe the current instruction and some other status indicators. I remember him saying that he could tell if his program was running normally by the patterns in the lights.

There were some telephone engineers I used to know who worked in telephone exchanges in the electromechanical era, and they could tell when they walked in if things were not working correctly by the sound...

Re: Home made GPU escalated quickly [video]

#56

Earlier quoted context omitted.

My fav job was an internship with a EW company that made these massive rackmount radar systems. My boss, the lead engineer, loved blue LED status lights. So the massive boards we made had hundreds on them. There were thousands total in the rack. GPIOs, busses, fpga signal outputs, all got an individual LED and mosfet. It was crazy when it was running. You could visually debug a bus, if you wanted to.

My dad programmed minicomputers, which normally had LEDs on the front panel that showed the binary contents of the program counter, and maybe the current instruction and some other status indicators. I remember him saying that he could tell if his program was running normally by the patterns in the lights.

If the system was used to control something it was obvious.

There would be some kind of main / idle loop. When this was running, you could see the pattern in the data buss and in the address buss. The loop of code would not be many instructions, they would all be together, so the pattern on the address buss would be quickly repeating.

If you had a slow peripheral, you could see the ISR getting called. It would be at a different address than the main loop, and you could see it flash periodically, like every time a line went to the printer.

As an operator of the system, you had a unique perspective of watching the program run as the system operated; if it was data processing it would not mean much but in a control system it was uncanny.

Re: Home made GPU escalated quickly [video]

#57
post #53
post #27

I love the absolute insanity of this design. And the Thinking Machines vibe. Computers need more blinking lights.

best we can do is Rainbow RGB

There are a number of things we can do that expose system status via a LED panel. Network traffic, IP x IP, IP x incoming port, core status, L1, L2, L3 cache faults per PID or core, and so on.

I would never use blue LEDs though.

Re: Home made GPU escalated quickly [video]

#58
post #43

Earlier quoted context omitted.

I would say it's funny but really it's just true. Thunderbolt cables and a lot of other high bandwidth cables have chips in either end, which some people don't realize. And yes, these chips have to handle IO in the gigabit range. So these microcontrollers are a bit apples to oranges (just like my WCH570 vs 486 comparison, to be clear) but even with all those caveats, yes it's basically true. And yeah, that's wild.

Not an area I'm familiar with, but do the chips really handle up in the gigabit IO? I thought they just handle connection setup. I had thought the gigabytes were not processed at all by the chips. They just flow through the wires. ?

"Cable" is loosely defined, but a cable with usb-c on one end and anything else (eg HDMI) on the other has to do some processing of the signal in order to do that.

Re: Home made GPU escalated quickly [video]

#59

Earlier quoted context omitted.

My dad programmed minicomputers, which normally had LEDs on the front panel that showed the binary contents of the program counter, and maybe the current instruction and some other status indicators. I remember him saying that he could tell if his program was running normally by the patterns in the lights.

There were some telephone engineers I used to know who worked in telephone exchanges in the electromechanical era, and they could tell when they walked in if things were not working correctly by the sound...

When I got my first Apple II+ (a Brazilian clone) I didn't have money for a floppy drive, so I used cassettes extensively (I learned to... erm... backup games to tape). By the sound of the tape you could identify code and bitmap segments, code being more random and data being more "rhythmic". Pattern recognition is a wonderful ability.

Re: Home made GPU escalated quickly [video]

#60
post #27

I love the absolute insanity of this design. And the Thinking Machines vibe. Computers need more blinking lights.

My fav job was an internship with a EW company that made these massive rackmount radar systems. My boss, the lead engineer, loved blue LED status lights. So the massive boards we made had hundreds on them. There were thousands total in the rack. GPIOs, busses, fpga signal outputs, all got an individual LED and mosfet. It was crazy when it was running. You could visually debug a bus, if you wanted to.

The Centre for Computing History, in Cambridge, UK, has a computer made from discrete components that occupies a wall. It can be downclocked to a point where you can see the code making its way through the machine, fetching data, pushing it through the ALU, and on to either the register file or back into the main memory.
Post reply on HN