The Future is Now!

Status
Not open for further replies.
Jump to Latest
#44
Was actually Samuel Langhorne Clemens, but why quibble.... they were so much alike.😉

Getting back on thread:

Computers.
In the mid eighties I had an Atari ST 520 that I bricklayered ram chips onto and made it a 1040 (A full Meg of RAM!). I added an Adaptec SCSI to IBM MFM drive adapter and to Lapine 10 Mb hard drives for a whopping 20 megabytes of online storage. I think the SCSI to IBM adapter was Adaptecs first product. Anyway, the resultant machine was full multitasking. A 16MHz Motorola 68000 with separate data and address buses. It ran a GEM GUI (looked and operated like a Mackintosh, but wasn't a copy. Both were originally licensed from Digital Research. Still the Atari was known as a Jackintosh after Jack Trammel). By full multitasking I mean that I could (and did) download a file over modem in the background (From Genie) while I listened to Motzart on the internal midi synth at the same time I doodled with Neopaint...... Such ability didn't reach the Micro$oft Window$ world untill at least Window$ 98.... 13 years later.

Sadly my Atari died.... likely that long CAS/RAS line running over the ram chips got zapped.... Had a unique system crash indicator. Little bomb symbols would show up on the screen. The more bombs the more severe the crash.
Brought up a NINE BOMB crash and never lived again.
Doc
 
#45
thaumaturge said:
A 16MHz Motorola 68000 with separate data and address buses.
Click to expand...

Respect🙂 All 68k family members had non-multiplexed data and address. Just the 68008 had only an 8 bit data bus so had to do two fetches for even the simplest of instructions (like NOP). 16MHz was pretty special in those days - I seem to recall the normal HMOS devices topped out at 12.5MHz, 16MHz would have been a 68HC000 in CMOS?
 
#46
68000

» 32-bit CPU
» 16-bit data bus
» Up to 20 MHz
» 16 MB RAM
» No I/O ports

64-pin DIP
68-pin LCC
68-pin PGA

Was the 68008 that had the narrow data bus. Early Motorola 68000 had a 10MHz clock, but the upped to 12.5 by release of the 68010 (had internal MMU). The 68020 had a screamin 25MHz with 32bit data/address. I thought sure my Atari ST clock was 16 MHz, but could have been slower. I was very fond of the 68000 series. Still have a number of HP 9000 workstations running up to a 68040. Have a MacSack Mackintosh ROM box to run Mac progs on my Atari, but I never got it to work right.
Doc
 
#47
thaumaturge said:
68010 (had internal MMU).
Click to expand...

OK since this is all great geeky stuff, I'm going to play my extreme anorak card here. The '010 had no internal MMU, but it was compatible with an external one, unlike the bog-standard '000. The problem with the original '000 was it wasn't guaranteed to recover from all bus faults - not enough status was stacked to re-try the operation. The '010 added extra info which meant it could fully recover if a memory access wasn't made to a valid area of memory. This meant the '010 could run Unix OS which depends on having virtual memory. Thinking back though I did have on my desk at one time a 'Torch' box (second processor for BBC micro) which contained a 68k running Unix. I just wonder how they implemented the virtual memory as I don't recall it being an '010.

I was very fond of the 68000 series.
Click to expand...

Me likewise 🙂 I've transferred some of my affections for it to the ARM Cortex M series now though.😱
 
#48
Now going back off-thread. 😀
abraxalito said:
Here's one for your crystal ball gazing skills. In which decade will a computer actively decline interaction with a human?
Click to expand...

abraxalito said:
Yes one might argue that way, hence I purposely inserted the word 'actively' to head off such interpretations at the pass (those examples you're referring to being passively).
Click to expand...
I take this to be the equivalent of the Turing Test. There's no way to know if a computer was actually declining interaction, or just "simulating: declining interaction but just appearing to do so by observers. But in practical terms, if you can't tell the difference, it doesn't make a difference.

There was some argument many decades ago between two AI researchers (or a researcher and a naysayer) in which one said in n a loud voice and in forceful terms that a computer would never convincingly show emotion. The other's response was that the first was showing a remarkably good simulation of anger.

With that in mind, I'll pick the category of "Kurzweil's human-brain-equivalent computer for $1000" and answer with "What are the 2020's?"
 
#49
The way the human brain functions is so different from a digital computer, that we would need quite some breakthroughs in the field of computer architecture before we come even close.

Don't forget there has not been any fundamental change in the basic operating structure of the freely programmable computer as we have learned to know it since it's inception.

vac
 
#50
benb said:
The other's response was that the first was showing a remarkably good simulation of anger.
Click to expand...

That hypothesis is a scientific one - meaning its testable. Just take a blood sample of the supposed 'angry' person and analyse it for the requisite neurotransmitters or neuromodulators. Real anger has a chemical fingerprint, simulated anger will not.

Same applies to your claim about the computer - it has an internal state which can be analysed to differentiate the two conditions.
 
#54
I just recall the 010 added MMU features. HP did a 9000 work station that used it. But the 020 was the CPU that first amazed me. I had a benchmark that ran about 45 seconds on a 68000 machine that returned 2.3 second on a new 020 machine... I remember saying "I don't care if computers ever get any faster!". Now I sigh in frustraition that my 1.6 GHz Dell 260 isn't quite fast enough to decode MP4's and flat latches up on MKV's.
Doc
 
#55
thaumaturge said:
I had a benchmark that ran about 45 seconds on a 68000 machine that returned 2.3 second on a new 020 machine...
Click to expand...

That's what's missing with the CPU developments of today. Long gone are the 'quantum leaps' in performance - makes me all misty-eyed.

I'm trying to work out just how you got that degree of speed up though because the clock speeds (say the '020 was 25MHz) instruction streamlining and bus width and the cache don't all add up to 20X improvement. Did it have floating point in the benchmark by any chance? If so, and your new machine carried a 68881 FPU then that's entirely understandable.
 
#57
Lets keep it going here...

Fuel Injection for internal combustion engines
Then: I think it was on my brothers 62 Vette that I first saw a fuel injected engine. I recall it being a PITA. The early Bosch systems used on the type 3 were also problematic.

Now: Non EFI cars are rare... and with good reason. I had my choice of a sweet old Toyota landcruiser or a wimpy 6cyl 4 runner for use as a mountaintop transmitter service vehicle. I sighed heavily but had to take the 4 runner... EFI was the reason. They start much more reliably in serious cold climates. The old landcruiser had a carb with a manual choke. (My best choice would have been the landcruiser with an EFI 350 engine, but that wasn't an option)
Doc
 
Status
Not open for further replies.
Top Bottom