What was the very first computer you owned or used ?

[bobcat2000]

I had one of those Sharps...   great little tool.   Had it programmed to calculate probabilities for Bridge (card game)!

I used to have a kind of Street Fighter game written in machine language for the PC-1350.  I had the game in a tape.  The tape was lost somewhere in the garage.  I think if I can covert the tape to MP3, I may still be able to load the game to the PC-1350.

[ebastler]

I used to have a kind of Street Fighter game written in machine language for the PC-1350.  I had the game in a tape.  The tape was lost somewhere in the garage.  I think if I can covert the tape to MP3, I may still be able to load the game to the PC-1350.

Better use a lossless audio format (.wav, .flac or such). MP3 is designed to avoid artefacts perceptible to human listeners, but your computer's demodulator might be offended anyway.

[Wilksey]

Acorn Electron

[bobcat2000]

I was watching someone on youtube testing the performance of his calculators.
He ran this calculation.



My PC-1350 does not have summation.  So, I run a for-loop to add up X^-2.  It takes my Sharp 2 mins and 25 seconds to finish.  Cool!
I then downloaded an app called SmallBasic to my phone.  Run the same for-loop.  My phone finishes the calculation before I can blink my eyes.  So fast compared to this old relic.

[Nusa]

I was watching someone on youtube testing the performance of his calculators.
He ran this calculation.

(Attachment Link)

My PC-1350 does not have summation.  So, I run a for-loop to add up X^-2.  It takes my Sharp 2 mins and 25 seconds to finish.  Cool!
I then downloaded an app called SmallBasic to my phone.  Run the same for-loop.  My phone finishes the calculation before I can blink my eyes.  So fast compared to this old relic.

Now see how well your phone does if you limit to the energy reserves of two CR-2032 lithium batteries, with a maximum consumption rate of 5 mA.

[brucehoult]

I was watching someone on youtube testing the performance of his calculators.
He ran this calculation.

(Attachment Link)

My PC-1350 does not have summation.  So, I run a for-loop to add up X^-2.  It takes my Sharp 2 mins and 25 seconds to finish.  Cool!
I then downloaded an app called SmallBasic to my phone.  Run the same for-loop.  My phone finishes the calculation before I can blink my eyes.  So fast compared to this old relic.

Now see how well your phone does if you limit to the energy reserves of two CR-2032 lithium batteries, with a maximum consumption rate of 5 mA.

The good old ATmega328 can run at 4 MHz at 1.8 V using about 1 mA. Or 8 MHz at 2.7 - 3.3 V is under 4 mA.

I timed summing 1.0/(i*i) for i from 1.0 to 500.0 using the AVR gcc single precision soft float library and got 112 ms at 4 MHz and 56 ms st 8 MHz.

The stm32l476 (ARM Cortex M4) datasheet says it can run at around 16 MHz to 24 MHz on under 5 mA depending on the exact operating conditions. And it's 32 bit with an FPU.

[ebastler]

The good old ATmega328 can run at 4 MHz at 1.8 V using about 1 mA. Or 8 MHz at 2.7 - 3.3 V is under 4 mA.

I timed summing 1.0/(i*i) for i from 1.0 to 500.0 using the AVR gcc single precision soft float library and got 112 ms at 4 MHz and 56 ms st 8 MHz.

Thanks for givig that a try! So technical progress is real, after all...   

Amazing how far we have come in a few decades. I will try that program on the LGP-30 at Technikum29 when that machine finally comes back to life. Be prepared for minutes of execution time and a kW of power intake...

By the way, better run that loop counting down from 500 to 1, for improved precision!

[bobcat2000]

Does the LGP-30 have enough memory to run this calculation? 

[ebastler]

Does the LGP-30 have enough memory to run this calculation? 

Sure does; 4096 words (31 bit each). The large memory was one of the big selling points of these magnetic drum memory computers -- besides the low cost and the compact, freezer-sized form factor. 

[bobcat2000]

Sure does; 4096 words (31 bit each). The large memory was one of the big selling points of these magnetic drum memory computers -- besides the low cost and the compact, freezer-sized form factor. 

I am reading from Wikipedia.  Do you load the program from the tape to the memory first?  Or the program is being run while the tape is being read like a BASIC interpreter?

So interesting.

[ebastler]

The program resides on the magnetic drum while it is executed. (And the data too, it's a von Neumann architecture.) A drum rotation takes 17 ms; with carefully optimized programs, up to 7 instructions per revolution can be executed. But that's a somewhat theoretical optimum, a couple of instructions per rev are probably more typical.

Edit: As a nice side effect of the magnetic drum architecture, all information is non-volatile, including the CPU registers (which reside on the magnetic drum too). Stop the program, switch the computer off, and restart execution in the exact same place tomorrow. "Hibernate" mode is not a 1990s Windows innovation!

The hardware supports 31-bit fixed-point arithmetic, including division and multiplication in hardware. But there are floating point libraries, and even a compiler for a basic Fortran-style language, ACT.

[bobcat2000]

Wow!  Cool!

So the whole machine is actually a CPU entirely. (kind of...)

[iMo]

DIY 8085 based, then ZX81..

[SiliconWizard]

My PC-1350 does not have summation.  So, I run a for-loop to add up X^-2.  It takes my Sharp 2 mins and 25 seconds to finish.  Cool!

Hehe, sure. Running intepreted Basic on a 768kHz CPU.

[brucehoult]

My PC-1350 does not have summation.  So, I run a for-loop to add up X^-2.  It takes my Sharp 2 mins and 25 seconds to finish.  Cool!

Hehe, sure. Running intepreted Basic on a 768kHz CPU.

BASIC is not sooo bad if the code is tokenized, the variable names are replaced with offsets into the symbol table, and the operations being performed are using a soft float library (which is going to be slow even with a native compiler).

BASIC's biggest fault is not the speed but the sheer pain of coding anything complex in it. I don't recall UCSD Pascal being any faster than AppleSoft, it was just far more pleasant to write and read.

[HobGoblyn]

Vic 20 followed soon after by C64

Then C128, next Amstrad 6128, Atari ST

Then started my never ending PC self build

[bobcat2000]

Vic 20 followed soon after by C64

Then C128, next Amstrad 6128, Atari ST

Then started my never ending PC self build

I wished I had an Atari ST or an Amigo.

I went straight from an Apple II to an IBM PC-XT clone skipping all the good stuffs in between.  Not a C64.  No IIe, IIc, IIgs or Mac.  Nothing in between.

Do you know how bad the XT was without any good sound or fancy color graphic?!!!  It was heavy and bulky.

[bobcat2000]

BASIC's biggest fault is not the speed but the sheer pain of coding anything complex in it. I don't recall UCSD Pascal being any faster than AppleSoft, it was just far more pleasant to write and read.

I like BASIC better.  I don't have to plan ahead too much when writing BASIC.  BASIC is so basic.

And I don't have to type BEGIN and END all over the places.

[SiliconWizard]

BASIC's biggest fault is not the speed but the sheer pain of coding anything complex in it. I don't recall UCSD Pascal being any faster than AppleSoft, it was just far more pleasant to write and read.

I don't know about UCSD Pascal on a 6502, never used that, but I did use Turbo Pascal back in the days on a Z80 CP/M system, and it sure did beat the crap out of any BASIC interpreter's pants in terms of speed.

As to the language - no contest here for sure.

[Tepe]

I don't know about UCSD Pascal on a 6502, never used that, but I did use Turbo Pascal back in the days on a Z80 CP/M system, and it sure did beat the crap out of any BASIC interpreter's pants in terms of speed.

Almost the same experience here but with Compas Pascal and later Poly Pascal on a Z80 CP/M system. They are both in the lineage of Turbo Pascal.

As for the 6502 it doesn't seem a great fit for Pascal like languages.

[SilverSolder]

I don't know about UCSD Pascal on a 6502, never used that, but I did use Turbo Pascal back in the days on a Z80 CP/M system, and it sure did beat the crap out of any BASIC interpreter's pants in terms of speed.

Almost the same experience here but with Compas Pascal and later Poly Pascal on a Z80 CP/M system. They are both in the lineage of Turbo Pascal.

As for the 6502 it doesn't seem a great fit for Pascal like languages.

UCSD P-system Pascal on Apple ][  ?

[SiliconWizard]

I don't know about UCSD Pascal on a 6502, never used that, but I did use Turbo Pascal back in the days on a Z80 CP/M system, and it sure did beat the crap out of any BASIC interpreter's pants in terms of speed.

Almost the same experience here but with Compas Pascal and later Poly Pascal on a Z80 CP/M system. They are both in the lineage of Turbo Pascal.

I actually still have the TP 3.0 manual in my library. I opened it recently just to take a look. Ah, memories! The amount of features you got with this (for a compiler that IIRC took about 30KB) was mind-boggling!

[brucehoult]

BASIC's biggest fault is not the speed but the sheer pain of coding anything complex in it. I don't recall UCSD Pascal being any faster than AppleSoft, it was just far more pleasant to write and read.

I don't know about UCSD Pascal on a 6502, never used that,

It compiled to a stack-oriented bytecode very very similar to the JVM 15 years later, which was then run by a simple interpreter. It needed a system with two floppy disk drives and the full 64 KB or RAM (48k plus 16k "language card" that overlaid the ROM). The environment had a graphical file manager, a reasonably decent full screen editor, compiler and assembler, linker. You could write external functions in asm. Performance was relatively usable, with changing between modes requiring a mass overlay swap to floppy which sounded about the same as and took the same time as racking a pump action shotgun twice.

I believe the bytecode was portable between different systems, and was not a very good fit to the 6502.

but I did use Turbo Pascal back in the days on a Z80 CP/M system, and it sure did beat the crap out of any BASIC interpreter's pants in terms of speed.

I used Turbo Pascal 1.0 on Z80 and 8088 systems in 1984. The generated code was abysmal but it generated it very very quickly, and anything compiled is massively better than anything interpreted...

[brucehoult]

I don't know about UCSD Pascal on a 6502, never used that, but I did use Turbo Pascal back in the days on a Z80 CP/M system, and it sure did beat the crap out of any BASIC interpreter's pants in terms of speed.

Almost the same experience here but with Compas Pascal and later Poly Pascal on a Z80 CP/M system. They are both in the lineage of Turbo Pascal.

As for the 6502 it doesn't seem a great fit for Pascal like languages.

It's not, but neither is the Z80.

The Z80 lures you into thinking it might have enough registers to be useful -- and it even has 16 bit registers! But it really doesn't, and you can't do much with them as 16 bit registers. It's enough for hand-written memcpy() or strcmp() etc, but not a lot more. And once you admit that you can't do much in registers, the 6502's 256 bytes of "zero page" pseudo-registers are just sooo much more useful. If you treat zero-page the same way registers are treated in a register-rich RISC CPU (or x86_64), with a similar ABI with argument/caller save registers, callee-save registers, temporary registers, special registers (e.g. a 16 bit stack pointer), and a whole bunch of permanently-allocated registers (i.e. frequently used globals) then you can generate pretty decent code for it -- and easily.

[KV6O]

TRS80 Model I, Level II was the first owned.  Used Wang 700 (dad worked for Wang), PDP 11 and PET's in high school.  Atari 800 with 88K floppy's was next at home, and added a 300baud acoustic modem - online in the early 80s!