Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - willrandship

Pages: 1 ... 5 6 [7] 8 9 ... 208
91
TI Z80 / Re: Wat *.*
« on: April 18, 2014, 12:17:54 am »
if you can make it only use safe RAM, no variables ever, it should be interrupt-safe. Then beg thepenguin to hook it for you :P

92
Humour and Jokes / Re: Funny #omnimaga quotes (NSFW)
« on: April 17, 2014, 11:57:03 pm »
willrandship>   how do you pull that? (I know magic)
willrandship>   oh man that needs to be quoted out of context

Is it impolite to post my own?

93
TI-Nspire / Re: nGL - a fast (enough) 3D engine for the nspire
« on: April 17, 2014, 10:12:20 pm »
Having the standard library would help ndless development immensely. More power to you.

94
News / Re: Imagine if the 84+C was 6 MHz (color 83+ coming soon?)
« on: April 16, 2014, 09:50:19 pm »
Well, I was probably wrong then. Wouldn't be the first time :P

For the record, ARM implementations are cheaper than Z80, especially at low speeds, and they take less power. The nspire's hardware does cost a fair amount more than the 84+, but keep in mind it runs code well over 200 times faster. (The Z80 has very poor cycle density when compared to ARM. I don't even have to take word size into account.) It also has significantly more RAM. 32 MB vs 32 KB addressable at any time.

Let me make a quick cost sheet.

MCU: $4.14 (Cortex-M3 w/ built-in 44 MHz oscillator, 32 KB RAM, 256 KB Flash, and it can execute from RAM, not to mention you can write for it in C without worrying as much about code bloat)

ROM: $0.52 for 4 MB of Flash in an SPI chip (the internal flash is small and inconvenient to write to, and we can just copy data from this to the RAM for execution, leaving the internal flash for the OS) An SD card slot would be a completely reasonable addition as well, since it is basically the same style of device. (SD cards can use SPI as a communication protocol, although it's slower than their normal protocol)

Since the ARM MCU implements internal timer interrupt mechanisms and its clock is about 3 times faster than the z80, there's no real need to put in the external crystal timers present on the 83+SE and above.

I'm not going to bother comparing things that would be almost exactly the same, like the keypad (which can just hook to 16 I/O pins) and the screen (the same screen would work fine, but other options would be cheaper and better)

Now, for the existing Z80 systems:

CPU: A SMD 20 MHz-capable CPU costs $10.12, and as it does not include an internal oscillator, it would need an external crystal (just a few cents, really) This is the cheapest 15 MHz capable Z80 I could find on my distribution site. (you don't ebay production parts)

RAM: Since the Z80 does not have internal RAM, you need to get parallel synchronous SRAM. A suitable 128 KB chip costs $0.94

ROM: Once again we need a parallel synchronous interface, meaning more expensive parts. $0.99 for a suitable 2 MB Flash ROM

So, assuming the architecture is otherwise similar, we come to $12.05 for the Z80 without including most of the unnecessary peripherals, vs $4.66 for a much faster, simpler ARM-based solution. Granted, that's no nspire, which would be significantly better, but even without considering the time saved on the math (Cortex M3 has hardware multiply and divide, handles 4x the math at once) the ARM version would vastly outperform the Z80.

Since TI clearly doesn't care about assembly compatibility, the best way I would see for them to use this style of design would be to rewrite the 83+ BASIC interpreter for ARM, such that programs are cross compatible. Then the vast majority of their users' programs would still function. To an extent they did this on the platforms they have. They're not 100% compatible, but if you know 84+ BASIC you won't be very lost in the 89, except perhaps by the new interface.

95
News / Re: Imagine if the 84+C was 6 MHz (color 83+ coming soon?)
« on: April 16, 2014, 06:17:55 pm »
I would bet it's for Canada more than France. No offense to France, but Canada is close to the US, a fair amount bigger, and depends heavily on US merchandise, especially in technology. (Wherever it doesn't come from China, anyway :P) Plus, Quebec has, from what I hear, a bunch of social rules about using French. That would make demand for French equipment (on the label, not just as a setting) higher demand than in France, which handles international language issues more, being a part of Europe.

Just a thought.

96
Humour and Jokes / Re: Is your son a computer hacker?
« on: April 13, 2014, 08:14:58 pm »
That article gave me an aneurysm.

97
TI Z80 / Re: PolyAOI v6?
« on: April 13, 2014, 07:14:48 pm »
FYI you can use ALL the TI-Basic floating point operations for normal OS floats through assembly with bcalls. You cannot do the same with axe. Its floats are less precise.

If you wrote this in Asm with bcalls, the actual math would probably be slower, but the results would be higher precision and any non-float logic would be a little bit faster. (Not tons faster because axe is really good at optimization for what it is) If you're familiar enough with assembly, don't ignore it as a potential platform for this project.

Although, looks like you've already started. Oh well. :P

98
News / Re: Imagine if the 84+C was 6 MHz (color 83+ coming soon?)
« on: April 13, 2014, 07:03:17 pm »
They already have a fully functional z80 emulator for ARM. Why aren't they using it to make new calculator designs, if they want to continue the 84+ series?

99
TI Z80 / Re: Wat *.*
« on: April 12, 2014, 12:54:39 am »
If you could make an interrupt-safe version of this, you could potentially hook it to the APD hook, then have a timer in it that runs its own power down.

100
Computer Programming / Re: MS-DOS source is officially released
« on: April 12, 2014, 12:40:11 am »
FYI all x86 CPUs support the 16 bit mode. Nowadays the common way to implement it is through hardware emulation (read: it looks like normal operation but it's slower than 32 or 64 bit modes would be) by using microcode instructions. They basically have a CPU inside the CPU that executes bigger instructions as a series of smaller instructions. As such, they take several cycles to complete, but since modern CPUs are several hundred times faster than a 33 MHz machine, it shouldn't be an issue.

101
Art / Re: Mockups "please say this is going to be a game"
« on: April 03, 2014, 07:20:15 pm »
I think the first would be better especially in a smooth scrolling game, since the bricks would blur the shop logos into quite a mess.

102
Forum Arcade Games / Re: Final Omni Arcade highscores (top 3)
« on: April 03, 2014, 01:04:47 am »
The games on it were all pretty terrible, especially when compared to today's flash games. The whole point was the high score system :P

There were 2-3 versions of frogger. My favorite was the one that ended at level 3 so you would max out your score. That meant the top scorer was always the last person who won, and all three scores were the same.

103
When you're using AC you don't need to have a complete circuit ;)

104
ASM / Re: Best App Compiler
« on: March 26, 2014, 10:21:51 pm »
The assemblers only differ as far as simple syntax, and even then not by much. Making apps is extremely similar to normal programs, with just a few extra steps.

I recommend Brass. It's a more modern assembler and it's cross-platform.

For making apps, as I said, there are a few extra steps. This means compiling to hex code in the right format, giving it the right header information, as well as signing the resulting output so calculators will accept it as a valid application. Brass has directives for all of these, so it's not much more complicated than making normal assembly programs.

105
Miscellaneous / Re: "What If" Discussion
« on: March 26, 2014, 10:12:25 pm »
Then Bing would be rich and Google would go the way of AOL

What if TI started selling calcs at-cost?

Pages: 1 ... 5 6 [7] 8 9 ... 208