Page 4 of 5 [ 69 posts ]  Go to page Previous  1, 2, 3, 4, 5  Next

Fogman
Veteran
Veteran

User avatar

Joined: 19 Jun 2005
Age: 57
Gender: Male
Posts: 3,986
Location: Frå Nord Dakota til Vermont

15 Aug 2016, 7:34 pm

The Motorola 68000 was the chip used in all of the standalone arcade game machines in the heyday of video games, from 1979 to about 1983.


_________________
When There's No There to get to, I'm so There!


marcb0t
Veteran
Veteran

User avatar

Joined: 9 Apr 2015
Gender: Male
Posts: 589
Location: Washington

15 Aug 2016, 8:11 pm

Fogman wrote:
The Motorola 68000 was the chip used in all of the standalone arcade game machines in the heyday of video games, from 1979 to about 1983.

Incorrect. It wasn't until 1982 with Atari's Food Fight, that they started using 68K chips in arcade games. And it was rare until later in the 80's and 90's. In 1985, games like Hang On, and Space Harrier were making good use of the Motorola 68K.

After Burner 2 and Outrun used a dual configuration. Galaxy Force and Power Drift used 3 Motorola 68K chips.

Street Fighter 2 was using that chip in the early 90's. SNK arcade games using then even into the late 90's.


_________________
The cutest most lovable little rob0t on Earth! (^.^)


mr_bigmouth_502
Veteran
Veteran

User avatar

Joined: 12 Dec 2013
Age: 31
Gender: Non-binary
Posts: 7,028
Location: Alberta, Canada

15 Aug 2016, 10:25 pm

Fogman wrote:
The Motorola 68000 was the chip used in all of the standalone arcade game machines in the heyday of video games, from 1979 to about 1983.

The Z80 and 6502 were far more popular during that time frame, and the 6809 was pretty popular too. In fact, aside from Food Fight, I can't think of any other particular games from that period that used the 68000.

This timeline seems pretty accurate:
http://triosdevelopers.com/jason.eckert ... eline.html

Also, here's a good article on the 68000 itself:
http://segaretro.org/Motorola_68000


_________________
Every day is exactly the same...


eric76
Veteran
Veteran

User avatar

Joined: 31 Aug 2012
Gender: Male
Posts: 10,660
Location: In the heart of the dust bowl

18 Aug 2016, 2:49 am

Aaendi wrote:
I don't understand the hype around this chip. It's basically a glorified 8086, but with more registers, and 32-bit instructions. The data bus and ALU are still 16-bit, and it still has the slow 4 cycle bus access. Most of the registers are wasted on getting around the slow data bus, and the jump between 8-bit and 16-bit instructions is a much larger performance leap than going from 16-bit to 32-bit.


The machine languages (and consequently the assembly languages) of the 68000 and the 8086 are fundamentally quite different.

I'd much rather program on a 68000 with its general purpose registers than an 80x86.



mr_bigmouth_502
Veteran
Veteran

User avatar

Joined: 12 Dec 2013
Age: 31
Gender: Non-binary
Posts: 7,028
Location: Alberta, Canada

18 Aug 2016, 4:45 am

eric76 wrote:
Aaendi wrote:
I don't understand the hype around this chip. It's basically a glorified 8086, but with more registers, and 32-bit instructions. The data bus and ALU are still 16-bit, and it still has the slow 4 cycle bus access. Most of the registers are wasted on getting around the slow data bus, and the jump between 8-bit and 16-bit instructions is a much larger performance leap than going from 16-bit to 32-bit.


The machine languages (and consequently the assembly languages) of the 68000 and the 8086 are fundamentally quite different.

I'd much rather program on a 68000 with its general purpose registers than an 80x86.

What is your favorite CPU architecture to do low-level programming on?


_________________
Every day is exactly the same...


mr_bigmouth_502
Veteran
Veteran

User avatar

Joined: 12 Dec 2013
Age: 31
Gender: Non-binary
Posts: 7,028
Location: Alberta, Canada

12 Sep 2016, 6:48 am

bump


_________________
Every day is exactly the same...


Aaendi
Deinonychus
Deinonychus

User avatar

Joined: 31 Aug 2013
Gender: Male
Posts: 363

23 Sep 2016, 12:36 pm

Another thing that really bothers me is when programmers think moving sprites slower saves CPU speed. It really makes no difference in CPU speed.

Scrolling the background faster does take away a little CPU time because it has to update the background more, but it is such a tiny amount that it's totally worth it. Also, the SNES does have a 16x16 tile mode that can update backgrounds 4 times faster than the more common 8x8 tiles mode.



BaalChatzaf
Veteran
Veteran

User avatar

Joined: 11 Mar 2008
Gender: Male
Posts: 1,050
Location: Monroe Twp. NJ

24 Sep 2016, 11:56 am

eric76 wrote:
Aaendi wrote:
I don't understand the hype around this chip. It's basically a glorified 8086, but with more registers, and 32-bit instructions. The data bus and ALU are still 16-bit, and it still has the slow 4 cycle bus access. Most of the registers are wasted on getting around the slow data bus, and the jump between 8-bit and 16-bit instructions is a much larger performance leap than going from 16-bit to 32-bit.


The machine languages (and consequently the assembly languages) of the 68000 and the 8086 are fundamentally quite different.

I'd much rather program on a 68000 with its general purpose registers than an 80x86.


I agree. The 68000 was a coder's computer. I also like the DEC assembly language and the IBM 360/370 assembly language


_________________
Socrates' Last Words: I drank what!! !?????


mr_bigmouth_502
Veteran
Veteran

User avatar

Joined: 12 Dec 2013
Age: 31
Gender: Non-binary
Posts: 7,028
Location: Alberta, Canada

24 Sep 2016, 1:24 pm

Aaendi wrote:
Another thing that really bothers me is when programmers think moving sprites slower saves CPU speed. It really makes no difference in CPU speed.

Scrolling the background faster does take away a little CPU time because it has to update the background more, but it is such a tiny amount that it's totally worth it. Also, the SNES does have a 16x16 tile mode that can update backgrounds 4 times faster than the more common 8x8 tiles mode.

Why are backgrounds more intensive to manipulate than sprites?


_________________
Every day is exactly the same...


Meistersinger
Veteran
Veteran

User avatar

Joined: 10 May 2012
Gender: Male
Posts: 3,700
Location: Beautiful(?) West Manchester Township PA

24 Sep 2016, 5:41 pm

IIRC, A good many developers, especially if they were developing for any series of VAX from DEC, found the the 68000 was very similar to some of the processors used in VAXen.



Aaendi
Deinonychus
Deinonychus

User avatar

Joined: 31 Aug 2013
Gender: Male
Posts: 363

24 Sep 2016, 6:01 pm

Hardware backgrounds are only 2 screens long, and they repeat when scrolled too far, so it has to change tiles offscreens so it doesn't keep repeating itself. Far backgrounds typically repeat themselves so no updating is needed.

It only takes 64 cycles to fill a collumn or row of 32 tiles with DMA, so it isn't a big problem.

Sprites can be placed anywhere onscreen at any time. They don't have to be incremented 1 pixel at a time.



Aaendi
Deinonychus
Deinonychus

User avatar

Joined: 31 Aug 2013
Gender: Male
Posts: 363

16 Dec 2016, 7:52 pm

One feature of the 65816 SNES programmers seriously misused has to be the direct page.

The direct page is a small area of memory that is 256 bytes, and is slightly faster than the rest of the memory. What most people did was use it as the 68000's register set, where (non-direct page) memory had to temporarily be stored to "speed" up the CPU.

What people didn't realize is that, accessing a word from non-DP memory takes 5 cycles, accessing DP memory takes 4 cycles, and moving words from one area to the other takes 9 cycles, and going back and forth takes 18. You completely defeat the purpose of using the direct page.

You can use direct page memory as registers, just as long as you're not using more instructions than you otherwise would.

If you really want to move something into the direct page, you can move the direct page instead with the direct page address register.



Aaendi
Deinonychus
Deinonychus

User avatar

Joined: 31 Aug 2013
Gender: Male
Posts: 363

18 Jan 2017, 11:01 am

mr_bigmouth_502 wrote:
Ichinin wrote:
Old thread, i know, but OP should google "CISC vs RISC".

It's funny you bring that up, because it has been argued that the 6502 is an early example of a RISC CPU. The 65816 is basically just a 16-bit 6502 with a few additional instructions.

I honestly think the SNES would've had a good edge over the Genesis if Nintendo had clocked the SNES' CPU higher and found a way to access the cart memory at full speed from the get-go. Running at the same clock speed, the 65816 should be able to outperform the 68000 for sure. It already trades blows with the Genesis 68000 at the speed it runs at.


The funny thing is that it was actually possible due to design flaw in the 65816 where the bus was only active for half a cycle. Nintendo fixed it with the SA-1, but the stock SNES had an almost plain vanilla 65816.



mr_bigmouth_502
Veteran
Veteran

User avatar

Joined: 12 Dec 2013
Age: 31
Gender: Non-binary
Posts: 7,028
Location: Alberta, Canada

19 Jan 2017, 4:33 am

Aaendi wrote:
mr_bigmouth_502 wrote:
Ichinin wrote:
Old thread, i know, but OP should google "CISC vs RISC".

It's funny you bring that up, because it has been argued that the 6502 is an early example of a RISC CPU. The 65816 is basically just a 16-bit 6502 with a few additional instructions.

I honestly think the SNES would've had a good edge over the Genesis if Nintendo had clocked the SNES' CPU higher and found a way to access the cart memory at full speed from the get-go. Running at the same clock speed, the 65816 should be able to outperform the 68000 for sure. It already trades blows with the Genesis 68000 at the speed it runs at.


The funny thing is that it was actually possible due to design flaw in the 65816 where the bus was only active for half a cycle. Nintendo fixed it with the SA-1, but the stock SNES had an almost plain vanilla 65816.

Wait, what would've been possible if Nintendo exploited that design flaw? I'm confused.


_________________
Every day is exactly the same...


Aaendi
Deinonychus
Deinonychus

User avatar

Joined: 31 Aug 2013
Gender: Male
Posts: 363

20 Jan 2017, 12:32 am

For some reason, the original 65816 and the one used in the SNES accesses memory on the second half of each cycle.



mr_bigmouth_502
Veteran
Veteran

User avatar

Joined: 12 Dec 2013
Age: 31
Gender: Non-binary
Posts: 7,028
Location: Alberta, Canada

20 Jan 2017, 10:38 am

Aaendi wrote:
For some reason, the original 65816 and the one used in the SNES accesses memory on the second half of each cycle.

I'm guessing that this levies a speed penalty?


_________________
Every day is exactly the same...