65C816 vs 68000

For discussing the 65xx hardware itself or electronics projects.
Post Reply
User avatar
cbmeeks
Posts: 1254
Joined: 17 Aug 2005
Location: Soddy-Daisy, TN USA
Contact:

Re: 65C816 vs 68000

Post by cbmeeks »

GARTHWILSON wrote:
True—but what was the price of those? It was getting into the thousand-dollar-processor era.
Yeah, I remember those days when I had my 68K Amiga and then was able to afford a 68020. But the 68030/40/60 were way out of my reach. By the time the 68030 started getting more affordable the market moved away from it. Once Apple dropped it and moved to PPC, it was doomed. I mean, you still had Commodore using them and I think Cisco/3M/etc. may have used them in routers and such...but their days were numbered.

GARTHWILSON wrote:
rather than wasting resources to get the next software product to market, just because "memory/GHz/whatever is cheap."
You know, that's still a very bad problem. I admit, I'm guilty a little. I don't mind wasting hard drive space because it's so cheap. But I cringe when I work with other developers who think 20-30 SECONDS is OK on a "should be simple" request from a server. When I know in my heart it could be done in <100 ms.
Cat; the other white meat.
Tor
Posts: 597
Joined: 10 Apr 2011
Location: Norway/Japan

Re: 65C816 vs 68000

Post by Tor »

My work PC has a million times more RAM (16GB) than the very first IBM PC (16KB), so why does it feel like I only have 16KB then.. sigh.
User avatar
BigDumbDinosaur
Posts: 9425
Joined: 28 May 2009
Location: Midwestern USA (JB Pritzker’s dystopia)
Contact:

Re: 65C816 vs 68000

Post by BigDumbDinosaur »

Tor wrote:
My work PC has a million times more RAM (16GB) than the very first IBM PC (16KB), so why does it feel like I only have 16KB then.. sigh.
I'm sure it has nothing to do with the operating system... :D
x86?  We ain't got no x86.  We don't NEED no stinking x86!
DerTrueForce
Posts: 483
Joined: 04 Jun 2016
Location: Australia

Re: 65C816 vs 68000

Post by DerTrueForce »

Oh, it has a lot to do with that, if you're on Windows. Win7 uses about 1.5-2GB of RAM when idle, IIRC.
Linux(Ubuntu MATE) uses only 600 MB, but it doesn't feel much more responsive to me. The difference is there; it's just not huge.
I haven't tried Kolibri yet, but I suspect that the difference will be immense. It's supposed to be really fast.
User avatar
Arlet
Posts: 2353
Joined: 16 Nov 2010
Location: Gouda, The Netherlands
Contact:

Re: 65C816 vs 68000

Post by Arlet »

I don't know about Windows, but Linux will use RAM for disk buffers when not using it for anything else.
User avatar
BigDumbDinosaur
Posts: 9425
Joined: 28 May 2009
Location: Midwestern USA (JB Pritzker’s dystopia)
Contact:

Re: 65C816 vs 68000

Post by BigDumbDinosaur »

DerTrueForce wrote:
Oh, it has a lot to do with that, if you're on Windows. Win7 uses about 1.5-2GB of RAM when idle, IIRC.
Linux(Ubuntu MATE) uses only 600 MB, but it doesn't feel much more responsive to me. The difference is there; it's just not huge.
I haven't tried Kolibri yet, but I suspect that the difference will be immense. It's supposed to be really fast.
That Microsoft operating systems are resource hogs is well-known. The one good thing that has come out of it is that modern computer hardware is now so powerful. I recall working with minis in the 1970s in which each user had 8KB maximum in which to work, and a disk access was an exercise in patience.

Speaking of Linux, we use SLES here.
x86?  We ain't got no x86.  We don't NEED no stinking x86!
User avatar
BigDumbDinosaur
Posts: 9425
Joined: 28 May 2009
Location: Midwestern USA (JB Pritzker’s dystopia)
Contact:

Re: 65C816 vs 68000

Post by BigDumbDinosaur »

Arlet wrote:
I don't know about Windows, but Linux will use RAM for disk buffers when not using it for anything else.
...which is an occasionally annoying feature. A common kernel tweak is to set a fixed size for buffers so the disk isn't thrashed if one or more applications suddenly demand a lot of RAM.
x86?  We ain't got no x86.  We don't NEED no stinking x86!
User avatar
ttlworks
Posts: 1464
Joined: 09 Nov 2012
Contact:

Re: 65C816 vs 68000

Post by ttlworks »

"640K ought to be enough for anybody."
Of course, Bill Gates denies to have said something like that back in 1981. :)

But seriously: when building a 65816 PC, please try to have 16MB RAM.
User avatar
BigDumbDinosaur
Posts: 9425
Joined: 28 May 2009
Location: Midwestern USA (JB Pritzker’s dystopia)
Contact:

Re: 65C816 vs 68000

Post by BigDumbDinosaur »

ttlworks wrote:
But seriously: when building a 65816 PC, please try to have 16MB RAM.
Unless you do it in DRAM, the aggregate cost of the SRAMs will be more than the cost of the rest of the system. :D You could do it with four of Garth's memory modules.
x86?  We ain't got no x86.  We don't NEED no stinking x86!
User avatar
GARTHWILSON
Forum Moderator
Posts: 8773
Joined: 30 Aug 2002
Location: Southern California
Contact:

Re: 65C816 vs 68000

Post by GARTHWILSON »

BigDumbDinosaur wrote:
ttlworks wrote:
But seriously: when building a 65816 PC, please try to have 16MB RAM.
Unless you do it in DRAM, the aggregate cost of the SRAMs will be more than the cost of the rest of the system. :D You could do it with four of Garth's memory modules.
There's a 2-megabyte SRAM IC but it's 3.3V (won't do 5V) and only comes in a BGA last I checked. This is all fine for production but not always for hobbyist builds. If someone wants to give me an order big enough to justify the set-up cost of automated assembly :D, I could reduce the price of the modules quite a bit, and even use the 2MB BGAs. As it is, my last volume purchase of 10ns 5V 512Kx8 SRAM ICs came out to less than a third as much as my first purchase almost five years ago (and they're genuine Cypress parts, from Cypress distributor in the U.S.), so the current price for fully-populated modules is $69.

With 90° pins, so the module goes perpendicular to the motherboard:
Image

with straight pins, so the module goes parallel to the motherboard:
Image
except if you order this way, you'll get gold-plated pins, not tin-plated. You'll get this one:
StraightPinHeader.jpg
StraightPinHeader.jpg (49.44 KiB) Viewed 1144 times
Here's the bare board, to show the size:
WM-1PCBinHand.jpg
WM-1PCBinHand.jpg (29.15 KiB) Viewed 1144 times
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?
User avatar
ttlworks
Posts: 1464
Joined: 09 Nov 2012
Contact:

Re: 65C816 vs 68000

Post by ttlworks »

Hmm... now this brings up an interesting question:

When the 65816 would be available in TQFP...
would it be possible to make a module with a similar form factor
and a somewhat similar pinout like the memory module containing:
  • 65816
  • reset circuitry
  • CPLD (address decoder, maybe SPI, maybe video signal generation, maybe a "background debugger" too)
  • ROM
  • UART
Altera MAXV: 5M570 looks interesting, 1kB of integrated Flash might do for a bootloader.
But this brings up another interesting question: 5V or 3.3V power supply ?
User avatar
GARTHWILSON
Forum Moderator
Posts: 8773
Joined: 30 Aug 2002
Location: Southern California
Contact:

Re: 65C816 vs 68000

Post by GARTHWILSON »

I'm working on a couple of related ideas. The '816 is indeed available in PQFP now. I just got three of them a couple of weeks ago. One of the ideas is to make a 65802 module but possibly add memory that is available to the programmer although not addressing it offboard (since the module goes into a 6502 socket), and also possibly adding some I/O by way of separate connectors on the module. Since I expect the demand will be too low to justify the set-up cost for automated assembly, it would be assembled by hand which will be very labor-intensive. I'm not looking for any kind of corner on the market. If someone else wants to do it, please do! Another thing I'm very slowly working on is an '816 board with Jeff Laughton's ultrafast bit I/O. I'll be doing it in 5V. After his circuits are published, I hope someone will implement them in programmable logic.
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?
User avatar
KC9UDX
Posts: 246
Joined: 07 Dec 2013
Location: The Kettle Moraine

Re: 65C816 vs 68000

Post by KC9UDX »

Maybe I missed it, I haven't seen it posted here, but this is the link to the QFP 65816 at Mouser:

http://www.mouser.com/ProductDetail/Wes ... ttV3O4k%3d

They are not in stock but can be backordered (really a good idea to do this, I think).
User avatar
BigEd
Posts: 11463
Joined: 11 Dec 2008
Location: England
Contact:

Re: 65C816 vs 68000

Post by BigEd »

Hmm, I'm seeing non-stocked as opposed to out-of-stock, so I'd read that as the distributor not taking any risk on volume and the supplier possibly even manufacturing to order. (It could be that WDC has stock of tested die and is not packaging until they have volume. There are various places in the production of chips that you can stall the process and wait.)
User avatar
ttlworks
Posts: 1464
Joined: 09 Nov 2012
Contact:

Re: 65C816 vs 68000

Post by ttlworks »

It's a pity, that the W65C265 microcontroller (65816 core) hasn't evolved to nowadays standards.

8kB ROM and 576 Bytes RAM ain't much compared to a PIC32 with 2MB Flash and 512kB RAM.

Since WDC seems to have sold quite a few licences for 65816 cores,
there are microcontrollers more powerful than the W65C265 supposed to be... somewhere...
but I can't remember to ever have seen something like that.

Hmm... I just wonder, what size a 65816 computer might have when bonding all the chips on a ceramic substrate. ;)
But machines for bonding wires on chips probably won't be cheap, and bonding wires to chips also requires some experience...
to prevent the wires from falling off the chip after half a year or such.

Edit: Spotted something neat at Singer Elektronik for 2320,50€. Have fun...
assemblyautoab510.jpg
Post Reply