Atari 8 Bit I/O Performance vs C64

[Chilly Willy]

You can probably reach the higher speeds if you remove some of the capacitors on your SIO port (they are sometimes referred to as "vampire" caps.) There is also a minor SIO mod you can do where you add a resistor.

 

http://sio2sd.gucio.pl/wiki/HighSpeed_en

 

The article claims 127K is possible.

I know I can, I just don't want to open up my 65XE and hack it.

 

 

The Apple II tape interface is 1500 baud, not the disk drive.

 

The book "Beneath Apple ProDOS" says the Apple II disk can transfer data at 8K per second.

The source I was quoting from has it wrong... I thought that looked a bit slow! Thanks for the correction.

[JamesD]

A Coco tape drive normally ran at 1500 baud. It could be sped-up (slightly), but at the risk of data reliability.

All it required was a few POKEs to set the machine into high speed mode and to adjust the tape timing values.

You had to use good quality tapes but it was reliable. It wasn't quite 3000 baud but it was close.

[Stefan Both]

Dear guys,

The toppic comes a little late?!

 

In late 80's, it was a common theme in

Bars. In germany, you could run into

deep trouble if you praise the wrong computer

to the wrong guy.

Fact is, there was never ever a slower drive

than the 1541. In stock config, it was a pain in the ass.

Of course, there where hacks / speeder / fastloader etc.

You could even mod the drive for parallel data transfer.

 

So what?

 

We all loved "our" computer, didnt we?

Compared with the later models, every computer

was slow.

But didnt we enjoyed all those wonderful games?

 

I mean - no matter of the loading time ?

 

Stefan

[Chilly Willy]

So what?

 

We all loved "our" computer, didnt we?

Compared with the later models, every computer

was slow.

But didnt we enjoyed all those wonderful games?

 

I mean - no matter of the loading time ?

 

Stefan

Yep! I played Necromancer off cassette tape on my A400 and loved every minute of it!

[kenjennings]

I had Hellcat Ace and Pogo Joe on cassette. Hated every minute of waiting for them to load.

[Shawn Jefferson]

I remember the C64 drives were SLOOOOOOW, and waiting for a game to load so you play Bruce Lee with your buddy was horrible. And not everyone had Fast Load carts until later on... when my Atari had a Happy drive already (but no new games unfortunately... ). Many other Atari folks had hard drives and were using SpartaDOS when the C64 guys were finally loading their games at a decent speed.

 

 

Still, even though I didn't like things about the C64, I was envious of the game support, especially later on. Why couldn't we get Bard's Tale, or U5? Sometimes I was envious of the port quality. Racing Destruction Set is one I remember being very disappointed with. The Atari version just looks so crappy compared to the C64, and that was a game we had lots of fun playing on the C64.

Edited January 24, 2015 by Shawn Jefferson

[+Larry]

Surprised that no one has mentioned Jiffy-Dos for the C64 & 1541. Jiffy-Dos (JD) was a game-changer IMO. It was very compatible and as fast as any fast loader that I can remember. In addition to speed, it gave the user much more useful commands for drive operations than the standard 1541 command set. Jim Brain (Retro Innovations) licensed the rights to JD when CMD went out of business. My observation (since I never actually timed the C64 vs Atari I/O) was that JD made the I/O pretty comparable to a 3X UltraSpeed drive on an Atari. It was/is amazingly faster than a stock 1541. It would be interesting to run something like RWTEST on both systems to really get some hard numbers. (I still have a bunch of C64 stuff here, but haven't had it out in years.)

 

-Larry

[Rybags]

I didn't mention it as it requires modification of both the drive and computer (new OS) - really, when you take things to that extent it becomes a bit extravegant.

 

The Atari equivalent would be like doing another OS, extra PIA and do parallel IO to the 1050 with custom firmware.

Actually... it'd probably be a pretty cool hack but sort of irrelevant now that we have nicer IDE types and SIO2xx types that do turbo speeds by default.

Edited January 25, 2015 by Rybags

[+Larry]

I didn't mention it as it requires modification of both the drive and computer (new OS) - really, when you take things to that extent it becomes a bit extravegant.

 

The Atari equivalent would be like doing another OS, extra PIA and do parallel IO to the 1050 with custom firmware.

Actually... it'd probably be a pretty cool hack but sort of irrelevant now that we have nicer IDE types and SIO2xx types that do turbo speeds by default.

 

No, not really. JD is just a 24 or 28-pin rom. One for the 64/128 and one for the 1541 (or other C= drive). It has a switch so that you can use the stock system if you run into compatibility issues (rare). These are about $20 apiece, but IIRC, they make use of unused connected serial lines to increase speed. It isn't parallel, but it isn't strictly serial, either. I think they referred to it as "multi-line serial."

 

I've asked several times over the years why no one ever did a multi-line SIO for our A8. (We've done about everything else!) But I guess the PBI took it's place.

 

Here is a good description from Jim Brain's site (pdf manuals are at the bottom of the page):

 

http://www.go4retro.com/products/jiffydos/

 

-Larry

[Stefan Both]

Found it in C64 Wiki:

 

1 Block is 256 Bytes.

I also forgot that kind of hardware. It's a good choice..

The JiffyDos made the timing between C64 and the 1541 as fast as

possible.

However, you had to open both parts, Computer AND Floppy.

But it was totally easy. Just remove the regarding Eprom and replace it.

So you have finally the information you need.

How much faster is the 1050?

 

Stefan

Edited January 25, 2015 by Stefan Both

[Lazarus]

The JiffyDos made the timing between C64 and the 1541 as fast as

possible.

However, you had to open both parts, Computer AND Floppy.

But it was totally easy. Just remove the regarding Eprom and replace it.

 

JiffyDOS was a pretty slow "fast"loader. There are several which are much faster (15 times faster than CBM loader for example) and are software only, no hardware modification required. A good C64+1541 software fastloader easily outperforms the 1050.

Edited January 25, 2015 by Lazarus

[kogden]

 

JiffyDOS was a pretty slow "fast"loader. There are several which are much faster (15 times faster than CBM loader for example) and are software only, no hardware modification required. A good C64+1541 software fastloader easily outperforms the 1050.

 

The Happy mod for the 1050 wasn't astoundingly expensive and was 3 times faster than a stock 1050. Out of the box, both were pretty slow but I'll take the Atari's more flexible I/O and better OS over the C64 any day.

 

Overall, I'd say the Atari was probably more capable overall.

 

For example I have a Happy 1050, an SIO2PC and an 850 (4xRS232, 1xparallel) sharing the same SIO bus. All of it works with standard software configs. SIO is more like a 1979 attempt at USB. My main 800XL rig with a PBI IDE interface and the SIO pile o' gear works with just about everything from a software standpoint. This would be harder to pull off with a C64 I'm pretty sure. Not saying impossible, though.

[Kyle22]

Well, kogden, you are right. SIO is the forerunner to USB. The same man made them both.

 

I just can't get past the Commie BASIC Load "*",8 stuff.

 

Why should we need BASIC enabled to load a disk (and need to type a cryptic command)?

 

I am Asperger's, and I see things in binary In other words, Atari Rulez, Commie Sucks. NO gray area.

 

I am Atari8 to the core (so please don't reply to this about how you like your Commie)...

 

(I did NOT mean you, personally).

Edited January 26, 2015 by Kyle22

[Muzz73]

As far as the compatability issues with JiffyDOS being rare, I'm not sure I'd agree. I don't use it, myself, mainly because our user group's C=128 uses it and almost nothing works with it at all. When someone goes to fire up a program to demo, the machine stops part way through the loading process and we have to power the 128 down, switch off the JiffyDOS ROM, power it back up and begin the SSSSLLLLOOOOOWWWW booting process all over again. I'm not kidding, this is like 4/5 programs.

 

I'm not just saying this because I prefer the A8, and I have never had anything but stock drives on any of my systems, but I have never seen anything on the C=64 or C=128 that was anywhere near the speed of a stock 1050 or a stock Apple Disk ][. If I had, I'd be on it like a hobo on a hot dog!

[Chilly Willy]

I had a Percom DD disk drive. It was awesome. I RE'd the BIOS in it, and they support the speed change command in the BIOS! Unfortunately, they didn't support it in the hardware. I got the schematics from Percom, and started looking into a mod to allow higher speeds, but right at that time, I finally moved up to the Amiga (500) and it got pushed to the side.

[Kyle22]

Chilly, did you mean that you found a way to speed up the Percom?

 

I want to learn as much as possible about all A8 drives...

 

My goal is an UltraSpeed ATR-8000, and a Happy compatible Indus GT.

 

 

NEED AS MUCH INFORMATION AS POSSIBLE......

[+Larry]

Found it in C64 Wiki:

 

1 Block is 256 Bytes.

I also forgot that kind of hardware. It's a good choice..

The JiffyDos made the timing between C64 and the 1541 as fast as

possible.

However, you had to open both parts, Computer AND Floppy.

But it was totally easy. Just remove the regarding Eprom and replace it.

So you have finally the information you need.

How much faster is the 1050?

 

Stefan

 

Hi Stefan-

 

I'm going to "crunch" some timing numbers, but it seems a bit difficult to know what to compare to get the best apples-to-apples comparisons. For instance, apples-to-apples would say that if you are going to compare the 1541, it should be compared to the 810 (not a 1050). Because these were both the first-generation drives from each. If we're going to use the 1050, then it should be compared to the 1571 -- the second generation drives. The 1571 is a bit faster than the 1541, even when used with a C64, just as the 1050 is a very slight bit faster than an 810 (track-to-track and seek of the mechs.)

 

Some questions for you or others that are more familiar with C= hardware than me:

 

1) Is the default from a C= drive "write with verify?" I remember that there was an article in Creative Computing (I think) that showed that the 810 under Dos 2.0s was actually slower writing than a 1541 when the 810 was used under write with verify.

 

2) It really makes a difference what file type you are reading or writing on the C=. Generally not so much on the Atari (other than the verify issue). (IIRC) if you were reading/writing random files on an Atari, then that is dead-slow.

 

Any other suggestions for the comparisons?

 

-Larry

[Rybags]

1571 generally performs best when used with the 128. I'd not segregate these drives into generations.

Truth of the matter is 1050 is more like half a generation in front of 810.

 

Fairly sure the 1541 doesn't do write+verify. The 1541 as we know can do some turbo modes in software alone but also has the advantage over stock Atari drives of having more Ram which means more than 1 sector could be buffered.

 

Doing random r/w on any floppy drive will make speed suffer. Even CD and DVD has the same trait, the seek time on all these devices is a killer.

[+Stephen]

Don't forget the one major advantage and ultra cool thing about the Commodore drives is that they can have code uploaded to them. You can even use the drive as a co-processor for the computer.

[Muzz73]

Don't forget the one major advantage and ultra cool thing about the Commodore drives is that they can have code uploaded to them. You can even use the drive as a co-processor for the computer.

 

That was really cool! I remember there being BBS parties back in the 80's where a bunch of Commodore people would bring in 1541s, daisy chain four of them off of a single C=64 and there was a simple copy program that could be loaded into the 64's RAM and removed from the drive. You would then put your source disk in drive 8, blanks in 9-11, wait for the copying to start, then you could unplug the 64 and go do something else with it, watching the drives copy to and from one another without having a computer attached.

 

Being that I was like 14 at the time and only had the datasette drive, I thought it was pretty neat.

Edited January 26, 2015 by Muzz73

[+Larry]

Don't forget the one major advantage and ultra cool thing about the Commodore drives is that they can have code uploaded to them. You can even use the drive as a co-processor for the computer.

 

Absolutely. Being programmable is a big plus under specific circumstances (and if you know how to do it!).

 

Now for some numbers...

 

Atari 1050, 1X SIO

Wr. 1038 B/sec.

Rd. 1042 B/Sec.

 

Happy Buffered Rd. 1X SIO

Wr. 1039 B/Sec. (771 with verify)

Rd. 1386 B/Sec.

 

Atari 1050, 3X SIO

Wr. 1687 B/Sec.

Rd. 2576 B/Sec.

 

C= SEQ. PRG.

Wr. 316 341

Rd. 380 417

 

JD.

Wr. 948 1066

Rd. 2133 4309

 

Now, some qualifications. I used a 1050 Happy for all the Atari tests. Slow mode for 1X SIO and under US Emulation for the 3X. If someone wants to check with a 1050 Doubler drive, that would be good, although I doubt that the numbers would change much.

 

The Atari numbers are from RWTEST.COM (V3.7). All the C= numbers are from the posted JiffyDos comparison table.

 

-Larry

[fujidude]

 

I'm going to "crunch" some timing numbers, but it seems a bit difficult to know what to compare to get the best apples-to-apples comparisons. For instance, apples-to-apples would say that if you are going to compare the 1541, it should be compared to the 810 (not a 1050). Because these were both the first-generation drives from each.

 

I would say apples to apples means comparing when the product was available and in common use. The 810 was already superceded or pretty close to it by the 1050 before C= introduced the 64 and 1541.

Edited January 26, 2015 by fujidude

[bbking67]

 

I would say apples to apples means comparing when the product was available and in common use. The 810 was already superceded or pretty close to it by the 1050 before C= introduced the 64 and 1541.

 

Well... the C64 was released in August 1982 when the 810 was in full production. From my recollection the 1050 was not released for some time. The 1200XL came and went before the 1050 was introduced in October 1983 alongside the 600XL/800XL--and when the 1050 was released it was in very short supply initially. So there was at least a one year gap between the C64/1541 and the 1050.

[fujidude]

 

Well... the C64 was released in August 1982 when the 810 was in full production. From my recollection the 1050 was not released for some time. The 1200XL came and went before the 1050 was introduced in October 1983 alongside the 600XL/800XL--and when the 1050 was released it was in very short supply initially. So there was at least a one year gap between the C64/1541 and the 1050.

According to Wikipedia, the early 1541s were white (and said VIC 1541) and in very short supply. The beige ones did not get introduced until 1983 and produced in larger numbers until 1984 with the Mitsumi mechs. So if you still want to compare the 1541 to the 810, you need to do so with the early white ones, which by the way had about a 50% failure rate.

 

The more familiar beige 1541s with the improved mechs came out AFTER the 1050.

[sack-c0s]

To be fair I think the BBC wiped the floor with the Atari and Commodore so far as loading speeds were concerned.

 

Although to be fair the C64 and Atari 8-bit had stuff that was actually worth the wait, so it worked out nice in the end....