ICD Multi I/O

[+Larry]

Speaking of functionality -- a very popular feature of the MIO was the 256KB/1MB ram for ramdisk(s), etc. But according to the docs, the KMK-JZ interface will control two IDE devices --Master and Slave. Has anyone set one up with a regular drive as the Master and a compactflash as the Slave? This would pretty well simulate the MIO interface + ram feature, especially since the CF is non-volatile.

Has anyone timed the compact flash with the KMK-JZ interface to compare against the MIO ramdisk? -Larry

[Almost Rice]

The price reflected rarity rather than functionality.

964894[/snapback]

 

You dont have to tell me. I was saving for one in the late 80s and they stopped making it. I did not want to settle for the 256k model. So, I have not bought one. I was hoping to bid on it but there was already 3 motivated bidders. It probably got a few bucks more for spartados X

[drac030]

Has anyone set one up with a regular drive as the Master and a compactflash as the Slave? This would pretty well simulate the MIO interface + ram feature, especially since the CF is non-volatile. Has anyone timed the compact flash with the KMK-JZ interface to compare against the MIO ramdisk?

 

Yes and yes. I know people who have such a setup. The CF card performs similarly to an ordinary hard disk drive - except that the access to random sectors is much faster (naturally, since the CF does not have real heads and doesn't need to move them using a motor).

[+Larry]

> Yes and yes. I know people who have such a setup. The CF card performs similarly to an ordinary hard disk drive - except that the access to random sectors is much faster (naturally, since the CF does not have real heads and doesn't need to move them using a motor). <

 

We are getting enough benchmark numbers to make a nice technical blurb -- perhaps good food for the FAQ's. Now that "Atari-time" is here for me (Winter, almost), I'll try to get these numbers and some additional ones organized. -Larry

[+Larry]

>Here are the results for the Black Box (3-1/2" Conner SCSI drive)

Dos Writing 7110.5967

Dos Reading 20373.8859

Dos Average 13742.2413 <

 

Here are some updated figures using my KMK-JZ interface:

 

Toshiba HD, MyDos 4.50

Dos Writing 8585.50

Dos Reading 26040.79

Dos Ave. 17313.15

 

Sandisk CF

Dos Writing 9273.96

Dos Reading 26040.79

Dos Ave. 17043.73

 

And for comparison, the MyDos Ramdisk

Dos Writing 47953.17

Dos Reading 44683.64

Dos Ave. 46318.40

 

Does anyone have figures for the original MIO using an MFM HD?

[+Larry]

I just made an interesting discovery. I wondered about the difference between my KMK-JZ interface RWTEST speeds and those posted by Drac030. The only thing that seemed logical was that the benchmark figures varied quite a bit from the Dos overhead for RW. Sure enough, I re-ran the tests on the hard drive using my SDX cart with quite different results:

Dos Writing 11331.87

Dos Reading 38560.59

Dos Ave. 24941.23

Long story short, apples to apples comparisons will need to be SDX or MyDos specific. I suspect (but don't know) that the file opening and closing is a major component of the difference.

-Larry

[drac030]

Long story short, apples to apples comparisons will need to be SDX or MyDos specific.  I suspect (but don't know) that the file opening and closing is a major component of the difference.

 

No, the opening and closing the file does not count. The program works as this:

 

1) open the file for write

2) get clock state

3) write, write, write, write

4) get clock state

5) close

6) calculate & display the result

 

And the same for reading.

 

I can only wonder, why the figures are so different. Unless the time measurement fails somehow (it is based on the VBL clock, 49.86 ticks/s in PAL and 59.92 in NTSC), it seems that SpartaDOS X is really this faster than MyDOS.

[+Larry]

> No, the opening and closing the file does not count. ... <

 

OK. I certainly would not have expected SDX to be that much faster than MyDos, but there it is. BTW, I did find the MIO HD benchmark figures.

 

I re-tested the Black Box using SDX, and as expected, the benchmarks improved quite a bit, although still not as fast as with KMK-JZ.

Dos Writing: 8436.12

Dos Reading 26568.65

Dos Ave. 17503.38

-Larry

[drac030]

I re-tested the Black Box using SDX, and as expected, the benchmarks improved quite a bit, although still not as fast as with KMK-JZ.

 

Yes, this is, I believe, where the IDE driver efficiency steps in

[+Larry]

I decided to re-run the Black Box RWTEST using my IDE laptop drive with the IDE/SCSI converter. For whatever reasons, it came out better than with my previous Conner 3-1/2" drive. New values:

Dos writing 11045

Dos reading 20587

Dos average 15816

This was with MyDos 4.50 -- same as original test. I repeated this several times, and the numbers were quite consistent. The previous numbers for comparison were

Dos Writing 7110.5967

Dos Reading 20373.8859

Dos Average 13742.2413

So the writing improved by more about 55%. That's quite an improvement -- any thoughts as to why?

-Larry

[MEtalGuy66]

Im building brand new MIOs. I plan to have them in production within the month. The 2 changes Im making are 1) PC standard serial & paralell ports instead of ATARI 850 standard ones. 2) Instead of all the impossible to find SIPP rams, there will be a single 72pin SIMM socket.

 

Im am shooting for a target price of $150.00.

 

If anyone knows of a source of 6551 UART chips, that would be helpful, otherwise I may have to change the design slightly to use more commonly available Intel 8250 or 16550 UART chips.

 

I am producing the board only. You have to get your own case, however I do guarantee it will exactly fit an original MIO case (with the execption of the paralell port which is now a DB25 instead of a DB15.)

[+Larry]

> Im building brand new MIOs... <

 

Very interesting. Still 256-byte sectors, SCSI?

-Larry

[atariportal]

@SpaceInvaders: Do you allow me to use the pictures for german description at ABBUC?

[MEtalGuy66]

> Im building brand new MIOs... <

 

Very interesting. Still 256-byte sectors, SCSI?

-Larry

 

Yeah. Its gonna be functionally identicle to the original. Im not going to bother making a 256k version. They will all be 1meg. Im also "beefing up" the regulator/rectifier stage of the board. It seems that a common failure on the original involves frying the diodes and regulator so bad that it often trashed the board (melted/delaminated the area).

 

Fact is, it's still one of the best supported interfaces, after all these years. It doesnt have any custom logic, (unless you count the now obsolete 6551, which used to be a fairly common "off the shelf" part.) So Im just taking the original design and updating it a bit. I'll probably make an initial run of 10 units, and make more, based on demand.

[Almost Rice]

II am producing the board only. You have to get your own case, however I do guarantee it will exactly fit an original MIO case (with the execption of the paralell port which is now a DB25 instead of a DB15.)

 

 

Who has bought the inventory at the demise of ICD/FTe? Didnt they have extra MIO cases?

[Almost Rice]

Yeah. Its gonna be functionally identicle to the original. Im not going to bother making a 256k version. They will all be 1meg. Im also "beefing up" the regulator/rectifier stage of the board. It seems that a common failure on the original involves frying the diodes and regulator so bad that it often trashed the board (melted/delaminated the area).

 

Fact is, it's still one of the best supported interfaces, after all these years. It doesnt have any custom logic, (unless you count the now obsolete 6551, which used to be a fairly common "off the shelf" part.) So Im just taking the original design and updating it a bit. I'll probably make an initial run of 10 units, and make more, based on demand.

 

 

When can I come by and see these beauties. And please count me in for 1

Edited April 6, 2006 by Almost Rice

[MEtalGuy66]

I should have the boards in a couple of weeks. I have to resolve the UART availability issue before I can have them manufactured. If I cant find a source of 6551s I need to change the design to accomadate a different part.

[kenfused]

I should have the boards in a couple of weeks. I have to resolve the UART availability issue before I can have them manufactured. If I cant find a source of 6551s I need to change the design to accomadate a different part.

You might check out bgmicro for the 6551. I used to buy a lot of stuff from them long ago. Don't know what quanties they would have since there 65xx and 68xx chips available seems to be decreasing as time goes on.

[danwinslow]

I am buying at least 2.

 

 

Then he has to make me a PBI thingy to map that IO from the Aetherboard!

[+warerat]

Im building brand new MIOs. I plan to have them in production within the month. The 2 changes Im making are 1) PC standard serial & paralell ports instead of ATARI 850 standard ones. 2) Instead of all the impossible to find SIPP rams, there will be a single 72pin SIMM socket.

 

Im am shooting for a target price of $150.00.

 

If anyone knows of a source of 6551 UART chips, that would be helpful, otherwise I may have to change the design slightly to use more commonly available Intel 8250 or 16550 UART chips.

 

I am producing the board only. You have to get your own case, however I do guarantee it will exactly fit an original MIO case (with the execption of the paralell port which is now a DB25 instead of a DB15.)

 

The board as in just the bare PCB, or the board as in PCB and fully populated?

[MEtalGuy66]

Fully populated, and tested functional.

[Almost Rice]

Fully populated, and tested functional.

 

In that case I'll take 2

[MEtalGuy66]

Fully populated, and tested functional.

 

In that case I'll take 2

 

Ok. I'll run them down the street to you, as soon as they're done. hahah.

 

But seriously, I'm estimating another week on the board layout. I'll make an initial run of 10 boards. When I get one asembled & tested, I'll post a new topic here in the forums, and start taking pre-orders. As soon as I get the money for the first 10, I'll order all the components and assemble/test the rest of them and ship them out.

 

Then from that point on, it just depends how many more people want one.

 

Im gonna be dead honest. This is a SH*TLOAD of work. I'm probably going to be making about $40.00 over parts/materials cost per unit, and we are talking about somewhere in the neighborhood of 500 solder joints per unit (a good bit of assembly time). I have devoted every spare moment of free time to this for the past 2 weeks, and Ive got at least a week to go just on the board layout, Im estimating. All of the money I get above the cost of materials for the first run will be spent on parts for the second run. I'm not going to see any compensation for my labor until Im well into the second production run.

 

I do this COMPLELTELY for the benefit of the ATARI community, and in the end, maybe we'll at least have a bit larger user base for the MIO.

 

When I was a kid, I wanted one of these so bad, I'd have given my left nut for one. Now maybe a few more of us can have one, at a reasonable price.

[Goochman]

Metalguy - though Im not on the market for one I appreciate your efforts on the 8bit no matter what.

 

Ideas like these can turn into other great things for our little workhorses

[+warerat]

Fully populated, and tested functional.

 

Put me down for one. Just need to find a XE adaptor now.