Jump to content
IGNORED

Fastmath OS?


Larry

Recommended Posts

I just checked, and the Fastmath OS gives about 15% or so faster execution time for Ahl's Benchmark than does the RamRod XE (aka OSN XL) OS. Does anyone know where the Fastmath OS came from? I know that quite a few folks have tinkered with the math pack going back to Wes Newell and Charles Marslett. Perhaps the Fastmath comes from Marslett's code?

 

-Larry

Link to comment
Share on other sites

To follow up -- I used Drac030's UBI Basic (1st version) which uses the C. Marslett fast math package, and it gives the same execution time as the Newell OS, so that means that the Fastmath OS has a different math pack in it. The actual times for Rev C Basic are

 

Fastmath OS 117 sec.

 

Newell/Marslett 144 sec.

 

XLOS 405 sec. (for comparison)

 

So Fastmath OS is actually right at 19% faster -- quite significant at least in this type benchmark.

 

-Larry

Link to comment
Share on other sites

If it is the same fastmath.rom I have here, it is probably faster just because it contains the same Marslett's FPP just with many (all?) loops unrolled. The FPP thus is faster, but occupies more ROM space: 3k instead of 2k. The additional code has been inserted instead of the CHARSET 2, do a POKE 756,204 and see what happens...

Edited by drac030
  • Like 3
Link to comment
Share on other sites

Fairly sure the original fast FP was marketed as a Newell product and available early 1980s as a 2K drop-in replacement for the 400/800 FP chip.

 

At some stage it was integrated into the Omnimon/Omniview OS - whether that OS ever came without it, unknown.

 

But both were fairly heavily advertised in the likes of Antic, Compute, Creative Computing, Byte etc.

Link to comment
Share on other sites

I had just been trying to piece together what I could about the OS replacement chip products from Computer Support in the UK. I think I can characterize most of them to a basic extent, but I really can only guess about their "Super-Chip" or "Superchip" chips for the 400/800, 600XL/800XL, or 130XE. My best guess so far is that these might have been fast math offerings, but that's really only speculation. Anyone know what they really were? They were advertised for quite a while so I think they were probably real. These ads are all I have to go on.

 

 

Computer Support Atari-User-Vol-1-No-12.pdf

Computer Support Atari-User-Vol-2-No-04.pdf

Link to comment
Share on other sites

  • 4 weeks later...

To follow up -- I used Drac030's UBI Basic (1st version) which uses the C. Marslett fast math package, and it gives the same execution time as the Newell OS, so that means that the Fastmath OS has a different math pack in it. The actual times for Rev C Basic are

 

Fastmath OS 117 sec.

 

Newell/Marslett 144 sec.

 

XLOS 405 sec. (for comparison)

 

So Fastmath OS is actually right at 19% faster -- quite significant at least in this type benchmark.

 

-Larry

Tom Hunt's Fast BASIC (FB.COM) does an Ahl's in 144 seconds. I tried MMG compiled on Ahl's and it got 360 seconds.

I see Newell/Marslett (what BASIC?)does 144 seconds. Hunt may have used the same code, or maybe they

are the same thing. I'm using Altirra.

FB.zip

MYAHL.BAS

Edited by russg
Link to comment
Share on other sites

MMG compiled in 360 seconds would be reasonable since it is using the stock floating point package. There is another whole thread on the Ahl benchmark with many timings but I don't remember MMG in the mix. And we do have a plethora of Basics!

 

Did you try MMG with another math pack? I don't know/remember if the MMG compiler f.p. code is stand-alone or still needs the math pack to execute. You don't hear much about MMG any more.

 

Larry

Link to comment
Share on other sites

Fairly sure the original fast FP was marketed as a Newell product and available early 1980s as a 2K drop-in replacement for the 400/800 FP chip.

 

At some stage it was integrated into the Omnimon/Omniview OS - whether that OS ever came without it, unknown.

 

But both were fairly heavily advertised in the likes of Antic, Compute, Creative Computing, Byte etc.

I put the fast math chip, along with a custom character set, in my custom 400/800 16K eprom that I have in my 600XL. It's Rev. A. Built in basic doesn't work, but the cart works ok, like it that way. Also, when I hit reset, it resets completely, just like turning it off and off!

 

I was hoping someone would hack the XL/XE OS so it doesn't do the power-up test, bypass if you will. Then I could make the same changes to it!

 

One day I plan to hack my XE-Game system OS and change out the built in Basic/Missile Command to something else. 2 other games, 8K, or one 16K game.

Link to comment
Share on other sites

MMG compiled in 360 seconds would be reasonable since it is using the stock floating point package. There is another whole thread on the Ahl benchmark with many timings but I don't remember MMG in the mix. And we do have a plethora of Basics!

 

Did you try MMG with another math pack? I don't know/remember if the MMG compiler f.p. code is stand-alone or still needs the math pack to execute. You don't hear much about MMG any more.

 

Larry

I did stock MMG. ABC compiler I like better, but it doesn't do floating point. I don't know where I'd get a different math pack for MMG.

Link to comment
Share on other sites

  • 1 year later...

I had just been trying to piece together what I could about the OS replacement chip products from Computer Support in the UK. I think I can characterize most of them to a basic extent, but I really can only guess about their "Super-Chip" or "Superchip" chips for the 400/800, 600XL/800XL, or 130XE. My best guess so far is that these might have been fast math offerings, but that's really only speculation. Anyone know what they really were? They were advertised for quite a while so I think they were probably real. These ads are all I have to go on.

 

 

 

The OS Image dumped in http://atariage.com/forums/topic/246862-more-questions-than-answers/contains the text "SUPER CHIP! ©1986 COMPUTER SUPPORT (UK) LTD" and is an OS hack with a built-in memory dumper.

Link to comment
Share on other sites

I put the fast math chip, along with a custom character set, in my custom 400/800 16K eprom that I have in my 600XL. It's Rev. A. Built in basic doesn't work, but the cart works ok, like it that way. Also, when I hit reset, it resets completely, just like turning it off and off!

 

I was hoping someone would hack the XL/XE OS so it doesn't do the power-up test, bypass if you will. Then I could make the same changes to it!

 

One day I plan to hack my XE-Game system OS and change out the built in Basic/Missile Command to something else. 2 other games, 8K, or one 16K game.

Try this http://atariage.com/forums/topic/229116-modification-of-xl-os-rom-triggered-memory-test/?hl=%2Bos+%2Bchecksum+%2Bbob&do=findComment&comment=3060464

 

At least a couple of other things you may want to do like make the left margin 0 and change the default color of the screen at startup so you know your mod is active. I think I ran into one or two seldom used BASIC programs that didn't allow for changes to the left margin but it wasn't critical.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...