68010 In A500+

[k4lmp]

I just found a pretty good deal on some 68010 CPUs, so I ordered a few. I am wanting to replace the 68000 in my A500+ with the 68010. Any problems, or considerations I should know about. I've read it is a direct replacement for the 68K, with the downside being some games not playing. I realize that I am not overclocking it, and it will run at the same speed as the 68K, but it is supposed to make the machine a bit faster. Any input before I make the mod? Thanks.

[s2325]

Maybe look here http://eab.abime.net/showthread.php?t=42122

[mombasajoe]

I think it depends on the system. I´d never place a 68010 into a A500 without harddisk. Increase of speed? Forget it. SysInfo says: about 7%. There are also some problems with some games. The problems can be solved by a software patch (decigel). So the patch hast to be loaded first (no problem with a harddisk).


E. g. I´ve got a 500+ with 2 MB chipmem, 8 MB fastmem and a harddisk as a whdload-machine. Here the 68010 makes much sense because it contains an opcode that is needed for "quit to workbench" from many whdload-installs.

[Adropac2]

I recall when I had an 68010 for my A500, the increase while there seemed barely noticable which was surprising at the time. Makes me wonder what the difference is that people see when they notice the 'faster' Atari ST version of 3d games

[robinsonb5]

Originally Posted by Adropac2

Makes me wonder what the difference is that people see they notice the 'faster' Atari ST version of 3d games

The placebo effect, maybe?

[k4lmp]

Well, I got 3 of them on the way. It should be a straight swap from what I've read. I'll let you know what SysInfo says when I get it changed.

[Adropac2]

Originally Posted by robinsonb5

The placebo effect, maybe?

Yeah pretty much.Loved my ST but I sure didn't see any noticable change to how these games ran

There is the slightest boost however with the 68010 and I was eventually pleased with it.Unless I imagined it too

[hooverphonique]

ironically, the 'extra speed' of the 010 is often what breaks old games/demos, since it has a feature for speeding up small loops, causing delay loops to be to fast compared to a vanilla 68k.

[b0lt-thrower]

I remember when I got a 14mhz AdSpeed...God I was in heaven Rendering in Vista flew! Games...broke, spectacularly!

Last Amiga I had was a 28mhz 030 A1200 with 4mb RAM.

[k4lmp]

I got my 68010's, and installed one in my A500+. It does seem a bit faster. Not blazing fast, but a bit. Using SysInfo 4, it shows a speed of 1.25. I don't remember what it showed with the 68000.

[mombasajoe]

Originally Posted by k4lmp

Using SysInfo 4, it shows a speed of 1.25. I don't remember what it showed with the 68000.

With or without 8 MB FastRam? If you had turned the GVP-RAM (seen in your sig) on you should try SysInfo again without FastRam.

[8bitbubsy]

Originally Posted by robinsonb5

The placebo effect, maybe?

The Atari ST ran at 8MHz while the Amiga 500/600/1000/2000 ran at 7.14~MHz. The Atari ST's CPU was clocked around 12.67% faster. This helped a tiny bit with 3D stuff, although the Amiga owned the Atari ST in the other factors.

[Hewitson]

In a number of ways the ST was inferior to the Commodore 64. It may have been clocked faster, but it didn't have a fraction of the capabilities. It was an absolutely terrible design.

[k4lmp]

Originally Posted by mombasajoe

With or without 8 MB FastRam? If you had turned the GVP-RAM (seen in your sig) on you should try SysInfo again without FastRam.

That was with the 8 mb of fast ram, and the extra 512k in the trap door.

[Adropac2]

Originally Posted by 8bitbubsy

The Atari ST ran at 8MHz while the Amiga 500/600/1000/2000 ran at 7.14~MHz. The Atari ST's CPU was clocked around 12.67% faster. This helped a tiny bit with 3D stuff, although the Amiga owned the Atari ST in the other factors.

Sure but it really doesn't show is more the point, at least going from one machine to the other. Never actually done a side to side though which would be interesting to see From my experience though it's not something you notice

[Geijer]

Originally Posted by k4lmp

That was with the 8 mb of fast ram, and the extra 512k in the trap door.

Then it is the presence of Fast Ram that makes most of the difference when compared to a A600 in SysInfo.

[mc6809e]

Originally Posted by 8bitbubsy

The Atari ST ran at 8MHz while the Amiga 500/600/1000/2000 ran at 7.14~MHz. The Atari ST's CPU was clocked around 12.67% faster. This helped a tiny bit with 3D stuff, although the Amiga owned the Atari ST in the other factors.

The key to making the Amiga competitive (or better) than the ST in 3D is to carefully time any blits so that they overlap with all the MULs and DIVs that you find in a lot of 3D code. Typically, though, programmers just replaced any software line-drawing or frame buffer clearing code with blits while the CPU just waited. Terrible waste of the Amiga's capabilities.

But even for typical codes the clock rate difference doesn't have as big an effect as the numbers suggest. The Atari ST blocks the CPU on every other memory access cycle, even during the blanking intervals and overscan areas. The Amiga will actually allow the CPU to get at these cycles if they're available. It doesn't happen very often, but sometimes the CPU can take advantage of these "odd" cycles. This closes the gap somewhat between the machines. I seem to remember someone finding that the CPU in the Atari ST was blocked 7% of the time while running one particular game waiting for access to memory during "odd" cycles.

All this is very code dependent, of course but there are even situations where (very improbable) code will run much faster on the Amiga. An unrolled loop of CLRs, for example, will often take fewer cycles to run than on an ST if the code is being executed during the top or bottom overscan areas.

I wonder if WinUAE can track of the number of times the CPU is allowed access to an "odd" cycle. It would be interesting to see just how often this happens when running typical codes.

My guess is: on average about 4% of the time, but better or worse depending on the code.

Tight loops might be interesting since branch instructions can take 10 cycles, though instruction pairing rules can make cycle timing analysis difficult.

Interesting thing: for some codes the Amiga might be faster or slower than the ST depending on whether or not some part of the code runs during blanking/overscan or during audio/sprite/disk/bitplane DMA.

[mombasajoe]

Originally Posted by mombasajoe

With or without 8 MB FastRam? If you had turned the GVP-RAM (seen in your sig) on you should try SysInfo again without FastRam.

Originally Posted by Geijer

Then it is the presence of Fast Ram that makes most of the difference when compared to a A600 in SysInfo.

That´s right. I´ve had some spare time today so I will show you that effect with my real A500+ I´ve used SysInfo 3.24 though.

[klx300r]

Originally Posted by mombasajoe

That´s right. I´ve had some spare time today so I will show you that effect with my real A500+ I´ve used SysInfo 3.24 though.

That's some cool stuff had a 68010 in my Amiga 1000 for years now and remember main reason I got it was for a large spreadsheet program I used for my University thesis & it definitely helped back then Never had any issues with any games too though I did install the 010 patch on aminet on my hard drive (whopping 33MB) just in case eh

[shaf]

I also never had any problems with the 68010 on the A1000.