Apollo 1220 - instructions and 68020 to 68030 upgrade

fc.studio · 20 November 2007, 10:32

This board will reach me next week without instructions.
If someone could upload them to zone or link me somewhere it will be much appreciated.

However this little accelerator has only two jumpers and shouldn't be a problem... what really I would like to know, since its CPU is PGA, if replacing 68020 CPU with a 68030 is possible.

The Apollo 1230 LC seems to be the same board. What is your opinion on this CPU upgrade?

RedskullDC · 20 November 2007, 11:35

Quote:

Originally Posted by fc.studio

This board will reach me next week without instructions.
If someone could upload them to zone or link me somewhere it will be much appreciated.

However this little accelerator has only two jumpers and shouldn't be a problem... what really I would like to know, since its CPU is PGA, if replacing 68020 CPU with a 68030 is possible.

The Apollo 1230 LC seems to be the same board. What is your opinion on this CPU upgrade?

Hi,

The boards do have similar layout, but do not appear to be identical (at least from the pics on BBoAH).

The 1220 will take a 4MB memory module.

68020 should be a 114pin PGA, where the 68030 should be a 128pin PGA from the chip docs I have here.

Cheers,
Red

fc.studio · 20 November 2007, 11:58

Quote:

Originally Posted by RedskullDC

68020 should be a 114pin PGA, where the 68030 should be a 128pin PGA from the chip docs I have here.

Therefore 68030 and 68020 are not pin compatible... Thanks, I didn't know.

About speed, is not a great loss... Apollo 1220 or even better Blizzard 1220/4 (that I already have) are some times faster than 68030 boards clocked at similar rate.

Obviously 68020 have limit to max 8 MB RAM and can't have MMU.

Zetr0 · 20 November 2007, 13:44

@fc.studio

hello there my friend, it does sound like a pickle of accelerator cards info.

just to help clear things up,

the 020 is a 32bit chip as is the 030, they are not pin compatible and there are a few more signals required to drive the 030.

the memory structure are non-consequential as it comes down to how the card is made not the processors. as there is only 8MB of configurable space on the A1200 one can use all that as a memory location (hence 8MB) or map information in and out of that space allowing for larger memory portions, this requires more elaborate electronic design and more costly electronics. Maximum memory has nothing to do with the chip.

An 030 and an 020 with identical amounts of memory and clock speed would be seriously different in performance for instance an 020@33mhz is about half two thirds the speed of an 030@33mhz.

Indeed, i believe you are correct about the MMU as i know not of an 020 that has one... but saying that the MMU is not used under workbench infact i think there only one piece of software that uses the MMU on an 030, and to be honest i don't know what it is.

fc.studio · 20 November 2007, 19:13

Quote:

Originally Posted by Zetr0

@fc.studio

hello there my friend, it does sound like a pickle of accelerator cards info.

just to help clear things up,

the 020 is a 32bit chip as is the 030, they are not pin compatible and there are a few more signals required to drive the 030.

the memory structure are non-consequential as it comes down to how the card is made not the processors. as there is only 8MB of configurable space on the A1200 one can use all that as a memory location (hence 8MB) or map information in and out of that space allowing for larger memory portions, this requires more elaborate electronic design and more costly electronics. Maximum memory has nothing to do with the chip.

An 030 and an 020 with identical amounts of memory and clock speed would be seriously different in performance for instance an 020@33mhz is about half two thirds the speed of an 030@33mhz.

Indeed, i believe you are correct about the MMU as i know not of an 020 that has one... but saying that the MMU is not used under workbench infact i think there only one piece of software that uses the MMU on an 030, and to be honest i don't know what it is.

Thanks for the info friend,

- Now I know that these CPUs are not pin compatible.

- You were clear about the memory limit. Probably the electronic costs didn't allow the hardware-houses to go over 8MB and prefered to build 68030 boards.

- CPU FASTROM is the Workbench command using MMU. It easily moves the Kickstart from ROM to RAM and therefore performances increase a bit.

- About speed I don't agree with you:
Well, look at AIBB tests in Amiga Resources. The performances of the Blizzard 1220/4 (EC020/28) VS. Apollo 1230 LC (030/25), Apollo 1230 (030/28), DKB 1240 (030/28), Viper/T1230 (030/28). These boards can be about 10% (media) faster OR SLOWER. Sometimes 68020 boards access to memory 100% faster than 68030.

But I can report personal experience since I own no. 2 A1200 with:
1st - a Blizzard 1230/IV (030/50) with FPU (68882/50) and 16MB
2nd - a Blizzard 1220/4 (EC020/28) with FPU (68882/28) and 4MB

About them Sysinfo 3.24 reports:
Blizzard 1230: 9486 Dhrystones, 9.90 Mips, 7.79 Vs. standard A1200 (with Kickstart in RAM)
Blizzard 1230: 9228 Dhrystones, 9.63 Mips, 7.58 Vs. standard A1200
Blizzard 1220: 4849 Dhrystones, 5.06 Mips, 3.98 Vs. standard A1200

Blizzard 1230 is about 100% faster than Blizzard 1220, but this has a clock rated at 50 MHz against 28 MHz and probably this is the best 68030 board available.
I have just sold a DKB Cobra 1230 @28 with 16MB. This is a reliable board with MMU, MapROM ability and can mount up to 128 MB, but about performances it is surely NOT so better than my Blizzard 1220/4. I have no more Sysinfo data but I can assure that results were very, very near to it.
Therefore in my opinion 68030 CPU, although data & inst cache, are not so faster Vs. 68020.
68040 and 68060 nearly double performances at the same clock rate.

Apollo 1220 is very similar to Blizzard 1220/4. I'm just waiting for it to test my own.

Zetr0 · 20 November 2007, 20:26

@fc.studio

you have done some good research my friend, and to assist your efforts I think i can explain why in some cases an 020 might access memory faster than an 030.

It comes down to wait states.. the time the cpu has to wait untill its safe to read / write to memory. from my research in developing an 020 board the wait state depends on the amount of 'address decoding' circuitry between the CPU and the memory matrix its self.

from my research too, FASTROM and KICKSTARTS to RAM, don't use an MMU, this is evident when an a1200 can map ROM to fast (an 68EC020) and with my A600 which has 4MB SRAM it can map the KS ROM to ram too and thats a basic 68k chip. virtually nothing on the amiga actually uses the MMU.

I am very interested in you comparison research some fine looking work there. I have been going over the efforts collected on amiga.resource.cx there are some interesting conclusions to be drawn over memory controllers.

ganralf · 21 November 2007, 00:03

Quote:

Originally Posted by Zetr0

from my research too, FASTROM and KICKSTARTS to RAM, don't use an MMU, this is evident when an a1200 can map ROM to fast (an 68EC020) and with my A600 which has 4MB SRAM it can map the KS ROM to ram too and thats a basic 68k chip.

Sorry Zetro you're wrong here. It's just that the CPU command does not give errors. You can execute the whoöe set of "cpu cache burst copyback fastrom" on any processor.
FASTROM can not work without MMU unless you have relocation tables, like zkick has. Sometimes CPU allocates memory for the rom image, though there's no MMU. Maybe your observation is based on that bug.
BTW CPU happily maps the Kickstart to Chip RAM when there is no Fast RAM but a MMU.

Zetr0 · 21 November 2007, 00:19

@ganralf

hmmm I will review what i have researched thus far. but you dont explain how a standard 68000 and a 68EC020 manage to allow Kickstarts in FastRam ?

anyway I will research some more...

fc.studio · 21 November 2007, 00:57

I can confirm that CPU FASTROM command doesn't work without MMU.
I remember that, about DKB Cobra, there is a version without MMU (68EC030 @40) that allows to move Kickstart to RAM, but only putting a little program in the WB-StartUp drawer.
This program can be found in the old supplied disk or in the update file dkb.ferret.DMS on Amiga Resources.

fc.studio · 19 December 2007, 21:40

I can update my table above with my (new) Apollo 1220 (EC020/28) with FPU (68882/28) and 4 MB:

SysInfo 3.24 reports:

Blizzard 1230: 9486 Dhrystones, 9.90 Mips, 7.79 Vs. 1,00 basic A1200 (with Kickstart in RAM)
Blizzard 1230: 9228 Dhrystones, 9.63 Mips, 7.58 Vs. 1,00 basic A1200
Blizzard 1220: 4849 Dhrystones, 5.06 Mips, 3.98 Vs. 1,00 basic A1200
Apollo 1220: 4788 Dhrystones, 4.99 Mips, 3.93 Vs. 1,00 basic A1200
Amiga 1200: 1283 Dhrystones, 1.33 Mips, 1.00 = basic A1200

Apollo 1220 is a bit bit slower than Blizzard 1220/4, but like it performs very near (a bit slower or a bit higher) to 68030/25 or 68030/28 competitors.
Surely 68030 cpu have better design so Motorola engineers could push it to higher speed Vs. 68020, but at the same clock rate, unlike 68040 or 68060, performances are almost the same.
Soon I'll have the opportunity to test the Pyramid TRA1200 accelerator so I can reach the end of this voyage.
Then probably moderators should rename this thread 68020 Vs. 68030.

fc.studio · 10 January 2008, 10:03

This is the last update, since now I have a TRA1200 full working. It has a 68EC020/28, no FPU, 8 MB RAM.
I remember that accelerators, to work at their full speed, NEED some Fast RAM, but size is not important (performances are the same with 4 MB, 8 MB or more).

FINAL CHART (SysInfo 3.24 reports):

Blizzard 1230: 9486 Dhrystones, 9.90 Mips, 7.79 Vs. 1,00 basic A1200 (with Kickstart in RAM)
Blizzard 1230: 9228 Dhrystones, 9.63 Mips, 7.58 Vs. 1,00 basic A1200
Blizzard 1220: 4849 Dhrystones, 5.06 Mips, 3.98 Vs. 1,00 basic A1200
TRA 1200: 4833 Dhrystones, 5,04 Mips, 3,97 Vs. 1,00 basic A1200
Apollo 1220: 4788 Dhrystones, 4.99 Mips, 3.93 Vs. 1,00 basic A1200
Amiga 1200: 1283 Dhrystones, 1.33 Mips, 1.00 = basic A1200

AGAIN: All 68EC020 accelerator @ 28 MHz boards are very near, (a bit slower or a bit higher) to 68030/25 or 68030/28 competitors. Access to CHIP RAM is incredible and often wins Vs. 68030 boards higher clocked.
Surely 68030 cpu has better design so Motorola engineers could push it to higher speed Vs. 68020, it can have MMU, usually 68030 boards can have more than 4 or 8 MB RAM... but at the same clock rate, unlike 68040 or 68060, speed performances are almost the same.

----------------------
NOTE TO MODERATORS: It would be better to change title thread to "Apollo 1220, 68020 Vs. 68030 upgrade & performances.

Zetr0 · 10 January 2008, 20:30

@fc.studio

Thats some good research!

I am looking on amiga.resource.cx for comparisons.

http://amiga.resource.cx/perf/aibb.p...lizzard1230mk3

very interesting.

20 November 2007, 10:32	#1
fc.studio Cat lover Join Date: May 2004 Location: Italy Age: 54 Posts: 804	Apollo 1220 - instructions and 68020 to 68030 upgrade This board will reach me next week without instructions. If someone could upload them to zone or link me somewhere it will be much appreciated. However this little accelerator has only two jumpers and shouldn't be a problem... what really I would like to know, since its CPU is PGA, if replacing 68020 CPU with a 68030 is possible. The Apollo 1230 LC seems to be the same board. What is your opinion on this CPU upgrade?

21 November 2007, 00:57	#9
fc.studio Cat lover Join Date: May 2004 Location: Italy Age: 54 Posts: 804	I can confirm that CPU FASTROM command doesn't work without MMU. I remember that, about DKB Cobra, there is a version without MMU (68EC030 @40) that allows to move Kickstart to RAM, but only putting a little program in the WB-StartUp drawer. This program can be found in the old supplied disk or in the update file dkb.ferret.DMS on Amiga Resources. Last edited by fc.studio; 21 November 2007 at 01:02.

10 January 2008, 10:03	#11
fc.studio Cat lover Join Date: May 2004 Location: Italy Age: 54 Posts: 804	This is the last update, since now I have a TRA1200 full working. It has a 68EC020/28, no FPU, 8 MB RAM. I remember that accelerators, to work at their full speed, NEED some Fast RAM, but size is not important (performances are the same with 4 MB, 8 MB or more). FINAL CHART (SysInfo 3.24 reports): Blizzard 1230: 9486 Dhrystones, 9.90 Mips, 7.79 Vs. 1,00 basic A1200 (with Kickstart in RAM) Blizzard 1230: 9228 Dhrystones, 9.63 Mips, 7.58 Vs. 1,00 basic A1200 Blizzard 1220: 4849 Dhrystones, 5.06 Mips, 3.98 Vs. 1,00 basic A1200 TRA 1200: 4833 Dhrystones, 5,04 Mips, 3,97 Vs. 1,00 basic A1200 Apollo 1220: 4788 Dhrystones, 4.99 Mips, 3.93 Vs. 1,00 basic A1200 Amiga 1200: 1283 Dhrystones, 1.33 Mips, 1.00 = basic A1200 AGAIN: All 68EC020 accelerator @ 28 MHz boards are very near, (a bit slower or a bit higher) to 68030/25 or 68030/28 competitors. Access to CHIP RAM is incredible and often wins Vs. 68030 boards higher clocked. Surely 68030 cpu has better design so Motorola engineers could push it to higher speed Vs. 68020, it can have MMU, usually 68030 boards can have more than 4 or 8 MB RAM... but at the same clock rate, unlike 68040 or 68060, speed performances are almost the same. ---------------------- NOTE TO MODERATORS: It would be better to change title thread to "Apollo 1220, 68020 Vs. 68030 upgrade & performances. Last edited by fc.studio; 10 January 2008 at 10:10.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Apollo Upgrade Question	moxavius	support.Hardware	48	24 February 2012 23:28
Apollo 1220 accelorator	Kakaboy	Hardware mods	9	23 March 2010 02:39
Wanted: Apollo 1240 060 upgrade kit	8bitbubsy	MarketPlace	1	16 January 2010 15:07
Serious Apollo 1240 Upgrade Pron	Dimlow	Hardware mods	54	15 June 2009 18:58
Help making a choice needed apollo or Blizz upgrade?	DoogUK	support.Hardware	2	16 February 2009 13:12

20 November 2007, 13:44	#4
Zetr0 Ya' like it Retr0? Join Date: Jul 2005 Location: United Kingdom Age: 49 Posts: 9,768	@fc.studio hello there my friend, it does sound like a pickle of accelerator cards info. just to help clear things up, the 020 is a 32bit chip as is the 030, they are not pin compatible and there are a few more signals required to drive the 030. the memory structure are non-consequential as it comes down to how the card is made not the processors. as there is only 8MB of configurable space on the A1200 one can use all that as a memory location (hence 8MB) or map information in and out of that space allowing for larger memory portions, this requires more elaborate electronic design and more costly electronics. Maximum memory has nothing to do with the chip. An 030 and an 020 with identical amounts of memory and clock speed would be seriously different in performance for instance an 020@33mhz is about half two thirds the speed of an 030@33mhz. Indeed, i believe you are correct about the MMU as i know not of an 020 that has one... but saying that the MMU is not used under workbench infact i think there only one piece of software that uses the MMU on an 030, and to be honest i don't know what it is.

20 November 2007, 20:26	#6
Zetr0 Ya' like it Retr0? Join Date: Jul 2005 Location: United Kingdom Age: 49 Posts: 9,768	@fc.studio you have done some good research my friend, and to assist your efforts I think i can explain why in some cases an 020 might access memory faster than an 030. It comes down to wait states.. the time the cpu has to wait untill its safe to read / write to memory. from my research in developing an 020 board the wait state depends on the amount of 'address decoding' circuitry between the CPU and the memory matrix its self. from my research too, FASTROM and KICKSTARTS to RAM, don't use an MMU, this is evident when an a1200 can map ROM to fast (an 68EC020) and with my A600 which has 4MB SRAM it can map the KS ROM to ram too and thats a basic 68k chip. virtually nothing on the amiga actually uses the MMU. I am very interested in you comparison research some fine looking work there. I have been going over the efforts collected on amiga.resource.cx there are some interesting conclusions to be drawn over memory controllers.

21 November 2007, 00:19	#8
Zetr0 Ya' like it Retr0? Join Date: Jul 2005 Location: United Kingdom Age: 49 Posts: 9,768	@ganralf hmmm I will review what i have researched thus far. but you dont explain how a standard 68000 and a 68EC020 manage to allow Kickstarts in FastRam ? anyway I will research some more...

19 December 2007, 21:40	#10
fc.studio Cat lover Join Date: May 2004 Location: Italy Age: 54 Posts: 804	I can update my table above with my (new) Apollo 1220 (EC020/28) with FPU (68882/28) and 4 MB: SysInfo 3.24 reports: Blizzard 1230: 9486 Dhrystones, 9.90 Mips, 7.79 Vs. 1,00 basic A1200 (with Kickstart in RAM) Blizzard 1230: 9228 Dhrystones, 9.63 Mips, 7.58 Vs. 1,00 basic A1200 Blizzard 1220: 4849 Dhrystones, 5.06 Mips, 3.98 Vs. 1,00 basic A1200 Apollo 1220: 4788 Dhrystones, 4.99 Mips, 3.93 Vs. 1,00 basic A1200 Amiga 1200: 1283 Dhrystones, 1.33 Mips, 1.00 = basic A1200 Apollo 1220 is a bit bit slower than Blizzard 1220/4, but like it performs very near (a bit slower or a bit higher) to 68030/25 or 68030/28 competitors. Surely 68030 cpu have better design so Motorola engineers could push it to higher speed Vs. 68020, but at the same clock rate, unlike 68040 or 68060, performances are almost the same. Soon I'll have the opportunity to test the Pyramid TRA1200 accelerator so I can reach the end of this voyage. Then probably moderators should rename this thread 68020 Vs. 68030.

10 January 2008, 20:30	#12
Zetr0 Ya' like it Retr0? Join Date: Jul 2005 Location: United Kingdom Age: 49 Posts: 9,768	@fc.studio Thats some good research! I am looking on amiga.resource.cx for comparisons. http://amiga.resource.cx/perf/aibb.p...lizzard1230mk3 very interesting.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)