Apollo4060 @ 100MHz

[Ratte]

Benchmarks:
http://www.a1k.org/forum/showthread.php?p=412295

3.5V regulator-mod:
http://www.amiga.org/gallery/index.php?u=4308

[mfilos]

OMG I'll be damned! That is indeed a serious overclockage!!!
Awesome info there my friend! Insane numbers :O

/respect \o/

[fitzsteve]

Impressive work, 130+ MIPS

[amigakit.com]

If only my A1200 could achieve that speed- well done!

[mfletcher]

Crikey!

Im looking at some of the other mods over on a1k.org, well impressed!

[alexh]

Nice job.

I'd be curious to know once and for all if you need Mach131 chips on an A4040 to convert to A4060? I know the general feeling is yes... but is there any evidence?

[kriz]

wow thats sweet, my A4000 got a Apollo 060 aswell....

[Chain]

my benchs with same card
http://chain.3dgrafika.cz/100mhz/AIBB/

whole card needs alot of cooling (see here http://www.amiga.org/gallery/index.php?n=2961 ), if you want to keep it stable

im also using newly built votage regulator with proper 3.5V chip

[Ratte]

I have to play with the frequencies and voltage ... 100MHz is a bit unstable here.
Sometimes the systems hangs, no matter what type of memory is used.
It happens in chipmem, zorram, fastmem, 32bitmem.
Tested with analyser, after some passes it hangs.
(I think i have to try different corevoltages or the cpu has reached the limit)
For now i have a rockstable 90mhz setup:
http://www.youtube.com/watch_popup?v...Vn5Mw&vq=large

[Cosmos]

95 Mhz will works too, I'm sure...

Some Atari geeks reached 105 Mhz with rev6 !

[Zetr0]

According to the Motorola Specification sheet on the 68060, as with all processors - the higher the frequency the higher the voltage you need.

the latest core's of 060 *(according to the the data sheet from free-scale) can handle up to 4.7 volts... I would suggest you would need a good 4 volts for stability at 100MHz - pending on mask revision there will likely be a play on Stability {VCore / Frequency}, however obviously the XC range leak like a sieve so avoid those.

Obviously there will be the need for some serious active cooling

[Cosmos]

How many A are required by the 060 at 90 Mhz or more ??

[Zetr0]

@Cosmos

According to the data sheet I have on the Motorola 68060 family CPU, Section 12: Electrical and Mechanical Characteristics > Page 343

Under normal operation of 3.465 volts the following power dissipation pattern

50MHz = output 3.9Watts = 1.126 Amps
60MHz = output 4.9Watts = 1.414 Amps

That is a factor of 0.288 Amps dissipation for a 10Mhz increase, what that is internally I am not 100% sure but I suspect it would be very close to those numbers.

Since we are altering the voltage (volume) of the electrical input, it will offset the needed amps (current) required to push the gates - thus -

if you increase the voltage between 3.6 - 3.8 volts with a maximum draw of 1.8 Amps you should be able to have a stable 060 at 90Mhz... (3.7v / 6.9Wd = 1.86 Amps)

But that's only half of one possible solution.. like the timing of an engine in a car if only one part is slightly off-beat then its just going to hang... calculating the right frequency so that everything is in step - that's all the logic, both in propagation and process delays, also the RAM timings, is very important.

I find writing it out on paper with a calculator is actually quicker than trying to plug at it on notepad++ as you soon see the patterns needed. That's why there are some non-working frequencies with the Amiga, where although the CPU can run, the rest is just out of step.

you can find the timings in both the A1200 schematics and you can find the CPU timings in the Motorola specification sheet on Section 7: Bus Operation - page 156,157 etc.

As you know my friend, the choice of RAM is very critical too... I would suggest that if you go above 66Mhz think about getting 50ns RAM - yes.... its rare, but these are going to get hammered and I have 60ns memory that fails even at 66Mhz.

I have only 1 stick of 32MB at 60ns that will run on my Apollo 060@80Mhz, unfortunately the last time I held 50ns RAM was a fair few years ago!

[Ratte]

@Zetr0:
According to the latest datasheet the max. Vcc is 4.0V.
I tried to push the 68060 up to 100Mhz with a modified 3.75V powerregulator.
But its more unstable.
I got the best results (but still unusable) at 3.5 V.

So I have to stay at 90Mhz with my 68060 ...

[Zetr0]

@Ratte

I had a thought that might be the case, the data-sheets I have are old and may need updating and getting them from free-scale might prove more accurate as opposed to a google search.

A thought maybe the processor itself, have you tired another E41J masked CPU?

Although I do fear that its quite likely the logic holding it all together that is another hurdle.

[Ratte]

Its the latest rev.6 (71E41J)
The Apollo has the Mach -131 chipset.
All components are cooled ... problem seems to be the CPU.
If you know the Falcon060-boards, they are running between 90 - 105 Mhz.
Only limited by the CPU ...
Nice feature on the Falcon060-board, it has a programmable CT-ClockModule.
A must have for upcoming Classic-Acc...
http://www.powerphenix.com/CT60/english/CTCM.htm

But now I reached a good state at 90Mhz .. system is nice like it is.

[roy bates]

hi ratte has anyone done this to a 1260 yet . ile give you a clue 132.81 mips

hi would anyone like to know how to make a 1260 run at 100mhz? because i have done it

[mfilos]

Don't ask for these things mate cause the answer is simple.

Share some info and pr0n pics!!!

[roy bates]

i will try to get some pics with my phone.btw this is stable at 100mhz i would like to know if anyone has done it.
if you guys are interested its 3.3 volts at around 5-6 amps:

[roy bates]

ok pictures are in my profile enjoy!!