Apollo 1260@100 crashes the system

[Kppot]

Hello everyone, and I am overclocker and I cannot help myself
So I do have 060 rev 6 which works perfectly fine @80 and loads OS @100 with passive cooling.
However after 2-3 mins of work my systems crash. Especially if I run Quake. It shows me some weird visuals in 5-7 secs and then it drops to guru.
The cooling is not a problem since the CPU temperature is low, it is barely warm. Again, it is not even remotely hot, just barely warm.
So what causes the crashes:
1. Low voltage?
2. 60ns RAM
Should I go for 50ns RAM may be? Cheers!

[AMIGASYSTEM]

If you can detect the Guru code it could already help, if you do not have time to read the code of the Guru use a video camera or phone and then analyze the video on the Computer.

[eXeler0]

You could read here:
http://eab.abime.net/showthread.php?...ferrerid=52972
But yea, fastest possible memory will usually help and the CPU will probably survive voltage increase to 4V :-)
Another theory is that - since the FPU is used extensively when you play Quake - the combined load on CPU/FPU simply is too much.
Try running different apps that use CPU only and something else than Quake that uses the FPU extensively and see if you can note a pattern..

[Daedalus]

Originally Posted by Kppot

Hello everyone, and I am overclocker and I cannot help myself

Hi Kppot, and welcome to Overclockers Anonymous. You're in a safe place now, together we will beat our addiction!

So I do have 060 rev 6 which works perfectly fine @80 and loads OS @100 with passive cooling.
However after 2-3 mins of work my systems crash. Especially if I run Quake. It shows me some weird visuals in 5-7 secs and then it drops to guru.
The cooling is not a problem since the CPU temperature is low, it is barely warm. Again, it is not even remotely hot, just barely warm.
So what causes the crashes:
1. Low voltage?
2. 60ns RAM
Should I go for 50ns RAM may be? Cheers!

FPU and RAM limitations as suggested are possibilities. It's also possible that some of the glue logic on the board is just not able to keep up. Are any of the other chips on the board getting hot?

[TjLaZer]

I have a Blizzard 1260 that was was only able to over clock to 66mhz using 50 ns ram. I’d try that and see if it helps.

[Kppot]

Originally Posted by Daedalus

Hi Kppot, and welcome to Overclockers Anonymous. You're in a safe place now, together we will beat our addiction!

Are any of the other chips on the board getting hot?

Thanks for accepting me with OA
Yes some chips, Paula and Alice in particular, are getting hot, but they do have some radiators glued as well

[Daedalus]

It shouldn't really affect the custom chips on the Amiga motherboard, I meant the GALs / Mach chips that are on the accelerator. These are likely overclocked too and might be running too fast to remain stable.

[trixster]

There’s a good chance the cpu is ok with 100mhz - I say this because Falcon CT60/60e/63 aceraator board users routinely manage 95+MHz with rev6 060s thanks to fast sdram and modern logic. So as others suggest, i would say it’s likely the glue logic and/or ram that stops you getting 100mhz on your Apollo.

[Kppot]

Originally Posted by trixster

it’s likely the glue logic and/or ram that stops you getting 100mhz on your Apollo.

Right, so I will start with re[lacing 64 with 128 Mb of RAM with 50ns
Regarding glue logic, which chips should I check in particular? To my feeling, the only chip which getting warm is CPU and the voltage regulator PCB, which mounted atop of Apollo, near CPU itself

[plasmab]

Its also likely you dont have enough decoupling for the speed you're going for.

[Kppot]

Originally Posted by plasmab

Its also likely you dont have enough decoupling for the speed you're going for.

How do I fix that, some capacitors?

[plasmab]

the best you can do is to replace the small decouping caps with multilayer caps. The surface mount caps i mean. the big guys wont be doing much at those speeds. its also possible that the caps arent close enough to the power pins for the speeds your after.

The symptoms you describe are random errors.. which is pretty much decoupling all over. if the results are predicable then its not caps.

EDIT: You are going to hit a hard limit of what the glue logic can do at some point. I have one of those cards actually.

[Kppot]

Aight, that makes sense. Will start with getting 50ns RAM, then check some caps.
What max frequency your card operates with?

[plasmab]

Originally Posted by Kppot

Aight, that makes sense. Will start with getting 50ns RAM, then check some caps.
What max frequency your card operates with?

ive not cranked it up in a few years but it was hitting 80Mhz.

The price i paid for it made me start building accelerators.

[Kppot]

Originally Posted by plasmab

ive not cranked it up in a few years but it was hitting 80Mhz.

The price i paid for it made me start building accelerators.

Very nice indeed, I am thinking of getting one of TF328 boards one day

[alexh]

The Mach chips on later cards are Mach 131 and are rumored to contain a design with better timing than those in the Mach 130 CPLD.

Check to see if you have a Mach 131 or 130.

[Hewitson]

It'll definitely a 131, the 130 doesn't support the 060.

I'm sure I remember reading that these chips were the reason these accelerators can't be overclocked to 100mhz.

[plasmab]

Originally Posted by Hewitson

It'll definitely a 131, the 130 doesn't support the 060.

I'm sure I remember reading that these chips were the reason these accelerators can't be overclocked to 100mhz.

What is the incompatibility? Both are just blogs of logic. Voltage perhaps?

[Kppot]

Mine is 131, just for the record
Sorry for a late response. At the end of the day, I am sticking to @90 MHZ at which machine is rock solid and stable. And requires no active cooling (still I mounted ultra silent 12V fan powered by +5v)