overclocking a propeller chip

chbrandl

Hi,
what is the highest speed, which can a propeller chip be driven without problems.
I tested it with 20Mhz and PLL=8x without problems, but what is recommended.
The chip will not warm up.

Best regards
Christopher
·

Baggers

Hi Christopher,
I'd also like to know the answer to this, as it'll be interesting [noparse]:)[/noparse]
I read a while back about using a 6Mhz and PLL=16x to 96Mhz, and other talk about taking it up to 128Mhz, no worries.
But 20Mhz at PLL8x that would be good. how long did you test it for? how many hours? constant usage? was it doing LOTS at that speed? like a tv driver, and graphics examples, with all cogs running full whack?

I'm curious, as this is double the current speed of 80Mhz, and would theoretically double how much a scanline renderer could do [noparse]:)[/noparse]

Cheers,
Baggers.

Lawson

chbrandl said...
I tested it with 20Mhz and PLL=8x without problems, but what is recommended.

The chip will not warm up.

did you measure how fast the chip was ACTUALLY running? There was an older thread on over-clocking that had lots of info. (well worth digging up) If I remember right 128MHz is the fastest the PLL is specified to run. (10MHz crystal and 8x works, so i guess it will keep a lock at 160MHz) Also from that thread, stuff like the zero flag and cary flag start failing at over 100MHz. Most likely you had the PLL running at it's top speed, and then were using 1/2 that as the system clock. (likely in the 90-100MHz range)

heh, If you want the chip to warm up, Paul (i think) posted a demo that caused the Prop to draw 1.7 watts or something. Basically he setup the counters on all the cogs to toggle all the pins as fast as they could.

Later,
Marty

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Lunch cures all problems! have you had lunch?

Paul Baker

This topic has been discussed a few times, this is one of the more complete discussions: http://forums.parallax.com/showthread.php?p=620404

I share Marty's skeptism of being able to run the chip at 20MHz x 8PLL. Neither the PLL was designed to run·at 320MHz (remember that regardless of what tap you choose, the PLL is always scaling it up by a factor of 16), nor was the Propeller designed to run at 160MHz.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Coley

I have run many apps at 96Mhz with absolutelty no problem whatsoever.

In fact the FTDI propstick type DIP board is designed to run at that speed.

I must admit that I haven't tried going any higher but why would you want to?

is 160Mips not enough these days? lol

Just waitfor PROP II.....

Regards,

Coley

Kevin Wood

But it would be cool to see how fast somebody can get one going.

After all, this IS the land of big-block Chevys...

Baggers

Just tried a few 20Mhz clocks, with a few tv drivers, and all failed, so I'd go with a NOPE that don't work, sorry, but it just doesn't run anything stable at 160mhz.

Nice thought though.

Baggers.

Mike Green

There were a few postings some time ago on this theme. If I remember correctly, the PLL stops working at a little over 100MHz (maybe 120MHz?). I think they used a direct clock input and refrigerated the chip and they got it to run around 200MHz (maybe 240MHz?). I think it was the JMP logic that stopped working first. Try searching with the Google engine (search.parallax.com).

Paul Baker

From the thread I linked to in my previous post:

Chip said...
The first thing to fail as speed increases is the hub memory access (~112MHz), particularly the hub SRAM read -- this is the critical path in the chip. The next thing to fail is the Z flag (~120MHz), then the C flag (~128 MHz).

The aerosol cooled chip wasn't that much faster.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 8/10/2007 7:42:33 PM GMT

Mike Green

Thanks Paul ... so much for my memory. I'll look first next time.

RinksCustoms

Kevin Wood said...
But it would be cool to see how fast somebody can get one going.

After all, this IS the land of big-block Chevys...

AMEN to that brother!!· And the only replacement for displacement is "mo-boost".

Baggers, i'm guessing you tried setting to _xtal2 or 3?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
E³ = Thought

http://folding.stanford.edu/·- Donating some CPU/GPU downtime just might lead to a cure for cancer! My team stats.

Baggers

yeah, Rinks, I didn't just leave it on the xtal1 setting [noparse]:)[/noparse]

Paul Baker

Because you guys asked for it, I just ran a chip -74 C at Vdd=3.6 with the PLL=16x and had it running on an input frequency of 9.45 MHz before it failed at 9.46. This means the Propeller was running at 151.2 MHz.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Fred Hawkins

Does that failure produce magic smoke? Or merely indigestion?

Paul Baker

No magic smoke, it just stopped working. Scaling the frequency back down and reseting causes it to run again.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Fred Hawkins

The lesser gods of bitdom smile and bless this chip.

Rob7

"151.2 MHz"
That's amazing !
I remember way back when I used to overclock the Rendition graphics card's.
Great info for maybe some overclocking projects in the future.

Rob 7

deSilva

You will need many ice cubes for -74 °C....

Fred Hawkins

Dry ice has a sublimation point of -78.5 C. Use in a well ventilated area, though.

And after you're done testing, toss it in pan of water and have a prom.