PDA

View Full Version : RamBlade3 running ZiCog and CPM2.2 at 114.5MHz ;)



Cluso99
10-19-2011, 08:05 AM
Today I succeeded in getting the RamBlade3 (stackable version) running.

Next I achieved getting ZiCog and CPM2.2 running at 5MHz. I then cranked up my boards xtal to see what it can do....
5.0MHz = 80MHz yes
6.5MHz = 104MHz yes
13.5MHz = 108MHz yes
14.31818MHz = 114.54544MHz yes
15MHz = 120 MHz no!
7.3728MHz = 117.9648MHz not quite

The 7.3728MHz maybe the SD driver failing because I get the occasional glitch in the text display but CPM keeps running.

Now, just to qualify this... I have not done any exhaustive testing. It is about 20C here atm. I especially designed my prop board for overclocking.

Now, here is a pic of the display ;) and the boards ;)

86088 86089

Sapieha
10-19-2011, 08:23 AM
Hi Cluso.

I have looked on 15MHz alternatives for Propeller.
As I said You I have one Original ProtoBoard that can run that -- But never considered WHY?.

Now as You write this I have looked on it again.

Conclusion of this IS. ----> For Propeller to be possible RUN 15MHz V-regulator NEED be adjusted to have about 3.5V instead of 3.3V.
As said it was bad V-regulator that give me possibility to run that frequency --- Life



Today I succeeded in getting the RamBlade3 (stackable version) running.

Next I achieved getting ZiCog and CPM2.2 running at 5MHz. I then cranked up my boards xtal to see what it can do....
5.0MHz = 80MHz yes
6.5MHz = 104MHz yes
13.5MHz = 108MHz yes
14.31818MHz = 114.54544MHz yes
15MHz = 120 MHz no!
7.3728MHz = 117.9648MHz not quite

The 7.3728MHz maybe the SD driver failing because I get the occasional glitch in the text display but CPM keeps running.

Now, just to qualify this... I have not done any exhaustive testing. It is about 20C here atm. I especially designed my prop board for overclocking.

Now, here is a pic of the display ;) and the boards ;)

86088 86089

Ale
10-19-2011, 09:55 AM
Hei Cluso, the title looks sort of messed up: 115=4.5MHz. I thought that the ZiCog was emulating the equivalent of a 4.5 MHz CPU... It would be great if they refine the propeller process for 120 MHz operation... one can surely dream! BTW, your boards are always very nice!!!

Dr_Acula
10-19-2011, 10:09 AM
Hey, I have trouble beating Pacman on the 2Mhz emulation. Now if you go and make it faster... *grin*

Cluso99
10-19-2011, 01:23 PM
Thanks Ale... fixed the subject typo :(

Sapieha: I didn't want to crank up the voltage but maybe it is worth a try sometime to see if that will work for 120MHz.

Drac: something I have always wanted to do was try and speed up ZiCog but never had the time.

Now I have to get the RTCmodule working too :-) Then Catalina.

jazzed
10-19-2011, 02:37 PM
Congrats on your latest CP/M platform.

Dr_Acula
10-19-2011, 11:32 PM
Drac: something I have always wanted to do was try and speed up ZiCog but never had the time.

Now I have to get the RTCmodule working too :-) Then Catalina.


Catalina is great, not just because you can write bigger C programs than you possibly can in CP/M, but also for all the hidden gems within the code. One of those gems I have just found is caching, and I am sure the Zicog can be sped up with caching. Catalina has an 8k cache and if Zicog had that as well it ought to run a lot faster.

The catch is that I think the hub is full, but again I think there are ways around that. For instance, one of the things Catalina can do is load a prop in two phases - load all the cogs up, then reload things into hub. That frees up the up-to 14k of wasted space in so many spin programs.

I'm particularly interested in caching, as most of my designs only leave 12 pins free for interfacing to external memory, so external memory access is much much slower than your ramblade.

Cluso99
10-20-2011, 12:27 AM
My main purpose for building the RamBlade3 is to run Catalina. I have a commercial project.
I still believe that a dual prop solution is more effective. Caching performance is specific to each application.