Here are the files needed for running P8X32A on the BeMicroCV
cgracey
Posts: 14,155
I got into the BeMicroCV board today and got things straightened out.
The problem with the cog-indicator LEDs was that, aside from being driven inversely, they were laid out in this order on the PCB: 7-1-0-2-3-4-6-5. That's fixed now, requiring changes to the top.qsf file (different pin order) and a change to the top.tdf file ("!" on the right-hand side of the 'ledg' assignment to invert signal polarity). The .png file was updated, too, to show the 3.3V I/O jumper placement. Here is a .zip containing the new files:
P8X32A_BeMicroCV_files.zip
As before, start off with the DE2-115 directory setup and get rid of all files that have DE2_115 in their names. Then, put these files into that folder and rename the folder P8X32A_BeMicroCV. After that you can follow directions in the readme.txt file.
Also, be sure to modify the 'dira' register assignment starting on line 317 of the cog.v file, so that it reads:
It turns out that only 'dira' was needing the reset, not 'outa'. This was an AHDL-to-Verilog transcription error I had made in the original files release.
This should be final for the BeMicroCV board.
The problem with the cog-indicator LEDs was that, aside from being driven inversely, they were laid out in this order on the PCB: 7-1-0-2-3-4-6-5. That's fixed now, requiring changes to the top.qsf file (different pin order) and a change to the top.tdf file ("!" on the right-hand side of the 'ledg' assignment to invert signal polarity). The .png file was updated, too, to show the 3.3V I/O jumper placement. Here is a .zip containing the new files:
P8X32A_BeMicroCV_files.zip
As before, start off with the DE2-115 directory setup and get rid of all files that have DE2_115 in their names. Then, put these files into that folder and rename the folder P8X32A_BeMicroCV. After that you can follow directions in the readme.txt file.
Also, be sure to modify the 'dira' register assignment starting on line 317 of the cog.v file, so that it reads:
always @(posedge clk_cog or negedge ena) if (!ena) dira <= 32'b0; else if (setdira) dira <= alu_r;
It turns out that only 'dira' was needing the reset, not 'outa'. This was an AHDL-to-Verilog transcription error I had made in the original files release.
This should be final for the BeMicroCV board.
Comments
The file is identifying as attachment.php
Is this correct???
Shouldn't we host these files on our web site along with the rest of the downloads so they can be easily found?
Let me know please.
- Ken
Well, that's what's in the link text:
http://forums.parallax.com/attachment.php?attachmentid=110303&d=1407717164
When I click on it, though, it downloads P8X32A_BeMicroCV_files.zip, as I'd expect.
I don't know why it's not working for you.
Yes, I have a new master file, but it needs to have that GPL text stuffed into the .v and .src files. I'm just putting patches on this forum. I'll send you the new files, sans GPL text.
Mine shows and downloads as a .zip
Mac OSX 10.9 and Firefox
I'll grab it on my Windows machine later....
It downloaded the file ok but called it attachment.php.
Tried with Firefox and no problems.
BTW I cannot post to the forum with IE11 either.
Wahnsinn !!
It says "propeller Version1 found on COM3" ;-)
Sadly it takes 25 minutes to re-compile, even for changes the pin assignment (I have my own prop-plug with 5 pins instead of 4). (i7 Quad-core 2.2 GHz), no multicore license.
Good work Chip !, Great work!
http://www.altera.com/literature/po/ss_quartussevswe.pdf
http://www.altera.com/support/kdb/talkbackfaq.html
Are you going to give that (talkback) a try when your board arrives today? I'm very interested to hear how that works for you.
I'll post compile times with/without talkback.
If I find it annoyingly slow, I may bump the memory to 32GB, and maybe go to a 4.4GHz eight core processor with an SSD (instead of 8 core 4GHz / 8GB / 3TB 7200rpm drive)
I installed Quartus 14... and I get :
Edit: I thought this talkback was enabled by default ?.. it is not. enable it per hand.
Compiled in 15 minutes wall time, 21 minutes CPU time, not the screamer by any length. I'm glad orca (www.thch.uni-bonn.de/tc/orca) does it right...
Utilization climbs to 90 % from 85 % in Quartus 13...
Set talkback feature on. It will enable parallel compilation which reduces compilation time.
What were the compile times with and without multicores?
Multicore 4 cores, fitter 15 wall time min, cpu time 21 min. Quartus 14. 90% used 65 MHz
Multicore 2 cores, fitter 18 wall time mins, cpu time 21 min. Q 14. 90% used 65 MHz
Back to q13 I suppose. Have to look at the reset issue now....
And yes, I would have expected better results from more cores. Guess they cannot divide the chores up that easily.
Downloading V13.1 now...
I'd be a little cautious about 'blaming' Q14, as sometimes the device speed files these tools use are revised.
IIRC those files set the corner cases, and they can sometimes revise down due to more experience in corner-case failures, and even varying FAB suppliers corner cases.
For bench testing, it is probably ok to focus on the faster/smaller case, (Q13) and do some of your own margin sanity tests. ( ie hike the clock, until test suites fail, and confirm there is plenty of headroom )
Downloading 13.1 ...
I downloaded your 8-11 zip, and followed you instructions - Quartus 14.0 compiled the BE MICRO CV image in ~15min!
Thank you!
Quartus pegged one core at 100%.
I'll try enabling Snoopy mode, and seeing what difference it makes on my AMD box.
Ok, I enabled talk back.
Compilation time dropped to 9 minutes.
I had the system monitor open, they do not make efficient use of multiple cores.
Looks like there is a LOT of memory still available on the BE/CV!
http://www.parallax.com/downloads/propeller-1-design
for the latest design files
And I agree, a github would be nice.
I've been able to re-compile the 8-11 code, now I just need to find the time to test it, and start making changes