84 lines
3.5 KiB
YAML
84 lines
3.5 KiB
YAML
---
|
|
title: CPU Usage Meter
|
|
url: /projects/cpu-usage-meter/
|
|
parent: Projects
|
|
---
|
|
Back in the day, there was a little obscure operating system called BeOS. The
|
|
company which made the OS was brave enough to put it on their own hardware, too.
|
|
This was dubbed the [BeBox](https://en.wikipedia.org/wiki/Bebox). Among all the
|
|
neat doohickies on the computer were two CPU load meters (one for each processor).
|
|
Now, how cool would it be to have a computer with those?
|
|
|
|
__Very.__
|
|
|
|
### Download
|
|
Note: Linux code makes use of libserial and libstatgrab. Both must be installed for
|
|
program to work/compile. The UM245R device uses the ftdi_sio driver. It's in the
|
|
2.6 kernel tree, so it should (hopefully) be detected when you plug the device in.
|
|
|
|
USB controller program:
|
|
|
|
* [Linux code and executable](/media/projects/cpu_meter.tar.gz)
|
|
* Windows code removed due to buginess
|
|
|
|
### Updates
|
|
|
|
#### Status Update - Nov 27 2007
|
|
|
|
Wow, it's been almost a year since I've put work on this. I decided to finally write
|
|
code to make it work under Linux. I ended up throwing out the notion of trying to
|
|
use libusb. After 4 hours of research and code hacking, it worked!
|
|
|
|
#### Status Update - Dec 08 2006
|
|
|
|
The hardware works and the software works (kinda). Once I get around to cleaning up
|
|
some of the code and adding some documentation, I'll start uploading stuff. Stay tuned!
|
|
|
|
#### Status Update - Dec 02 2006
|
|
|
|
Got the hardware working on a breadboard. Using DLP's test program, various patterns
|
|
were able to be sent. Video coming soon!
|
|
|
|
### Statistics:
|
|
* Cost: $27.69
|
|
* Lines of code (Linux): 83
|
|
* Lines of code (Windows): 1503
|
|
* Sleep lost: Unknown
|
|
|
|
### Hardware Design
|
|
The hardware is pretty simple. Using an UM245R, most of the work is done for you. The
|
|
[UM245R](https://www.ftdichip.com/Products/EvaluationKits/UM245R.htm) takes in USB data
|
|
and outputs it on the 8 data pins, and those 8 pins directly drive the LEDs. It's not
|
|
quite as simple as that, since there's all sorts of protocol with Ready-to-Read and
|
|
Ready-to-Write and Read and Write pins that go high and low. I just cheated and used
|
|
at 555 timer to generate a clock signal on the RD pin to give me the data. I just
|
|
lucked out and the UM245R outputs the last data if there's no new data available.
|
|
|
|
[Circuit Design](/media/img/cpu_meter/circuit.png)
|
|
|
|
[Testing the circuit](/media/img/cpu_meter/testing.jpg)
|
|
|
|
### Software
|
|
As with any hardware, there needs to be software which controls it. For the Windows
|
|
code, I decided to use [LibUSB](https://libusb.sf.net/) to help me with this project. Programming with LibUSB
|
|
is fairly straightforward, which helps since the documentation is rather spotty. Along
|
|
with LibUSB, I also took the Queue class from [nicklib](/projects/) and wrote a UsbDevice class to
|
|
help handle failures better. This stuff can be found in the source package above.
|
|
|
|
After wrangling with libusb on the Windows side of things, I decided to throw out
|
|
that idea on the Linux client. It turns out that FTDI makes a driver for the UM245R
|
|
called ftdi_sio which creates a virtual serial interface. I used this along with
|
|
libserial and libstatgrab to get it working.
|
|
|
|
The Linux code is rather simple and only does CPU usage. I'm planning on extending
|
|
it to do things such as music visualization. This of course means writing some sort
|
|
of user interface and probably using threads.
|
|
|
|
### Pictures
|
|
Some pictures of the final product:
|
|
|
|
 
|
|
|
|
 
|
|
|
|
|