Over a year ago, my dusty old server at Caltech finally went down for the last time. When I flew back to visit, I saw the hard drive had crashed and was beyond repair. RIP, faithful Pentium III box. Your years of free, tireless service will never be forgotten.
I've finally settled on a hosting service and have migrated over what survived. I'll be updating a redesigning this site to reflect things that I've done in the two years post-Tech.
I don't like running MATLAB's bloated java-based GUI, so I try to do everything from the command line (matlab -nojvm) when I can get away with it. Unfortunately, with R2010a for OS X, I've found MATLAB messes with my tty settings to where I can't see anything I type on the command line after quitting it. Thankfully, running:
$ stty sane
will undo the damage for now, and I hope MATLAB will fix this in the future.
I've been using some ATmega32U2s for a project I'm working on, and I was sick of running Windows just to program them using my AVRISP MKII in AVR Studio. Unfortunately, even the latest svn version of AVRDUDE doesn't yet have support for the 32U2. I put together a modified avrdude.conf
file that works for me, though. I made a few guesses on constants I couldn't find immediately from the datasheet, but the file works just fine for my AVRISP MKII.
You can download the file here
I was helping a friend with CUDA "out of memory" problems, and it reminded me of a trick I discovered by chance last year.
Of course, the amount of VRAM you have available to CUDA code on your system can be disappointingly small, so you might be getting these problems legitimately. But on several occasions I've had my code run out of memory before any of my own cuda_mallocs, or on my first, smallish cuda_malloc. I had a hunch that the OS might be caching way more crap in VRAM than it needs to, and that it would help to flush the cache somehow.
I ended up trying to change and reset my color depth or screen resolution.
After doing this, the CUDA code that really should have run to begin with ran perfectly!
I recently returned to CUDA for a project and found the new CUDA SDK from NVIDIA has some 32/64-bit library linking problems. It seems to be trying to compile projects as 64-bit even though the libraries are compiled for 32-bit.
You have a few options for how to fix this. At first I was just adding i386=1 to the end of all my make commands to force it into 32-bit compilation, but this was annoying. Then, I found out from http://docteurseb.com
that "Forcing $SNOWLEOPARD to a non null string in the Makefile should work as well."
I thought this was a bit strange, so I took a look at the NVIDIA-supplied common.mk and saw:
SNOWLEOPARD = $(strip $(findstring 10.6, $(shell egrep "<string>10\.6</string>" \
Silly NVIDIA! Updated systems will be version "10.6.*" not "10.6". So I fixed their regular expression and now everything works:
SNOWLEOPARD = $(strip $(findstring 10.6, $(shell egrep "<string>10\.6(.[0-9]+)?</string>" \
I just released a small open source utility for POSIX-compliant systems called into. This utility sidesteps the annoying impossibility of using a pipeline to update a file without first explicitly dumping to a temporary file.
Read more about into here
or skip straight to downloading it here
Were it not for the campus police, Caltech would have converted MIT into our East Campus--a place for techers to study the humanities. Nevertheless, the website I made for Caltech's East campus can be found at east.caltech.edu.
A mirror can be found on this server here
The PNM image format is really useful when developing quick little image processing utilities, but there's very little support for it built into Mac OS X. To help remedy this problem, I have just released a quick look plugin for the PNM image format.
Check it out on the software page
, or jump straight to downloading it here
I just added a blog
to this site. I intend to use it to document solutions to various computer and programming-related problems I encounter or just cool things I come across. It happens too many times that I'll finally solve a problem only to run into it again a year or so later after forgetting the solution.
Who knows? Maybe with a little help from Google, others can find some use in it!
I finally got around to installing 10.6 on my laptop after having run it on my mini for some time. But this time, things didn't go quite as smoothly. The installer claimed it couldn't install on my hard drive because the drive couldn't be used to boot my computer.
I've seen this problem reported elsewhere, but for me, fixing it turned out to be pretty easy. I booted from the install DVD and repaired my boot partition with Disk Utility. While Disk Utility didn't notice any problems, at the end of its tests it said it was "updating boot support partitions for the volume as required." That sounded promising, and sure enough the install worked when I tried it again.
I have made a few changes to this website that hopefully will do something about the comment spam that has been plaguing it. I guess I'll just have to wait and see.
I spent a few days over winter break updating Graphite. I implemented several new features, but I still have a bit more testing I want to do. Unfortunately classes started a few weeks ago, and this term is shaping up to be quite difficult. Hopefully I'll get a chance one of these weekends.
One of the classes I'm taking, CS176
, involves weekly implementations of various graphics algorithms. I have posted a webpage for each of the first two assignments that I've finished:
CS176 HW1: Seam Carving
CS176 HW2: Texture Synthesis
The remaining assignments will be linked from my school page
when they are completed.
Apple just approved my graphing calculator, Graphite, for sale on the iTunes App Store. Read more about it here
I've finally finished Graphite, my graphing calculator for the iPhone and iPod touch. I'm just waiting now for it to be accepted to the iTunes application store. Check out the page for it
in the software section!
Ever since Caltech Housing/IMSS's epic failure at upgrading our ethernet cables, my server has been down. Lloyd's radio (server) room was without internet for several weeks. But, when the internet came back, my server didn't come back with it. I finally had some time tonight to investigate, and it turns out there were two problems: my server had been turned off (oops!), and the network switch to which it was connected wasn't plugged into the wall. At least it's back up now!
I finally have gotten around to writing a new website for myself. There is basically no content at the moment, but I will eventually add information on some (EE) projects I have done. I also plan to upload software I have written and interesting (?) school assignments.
Design and code ©2012 Julian Panetta