Wait no longer! We’ve opened up PiCloud to another batch of users today, and from now onward, we promise to accelerate the roll out of PiCloud to new users. For users, both new and old, I wanted to highlight some of the many changes we’ve made in the past month that haven’t necessarily been the most visible.
Variable Compute Units
We had customers asking us for more powerful CPUs, and so we’ve delivered. With a simple keyword argument change, you can now switch between using 1 Compute Unit (1-1.2 ghz Xeon) to 2.5 Compute Units (2.5-3ghz Xeon). Check it out (code):
cloud.call(cpu_intensive_func, _high_cpu=True) # uses 2.5 compute units
Profiler Option
While we’ve gotten great feedback for profiling functions that run on PiCloud, we’ve also received requests to have the ability to turn off the profiler. After all, the deterministic profiler does have overhead that scales with the number of function calls in a script. To turn off the profiler, it’s simply another keyword argument _profile.
cloud.call(foo, _profiler=False)
Drop in for multiprocessing
If you’re already using Python multiprocessing, but want to run your computation across our cluster, now you can. Check out our docs to see how.
cloud library is now open source
We told users before that the client library was not open sourced, because frankly, we didn’t believe it was stable enough to deserve the attention of developers in the community. We are now at that point, so the client library has been released with an LGPL license.
Inclusion in the Enthought Python Distribution (EPD)
EPD is ideal for scientists and engineers looking for an easy, standardized way to deploy a powerful set of scientific tools on their own computer or across a whole organization. As of the latest EPD release, 6.0, the cloud library is now included in the distribution. Welcome EPD customers!
Bug fixes
Having hundreds of users using our platform is the easiest way to expose all the nitty-gritty bugs and race conditions that are lurking in our system. We would like to thank our ever-growing community for the many bug reports and critical fixes we have had over the past month.
Lastly, our CTO, Aaron Staley, and I will be at PyCon this weekend. Hope to see you all there!