PhoenixMiner 2.7c is officially released. In addition to the changes in 2.7a, and 2.7b, we have added support for
secure stratum connections using SSL (supported by ethermine.org) to prevent the increasing IP hijacking attacks. To connect to a secured pool, use the
ssl:// prefix (e.g.
-pool ssl://eu1.ethermine.org:5555). We have also Added support for
solo mining (HTTP GetWork protocol).
We are switching our attention to 2.8, which will include a lot of small improvements, as well as some big ones but we prefer to keep them confidential for now in order to avoid giving any ideas to the competition

We also continue working towards dual mining and Linux support as our next major milestones.
By "competition" you mean those who you steal code from?
Let's sum up. Claymore claims phoenix stole the claymore V10 (if I remember correctly) open GL kernels. Then Claymore has updated his miner to 10.3, 10.6, 11, 11.1, 11.2 (did I miss a release here ?), claiming a speed increase for each new release. Still, phoenix 2.6/2.7c seems as fast as Claymore 11.2, despite outdated kernels, because claymore claims the phoenixminer 2.6 still has Claymore V10 unmodified kernels.
I find this weird that a miner with alledged outdated kernels can be as fast as the latest Claymore. If it was just a "stolen" code with no dev, it wouldn't be as fast IMO.
There was not speed update every release, most were feature or bug fixes. Claymore Eth speed improvements are rare, speed has barely changed in over a year, the recent -asm 2, slightly less stales, and an Nvidia speedup are all I can remember, and I have followed it closely. Most releases are features, or different dual mine coins. And displayed speed isn't necessarily effective speed.
How do you explain kernel dump confirming it?
Also, it's OpenCL, not open GL, which is a graphics library for games.