Advanced LLVM : compile, link, optimize


Yesterday and Today were interesting days for Clover : I read big parts of the LLVM documentation and implemented the first bits of the OpenCL standard library.

Clover is now able to compile a program and to link it with the standard library (if the device needs that, like CPUDevice). Some work was also done to make Clover more standard-compliant regarding the devices and the programs. The code is now relatively clean I think.

Now that I am on holidays, I can work hours and hours a day (I worked today from 1 pm to now, 10 pm, with some pauses. I think it’s nearly 6 hours a day) and Clover advances fast.

The next step now that compiling and linking work is to implement the kernel functions (a kernel is a simple function with a __kernel attribute, and means to set its arguments for execution on a GPU). Then, it will be the first main target : having some small compiled kernels running using Clang.

After that, I will have to implement the standard library and all its functions (from the simple geometric ones to the tricky ones like barrier() that need tight integration with the host CPU). I have more than a month and a half to do that, so I hope it will suffice.


One response to “Advanced LLVM : compile, link, optimize

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: