msj
Researcher
- Joined
- 5/1/07
- Messages
- 27
- Points
- 13
I have now done release 0.3 of Kooderive.
For those not familiar, this is an open source project for pricing derivatives using Monte Carlo simulation.
It contains implementations in both C++ and CUDA to allow running on either CPU or GPU.
The flagship example is the pricing of a 40 rate cancellable swap using the LMM with a million paths in roughly a second.
There is a paper at
http://ssrn.com/abstract=2388415
We provisionally plan to run a course later in the year.
The main innovation in the new release is that the code adapts to the user's GPU. This should make it easier to use for those who don't have access to expensive GPUs. It is also no longer dependent on CULA.
The main requirement is to have a GPU of compute capability at least 1.3 and to have the NVIDIA CUDA 5.5 toolkit installed.
For those not familiar, this is an open source project for pricing derivatives using Monte Carlo simulation.
It contains implementations in both C++ and CUDA to allow running on either CPU or GPU.
The flagship example is the pricing of a 40 rate cancellable swap using the LMM with a million paths in roughly a second.
There is a paper at
http://ssrn.com/abstract=2388415
We provisionally plan to run a course later in the year.
The main innovation in the new release is that the code adapts to the user's GPU. This should make it easier to use for those who don't have access to expensive GPUs. It is also no longer dependent on CULA.
The main requirement is to have a GPU of compute capability at least 1.3 and to have the NVIDIA CUDA 5.5 toolkit installed.