model.bricks
we are pleased to announce that model.bricks v 3.0 is now in limited beta! Enriched functionality, multi-core and GPU accelerated performance and an open, accelerator-independent API to support bespoke FPGA's / specialized accelerators..
the open modeling framework provides a simple path to integrating existing or bespoke models . Securities valuation in addition to signal generating models can be implemented for use in risk applications, or to implement HF or lower frequency systematic strategies.
the 3.0 feature set includes:
- Thread-safe, extensible Spline & surface patterns - NEW
- Root finding, parallel numerical integration
- Linear Algebra (vector / matrix operations, SVD)
- Parallel convolution, DFT, auto-/cross-correlation, pattern searching - NEW
- Parallel Statistical methods
- Ordinary Least Squares (linear, logarithmic, exponential) - NEW
- Parallel monte-carlo framework - NEW
- Schedule generation
- Extensible derivatives model framework
- A rich and growing set of instruments / asset classes
speed
with CPU and GPU accelerated computations within a single framework, v 3.0 opens up an organic path to GPU acceleration for existing custom analytics. Depending on context and problem size, different parallelization strategies can be chosen for fastest performance. To illustrate, at 50,000 yield curve fits / second; model.bricks brings high-frequency to fixed-income.
future-proof
growing performance demands, and power constraints, have brought about an emerging diversity of hardware choices. The challenge for the writers of high performance software is daunting - taking advantage of any particular hardware engine often requires significant re-writes. With a simple and elegant template framework, model.bricks takes a step towards making the implementation of mixed-mode algorithms easy, dramatically reducing time to market.
evaluate model.bricks today; unleash your imagination. Complte with an unfair advantage.
download model.bricks pdf