Direct execution? With GPUDirect from NVIDIA you could have an FPGA get a memcpy() from graphics RAM to its data has. Still, not sure how you plan on using this. I would hope that NVIDIA could give us programmable fabric on the GPU silicon, and it could somehow access each compute units instruction execution pipeline. Would be interesting indeed. Or just massive, fast FPGAs with included highly parallel compute IP built in. Or all new NVIDIA cards just ship a very fast FPGA that runs 3D graphics by default, but can be reprogrammed to do CUDA compatible computation and its open source so users can add new instructions or modify existing ones or even reconfigure the number or arrangement of cores.
Thinking something along the lines of, implementing/modelling SQL operators in FPGA hardware, using massive parallelization through the FPGA. That needs direct I/O access to very fast storage to be optimal, and probably still additional processing via a CPU or GPU.
Expensive custom data warehouse appliances like Netezza already do something like this, but I'm hoping for an implementation that is less expensive and more generally available, that can be exploited to accelerate any database's operations. Bonus points if it can scale up to a distributed model somehow.
