For an overview of the research space, see:

• this survey article
• this blog post
• this video (this is a session on verifiable computation at the 2013 Microsoft Research Faculty Summit, hosted by Bryan Parno)

We know of three projects that are related to ours:

• CMT [see here and here]
  

  Refines a non-cryptographic interactive proof protocol. Can achieve much lower overhead than the other approaches but requires that computations be naturally parallelizable.
• GGPR and Pinocchio [project page]
  

  Introduces the remarkable QAP/QSP encoding, which is borrowed by many of the projects in the area. Pinocchio and Zaatar have similar performance except that Pinocchio has much better amortization behavior.
• BCTV, BCGTV, and TinyRAM [project page]
  

  Introduces an innovative program representation and combines it with an optimized implementation of Pinocchio's verification engine. This approach brings excellent programmability (enabling support of the full C programming language) and the best amortization behavior in the literature but overhead is very high. Compared to Buffet, for example, programmability is slightly better but performance is orders of magnitude worse.

For a quick performance comparison, based on re-implementations of many of the systems, see this presentation (accompanies the Microsoft Faculty Summit video above). Also, the Buffet paper contains a careful, apples-to-apples performance comparison of BCTV and Pantry (and by extension, Pinocchio and Zaatar).