Fix8 is a modern open source C++ FIX framework featuring complete schema driven customisation, high performance, and fast application development. The system is comprised of a compiler for generating C++ message and field encoders, decoders, and instantiation tables, a runtime library to support the generated code and framework, and a set of complete client/server test applications. For the same message, Fix8 encodes 2.8 times faster and decodes 1.6 times faster than Quickfix, an average of 2 times. In other words, it reduces encode latency by 64% and reduces decode latency by 36%.
|Tags||framework C++ Finance API high-performance FIX protocol adaptor|
|Operating Systems||Linux (32 and 64 bit)|
We're planning to go GA in a few months. There will probably be one or two more releases as beta and then version 1.0.0 will follow.
Release Notes: Performance and reliability were improved. This release is 50% faster than QuickFix for the same messages. Gtest unit testing is now built in. A scriptable test application that permits test data playback was included. Numerous bugs were fixed, and other improvements were made.
Release Notes: This release removes a dependency on TBB. It will supply and use FastFlow. It should run on non-intel hardware. Stack based messages. Framework pipelining control, which can result in significant improvements in performance. A persistence sequence number edit utility 'seqedit'. Bugfixes and code changes. Jira tickets have been fixed.
Release Notes: Indexed file based message persistence is now the default. BerkeleyDB no longer required to build. A fix8.spec.in file which configure will use to generate fix8.spec was added. RPM and Deb binaries will now be released with each tagged release. Persistence and logfiles now use SessionIDs to create unique sessions. Server files are created on logon. Bugs were fixed.
Release Notes: The big news here is performance. Testing shows that Fix8 is on average 29% faster at encoding/decoding the same message than Quickfix. Typical client NewOrderSingle encode latency is now 9.8µs, and ExecutionReport decode is 11.6µs.
Release Notes: Support for clang (LLVM) compilation, and fixes for bugs that may have prevented users from executing performance tests.