aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2015-06-02turn off eliminate-temps until improvedjackbackrack
2015-06-02merge + fix trim to use correct bits operandsjackbackrack
2015-05-29fix concat, as-sint, turn off temp-eliminationjackbackrack
2015-05-29Fixed bugs in when-coverage pass. Works but has not been thoroughly testedazidar
2015-05-29Added new stanzaazidar
2015-05-29Added custom pass. Does not correctly run, stanza just spins. Requires ↵azidar
debugging.
2015-05-27Added sequential memories. mem no longer exists, must declare either cmem or ↵azidar
smem. Added firrtl-gensym utility to generate a hashmap of names
2015-05-27Added external modules. Switched lower firrtl back to wire r; r := Register, ↵azidar
instead of using nodes. Added a renaming pass for different backends. This will likely get deprecated, as a more robust name mangling scheme could be needed
2015-05-26Added <>. Added additional checks for primops. Added new chisel3 files.azidar
2015-05-21fix pad/trim pass and fix bug in bits-select width inferencejackbackrack
2015-05-21Added pad pass, used for flo backendazidar
2015-05-20Merge branch 'master' of github.com:ucb-bar/firrtlazidar
2015-05-20Merge pull request #9 from jackbackrack/masterAdam Izraelevitz
fix flo writeport emission to new style
2015-05-20fix writeport emission for flojackbackrack
2015-05-20mergejackbackrack
2015-05-20Added Pad pass to flo.stanza, which pads widths to make := and primops ↵azidar
strict. Have not tested this
2015-05-19Merge branch 'master' of github.com:ucb-bar/firrtlazidar
2015-05-19Updated testsazidar
2015-05-19Merge pull request #8 from jackbackrack/masterAdam Izraelevitz
new flo backend without pads but with generic ops
2015-05-19mergejackbackrack
2015-05-19get flo backend running again with no pads and generic operatorsjackbackrack
2015-05-19Added support for non-inlined modules in verilog backendazidar
2015-05-18get coercion running for flo backend and disable negative lit checkjackbackrack
2015-05-18First pass at a Verilog Backend. Not tested, but compiles and generates ↵azidar
reasonable verilog. Requires inlining, future versions will instantiate modules
2015-05-18Big API Change. Pad is no longer supported. Widths of primops can be ↵azidar
flexible, and the output is usually the max of the inputs. Removed all u/s variants, which need to be dealt with in backends where it matters
2015-05-15Updated firrtl for its passes to be a bit more modular, and to enable ↵azidar
plugging in other backends. Also updated a lot of tests, but not all of them because its annoying.
2015-05-14mergejackbackrack
2015-05-13Added source indicators from FIRRTL files. Pass in -p i to get them printed. ↵azidar
Should show up with check passes
2015-05-13Updated Spec. Added scoped-reg which exposes on-reset bug. Fixed lowering bugazidar
2015-05-12mergejackbackrack
2015-05-07do signed padding as welljackbackrack
2015-05-05Added a bunch of tests. In the middle of implementing check kinds and check ↵azidar
types. Does not compile
2015-05-04Added new Control.fir with reduced paddingazidar
2015-05-04mergejackbackrack
2015-05-04Added new stanzaazidar
2015-05-04Fixed bug where instance types were not loweredazidar
2015-05-04mergejackbackrack
2015-05-04Merge branch 'master' of github.com:ucb-bar/firrtlazidar
2015-05-04Updated stuffazidar
2015-05-04add reduction operatorsjackbackrack
2015-05-04Merge pull request #6 from jackbackrack/masterAdam Izraelevitz
firrtl + flo now working with tutorial examples
2015-05-04Added a few more error checks. Not tested yet. Fixed bug in pad type inferenceazidar
2015-05-04mergejackbackrack
2015-05-04Fixed change where type of mux-ss was incorrectazidar
2015-05-02mergejackbackrack
2015-05-02Added a infrastructure for check passes, and wrote a fewazidar
2015-05-02Now when expanding ConnectFrom/ToIndex, create a node for the index so it ↵azidar
isn't duplicated for all the whens
2015-05-01add arsh refsjackbackrack
2015-05-01Bug fix. ExpWidth was improperly evaluated during simplify (not subtracting one)azidar
2015-05-01Fixed performance bug where PlusWidth, MinusWidth, and ExpWidth could be ↵azidar
simplified earlier, and also now have equal? defined so mMaxWidth doesn't blow up during width inference