Lava: Change Recipe to leverage the "do" syntax

Fx, before:

shiftAndAdd s =
  While (s.b.val =/= 0) $
    Seq [ s.a <== low `vshr` (s.a.val)
        , s.b <== (s.b.val) `vshl` low
        , s.b.val.vhead |>
            s.result <== s.result.val + s.a.val
        , Tick
        ]

now

shiftAndAdd s = do
    while (s.b.val =/= 0) $do
       s.a <== low `vshr` (s.a.val)
       s.b <== (s.b.val) `vshl` low
       iff (s.b.val.vhead) $do
           s.result <== s.result.val + s.a.val
       tick

Note, this is a purely syntactic change, the effective code is exactly
the same as before.  This is but a step on the path to making
sequential logic more readable and mallable.

README.md

@@ -86,44 +86,43 @@ To build a hardware version of a given test
     cd fpga; make && flite -r ../programs/$P | ./Red -v
 
 where $P is one of the programs (.hs).  Next, build a Reduceron system
-for DE2-115:
+for an FPGA board, fx the BeMicroCV A9:
 
-    make -C Reduceron/DE2-115
+    make -C Reduceron/BeMicroCV-A9
 
 Unfortunately programs can't currently be loaded dynamically but are
 baked into the FPGA image.  It's a high priority goal to change that.
 
 ## WHERE IS THIS GOING?
 
-Plan:
+### Plan ###
 
-  1. Port to Verilog and Altera. DONE!
+  1. Port to Verilog and remove Xilinx-isms. DONE!
 
-  2. Shrink to fit mid-sized FPGA kits (eg. DE2-115). DONE!
+  2. Shrink to fit mid-sized FPGA kits (eg. DE2-115 and BeMicroCV-A9).
+     DONE!
 
-  3. Support load/store to an external bus (the key difficulty is stalling while waiting on the bus).
+  3. Rework Lava and the Reduceron implementation to be more
+     composable and elastic; this means fewer or no global assumptions
+     about timing.  ONGOING!
+
+  4. Support load/store to an external bus (the key difficulty is
+     stalling while waiting on the bus).
 
-  4. Use the program memory as a cache, making programs dynamically loadable and dramatically raise the size limits.
+  5. Use the program memory as a cache, making programs dynamically
+     loadable and dramatically raise the size limits.
 
-In some order:
-
-  - Add basic primitives (`(*)`, `(.&.)`, `(.|.)`, `(.^.)`, `(.<<.)`, `(.>>.)`, ...)
-    DONE: `(.&.)`
-
-  - Improve Lava simulation and RTL gen
-       - Add a C backend for much faster simulation
-       - Use names from the design
-       - generate busses where possible
-       - Make the design resettable
+### Eventual Plan ###
 
   - Move the heap [and tospace] to external memory
   - Add a heap cache/newspace memory
   - Implement the emu-32.c representation for the external heap
-
+  - Much richer primitives
   - Haskell front-end
 
-Finally: Performance improvements
+### Long Term Plan ###
 
+  - Research the design space; explore parallelism
 
 ## OPEN QUESTIONS, with answers from Matthew:
 

fpga/CachingOctostack.hs

@@ -24,6 +24,7 @@ import Lava
 import Recipe
 import Data.List
 import Bytecode
+import Control.Monad (zipWithM_)
 
 -- Parameters
 
@@ -125,33 +126,33 @@ rotLeft :: [Bit] -> [Word n] -> [Word n]
 rotLeft n xs = map Vec $ rotateLeft n $ map velems xs
 
 update :: Word N4 -> Word N8 -> [Word n] -> Octostack n -> Recipe
-update o en xs s = Seq ([s.offset <== o, s.writeEn <== en] ++ assigns)
-  where assigns = zipWith (<==) (s.dataIn.velems) xs
+update o en xs s = do s.offset <== o
+                      s.writeEn <== en
+                      zipWithM_ (<==) (s.dataIn.velems) xs
 
 -- Example usage
 
 example :: Octostack N18 -> Recipe
-example s =
-  Seq [ s.update 3 (high +> high +> high +> 0) [4, 5, 6]
-      , Tick -- PUSH [4, 5, 6]
-      , s.update 2 (high +> high +> 0) [8, 9]
-      , Tick -- PUSH [8, 9]
-      , s.update (-1) 0 []
-      , Tick -- POP 1
-      , s.update 4 (high +> high +> high +> high +> 0) [1, 2, 1, 2]
-      , Tick -- PUSH [1, 2, 1, 2]
-      , s.update 4 (high +> high +> high +> high +> 0) [6, 7, 6, 7]
-      , Tick -- PUSH [6, 7, 6, 7]
-      , s.update (-3) 0 []
-      , Tick -- POP 3
-      , s.update (-2) (high +> 0) [10]
-      , Tick -- POP 3, PUSH [10]
-      , s.update 2 (high +> high +> 0) [8, 9]
-      , Tick -- PUSH [8, 9]
-      , s.update 0 (high +> 0) [55]
-      , Tick -- PUSH [8, 9]
-      , Tick
-      , Tick
-      ]
+example s = do s.update 3 (high +> high +> high +> 0) [4, 5, 6]
+               tick -- PUSH [4, 5, 6]
+               s.update 2 (high +> high +> 0) [8, 9]
+               tick -- PUSH [8, 9]
+               s.update (-1) 0 []
+               tick -- POP 1
+               s.update 4 (high +> high +> high +> high +> 0) [1, 2, 1, 2]
+               tick -- PUSH [1, 2, 1, 2]
+               s.update 4 (high +> high +> high +> high +> 0) [6, 7, 6, 7]
+               tick -- PUSH [6, 7, 6, 7]
+               s.update (-3) 0 []
+               tick -- POP 3
+               s.update (-2) (high +> 0) [10]
+               tick -- POP 3, PUSH [10]
+               s.update 2 (high +> high +> 0) [8, 9]
+               tick -- PUSH [8, 9]
+               s.update 0 (high +> 0) [55]
+               tick -- PUSH [8, 9]
+               tick
+               tick
+
 
 simExample = simRecipe (newOctostack "example_" id) example tops