cl-waffe2 has established the following step to achieve DAG-specific acceleration.
1. [Constructing DAG networks by defnode/call/forward]
-> when called with build
2. [Applying a topological sort to the given forward/backward networks, later applying in-place mutation]
3. [Generated cl-waffe2 IR for forward/reverse mode]
4. [Compiles each nodes with using cache.lisp, to each rank/type/layout of matrices]
5. [If any, applying JITxxTensor devices] (Undone)
-> when called with proceed
2. [Applying In-place mutation within small overheads]
3. [Evaluates the computation node directly]
Things undone
define-composite-node is no longer working with the new VM. Keep using old one?
With according to the changes of IR, JITCPUTensor/JITLispTensor is no longer working. (with a few adjustments, it should be work.)
It is possible to do FuseOps without JITxxTensor devices.
Changes
Introducing cl-waffe2 IR
Now,
Compiled-Composite
is working undercl-waffe2 VM
which interpretscl-waffe2 IR
.cl-waffe2 has established the following step to achieve DAG-specific acceleration.
Things undone
define-composite-node
is no longer working with the new VM. Keep using old one?JITCPUTensor
/JITLispTensor
is no longer working. (with a few adjustments, it should be work.)FuseOps
without JITxxTensor devices.call-with-view