Jakob> David Gregg (and friends) recently published a paper comparing
Jakob> stack based and register based VMs for Java and found that
Jakob> register based VMs were substantially faster. The main reason for
Jakob> this appears to be the absence of the various LOAD_ instructions
Jakob> in a register VM.
Back in the day (2001?) with the help of Tim Peters and others here I wrote
most of what was necessary to convert from a stack-based to a register-based
virtual machine for CPython. I had trouble with a couple VM instructions.
If memory serves correctly, exec was a problem because it could push a
variable number of results on the stack. I never completed the project
because the instruction set kept growing and I couldn't keep up. Still, it
looked promising because it avoided lots of intermediate loads and stores.
Jakob> ``Cell'' is, unfortunately, hardcoded into Vmgen. I could either
Jakob> patch that or run ceval-vm.i through sed or something.
If the patch is to work and eventually be accepted it will have to be
changed, probably with a little downstream find/replace. You might want to
also file a bug report to the Vmgen folks. If Vmgen is generating these
names the authors should really know to try to avoid name clashes.
Skip
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4