From: vmakarov@... Date: 2018-02-19T21:32:11+00:00 Subject: [ruby-core:85657] [Ruby trunk Feature#12589] VM performance improvement proposal Issue #12589 has been updated by vmakarov (Vladimir Makarov). Last 4 months I've been working on generation of RTL from stack insns. The reason for this is that stack insns are already a part of CRuby. The current generation of RTL directly from the nodes actually would remove this interface. Another reason for this work is to simplify future merging RTL and MJIT branches with the trunk. I think I've reached a project state when I can make the branch public. But still there are lot of things to do for this project. Generation of RTL from stack insns is a harder task than one from nodes. When we generate RTL from nodes we have a lot of context. When we generate RTL from stack insns we need to reconstruct this context (from different CFG paths in a stack insn sequence). To reconstruct the context we emulate VM stack and can pass a stack insn sequence several times. First we calculate possible stack values on the label. It is a typical forward data flow problem in compilers (the final fixed point is only temporaries on the emulated stack). Then using this info we actually generate RTL insns on the last pass. I was afraid that stack insn -> RTL generation might considerably slow done CRuby. Fortunately, it is not the case. This generation for optcarrot with one frame (it means practically no execution) takes about 0.2% of all CRuby run time. Running empty script takes about 2% more in comparison with using direct generation of RTL from the nodes. I created a new branch in my repository for this project. The branch name is `stack_rtl_mjit` (https://github.com/vnmakarov/ruby/tree/stack-rtl-mjit-base). All my work including MJIT will be done on this branch. The previous branch `rtl_mjit_branch` is frozen. The major code to generate RTL from stack insns is placed in a new file rtl_gen.c. I am going to continue work on this branch. My next plans will be a merge with the trunk and fixing bugs. It is a big job as the branch is based on more than one year old trunk. There were a lot of changes since then which will affect the code I am working on. The biggest one is Takashi Kokubun's work on MJIT for YARV. Another one is trace insns removal by Koichi Sasada. I am planning to work on merging with trunk, unification of MJIT code on trunk and the branch, and fixing bugs till April/May. Sorry for a slow pacing but I have no much time for this work until gcc-8 release (probably middle of April). After that I am going to work on MJIT optimizations including method inlining. ---------------------------------------- Feature #12589: VM performance improvement proposal https://bugs.ruby-lang.org/issues/12589#change-70448 * Author: vmakarov (Vladimir Makarov) * Status: Open * Priority: Normal * Assignee: * Target version: ---------------------------------------- Hello. I'd like to start a big MRI project but I don't want to disrupt somebody else plans. Therefore I'd like to have MRI developer's opinion on the proposed project or information if somebody is already working on an analogous project. Basically I want to improve overall MRI VM performance: * First of all, I'd like to change VM insns and move from **stack-based** insns to **register transfer** ones. The idea behind it is to decrease VM dispatch overhead as approximately 2 times less RTL insns are necessary than stack based insns for the same program (for Ruby it is probably even less as a typical Ruby program contains a lot of method calls and the arguments are passed through the stack). But *decreasing memory traffic* is even more important advantage of RTL insns as an RTL insn can address temporaries (stack) and local variables in any combination. So there is no necessity to put an insn result on the stack and then move it to a local variable or put variable value on the stack and then use it as an insn operand. Insns doing more also provide a bigger scope for C compiler optimizations. The biggest changes will be in files compile.c and insns.def (they will be basically rewritten). **So the project is not a new VM machine. MRI VM is much more than these 2 files.** The disadvantage of RTL insns is a bigger insn memory footprint (which can be upto 30% more) although as I wrote there are fewer number of RTL insns. Another disadvantage of RTL insns *specifically* for Ruby is that insns for call sequences will be basically the same stack based ones but only bigger as they address the stack explicitly. * Secondly, I'd like to **combine some frequent insn sequences** into bigger insns. Again it decreases insn dispatch overhead and memory traffic even more. Also it permits to remove some type checking. The first thing on my mind is a sequence of a compare insn and a branch and using immediate operands besides temporary (stack) and local variables. Also it is not a trivial task for Ruby as the compare can be implemented as a method. I already did some experiments. RTL insns & combining insns permits to speed the following micro-benchmark in more 2 times: ``` i = 0 while i<30_000_000 # benchmark loop 1 i += 1 end ``` The generated RTL insns for the benchmark are ``` == disasm: #@while.rb>====================================== == catch table | catch type: break st: 0007 ed: 0020 sp: 0000 cont: 0020 | catch type: next st: 0007 ed: 0020 sp: 0000 cont: 0005 | catch type: redo st: 0007 ed: 0020 sp: 0000 cont: 0007 |------------------------------------------------------------------------ local table (size: 2, temp: 1, argc: 0 [opts: 0, rest: -1, post: 0, block: -1, kw: -1@-1, kwrest: -1]) [ 2] i 0000 set_local_val 2, 0 ( 1) 0003 jump 13 ( 2) 0005 jump 13 0007 plusi , 2, 2, 1, -1 ( 3) 0013 btlti 7, , -1, 2, 30000000, -1 ( 2) 0020 local_ret 2, 0 ( 3) ``` In this experiment I ignored trace insns (that is another story) and a complication that a integer compare insn can be re-implemented as a Ruby method. Insn bflti is combination of LT immediate compare and branch true. A modification of fib benchmark is sped up in 1.35 times: ``` def fib_m n if n < 1 1 else fib_m(n-1) * fib_m(n-2) end end fib_m(40) ``` The RTL code of fib_m looks like ``` == disasm: #========================================== local table (size: 2, temp: 3, argc: 1 [opts: 0, rest: -1, post: 0, block: -1, kw: -1@-1, kwrest: -1]) [ 2] n 0000 bflti 10, , -1, 2, 1, -1 ( 2) 0007 val_ret 1, 16 0010 minusi , -2, 2, 1, -2 ( 5) 0016 simple_call_self , , -1 0020 minusi , -3, 2, 2, -3 0026 simple_call_self , , -2 0030 mult , -1, -1, -2, -1 0036 temp_ret -1, 16 ``` In reality, the improvement of most programs probably will be about 10%. That is because of very dynamic nature of Ruby (a lot of calls, checks for redefinition of basic type operations, checking overflows to switch to GMP numbers). For example, integer addition can not be less than about x86-64 17 insns out of the current 50 insns on the fast path. So even if you make the rest (33) insns 2 times faster, the improvement will be only 30%. A very important part of MRI performance improvement is to make calls fast because there are a lot of them in Ruby but as I read in some Koichi Sasada's presentations he pays a lot of attention to it. So I don't want to touch it. * Thirdly. I want to implement the insns as small inline functions for future AOT compiler, of course, if the projects described above are successful. It will permit easy AOT generation of C code which will be basically calls of the functions. I'd like to implement AOT compiler which will generate a Ruby method code, call a C compiler to generate a binary shared code and load it into MRI for subsequent calls. The key is to minimize the compilation time. There are many approaches to do it but I don't want to discuss it right now. C generation is easy and most portable implementation of AOT but in future it is possible to use GCC JIT plugin or LLVM IR to decrease overhead of C scanner/parser. C compiler will see a bigger scope (all method insns) to do optimizations. I think using AOT can give another 10% improvement. It is not that big again because of dynamic nature of Ruby and any C compiler is not smart enough to figure out aliasing for typical generated C program. The life with the performance point of view would be easy if Ruby did not permit to redefine basic operations for basic types, e.g. plus for integer. In this case we could evaluate types of operands and results using some data flow analysis and generate faster specialized insns. Still a gradual typing if it is introduced in future versions of Ruby would help to generate such faster insns. Again I wrote this proposal for discussion as I don't want to be in a position to compete with somebody else ongoing big project. It might be counterproductive for MRI development. Especially I don't want it because the project is big and long and probably will have a lot of tehcnical obstacles and have a possibilty to be a failure. -- https://bugs.ruby-lang.org/ Unsubscribe: