Researchers from the Technical University of Munich (TUM) have announced TPDE as a fast and adaptable compiler back-end framework. [image] TPDE: A Fast Adaptable Compiler Back-End Framework Fast machine code generation is especially important for fast start-up just-in-time compilation, w…

Beat me to it… had the same thought. Since something like this works with LLVM at some level, how much effort would be required to “drop it in?”

The primary goal is low-latency compilation while maintaining reasonable (-O0 ) code quality, e.g., as baseline compiler for JIT compilation or unoptimized builds. Currently, TPDE only targets ELF-based x86-64 and AArch64 (Armv8.1) platforms. TPDE seems to be a research tool/benchmark, not aimed…

[image] xiaoxi: TPDE-LLVM: 10-20x Faster LLVM -O0 Back-End - Code Generation - LLVM Discussion Forums [image] xiaoxi: I wonder if it could be useful to Julia in the future. not really if -O0 stays where it is relative to -O3 in terms of performance.

In addition to the limited number of targets, do other unsupported features affect Julia? Targets other than x86-64-v1/AArch64 (ARMv8.1) (Linux) ELF. Code models other than Small-PIC. Scalar types: integer types larger than i64 except i128 (i128 is supported), pointers with non-zero address space,…

Of these the ones that would be an issue: Scalar types: integer types larger than i64 except i128 (i128 is supported), pointers with non-zero address space, half, bfloat Calling conventions other than the C calling convention ( Many intrinsics Non-empty inline assembly.

not really if -O0 stays where it is relative to -O3 in terms of performance. What about -O1? The report claims similar speedups for that optimization level aswell.

Also cross-posting from the other thread I opened ( TPDE-LLVM for 10-20x faster -O1 compilation ) to provide some more links: I just saw this post on hackernews, where an LLVM compiler backend ( github ) was posted that claims to get 10-20x compile time speedup compared to an LLVM 19 baseline. Quotin…

-O1 doesn’t vectorise: % julia -O1 -q julia> code_llvm((Float64, Vector{Float64}, Vector{Float64}); debuginfo=:none) do a, x, y for idx in eachindex(x, y) y[idx] = muladd(a, x[idx], y[idx]) end end ; Function Signature: var"#2"(Float64, Array{Float64, 1},…

Okay, but I think there are many uses cases where vectorization doesn’t matter as much, and TTFX dominates the overall user time.

TPDE: A Fast Adaptable Compiler Back-End Framework

Offtopic

Palli September 3, 2025, 3:21pm 12

It might be easier to use this TPDE-LLVM (for LLVM compatibility), or as I suggest, bypass LLVM (like Zig is doing; there no longer default):

Topic		Replies	Views
TPDE-LLVM for 10-20x faster -O1 compilation Internals & Design	3	381	September 3, 2025
Would it be possible to write LLVM in Julia? General Usage	6	877	September 3, 2025
Cranelift: a faster alternative to LLVM Offtopic	15	12033	March 19, 2024
Why are compiler devs in such high demand for Julia? General Usage question , compilation	41	3389	November 13, 2024
Julia LLVM projects? Internals & Design	31	7412	April 22, 2020

TPDE: A Fast Adaptable Compiler Back-End Framework

Related topics