similar to: [LLVMdev] Passing structures by value on Windows

---------- Forwarded message ---------- From: lost <lostfreeman at gmail.com> Date: 2010/6/5 Subject: Re: [LLVMdev] Passing structures by value on Windows To: Anton Korobeynikov <anton at korobeynikov.info> Ok, but how to bitcast structure? Documentation<http://llvm.org/docs/LangRef.html#i_bitcast>says that bitcast does not applicable to aggregate values. 2010/6/3 Anton

Hello! I'm looking at "Overview of available features" here: http://llvm.org/docs/GarbageCollection.html#collector-algos and can't understand something. First, does table header mean that there are already some GC's implemented besides "shadow stack"? E.g. can I just use them? Second, does row "JIT", "NO" mean that GC is not compatible with

Hi Victor, You can write your own GC or use other's GC with LLVM. What LLVM provides is a framework to generate a representation of objects locations in a method's frames. Right now, LLVM can emit the shadow stack (which dynamically updates the locations), or the ocaml format. If you have implemented a GC, you can parse the ocaml format to locate the objects. I think the web page needs

Hi Rotem, Thank to you, and especially to Marina! The problem gone. I'm a bit interested, what is the reason it worked in Win7, and not in Win8. I've recently used Process Explorer to discover, that the call was to ntdll.dll, which in Win8 is loaded to the totally different address. Best regards, Victor Milovanov Moscow State University graduate student 2012/3/3 Rotem, Nadav

On Windows, the LLVM JIT runner looks for the '_chkstk' symbol by enumerating all of the loaded DLLs. On Win8, NTDLL.DLL (where _chkstk is defined) is found in a location that is more than 32bits bytes away from the jitted code. Marina's patch changes the code that generates a call to '_chkstk' from PCREL32 (which uses a 32bit offset) to an indirect call (which uses a 64bit

On Mon, Aug 31, 2015 at 5:52 PM, Jake VanAdrighem <jvanadrighem at gmail.com> wrote: > Do you have some specific performance measurements? Averaging 4 runs of 10000 iterations each of Coremark on my X86_64 desktop showed: -O2 performance: +2.9% faster with the L.E.V. pass -Os size: 1.5% smaller with the L.E.V. pass In the case of Coremark, the benefit comes mainly from the matrix

Hello LLVM, This is a proposal for a new pass that improves performance and code size in some nested loop situations. The pass is target independent. >From the description in the file header: This optimization finds loop exit values reevaluated after the loop execution and replaces them by the corresponding exit values if they are available. Such sequences can arise after the

Hi, I want to use rsync to make daily backups using hard-links in this way: rsync -av --delete --link-dest=../backup_old ./source/. ./backup_new This works great, but the problem is, that rsync does not show correctly what action it performs (-v option). It shows correctly all new and changed files/folders, but it does not show files that were deleted. This works only if I do not use the

Hi, The LLVM lang ref (http://llvm.org/docs/LangRef.html#i_bitcast) notes that bitcast of vectors to vectors is OK, provided the vectors are of the same size. In attempting this on large vector types (e.g. 16xi64 -> 16xf64), I get assertions related to Extended Value Types when attempting to perform code generation, but no errors at the BC level. Is it simply unsupported to stray outwith

I submitted the problem report to clang's bugzilla but no one seems to care so I have to send it to the mailing list. clang 3.7 svn (trunk 229055 as the time I was to report this problem) generates slower code than 3.5 (Apple LLVM version 6.0 (clang-600.0.56) (based on LLVM 3.5svn)) for the following code. It is a "8 queens puzzle" solver written as an educational example. As

The regressions in the performance of generated code, introduced by the llvm 3.6 release, don't seem to be limited to this 8 queens puzzle" solver test case. See... http://www.phoronix.com/scan.php?page=article&item=llvm-clang-3.5-3.6-rc1&num=1 where a bit hit in the performance of the Sparse Matrix Multiply test of the SciMark v2.0 benchmark was observed as well as others.

Using the SciMark 2.0 code from http://math.nist.gov/scimark2/scimark2_1c.zip compiled with the same... make CFLAGS="-O3 -march=native" I am able to reproduce the 22% performance regression in the run time of the Sparse matmult benchmark. For 10 runs of the scimark2 benechmark, I get 998.439+/-0.4828 with the release llvm clang 3.5.1 compiler and 1217.363+/-1.1004 for the current

I've a question about the bandwidth consumed by IAX2/GSM. According to the wiki page, the GSM codec should run about 13 kilo-bits/sec for a voice encoding. However, watching gkrellm when I initiate a call to Digium, it looks like the channel is taking a consistent 5-6 kilo-bytes/sec. That's a lot more bandwidth than it should take. Is there perhaps a setting I have wrong somethere in

Hi all, These days I am working on a feature designed to insert nops for IA code generation that will provide performance improvements and performance stability. This feature will not affect other architectures. It will, however, set up an infrastructure for other architectures to do the same, if ever needed. Here are some examples for cases in which nops can improve performance: 1. DSB

hi, i found this code in X86DisassemblerDecoder.h #define EA_BASES_32BIT \ ENTRY(EAX) \ ENTRY(ECX) \ ENTRY(EDX) \ ENTRY(EBX) \ ENTRY(sib) \ ENTRY(EBP) \ ENTRY(ESI) \ ENTRY(EDI) \ ENTRY(R8D) \ ENTRY(R9D) \ ENTRY(R10D) \ ENTRY(R11D) \

Greetings CentOS users, I have run into a hardware issue with the Intel 965 chipset. It appears that the latest kernel doesn't support the Marvell PATA controller. I can kickstart the computer, but after the OS is installed the DVD drive can't be accessed. There are patches against 2.6.18 for the Marvell PATA controller here: http://lkml.org/lkml/2006/10/16/157 There is bug

I'm trying to use LLVM to create compiler-rt.o on Windows. I use this command from the compiler-rt project: [nix-shell:~/downloads/llvm-project/compiler-rt]$ clang -nostdlib -S -emit-llvm lib/builtins/udivti3.c -g -target x86_64-windows -DCRT_HAS_128BIT The resulting LLVM IR is: ================================================================= ; ModuleID = 'lib/builtins/udivti3.c'

> Is this documented somethere? This is the standard assumption. The calling convention can be complex enough and operate with the language-dependent definitions (e.g. "C" structs, etc.), thus frontend should lower such constructs as needed. For reference, normal x86-64 ABI has pretty complex rules wrt passing aggregates by value. You might want to look into llvm-gcc as an example.

Ok, another explanation. In my runtime that uses LLVM as JIT engine, I have an instnace of GlobalVariable. The program I run in it writes some value to this variable. How can I read it? 2011/5/18 Duncan Sands <baldrick at free.fr>: > Hi, > >> I need to define the global variable which would be modified by JITted >> code. Can I read it from the JITter itself? It is of

Most probably you need to properly specify the calling convention the backend is using for calling the runtime functions. Or implement the stub for udivti3 that performs the necessary argument lifting. I guess there is no standard ABI document describing the intended calling convention here, so I'd just do what mingw64 does here and make everything here compatible. On Thu, Apr 26, 2018 at