search for: eltoder

On Aug 19, 2013, at 18:47 , Eugene Toder <eltoder at gmail.com> wrote: > Have you tried running SLP vectorizer pass (-vectorize-slp)? Yes. That was the first thing i tried, and it didn't do anything. I was looking the vectorizer, but then I saw some things that made me wonder if it was even supposed to do this

Oh yes you're right, I missed that :) But the point still hold. Amaury Pouly 2010/8/8 Eugene Toder <eltoder at gmail.com> > > Not only this code does not compile with NDEBUG set > > I may be missing something, but why does it not compile with -DNDEBUG? > assert() macro expands to noop when NDEBUG is set. > > Eugene > > On Sun, Aug 8, 2010 at 2:19 PM, Amaury Pouly <amaury...

...possible) or use a container that doesn't reallocate on insert. Eugene On Sun, Aug 8, 2010 at 3:11 PM, Amaury Pouly <amaury.pouly at gmail.com> wrote: > Oh yes you're right, I missed that :) But the point still hold. > > Amaury Pouly > > > 2010/8/8 Eugene Toder <eltoder at gmail.com> >> >> > Not only this code does not compile with NDEBUG set >> >> I may be missing something, but why does it not compile with -DNDEBUG? >> assert() macro expands to noop when NDEBUG is set. >> >> Eugene >> >> On Sun, Aug 8,...

On 15 June 2010 23:08, Eugene Toder <eltoder at gmail.com> wrote: > Subsections is a very good idea. You can even do without > post-processing by using carefully crafted section names, e.g. > > __attribute__((section(".text,\"ax\", at progbits\n\t.subsection 1 #"))) > void foo() > { > } There is on...

> The clang driver generally uses an installed gcc to do the assemble > and link steps at the moment. gcc? I thought it uses binutils. Eugene

On Wed, Jul 7, 2010 at 1:55 PM, Eugene Toder <eltoder at gmail.com> wrote: >> The clang driver generally uses an installed gcc to do the assemble >> and link steps at the moment. > > gcc? I thought it uses binutils. It does. But clang uses the gcc driver to invoke the binutils tools anyway. Presumably it's easier that way....

Can you send the IR of the function ? On Aug 20, 2013, at 8:36 AM, Matt Arsenault <arsenm2 at gmail.com> wrote: > > On Aug 19, 2013, at 18:47 , Eugene Toder <eltoder at gmail.com> wrote: > >> Have you tried running SLP vectorizer pass (-vectorize-slp)? > Yes. That was the first thing i tried, and it didn't do anything. I was looking the vectorizer, but then I saw some things that made me wonder if it was even supposed to do this > _______...

...e.g. a load. Isn't a read-only call effectively just a series of loads from a memory-dependence perspective? In other words, why does this code fragment return MemDepResult::getClobber() instead of MemDepResult::getDef() for a read-only call? On Sat, Jul 17, 2010 at 6:18 PM, Eugene Toder <eltoder at gmail.com> wrote: > Since isLoad == false means we're looking at a store, what this does > is making the store *p depend on the load *p. This is correct -- you > can't move store before load, otherwise load will start returning a > different value. > > Eugene > &...

...a read-only call effectively just a series of loads > from a memory-dependence perspective? > In other words, why does this code fragment return > MemDepResult::getClobber() instead of MemDepResult::getDef() for a read-only > call? > On Sat, Jul 17, 2010 at 6:18 PM, Eugene Toder <eltoder at gmail.com> wrote: >> >> Since isLoad == false means we're looking at a store, what this does >> is making the store *p depend on the load *p. This is correct -- you >> can't move store before load, otherwise load will start returning a >> different value....

On Sun, Apr 18, 2010 at 4:36 AM, Eugene Toder <eltoder at gmail.com> wrote: > According to documentation Twines should be used only for temporary > values and not stored, so allocating the in heap sounds wrong. Yes, in general you should never be naming Twine directly, except in the case where you need to make a Twine for an integer. All othe...

...t any changes to LLVM, so the main points for having a direct support for side-tables in LLVM is 1) Not relying on external tool 2) Optimizer issues Eugene On Tue, Jun 8, 2010 at 10:35 PM, Andrew Lenharth <andrewl at lenharth.org> wrote: > On Tue, Jun 8, 2010 at 4:15 PM, Eugene Toder <eltoder at gmail.com> wrote: >> Let me point out that projects using standard toolchain (e.g. >> binutils) can already reorder code and data pretty much arbitrary >> using sections and linker scripts. >> I think it's still attractive to have reordering in LLVM to be >>...

On Wed, Jun 9, 2010 at 4:16 PM, Eugene Toder <eltoder at gmail.com> wrote: > Yes, the global structure is constant. This is indeed a side-table. > Overriding section of global to be in text is simple -- LLVM already > supports section attribute on globals and functions. However we also > need a specific ordering in text. > With some...

Subsections is a very good idea. You can even do without post-processing by using carefully crafted section names, e.g. __attribute__((section(".text,\"ax\", at progbits\n\t.subsection 1 #"))) void foo() { } (Note that you need ".subsection n" commands on ELF targets and ".section name, n" commands on COFF targets; seems that the latter was supported on

On 26 September 2010 13:08, Eugene Toder <eltoder at gmail.com> wrote: >  %s = invoke i32 @v(i32 %o) to label %ok >                   unwind %x to label %catch > ok: >  ret i32 %s > > catch: >  %type = call i32 @exception_type(i8* %x) >  %r = icmp eq i32 %type, 255 ; 255 is DivisionByZeroException type >  br i1 %r, la...

On Tue, Jun 8, 2010 at 4:15 PM, Eugene Toder <eltoder at gmail.com> wrote: > Let me point out that projects using standard toolchain (e.g. > binutils) can already reorder code and data pretty much arbitrary > using sections and linker scripts. > I think it's still attractive to have reordering in LLVM to be > independent from ext...

I may me wrong, but I think Nathan used ints for demonstration purposes only. unwind always takes i8* argument that ideally should be a pointer to exception structure, variable %x in invoke is also typed i8*, it's not "untyped". Probably more llvm-ish syntax would be unwind i8* %x to label %catch to show the type explicitly. However throwing a pointer to a structure raises

That and the possibility of differently sized pointers made me hesitate to dive into implementing this. I guess nailing it down to be the platform equivalent of size_t would be sensible here. On Mon, Aug 9, 2010 at 1:37 PM, Eugene Toder <eltoder at gmail.com> wrote: > Small nitpick: size_t is not guaranteed to be large enough to hold a > pointer (only an array index, which can be less; though such platforms > are pretty exotic now). [u]intptr_t are the types. > > Eugene > > On Mon, Aug 9, 2010 at 5:44 PM, Joshua War...

Let me point out that projects using standard toolchain (e.g. binutils) can already reorder code and data pretty much arbitrary using sections and linker scripts. I think it's still attractive to have reordering in LLVM to be independent from external toolchain. This will allow reordering in JIT and other interesting things. I agree with John that special global with ordered list looks like

On Fri, Jun 4, 2010 at 1:35 PM, Eugene Toder <eltoder at gmail.com> wrote: > It should be relatively simple to write a pass that turns each call > that has constant argument(s) into a call to specialized version of > the callee. To devirtualize C++ calls it needs to be smarter, since > the argument is not a constant, but a pointer to a...

Hi, I've found a case I would expect would optimize easily, but it doesn't. A simple implementation of vector select: float4 simple_select(float4 a, float4 b, int4 c) { float4 result; result.x = c.x ? a.x : b.x; result.y = c.y ? a.y : b.y; result.z = c.z ? a.z : b.z; result.w = c.w ? a.w : b.w; return result; } I would expect this would be optimized to %bool