search for: apfloats

Hello All, the patch below adds a target lowering hook to state that the target supports (or not) floating point exception behavior. The patch is small and contains one possible use for the hook (folding potentially exception raising fp operations). Any comments? Thanks Pedro -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: opensource1.txt

...e create a new APFloat from the string we get a slightly smaller number. I have observed that if I use ``APFloat::rmNearestTiesToEven`` when creating ``newF`` that I do get an APFloat instance which has the same value as the original APFloat. So I'm wondering if the comment about round tripping APFloats only holds for certain rounding modes. Any thoughts on this? Thanks, Dan

I assume you meant to ask for ports that *don’t* support floating point exceptions. To my knowledge, neither R600 nor NVPTX support floating point exceptions. —Owen > On Aug 8, 2014, at 2:41 PM, Eric Christopher <echristo at gmail.com> wrote: > > There's a lot of cut and paste in those routines. Can you do something > to unify it a bit? Also, do we have any ports that

This patch supplies software IEEE floating point support. The comment from the patch reproduced below says all there is to say. This patch contains the prior "cleanup" patch; please don't apply that one. Please let me know of any bugs. It is tested reasonably well, but until I put together random tests it's hard to have 100% confidence. Neil. /* A self-contained host- and

>> >> I wonder why APFloat::convertToDouble asserts if it is not an >> APFloat::IEEEdouble while the name "convertToDouble" suggests that it >> converts when it is not a double. > > The conversion is to (host) double, as the name implies. Among > APFloat types, that is only implemented for APFloat::IEEEdouble. > yes, it's to host double

Hi Hal, On Sun, Oct 2, 2016 at 7:43 PM Hal Finkel <hfinkel at anl.gov> wrote: > Hi Tim, > > How, in general, are you thinking about doing this? I ask because, as you > clearly know, the double-double format is formed by the sum of two > double-precision numbers, and the various arithmetic operations are formed > mostly in terms of double-precision arithmetic on the

I have found some internal test failures due to the wrong constant folding on ppc_fp128. As documented in APFloat::PPCDoubleDouble, APFloat doesn't support PowerPC double-double correctly < https://github.com/llvm-mirror/llvm/blob/492acdd450bcdf9837494d6da029ed064f14fc33/lib/Support/APFloat.cpp#L74 >. To support this, we need to add a second tuple of (sign, exponent, significand) to

Hi, I'm trying to separate the Support, System, ADT and Config header files into the support module, per previous plans. However, APFloat.h is not self-contained: APFloat.h:105:43: error: llvm/Bitcode/SerializationFwd.h: No such file or directory APFloat.h:106:37: error: llvm/CodeGen/ValueTypes.h: No such file or directory As you can see, APFloat.h depends on things that are not in the ADT,

...lizer; Deserializer; If this doesn't seem like a clean enough solution, there are other ways to potentially decouple APFloat more from the Serialization "library." The tradeoff is that it would require some code rewriting/ addition, and could potentially make deserialization of APFloats (as done in the new C frontend) slightly less efficient.

llvm[1]: Compiling APFloat.cpp for Release+Asserts build clang: TargetInfo.cpp:1778: llvm::Type *GetX86_64ByValArgumentPair(llvm::Type *, llvm::Type *, const llvm::TargetData &): Assertion `Lo->isIntegerTy() && "Invalid/unknown lo type"' failed. 0 clang 0x0000000001c132ef 1 clang 0x0000000001c13804 2 libpthread.so.0 0x00002ba7d7eaec60 3

...> If this doesn't seem like a clean enough solution, there are other > ways to potentially decouple APFloat more from the Serialization > "library." The tradeoff is that it would require some code rewriting/ > addition, and could potentially make deserialization of APFloats (as > done in the new C frontend) slightly less efficient. I doubt the light performance loss will make any significant difference. Further, putting all the serialization code in one translation unit may have some benefit for a) clarity (one place to go instead of all over the place) and b) m...

On Jul 9, 2010, at 12:59 PMPDT, Jochen Wilhelmy wrote: >>> I wonder why APFloat::convertToDouble asserts if it is not an >>> APFloat::IEEEdouble while the name "convertToDouble" suggests that >>> it >>> converts when it is not a double. >> >> The conversion is to (host) double, as the name implies. Among >> APFloat types, that is

Hi! I wonder why APFloat::convertToDouble asserts if it is not an APFloat::IEEEdouble while the name "convertToDouble" suggests that it converts when it is not a double. -Jochen

> I'd rather not. The functionality you want is there, feed another > APFloat type through APFloat::convert first. > > Using host FP is not something that should be encouraged; the main > point of APFloat is so people don't have to do that. Why do you want > to, btw? > Yes, i got it working using APFloat::convert. I need host float to output to my backend which

On Jul 9, 2010, at 7:59 AMPDT, Jochen Wilhelmy wrote: > Hi! > > I wonder why APFloat::convertToDouble asserts if it is not an > APFloat::IEEEdouble while the name "convertToDouble" suggests that it > converts when it is not a double. The conversion is to (host) double, as the name implies. Among APFloat types, that is only implemented for APFloat::IEEEdouble.

...sn't seem like a clean enough solution, there are other >> ways to potentially decouple APFloat more from the Serialization >> "library." The tradeoff is that it would require some code >> rewriting/ >> addition, and could potentially make deserialization of APFloats (as >> done in the new C frontend) slightly less efficient. > > I doubt the light performance loss will make any significant > difference. > Further, putting all the serialization code in one translation unit > may > have some benefit for a) clarity (one place to go inst...

I don't know if the toy program in chapter 4 of the tutorial implementing Kaleidoscope in llvm with C++ is part of your regression suite, but with the version of llvm I installed last weekend, it does not compile: hendrik at lovesong:~/dv/llvm/tut$ g++ -g toy.cpp `llvm-config --cppflags --ldflags --libs core jit native` -O3 -o toy toy.cpp: In member function ‘virtual llvm::Value*

On Jul 9, 2010, at 1:20 PMPDT, Jochen Wilhelmy wrote: >> I'd rather not. The functionality you want is there, feed another >> APFloat type through APFloat::convert first. >> >> Using host FP is not something that should be encouraged; the main >> point of APFloat is so people don't have to do that. Why do you >> want to, btw? > > Yes, i got

Hi @llvm, I stumbled over a strange behavior if a float containing a NaN is printed (e.g. in the clang rewriter). The local template method "append" in APFloat.cpp deduces the size from the char array, which for "NaN" is 4 (including the trailing zero). If APFloat::toString is called with a SmallString and then SmallString::str() is called, it returns "NaN\0". I

Hello, I'm playing with llvm/clang and I have a problem when I try to use the -ast-print options for a file that uses long double. The Basic/Targets.cpp file defined the x86 (and x86_64) long double type as APFloat::x87DoubleExtended. Then, when the AST printer try to write a long double literal value (using VisitFloatingLiteral()), it call FloatingLiteral- >getValueAsDouble()