Support for subnormal numbers

[Arjen_Markus]

Hello,

recently, on comp.lang.fortran the behaviour of various compilers with respect to functions like log() and atan() with complex arguments was discussed. I tried the test programs that were mentioned in that thread (cmplx.sourceforge.io, if you are interested) and found that Intel Fortran returns false for the support of subnormal numbers for all three types of reals. I verified this with the Intel Fortran compiler, version 18, on both Windows and Linux and I wondered what features are missing. Why is it reported that subnormals are not supported, as you do get subnormal numbers. It may simply be some detail in the IEEE standard that is not supported, but I am curious.

Regards,

Arjen

[jimdempseyatthecove]

Arjen,

In the MXCSR Control/Status Register, bit 6, is DAZ (Denormals Are Zero). FromIntel 64-ia-32 Instruction Reference (Combined Volumes):

10.2.3.4 Denormals-Are-Zeros
Bit 6 (DAZ) of the MXCSR register enables the denormals-are-zeros mode, which controls the processor’s response
to a SIMD floating-point denormal operand condition. When the denormals-are-zeros flag is set, the processor
converts all denormal source operands to a zero with the sign of the original operand before performing any
computations on them. The processor does not set the denormal-operand exception flag (DE), regardless of the
setting of the denormal-operand exception mask bit (DM); and it does not generate a denormal-operand exception
if the exception is unmasked.
The denormals-are-zeros mode is not compatible with IEEE Standard 754 (see Section 4.8.3.2, “Normalized and
Denormalized Finite Numbers”). The denormals-are-zeros mode is provided to improve processor performance for
applications such as streaming media processing, where rounding a denormal operand to zero does not appreciably
affect the quality of the processed data.
The denormals-are-zeros flag is cleared upon a power-up or reset of the processor, disabling the denormals-arezeros
mode.
The denormals-are-zeros mode was introduced in the Pentium 4 and Intel Xeon processor with the SSE2 extensions;
however, it is fully compatible with the SSE SIMD floating-point instructions (that is, the denormals-arezeros
flag affects the operation of the SSE SIMD floating-point instructions). In earlier IA-32 processors and in
some models of the Pentium 4 processor, this flag (bit 6) is reserved. See Section 11.6.3, “Checking for the DAZ
Flag in the MXCSR Register,” for instructions for detecting the availability of this feature.
Attempting to set bit 6 of the MXCSR register on processors that do not support the DAZ flag will cause a generalprotection
exception (#GP). See Section 11.6.6, “Guidelines for Writing to the MXCSR Register,” for instructions for
preventing such general-protection exceptions by using the MXCSR_MASK value returned by the FXSAVE instruction.

The above is for the SIMD instructions. The FPU appears to differ in that it will accept denormal numbers (but not as operand values), but will not generate them.

Jim Dempsey