[Beignet] [PATCH] backend: refine exp function with float input

[Pan, Xiuli]

I only check the correctness, there is no problem
It seems mix using tab and space in the function __gen_ocl_internal_simple_exp and there some intend problem.


-----Original Message-----
From: Beignet [mailto:beignet-bounces at lists.freedesktop.org] On Behalf Of rander.wang
Sent: Wednesday, June 7, 2017 13:49
To: beignet at freedesktop.org
Cc: Wang, Rander <rander.wang at intel.com>
Subject: [Beignet] [PATCH] backend: refine exp function with float input

	remove some corner cases check for these path can not be
	reached.And refine branch code to select. These improvements
	get 20% performance. and the performance of  OCL_ExpFixture_Exp
        in opencv can match up to other Gen driver

Signed-off-by: rander.wang <rander.wang at intel.com>
---
 backend/src/libocl/tmpl/ocl_math_common.tmpl.cl | 60 ++++++++++++++++++++++++-
 1 file changed, 58 insertions(+), 2 deletions(-)

diff --git a/backend/src/libocl/tmpl/ocl_math_common.tmpl.cl b/backend/src/libocl/tmpl/ocl_math_common.tmpl.cl
index 6b942db..16bbc82 100644
--- a/backend/src/libocl/tmpl/ocl_math_common.tmpl.cl
+++ b/backend/src/libocl/tmpl/ocl_math_common.tmpl.cl
@@ -1418,6 +1418,62 @@ OVERLOADABLE float __gen_ocl_internal_exp(float x) {
   }
 }
 
+OVERLOADABLE float __gen_ocl_internal_simple_exp(float x) {
+  float o_threshold = 8.8721679688e+01,  /* 0x42b17180 */
+  u_threshold = -1.0397208405e+02,  /* 0xc2cff1b5 */
+  twom100 = 7.8886090522e-31, 	 /* 2**-100=0x0d800000 */
+  ivln2	 =	1.4426950216e+00, /* 0x3fb8aa3b =1/ln2 */
+  one = 1.0,
+  P1 = 1.6666667163e-01, /* 0x3e2aaaab */
+  P2 = -2.7777778450e-03; /* 0xbb360b61 */
+  float y,hi=0.0,lo=0.0,c,t;
+  int k=0;
+  unsigned hx;
+  float ln2HI_0 = 6.9313812256e-01;	/* 0x3f317180 */
+  float ln2HI_1 = -6.9313812256e-01;	/* 0xbf317180 */
+  float ln2LO_0 = 9.0580006145e-06;  	/* 0x3717f7d1 */
+  float ln2LO_1 = -9.0580006145e-06; /* 0xb717f7d1 */
+  float half_0 = 0.5;
+  float half_1 =	-0.5;
+  float retVal = -1.0f;
+  hx = as_uint(fabs(x));
+
+  /* filter out non-finite argument */
+  /* if |x|>=88.721... */
+  if(hx >= 0x42b17218)
+  {
+    float tmp = (x > 0)? x:0.0;
+    retVal = (x > 0)? INFINITY:retVal; /* overflow */
+    retVal = (hx>0x7f800000)? NAN:retVal;
+    retVal = (hx==0x7f800000)? tmp:retVal;
+    retVal = (x < u_threshold)? 0.0:retVal; /* underflow */
+
+    if(retVal != -1.0f)
+      return retVal;
+  }
+
+  /* argument reduction */
+  float tmp = (x < 0) ? half_1 : half_0;  k  = mad(ivln2, x, tmp);  t  
+ = k;
+  hi = mad(t, -ln2HI_0, x);	/* t*ln2HI is exact here */
+  lo = t*ln2LO_0;
+  x  = hi - lo;
+
+  /* x is now in primary range */
+  t  = x*x;
+  c  = mad(t, mad(t, -P2, -P1), x);
+  y = one-((lo + (x*c)/(c-2.0f))-hi);
+
+  unsigned hy;
+  GEN_OCL_GET_FLOAT_WORD(hy,y);
+
+  float factor = (k >= -125)? 1.0f:twom100;
+  k = (k >= -125)? k:k+100;
+  GEN_OCL_SET_FLOAT_WORD(y,hy+(k<<23));	/* add k to y's exponent */
+  return y*factor;
+}
+
 /* erf,erfc from glibc s_erff.c -- float version of s_erf.c.
  * Conversion to float by Ian Lance Taylor, Cygnus Support, ian at cygnus.com.
  */
@@ -3105,10 +3161,10 @@ OVERLOADABLE float exp(float x) {
     return __gen_ocl_internal_fastpath_exp(x);
 
   /* Use native instruction when it has enough precision */
-  if (x > -0x1.6p1 && x < 0x1.6p1)
+  if (fabs(x) < 0x1.6p1)
     return __gen_ocl_internal_fastpath_exp(x);
 
-  return  __gen_ocl_internal_exp(x);
+  return  __gen_ocl_internal_simple_exp(x);
 }
 
 OVERLOADABLE float exp2(float x) {
--
2.7.4

_______________________________________________
Beignet mailing list
Beignet at lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/beignet