[Mesa-dev] [PATCH] gallivm, llvmpipe: fix float->srgb conversion to handle NaNs
sroland at vmware.com
sroland at vmware.com
Mon Nov 11 06:30:23 PST 2013
From: Roland Scheidegger <sroland at vmware.com>
d3d10 requires us to convert NaNs to zero for any float->int conversion.
We don't really do that but mostly seems to work. In particular I suspect the
very common float->unorm8 path only really passes because it relies on sse2
pack intrinsics which just happen to work by luck for NaNs (float->int
conversion in hw gives integer indeterminate value, which just happens to be
-0x80000000 hence gets converted to zero in the end after pack intrinsics).
However, float->srgb didn't get so lucky, because we need to clamp before
blending and clamping resulted in NaN behavior being undefined (and actually
got converted to 1.0 by clamping with sse2). Fix this by using a zero/one clamp
with defined nan behavior as we can handle the NaN for free this way.
I suspect there's more bugs lurking in this area (e.g. converting floats to
snorm) as we don't really use defined NaN behavior everywhere but this seems
to be good enough.
While here respecify nan behavior modes a bit, in particular the return_second
mode didn't really do what we wanted. From the caller's perspective, we really
wanted to say we need the non-nan result, but we already know the second arg
isn't a NaN. So we use this now instead, which means that cpu architectures
which actually implement min/max by always returning non-nan (that is adhering
to ieee754-2008 rules) don't need to bend over backwards for nothing.
---
src/gallium/auxiliary/gallivm/lp_bld_arit.c | 44 +++++++++++++-------
src/gallium/auxiliary/gallivm/lp_bld_arit.h | 12 ++++--
src/gallium/auxiliary/gallivm/lp_bld_format_srgb.c | 2 +-
src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c | 11 ++---
src/gallium/drivers/llvmpipe/lp_state_fs.c | 4 +-
5 files changed, 45 insertions(+), 28 deletions(-)
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_arit.c b/src/gallium/auxiliary/gallivm/lp_bld_arit.c
index 00052ed..70929e7 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_arit.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_arit.c
@@ -123,8 +123,10 @@ lp_build_min_simple(struct lp_build_context *bld,
}
}
else if (type.floating && util_cpu_caps.has_altivec) {
- debug_printf("%s: altivec doesn't support nan behavior modes\n",
- __FUNCTION__);
+ if (nan_behavior == GALLIVM_NAN_RETURN_NAN) {
+ debug_printf("%s: altivec doesn't support nan return nan behavior\n",
+ __FUNCTION__);
+ }
if (type.width == 32 && type.length == 4) {
intrinsic = "llvm.ppc.altivec.vminfp";
intr_size = 128;
@@ -159,8 +161,6 @@ lp_build_min_simple(struct lp_build_context *bld,
}
} else if (util_cpu_caps.has_altivec) {
intr_size = 128;
- debug_printf("%s: altivec doesn't support nan behavior modes\n",
- __FUNCTION__);
if (type.width == 8) {
if (!type.sign) {
intrinsic = "llvm.ppc.altivec.vminub";
@@ -191,7 +191,7 @@ lp_build_min_simple(struct lp_build_context *bld,
*/
if (util_cpu_caps.has_sse && type.floating &&
nan_behavior != GALLIVM_NAN_BEHAVIOR_UNDEFINED &&
- nan_behavior != GALLIVM_NAN_RETURN_SECOND) {
+ nan_behavior != GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN) {
LLVMValueRef isnan, max;
max = lp_build_intrinsic_binary_anylength(bld->gallivm, intrinsic,
type,
@@ -227,7 +227,7 @@ lp_build_min_simple(struct lp_build_context *bld,
return lp_build_select(bld, cond, a, b);
}
break;
- case GALLIVM_NAN_RETURN_SECOND:
+ case GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN:
cond = lp_build_cmp_ordered(bld, PIPE_FUNC_LESS, a, b);
return lp_build_select(bld, cond, a, b);
case GALLIVM_NAN_BEHAVIOR_UNDEFINED:
@@ -299,8 +299,10 @@ lp_build_max_simple(struct lp_build_context *bld,
}
}
else if (type.floating && util_cpu_caps.has_altivec) {
- debug_printf("%s: altivec doesn't support nan behavior modes\n",
- __FUNCTION__);
+ if (nan_behavior == GALLIVM_NAN_RETURN_NAN) {
+ debug_printf("%s: altivec doesn't support nan return nan behavior\n",
+ __FUNCTION__);
+ }
if (type.width == 32 || type.length == 4) {
intrinsic = "llvm.ppc.altivec.vmaxfp";
intr_size = 128;
@@ -336,8 +338,6 @@ lp_build_max_simple(struct lp_build_context *bld,
}
} else if (util_cpu_caps.has_altivec) {
intr_size = 128;
- debug_printf("%s: altivec doesn't support nan behavior modes\n",
- __FUNCTION__);
if (type.width == 8) {
if (!type.sign) {
intrinsic = "llvm.ppc.altivec.vmaxub";
@@ -362,7 +362,7 @@ lp_build_max_simple(struct lp_build_context *bld,
if(intrinsic) {
if (util_cpu_caps.has_sse && type.floating &&
nan_behavior != GALLIVM_NAN_BEHAVIOR_UNDEFINED &&
- nan_behavior != GALLIVM_NAN_RETURN_SECOND) {
+ nan_behavior != GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN) {
LLVMValueRef isnan, min;
min = lp_build_intrinsic_binary_anylength(bld->gallivm, intrinsic,
type,
@@ -398,7 +398,7 @@ lp_build_max_simple(struct lp_build_context *bld,
return lp_build_select(bld, cond, a, b);
}
break;
- case GALLIVM_NAN_RETURN_SECOND:
+ case GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN:
cond = lp_build_cmp_ordered(bld, PIPE_FUNC_GREATER, a, b);
return lp_build_select(bld, cond, a, b);
case GALLIVM_NAN_BEHAVIOR_UNDEFINED:
@@ -1399,6 +1399,7 @@ lp_build_max_ext(struct lp_build_context *bld,
/**
* Generate clamp(a, min, max)
+ * NaN behavior (for any of a, min, max) is undefined.
* Do checks for special cases.
*/
LLVMValueRef
@@ -1418,6 +1419,20 @@ lp_build_clamp(struct lp_build_context *bld,
/**
+ * Generate clamp(a, 0, 1)
+ * A NaN will get converted to zero.
+ */
+LLVMValueRef
+lp_build_clamp_zero_one_nanzero(struct lp_build_context *bld,
+ LLVMValueRef a)
+{
+ a = lp_build_max_ext(bld, a, bld->zero, GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN);
+ a = lp_build_min(bld, a, bld->one);
+ return a;
+}
+
+
+/**
* Generate abs(a)
*/
LLVMValueRef
@@ -3029,9 +3044,8 @@ lp_build_exp2(struct lp_build_context *bld,
/* We want to preserve NaN and make sure than for exp2 if x > 128,
* the result is INF and if it's smaller than -126.9 the result is 0 */
x = lp_build_min_ext(bld, lp_build_const_vec(bld->gallivm, type, 128.0), x,
- GALLIVM_NAN_RETURN_SECOND);
- x = lp_build_max_ext(bld, lp_build_const_vec(bld->gallivm, type, -126.99999), x,
- GALLIVM_NAN_RETURN_SECOND);
+ GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN);
+ x = lp_build_max(bld, lp_build_const_vec(bld->gallivm, type, -126.99999), x);
/* ipart = floor(x) */
/* fpart = x - ipart */
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_arit.h b/src/gallium/auxiliary/gallivm/lp_bld_arit.h
index 49d4e2c..75bf89e 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_arit.h
+++ b/src/gallium/auxiliary/gallivm/lp_bld_arit.h
@@ -142,9 +142,11 @@ enum gallivm_nan_behavior {
GALLIVM_NAN_RETURN_NAN,
/* If one of the inputs is NaN, the other operand is returned */
GALLIVM_NAN_RETURN_OTHER,
- /* If one of the inputs is NaN, the second operand is returned.
- * In min/max it will be as fast as undefined with sse opcodes */
- GALLIVM_NAN_RETURN_SECOND
+ /* If one of the inputs is NaN, the other operand is returned,
+ * but we guarantee the second operand is not a NaN.
+ * In min/max it will be as fast as undefined with sse opcodes,
+ * and archs having native return_other can benefit too. */
+ GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN
};
LLVMValueRef
@@ -176,6 +178,10 @@ lp_build_clamp(struct lp_build_context *bld,
LLVMValueRef max);
LLVMValueRef
+lp_build_clamp_zero_one_nanzero(struct lp_build_context *bld,
+ LLVMValueRef a);
+
+LLVMValueRef
lp_build_abs(struct lp_build_context *bld,
LLVMValueRef a);
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_srgb.c b/src/gallium/auxiliary/gallivm/lp_bld_format_srgb.c
index 2b1fe64..6645151 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_format_srgb.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_format_srgb.c
@@ -326,7 +326,7 @@ lp_build_float_to_srgb_packed(struct gallivm_state *gallivm,
* can't use lp_build_conv since we want to keep values as 32bit
* here so we can interleave with rgb to go from SoA->AoS.
*/
- alpha = lp_build_clamp(&f32_bld, src[3], f32_bld.zero, f32_bld.one);
+ alpha = lp_build_clamp_zero_one_nanzero(&f32_bld, src[3]);
alpha = lp_build_mul(&f32_bld, alpha,
lp_build_const_vec(gallivm, src_type, 255.0f));
tmpsrgb[3] = lp_build_iround(&f32_bld, alpha);
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c
index 5f81066..5fc47ed 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c
@@ -1384,21 +1384,18 @@ emit_store_chan(
assert(dtype == TGSI_TYPE_FLOAT ||
dtype == TGSI_TYPE_UNTYPED);
value = LLVMBuildBitCast(builder, value, float_bld->vec_type, "");
- value = lp_build_max_ext(float_bld, value, float_bld->zero,
- GALLIVM_NAN_RETURN_SECOND);
- value = lp_build_min_ext(float_bld, value, float_bld->one,
- GALLIVM_NAN_BEHAVIOR_UNDEFINED);
+ value = lp_build_clamp_zero_one_nanzero(float_bld, value);
break;
case TGSI_SAT_MINUS_PLUS_ONE:
assert(dtype == TGSI_TYPE_FLOAT ||
dtype == TGSI_TYPE_UNTYPED);
value = LLVMBuildBitCast(builder, value, float_bld->vec_type, "");
+ /* This will give -1.0 for NaN which is probably not what we want. */
value = lp_build_max_ext(float_bld, value,
lp_build_const_vec(gallivm, float_bld->type, -1.0),
- GALLIVM_NAN_RETURN_SECOND);
- value = lp_build_min_ext(float_bld, value, float_bld->one,
- GALLIVM_NAN_BEHAVIOR_UNDEFINED);
+ GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN);
+ value = lp_build_min(float_bld, value, float_bld->one);
break;
default:
diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c
index 8223d2a..b5816e0 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_fs.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c
@@ -1760,11 +1760,11 @@ generate_unswizzled_blend(struct gallivm_state *gallivm,
assert(row_type.floating);
lp_build_context_init(&f32_bld, gallivm, row_type);
for (i = 0; i < src_count; i++) {
- src[i] = lp_build_clamp(&f32_bld, src[i], f32_bld.zero, f32_bld.one);
+ src[i] = lp_build_clamp_zero_one_nanzero(&f32_bld, src[i]);
}
if (dual_source_blend) {
for (i = 0; i < src_count; i++) {
- src1[i] = lp_build_clamp(&f32_bld, src1[i], f32_bld.zero, f32_bld.one);
+ src1[i] = lp_build_clamp_zero_one_nanzero(&f32_bld, src1[i]);
}
}
/* probably can't be different than row_type but better safe than sorry... */
--
1.7.9.5
More information about the mesa-dev
mailing list