From 6d1cb6fe3a9a09ca3ebef1ee72a1f89974ca876a Mon Sep 17 00:00:00 2001
From: Mario Juric <mjuric@ias.edu>
Date: Thu, 20 May 2010 02:52:42 -0400
Subject: [PATCH] Add extra info to assertions in cux.h

---
 galaxy.kdevelop  | 26 +++++++++++++-------------
 src/common/cux.h | 32 +++++++++++++++++---------------
 2 files changed, 30 insertions(+), 28 deletions(-)

diff --git a/galaxy.kdevelop b/galaxy.kdevelop
index 922ebd1..505587a 100644
--- a/galaxy.kdevelop
+++ b/galaxy.kdevelop
@@ -17,12 +17,12 @@
     <secondaryLanguages/>
     <versioncontrol/>
     <projectname>galaxy</projectname>
-    <defaultencoding/>
+    <defaultencoding></defaultencoding>
   </general>
   <kdevautoproject>
     <general>
       <activetarget>src/galfast.x</activetarget>
-      <useconfiguration>debug</useconfiguration>
+      <useconfiguration>optimized</useconfiguration>
     </general>
     <run>
       <mainprogram>/home/mjuric/projects/galaxy/debug/src/galfast.x</mainprogram>
@@ -81,14 +81,14 @@
         <f77compiler>kdevg77options</f77compiler>
         <cxxflags>-O2 -Wstrict-aliasing -fno-strict-aliasing</cxxflags>
         <configargs>--with-cuda --enable-cuda-devemuX --with-sm=/usr/local --with-libpeyton=$HOME/projects/libpeyton --prefix=$HOME/projects/galaxy/workspace/staging --with-cfitsio-include=/usr/include/cfitsio --with-libpeyton-libdir=$HOME/projects/libpeyton/lib-optimized</configargs>
-        <topsourcedir/>
+        <topsourcedir></topsourcedir>
         <cppflags>-DDEBUGMODE=0</cppflags>
         <ldflags>-rdynamic -L/opt/cudatools/2.3/lib</ldflags>
-        <ccompilerbinary/>
-        <cxxcompilerbinary/>
-        <f77compilerbinary/>
-        <cflags/>
-        <f77flags/>
+        <ccompilerbinary></ccompilerbinary>
+        <cxxcompilerbinary></cxxcompilerbinary>
+        <f77compilerbinary></f77compilerbinary>
+        <cflags></cflags>
+        <f77flags></f77flags>
         <envvars/>
       </optimized>
       <debug>
@@ -206,7 +206,7 @@
     </codecompletion>
     <references/>
     <creategettersetter>
-      <prefixGet/>
+      <prefixGet></prefixGet>
       <prefixSet>set</prefixSet>
       <prefixVariable>m_,_</prefixVariable>
       <parameterName>theValue</parameterName>
@@ -232,11 +232,11 @@
   <kdevdebugger>
     <general>
       <programargs>postprocess postprocess.conf skygen.conf</programargs>
-      <gdbpath/>
+      <gdbpath></gdbpath>
       <dbgshell>libtool --mode=execute</dbgshell>
-      <configGdbScript/>
-      <runShellScript/>
-      <runGdbScript/>
+      <configGdbScript></configGdbScript>
+      <runShellScript></runShellScript>
+      <runGdbScript></runGdbScript>
       <breakonloadinglibs>true</breakonloadinglibs>
       <separatetty>false</separatetty>
       <floatingtoolbar>false</floatingtoolbar>
diff --git a/src/common/cux.h b/src/common/cux.h
index 3dfcde2..c9dd33f 100644
--- a/src/common/cux.h
+++ b/src/common/cux.h
@@ -21,11 +21,14 @@
 #ifndef cuda_cux__
 #define cuda_cux__
 
-#include <assert.h>
+#include <cassert>
+#include <cstdio>
 #include <map>
 #include <set>
 #include <algorithm>
 
+#include <astro/assert.h>
+
 #include "cux_lowlevel.h"
 
 /**
@@ -143,7 +146,7 @@ template<typename T>
 #elif BUILD_FOR_CPU
 		memcpy(&dest, &source, size);
 #else
-		assert(0);
+		ASSERT(0);
 		//#error cuxUploadConst can be used only in .cu files, or when BUILD_FOR_CPU is defined
 #endif
 	}
@@ -243,7 +246,7 @@ template<typename T, int dim = 1, int align = 128>
 			if(!this->ptr)
 			{
 				if(dim == 1) { alloc(lastdim); }	// auto-allocation allowed only for 1D arrays (convenience)
-				else { assert(this->ptr); }
+				else { ASSERT(this->ptr); }
 			}
 
 			size_t size = memsize(lastdim);
@@ -255,7 +258,7 @@ template<typename T, int dim = 1, int align = 128>
 			if(!this->ptr)
 			{
 				if(dim == 1) { alloc(lastdim); }	// auto-allocation allowed only for 1D arrays (convenience)
-				else { assert(this->ptr); }
+				else { ASSERT(this->ptr); }
 			}
 
 			size_t size = memsize(lastdim);
@@ -269,7 +272,7 @@ template<typename T, int dim = 1, int align = 128>
 		}
 		void alloc(size_t nx, size_t ny = 1, size_t nz = 1)
 		{
-			assert(dim <= 4); // not implemented for dim > 4
+			ASSERT(dim <= 4); // not implemented for dim > 4
 
 			if(dim >  1) { this->extent[0] = roundUpModulo(nx*sizeof(T), align); }
 			if(dim >  2) { this->extent[1] = ny; }
@@ -512,22 +515,23 @@ struct cuxSmartPtr
 	// host data accessors (note: use hptr<> interface if maximum speed is needed)
 	T& operator()(const uint32_t x, const uint32_t y, const uint32_t z) const	// 3D accessor (valid only if dim = 3)
 	{
-		assert(x < width());
-		assert(y < height());
-		assert(z < depth());
+		ASSERT(x < width())  { fprintf(stderr, "x=%u, width()=%u\n", x, width()); }
+		ASSERT(y < height()) { fprintf(stderr, "y=%u, height()=%u\n", y, height()); }
+		ASSERT(z < depth())  { fprintf(stderr, "z=%u, depth()=%u\n", z, depth()); }
 		if(m_impl->onDevice || !m_impl->m_data.ptr) { syncToHost(); }
 		return (*(arrayPtr<T, 3> *)(&m_impl->m_data))(x, y, z);
 	}
 	T& operator()(const uint32_t x, const uint32_t y) const	// 2D accessor (valid only if dim >= 2)
 	{
-		assert(x < width());
-		assert(y < height());
+		ASSERT(x < width())  { fprintf(stderr, "x=%u, width()=%u\n", x, width()); }
+		ASSERT(y < height()) { fprintf(stderr, "y=%u, height()=%u\n", y, height()); }
 		if(m_impl->onDevice || !m_impl->m_data.ptr) { syncToHost(); }
 		return (*(arrayPtr<T, 2> *)(&m_impl->m_data))(x, y);
 	}
 	T& operator()(const uint32_t x) const			// 1D accessor (valid only if dim >= 2)
 	{
-		assert(x < width());
+		ASSERT(x < width()) { fprintf(stderr, "x=%u, width()=%u\n", x, width()); }
+
 		if(m_impl->onDevice || !m_impl->m_data.ptr) { syncToHost(); }
 		return (*(arrayPtr<T, 1> *)(&m_impl->m_data))(x);
 	}
@@ -667,7 +671,7 @@ struct cuxTexture : public cuxSmartPtr<T>
 	}
 	void set(const cuxSmartPtr<T>& a, const float2 tcx, const float2 tcy)
 	{
-		assert(dim == 2);
+		ASSERT(dim == 2) { fprintf(stderr, "dim=%d\n", dim); }
 
 		*this = a;
 		coords[0] = tcx;
@@ -675,7 +679,7 @@ struct cuxTexture : public cuxSmartPtr<T>
 	}
 	void set(const cuxSmartPtr<T>& a, const float2 tcx, const float2 tcy, const float2 tcz)
 	{
-		assert(dim == 3);
+		ASSERT(dim == 3) { fprintf(stderr, "dim=%d\n", dim); }
 
 		*this = a;
 		coords[0] = tcx;
@@ -729,8 +733,6 @@ template<typename T, int dim, enum cudaTextureReadMode mode>
 	struct cuxTextureReference : public cuxTextureReferenceInterface
 	{
 	public:
-/*		cuxSmartPtr<T>	data;
-		cuxTexCoords<dim> tc;*/
 		cuxTexture<T, dim> tex;		// the texture from which we're going to sample
 		textureReference &texref;	// the CUDA texture reference to which the current texture will be bound
 		const char *tcSymbolName;	// the symbol name of the __constant__ variable that holds the texture