A bunch of maintenance fixes, the largest of which is getting the PTX simulation to work with CUDA 12. (#95)

* Fixing the formatter to always use a consistent format and running it on the codebase

* Update linux-so-version.txt

* Update Makefile

* A couple of unnecessary files that are lingering around

* Support CUDA 12

* Getting the PTX simulations to work with CUDA 12. The issue is that ptxas added more information (number of barriers and compile time). We have to parse these or lexx/yacc fail.

* Update ptxinfo.l

debug MACRO was ineffective

* Update gpgpusim_check.cmake

Update to make the CUDA version print a warning, not an error and updating the print to be more reflective of what the actual problem is.

7 files changed, 28 insertions, 27 deletions

diff --git a/src/cuda-sim/cuda-sim.cc b/src/cuda-sim/cuda-sim.cc
index d05549c..69d1eb7 100644
--- a/src/cuda-sim/cuda-sim.cc
+++ b/src/cuda-sim/cuda-sim.cc
@@ -2782,9 +2782,7 @@ void print_ptxinfo() {
   }
 }
 
-struct gpgpu_ptx_sim_info get_ptxinfo() {
-  return g_ptxinfo;
-}
+struct gpgpu_ptx_sim_info get_ptxinfo() { return g_ptxinfo; }
 
 std::map<unsigned, const char *> get_duplicate() { return g_duplicate; }
 
@@ -2801,6 +2799,8 @@ void ptxinfo_function(const char *fname) {
 
 void ptxinfo_regs(unsigned nregs) { g_ptxinfo.regs = nregs; }
 
+void ptxinfo_barriers(unsigned barriers) { g_ptxinfo.barriers = barriers; }
+
 void ptxinfo_lmem(unsigned declared, unsigned system) {
   g_ptxinfo.lmem = declared + system;
 }
diff --git a/src/cuda-sim/cuda-sim.h b/src/cuda-sim/cuda-sim.h
index 21e1ca0..b1caf0c 100644
--- a/src/cuda-sim/cuda-sim.h
+++ b/src/cuda-sim/cuda-sim.h
@@ -101,8 +101,8 @@ class functionalCoreSim : public core_t {
   bool *m_warpAtBarrier;
 };
 
-#define RECONVERGE_RETURN_PC ((address_type)-2)
-#define NO_BRANCH_DIVERGENCE ((address_type)-1)
+#define RECONVERGE_RETURN_PC ((address_type) - 2)
+#define NO_BRANCH_DIVERGENCE ((address_type) - 1)
 address_type get_return_pc(void *thd);
 const char *get_ptxinfo_kname();
 void print_ptxinfo();
diff --git a/src/cuda-sim/half.h b/src/cuda-sim/half.h
index fab1a22..67e607b 100644
--- a/src/cuda-sim/half.h
+++ b/src/cuda-sim/half.h
@@ -846,10 +846,8 @@ uint16 int2half_impl(T value) {
       bits |= 0x7BFF + (R != std::round_toward_zero);

   } else if (value) {

     unsigned int m = value, exp = 24;

-    for (; m < 0x400; m <<= 1, --exp)

-      ;

-    for (; m > 0x7FF; m >>= 1, ++exp)

-      ;

+    for (; m < 0x400; m <<= 1, --exp);

+    for (; m > 0x7FF; m >>= 1, ++exp);

     bits |= (exp << 10) + m;

     if (exp > 24) {

       if (R == std::round_to_nearest)

@@ -1274,8 +1272,7 @@ inline double half2float_impl(uint16 value, double, true_type) {
   int abs = value & 0x7FFF;

   if (abs) {

     hi |= 0x3F000000 << static_cast<unsigned>(abs >= 0x7C00);

-    for (; abs < 0x400; abs <<= 1, hi -= 0x100000)

-      ;

+    for (; abs < 0x400; abs <<= 1, hi -= 0x100000);

     hi += static_cast<uint32>(abs) << 10;

   }

   uint64 bits = static_cast<uint64>(hi) << 32;

@@ -2116,8 +2113,7 @@ struct functions {
   static half frexp(half arg, int *exp) {

     int m = arg.data_ & 0x7FFF, e = -14;

     if (m >= 0x7C00 || !m) return *exp = 0, arg;

-    for (; m < 0x400; m <<= 1, --e)

-      ;

+    for (; m < 0x400; m <<= 1, --e);

     return *exp = e + (m >> 10),

            half(binary, (arg.data_ & 0x8000) | 0x3800 | (m & 0x3FF));

   }

@@ -2135,8 +2131,7 @@ struct functions {
     unsigned int mask = (1 << (25 - e)) - 1, m = arg.data_ & mask;

     iptr->data_ = arg.data_ & ~mask;

     if (!m) return half(binary, arg.data_ & 0x8000);

-    for (; m < 0x400; m <<= 1, --e)

-      ;

+    for (; m < 0x400; m <<= 1, --e);

     return half(binary, static_cast<uint16>((arg.data_ & 0x8000) | (e << 10) |

                                             (m & 0x3FF)));

   }

@@ -2148,8 +2143,7 @@ struct functions {
   static half scalbln(half arg, long exp) {

     unsigned int m = arg.data_ & 0x7FFF;

     if (m >= 0x7C00 || !m) return arg;

-    for (; m < 0x400; m <<= 1, --exp)

-      ;

+    for (; m < 0x400; m <<= 1, --exp);

     exp += m >> 10;

     uint16 value = arg.data_ & 0x8000;

     if (exp > 30) {

@@ -2191,8 +2185,7 @@ struct functions {
     if (abs < 0x7C00) {

       int exp = (abs >> 10) - 15;

       if (abs < 0x400)

-        for (; abs < 0x200; abs <<= 1, --exp)

-          ;

+        for (; abs < 0x200; abs <<= 1, --exp);

       return exp;

     }

     if (abs > 0x7C00) return FP_ILOGBNAN;

@@ -2208,13 +2201,11 @@ struct functions {
     if (abs < 0x7C00) {

       int exp = (abs >> 10) - 15;

       if (abs < 0x400)

-        for (; abs < 0x200; abs <<= 1, --exp)

-          ;

+        for (; abs < 0x200; abs <<= 1, --exp);

       uint16 bits = (exp < 0) << 15;

       if (exp) {

         unsigned int m = std::abs(exp) << 6, e = 18;

-        for (; m < 0x400; m <<= 1, --e)

-          ;

+        for (; m < 0x400; m <<= 1, --e);

         bits |= (e << 10) + m;

       }

       return half(binary, bits);

diff --git a/src/cuda-sim/ptx_ir.h b/src/cuda-sim/ptx_ir.h
index b08a692..4a7d39b 100644
--- a/src/cuda-sim/ptx_ir.h
+++ b/src/cuda-sim/ptx_ir.h
@@ -39,7 +39,7 @@
 #include <string>
 #include <vector>
 
-//#include "ptx.tab.h"
+// #include "ptx.tab.h"
 #include "ptx_sim.h"
 
 #include "memory.h"
diff --git a/src/cuda-sim/ptx_sim.h b/src/cuda-sim/ptx_sim.h
index 8eec922..7128f8e 100644
--- a/src/cuda-sim/ptx_sim.h
+++ b/src/cuda-sim/ptx_sim.h
@@ -340,9 +340,7 @@ class ptx_thread_info {
   dim3 get_ctaid() const { return m_ctaid; }
   dim3 get_tid() const { return m_tid; }
   dim3 get_ntid() const { return m_ntid; }
-  class gpgpu_sim *get_gpu() {
-    return (gpgpu_sim *)m_gpu;
-  }
+  class gpgpu_sim *get_gpu() { return (gpgpu_sim *)m_gpu; }
   unsigned get_hw_tid() const { return m_hw_tid; }
   unsigned get_hw_ctaid() const { return m_hw_ctaid; }
   unsigned get_hw_wid() const { return m_hw_wid; }
diff --git a/src/cuda-sim/ptxinfo.l b/src/cuda-sim/ptxinfo.l
index 51371e3..db81bca 100644
--- a/src/cuda-sim/ptxinfo.l
+++ b/src/cuda-sim/ptxinfo.l
@@ -58,6 +58,8 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 "Compiling entry function"	TC; return FUNC;
 "Used"		TC; return USED;
 "registers"	TC; return REGS;
+"used"		TC; return USED;
+"barriers"	TC; return REGS;
 "bytes"		TC; return BYTES;
 "lmem"		TC; return LMEM;
 "smem"		TC; return SMEM;
@@ -65,11 +67,13 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 "gmem"		TC; return GMEM;
 "line"		TC; return LINE;
 "for"		TC; return FOR;
+"ms"		TC; return MS;
 "textures"	TC; return TEXTURES;
 "error   : Duplicate definition of"	TC; return DUPLICATE;
 "function"	TC; yylval->string_value = strdup(yytext); return FUNCTION;
 "variable"	TC; yylval->string_value = strdup(yytext); return VARIABLE;
 "fatal   : Ptx assembly aborted due to errors"	TC; return FATAL;
+"Compile time = "	TC; return COMPILETIME;
 
 [_A-Za-z$%][_0-9A-Za-z$]*  TC; yylval->string_value = strdup(yytext); return IDENTIFIER;
 [-]{0,1}[0-9]+	 TC; yylval->int_value =  atoi(yytext); return INT_OPERAND;
@@ -79,6 +83,7 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 "["     TC; return LEFT_SQUARE_BRACKET;
 "]"	TC; return RIGHT_SQUARE_BRACKET;
 ":"	TC; return COLON;
+"."	TC; return PERIOD;
 ";"	TC; return SEMICOLON;
 "'"	TC; return QUOTE;
 " " TC;
diff --git a/src/cuda-sim/ptxinfo.y b/src/cuda-sim/ptxinfo.y
index b303958..7225207 100644
--- a/src/cuda-sim/ptxinfo.y
+++ b/src/cuda-sim/ptxinfo.y
@@ -49,6 +49,8 @@ typedef void * yyscan_t;
 %token FUNC
 %token USED
 %token REGS
+%token BARRIERS
+%token COMPILETIME
 %token BYTES
 %token LMEM
 %token SMEM
@@ -70,6 +72,8 @@ typedef void * yyscan_t;
 %token <string_value> FUNCTION
 %token <string_value> VARIABLE
 %token FATAL
+%token PERIOD
+%token MS
 
 %{
 	#include <stdlib.h>
@@ -81,6 +85,7 @@ typedef void * yyscan_t;
 	void yyerror(yyscan_t yyscanner, ptxinfo_data* ptxinfo, const char* msg);
 	void ptxinfo_function(const char *fname );
 	void ptxinfo_regs( unsigned nregs );
+	void ptxinfo_barriers( unsigned barriers );
 	void ptxinfo_lmem( unsigned declared, unsigned system );
 	void ptxinfo_gmem( unsigned declared, unsigned system );
 	void ptxinfo_smem( unsigned declared, unsigned system );
@@ -126,8 +131,10 @@ info: 	  USED INT_OPERAND REGS { ptxinfo_regs($2); }
 	| INT_OPERAND BYTES LMEM { ptxinfo_lmem($1,0); }
 	| INT_OPERAND BYTES SMEM { ptxinfo_smem($1,0); }
 	| INT_OPERAND BYTES CMEM { ptxinfo_cmem($1,0); }
+	| USED INT_OPERAND BARRIERS { ptxinfo_barriers($2); }
 	| INT_OPERAND REGS { ptxinfo_regs($1); }
 	| INT_OPERAND TEXTURES {}
+	| COMPILETIME INT_OPERAND PERIOD INT_OPERAND MS {}
 	;
 
 tuple: INT_OPERAND PLUS INT_OPERAND BYTES { g_declared=$1; g_system=$3; }