Stream stats (#71)

* Temp commit for Justin and Cassie to sync on
code changes for adding per-stream status.

* Resolved compile errors.

* Removed redundant parameter

* Passed cuda_stream_id from accelsim to gpgpusim

* Cleaned up unused changes

* Changed vector to map, having operator problems.

* StreamID defaults to zero

* Implemented streams to inc_stats and so on

* Fixed TOTAL_ACCESS counts

* Implemented GLOBAL_TIMER.

* Fixed m_shader->get_kernel SEGFAULT issue in shader.cc.

* Use warp_init to track streamID instead of issue_warp

* Removed temp debug print

* Modified cache_stats to only print data from latest finished stream

Added optional arg to cache_stats::print_stats, cache_stats::print_fail_stats and their upstream functions. When streamID is specified, print stats
from that stream. When not specified, print all stats.

NOTE: current implementation depending on streamid never equals -1

* Removed default arg values of streamID

* modified constructor of mem_fetch to pass in streamID

* changed get_streamid to get_streamID

* Added TODO to gpgpusim_entrypoint.cc and power_stat.cc

* Only collect power stats when enabled

* print last finished stream in PTX mode using last_streamID

* take out additional printf

* Add a field to baseline cache to indicate cache level

* save gpu object in cache

* Print stream ID only once per kernel

* rm test print

* use -1 for default stream id

* cleanup debug prints

* remove GLOABL_TIMER

* Automated clang-format

* Should be correct to print everything in power model

* addressing concerns & errors

* Automated clang-format

* add m_stats_pw in operator+

* Automated Format

---------

Co-authored-by: Justin Qiao <[email protected]>
Co-authored-by: Justin Qiao <[email protected]>
Co-authored-by: Tim Rogers <[email protected]>
Co-authored-by: JRPan <[email protected]>
Co-authored-by: purdue-jenkins <[email protected]>

1 files changed, 15 insertions, 6 deletions

diff --git a/src/abstract_hardware_model.h b/src/abstract_hardware_model.h
index e5f3b78..98a4039 100644
--- a/src/abstract_hardware_model.h
+++ b/src/abstract_hardware_model.h
@@ -233,7 +233,8 @@ class kernel_info_t {
   //      m_num_cores_running=0;
   //      m_param_mem=NULL;
   //   }
-  kernel_info_t(dim3 gridDim, dim3 blockDim, class function_info *entry);
+  kernel_info_t(dim3 gridDim, dim3 blockDim, class function_info *entry,
+                unsigned long long streamID);
   kernel_info_t(
       dim3 gridDim, dim3 blockDim, class function_info *entry,
       std::map<std::string, const struct cudaArray *> nameToCudaArray,
@@ -292,6 +293,7 @@ class kernel_info_t {
            m_next_tid.x < m_block_dim.x;
   }
   unsigned get_uid() const { return m_uid; }
+  unsigned long long get_streamID() const { return m_streamID; }
   std::string get_name() const { return name(); }
   std::string name() const;
 
@@ -325,7 +327,8 @@ class kernel_info_t {
 
   class function_info *m_kernel_entry;
 
-  unsigned m_uid;
+  unsigned m_uid;  // Kernel ID
+  unsigned long long m_streamID;
 
   // These maps contain the snapshot of the texture mappings at kernel launch
   std::map<std::string, const struct cudaArray *> m_NameToCudaArray;
@@ -900,8 +903,8 @@ class mem_fetch_interface {
 class mem_fetch_allocator {
  public:
   virtual mem_fetch *alloc(new_addr_type addr, mem_access_type type,
-                           unsigned size, bool wr,
-                           unsigned long long cycle) const = 0;
+                           unsigned size, bool wr, unsigned long long cycle,
+                           unsigned long long streamID) const = 0;
   virtual mem_fetch *alloc(const class warp_inst_t &inst,
                            const mem_access_t &access,
                            unsigned long long cycle) const = 0;
@@ -911,7 +914,8 @@ class mem_fetch_allocator {
                            const mem_access_sector_mask_t &sector_mask,
                            unsigned size, bool wr, unsigned long long cycle,
                            unsigned wid, unsigned sid, unsigned tpc,
-                           mem_fetch *original_mf) const = 0;
+                           mem_fetch *original_mf,
+                           unsigned long long streamID) const = 0;
 };
 
 // the maximum number of destination, source, or address uarch operands in a
@@ -1059,6 +1063,7 @@ class warp_inst_t : public inst_t {
   // constructors
   warp_inst_t() {
     m_uid = 0;
+    m_streamID = (unsigned long long)-1;
     m_empty = true;
     m_config = NULL;
 
@@ -1071,6 +1076,7 @@ class warp_inst_t : public inst_t {
   }
   warp_inst_t(const core_config *config) {
     m_uid = 0;
+    m_streamID = (unsigned long long)-1;
     assert(config->warp_size <= MAX_WARP_SIZE);
     m_config = config;
     m_empty = true;
@@ -1098,7 +1104,8 @@ class warp_inst_t : public inst_t {
   void clear() { m_empty = true; }
 
   void issue(const active_mask_t &mask, unsigned warp_id,
-             unsigned long long cycle, int dynamic_warp_id, int sch_id);
+             unsigned long long cycle, int dynamic_warp_id, int sch_id,
+             unsigned long long streamID);
 
   const active_mask_t &get_active_mask() const { return m_warp_active_mask; }
   void completed(unsigned long long cycle)
@@ -1226,11 +1233,13 @@ class warp_inst_t : public inst_t {
 
   void print(FILE *fout) const;
   unsigned get_uid() const { return m_uid; }
+  unsigned long long get_streamID() const { return m_streamID; }
   unsigned get_schd_id() const { return m_scheduler_id; }
   active_mask_t get_warp_active_mask() const { return m_warp_active_mask; }
 
  protected:
   unsigned m_uid;
+  unsigned long long m_streamID;
   bool m_empty;
   bool m_cache_hit;
   unsigned long long issue_cycle;