From ee69cffdbbd7399b2981507974d3b407e1f789fd Mon Sep 17 00:00:00 2001
From: Dr-Noob <peibolms@gmail.com>
Date: Sun, 8 Sep 2024 13:46:02 +0100
Subject: [PATCH] WIP: Move accurate-pp to get_cpu_info. Use a vector as
 argument to store freq

---
 src/common/cpu.c    |  6 ++++++
 src/common/cpu.h    | 10 ++++++++++
 src/x86/cpuid.c     | 31 +++++++++++++++++++++++--------
 src/x86/cpuid.h     |  2 +-
 src/x86/freq/freq.c | 21 ++++++++++++---------
 src/x86/freq/freq.h |  2 +-
 6 files changed, 53 insertions(+), 19 deletions(-)

diff --git a/src/common/cpu.c b/src/common/cpu.c
index 946c990..e3f6bcb 100644
--- a/src/common/cpu.c
+++ b/src/common/cpu.c
@@ -34,6 +34,12 @@ int64_t get_freq(struct frequency* freq) {
   return freq->max;
 }
 
+#ifdef ARCH_X86
+int64_t get_freq_pp(struct frequency* freq) {
+  return freq->max_pp;
+}
+#endif
+
 #if defined(ARCH_X86) || defined(ARCH_PPC)
 char* get_str_cpu_name(struct cpuInfo* cpu, bool fcpuname) {
   #ifdef ARCH_X86
diff --git a/src/common/cpu.h b/src/common/cpu.h
index aabcee7..3e63e8b 100644
--- a/src/common/cpu.h
+++ b/src/common/cpu.h
@@ -60,6 +60,11 @@ struct frequency {
   int32_t max;
   // Indicates if max frequency was measured
   bool measured;
+#ifdef ARCH_X86
+  // Max frequency when running vectorized code.
+  // Used only for peak performance computation.
+  int32_t max_pp;
+#endif
 };
 
 struct hypervisor {
@@ -188,6 +193,8 @@ struct cpuInfo {
 #ifdef ARCH_X86
   // The index of the first core in the module
   uint32_t first_core_id;
+  // The index of this module
+  uint32_t module_id;
 #endif
 #endif
 };
@@ -200,6 +207,9 @@ uint32_t get_nsockets(struct topology* topo);
 
 VENDOR get_cpu_vendor(struct cpuInfo* cpu);
 int64_t get_freq(struct frequency* freq);
+#ifdef ARCH_X86
+int64_t get_freq_pp(struct frequency* freq);
+#endif
 
 char* get_str_aes(struct cpuInfo* cpu);
 char* get_str_sha(struct cpuInfo* cpu);
diff --git a/src/x86/cpuid.c b/src/x86/cpuid.c
index dd67ee0..0753399 100644
--- a/src/x86/cpuid.c
+++ b/src/x86/cpuid.c
@@ -210,18 +210,14 @@ int64_t get_peak_performance(struct cpuInfo* cpu, bool accurate_pp) {
 
   for(int i=0; i < cpu->num_cpus; ptr = ptr->next_cpu, i++) {
     struct topology* topo = ptr->topo;
-    int64_t max_freq = get_freq(ptr->freq);
+    int64_t freq = get_freq(ptr->freq);
 
-    int64_t freq;
   #ifdef __linux__
     if(accurate_pp)
-      freq = measure_frequency(ptr);
-    else
-      freq = max_freq;
+      freq = get_freq_pp(ptr->freq);
   #else
     // Silence compiler warning
     (void)(accurate_pp);
-    freq = max_freq;
   #endif
 
     //First, check we have consistent data
@@ -518,6 +514,7 @@ struct cpuInfo* get_cpu_info(void) {
 
   if(cpu->hybrid_flag) cpu->num_cpus = 2;
 
+  int32_t *max_freq_pp_vec = NULL;
   struct cpuInfo* ptr = cpu;
   for(uint32_t i=0; i < cpu->num_cpus; i++) {
     int32_t first_core;
@@ -546,10 +543,14 @@ struct cpuInfo* get_cpu_info(void) {
       ptr->core_type = get_core_type();
     }
     ptr->first_core_id = first_core;
+    ptr->module_id = i;
     ptr->feat = get_features_info(ptr);
 
     ptr->arch = get_cpu_uarch(ptr);
-    ptr->freq = get_frequency_info(ptr);
+    // If accurate_pp is requested, we need to get the frequency
+    // after fetching the topology. Otherwise we can do it now.
+    if (!accurate_pp())
+      ptr->freq = get_frequency_info(ptr, accurate_pp(), max_freq_pp_vec);
 
     if (cpu->cpu_name == NULL && ptr == cpu) {
       // If we couldnt read CPU name from cpuid, infer it now
@@ -568,6 +569,11 @@ struct cpuInfo* get_cpu_info(void) {
     // If topo is NULL, return early, as get_peak_performance
     // requries non-NULL topology.
     if(ptr->topo == NULL) return cpu;
+
+    // If accurate_pp is requested, we need to get the frequency
+    // after fetching the topology
+    if (accurate_pp())
+      ptr->freq = get_frequency_info(ptr, accurate_pp(), max_freq_pp_vec);
   }
 
   cpu->peak_performance = get_peak_performance(cpu, accurate_pp());
@@ -935,7 +941,7 @@ struct cache* get_cache_info(struct cpuInfo* cpu) {
   return cach;
 }
 
-struct frequency* get_frequency_info(struct cpuInfo* cpu) {
+struct frequency* get_frequency_info(struct cpuInfo* cpu, bool accurate_pp, int32_t *max_freq_pp_vec) {
   struct frequency* freq = emalloc(sizeof(struct frequency));
   freq->measured = false;
 
@@ -1005,6 +1011,15 @@ struct frequency* get_frequency_info(struct cpuInfo* cpu) {
     }
   #endif
 
+  freq->max_pp = UNKNOWN_DATA;
+  #ifdef __linux__
+    if(accurate_pp)
+      freq->max_pp = measure_frequency(cpu, max_freq_pp_vec);
+  #else
+    // Silence compiler warning
+    (void)(accurate_pp);
+  #endif
+
   return freq;
 }
 
diff --git a/src/x86/cpuid.h b/src/x86/cpuid.h
index a0a6a5c..0b57bb6 100644
--- a/src/x86/cpuid.h
+++ b/src/x86/cpuid.h
@@ -5,7 +5,7 @@
 
 struct cpuInfo* get_cpu_info(void);
 struct cache* get_cache_info(struct cpuInfo* cpu);
-struct frequency* get_frequency_info(struct cpuInfo* cpu);
+struct frequency* get_frequency_info(struct cpuInfo* cpu, bool accurate_pp, int32_t *max_freq_pp_vec);
 struct topology* get_topology_info(struct cpuInfo* cpu, struct cache* cach, int module);
 
 char* get_str_avx(struct cpuInfo* cpu);
diff --git a/src/x86/freq/freq.c b/src/x86/freq/freq.c
index 10171c0..de57a0a 100644
--- a/src/x86/freq/freq.c
+++ b/src/x86/freq/freq.c
@@ -26,7 +26,7 @@ struct freq_thread {
   bool end;
   bool measure;
   // Output
-  double freq;
+  int32_t *max_pp;
 };
 
 double vector_average_harmonic(double* v, int len) {
@@ -88,34 +88,37 @@ void* measure_freq(void *freq_ptr) {
     double* freq_vector_ptr = freq_vector;
 
     for (int i=0; i < cpu->num_cpus; ptr = ptr->next_cpu, i++) {
-      ptr->freq->max_pp = vector_average_harmonic(freq_vector_ptr, ptr->topo->total_cores_module);
-      printWarn("AVX2 measured freq=%d (module %d)", ptr->freq->max_pp, i);
+      freq->max_pp[i] = vector_average_harmonic(freq_vector_ptr, ptr->topo->total_cores_module);
+      printWarn("AVX2 measured freq=%d (module %d)", freq->max_pp[i], i);
 
       freq_vector_ptr = freq_vector_ptr + ptr->topo->total_cores_module;
     }
   }
   else {
-    cpu->freq->max_pp = vector_average_harmonic(freq_vector, v);
-    printWarn("AVX2 measured freq=%d\n", cpu->freq->max_pp);
+    freq->max_pp[0] = vector_average_harmonic(freq_vector, v);
+    printWarn("AVX2 measured freq=%d\n", freq->max_pp[0]);
   }
 
   return NULL;
 }
 
-int64_t measure_frequency(struct cpuInfo* cpu) {
-  if (cpu->hybrid_flag && cpu->first_core_id > 0) {
+int64_t measure_frequency(struct cpuInfo* cpu, int32_t *max_freq_pp_vec) {
+  if (cpu->hybrid_flag && cpu->module_id > 0) {
     // We have a hybrid architecture and we have already
     // measured the frequency for this module in a previous
     // call to this function, so now just return it.
-    return get_freq_pp(cpu->freq);
+    return max_freq_pp_vec[cpu->module_id];
   }
 
+  max_freq_pp_vec = malloc(sizeof(int32_t) * cpu->num_cpus);
+
   int ret;
   int num_spaces;
   struct freq_thread* freq_struct = malloc(sizeof(struct freq_thread));
   freq_struct->end = false;
   freq_struct->measure = false;
   freq_struct->cpu = cpu;
+  freq_struct->max_pp = max_freq_pp_vec;
 
   void* (*compute_function)(void*);
 
@@ -187,5 +190,5 @@ int64_t measure_frequency(struct cpuInfo* cpu) {
   }
 
   printf("\r%*c", num_spaces, ' ');
-  return cpu->freq->max;
+  return max_freq_pp_vec[0];
 }
diff --git a/src/x86/freq/freq.h b/src/x86/freq/freq.h
index de62916..8d79db3 100644
--- a/src/x86/freq/freq.h
+++ b/src/x86/freq/freq.h
@@ -8,6 +8,6 @@
 #define MEASURE_TIME_SECONDS         5
 #define LOOP_ITERS           100000000
 
-int64_t measure_frequency(struct cpuInfo* cpu);
+int64_t measure_frequency(struct cpuInfo* cpu, int32_t *max_freq_pp_vec);
 
 #endif