From 5b41cd7e5db78616387392872de15299a2432617 Mon Sep 17 00:00:00 2001
From: LoganDark <git@logandark.mozmail.com>
Date: Sat, 3 Jun 2023 03:44:50 -0700
Subject: [PATCH] Add capability for extra binaries to be built with rwkv.cpp
 (#87)

* Add capability for examples

This also adds a quantizer that works without python.
in the future, we might be able to convert from pytorch as well,
without python.

* example implied code style

* rename examples to tools

* rename cpuinfo.c to cpu_info.c

* include ggml header again

* Return EXIT_FAILURE on help

* done with this

* final name: extras

* going To have a seizure

* wait literal double n
---
 CMakeLists.txt        |  1 +
 extras/CMakeLists.txt | 10 ++++++++
 extras/cpu_info.c     |  7 ++++++
 extras/quantize.c     | 58 +++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 76 insertions(+)
 create mode 100644 extras/CMakeLists.txt
 create mode 100644 extras/cpu_info.c
 create mode 100644 extras/quantize.c

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 38931e5..38497c4 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -286,3 +286,4 @@ endif()
 
 enable_testing()
 add_subdirectory(tests)
+add_subdirectory(extras)
diff --git a/extras/CMakeLists.txt b/extras/CMakeLists.txt
new file mode 100644
index 0000000..ff634df
--- /dev/null
+++ b/extras/CMakeLists.txt
@@ -0,0 +1,10 @@
+function(rwkv_add_extra source)
+	get_filename_component(EXTRA_TARGET ${source} NAME_WE)
+	add_executable(rwkv_${EXTRA_TARGET} ${source})
+	target_link_libraries(rwkv_${EXTRA_TARGET} PRIVATE ggml rwkv)
+endfunction()
+
+file(GLOB extras *.c)
+foreach (extra ${extras})
+	rwkv_add_extra(${extra})
+endforeach()
\ No newline at end of file
diff --git a/extras/cpu_info.c b/extras/cpu_info.c
new file mode 100644
index 0000000..115caf8
--- /dev/null
+++ b/extras/cpu_info.c
@@ -0,0 +1,7 @@
+#include "rwkv.h"
+
+#include <stdio.h>
+
+int main() {
+    printf("%s", rwkv_get_system_info_string());
+}
\ No newline at end of file
diff --git a/extras/quantize.c b/extras/quantize.c
new file mode 100644
index 0000000..ed98ac3
--- /dev/null
+++ b/extras/quantize.c
@@ -0,0 +1,58 @@
+#include "ggml.h"
+#include "rwkv.h"
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+
+enum ggml_type type_from_string(const char* string) {
+    if (strcmp(string, "Q4_0") == 0) return GGML_TYPE_Q4_0;
+    if (strcmp(string, "Q4_1") == 0) return GGML_TYPE_Q4_1;
+    if (strcmp(string, "Q5_0") == 0) return GGML_TYPE_Q5_0;
+    if (strcmp(string, "Q5_1") == 0) return GGML_TYPE_Q5_1;
+    if (strcmp(string, "Q8_0") == 0) return GGML_TYPE_Q8_0;
+    return GGML_TYPE_COUNT;
+}
+
+#ifdef _WIN32
+bool QueryPerformanceFrequency(uint64_t* lpFrequency);
+bool QueryPerformanceCounter(uint64_t* lpPerformanceCount);
+
+#define time_t uint64_t
+#define time_calibrate(freq) do { QueryPerformanceFrequency(&freq); freq /= 1000; } while (0)
+#define time_measure(x) QueryPerformanceCounter(&x)
+#define TIME_DIFF(freq, start, end) (double) ((end - start) / freq) / 1000.
+#else
+#include <time.h>
+
+#define time_t struct timespec
+#define time_calibrate(freq) (void) freq
+#define time_measure(x) clock_gettime(CLOCK_MONOTONIC, &x)
+#define TIME_DIFF(freq, start, end) (double) ((end.tv_nsec - start.tv_nsec) / 1000000) / 1000
+#endif
+
+int main(int argc, char* argv[]) {
+    if (argc != 4 || type_from_string(argv[3]) == GGML_TYPE_COUNT) {
+        fprintf(stderr, "Usage: %s INPUT OUTPUT FORMAT\n\nAvailable formats: Q4_0 Q4_1 Q5_0 Q5_1 Q8_0\n", argv[0]);
+        return EXIT_FAILURE;
+    }
+
+    time_t freq, start, end;
+    time_calibrate(freq);
+
+    fprintf(stderr, "Quantizing ...\n");
+
+    time_measure(start);
+    bool success = rwkv_quantize_model_file(argv[1], argv[2], argv[3]);
+    time_measure(end);
+
+    double diff = TIME_DIFF(freq, start, end);
+
+    if (success) {
+        fprintf(stderr, "Succeeded in %.3fs\n", diff);
+        return EXIT_SUCCESS;
+    } else {
+        fprintf(stderr, "Error in %.3fs: 0x%.8X\n", diff, rwkv_get_last_error(NULL));
+        return EXIT_FAILURE;
+    }
+}
\ No newline at end of file