ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-01-26 17:20:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

Remove unused file

Browse Source
This commit is contained in:
Kawrakow
2024-07-24 19:33:19 +03:00
parent 6fd0a92cb0
commit e0b2dd511c
4 changed files with 1 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
Makefile
View File

@@ -708,7 +708,7 @@ OBJS += ggml-alloc.o ggml-backend.o ggml-quants.o unicode.o unicode-data.o
COMMON_H_DEPS = common/common.h common/sampling.h common/log.h llama.h
COMMON_DEPS = common.o sampling.o grammar-parser.o build-info.o json-schema-to-grammar.o
iqk-quantize.o: iqk-quantize.cpp iqk-quantize.h iqk_mul_mat.h ggml-quants.h ggml-common.h ggml.h ggml-impl.h
iqk-quantize.o: iqk-quantize.cpp iqk_mul_mat.h ggml-quants.h ggml-common.h ggml.h ggml-impl.h
$(CXX) $(CXXFLAGS) -c $< -o $@
ifndef LLAMA_NO_IQK_MULMAT
iqk_mul_mat.o: iqk_mul_mat.cpp ggml-impl.h ggml.h ggml-quants.h ggml-common.h iqk_mul_mat.h

1
iqk-quantize.cpp
View File

@@ -13,7 +13,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
#include "iqk-quantize.h"
#if GGML_USE_IQK_MULMAT
#include "iqk_mul_mat.h"
#endif

46
iqk-quantize.h
View File

@@ -1,46 +0,0 @@
#pragma once
#include <stdint.h>
typedef union {
float f;
uint32_t i;
} iq1bn_scale_t;
#ifdef __cplusplus
extern "C" {
#endif
#ifdef BITNET_IQ1BN_4x4
static inline float iq1bn_min_value(void) { return 1.9074e-06f; }
static inline float iq1bn_max_value(void) { return 0.12109f; }
#else
static inline float iq1bn_min_value(void) { return 0.000488281f; }
static inline float iq1bn_max_value(void) { return 0.123047f; }
#endif
static inline uint8_t iq1bn_float_to_fp8(float f) {
if (f <= iq1bn_min_value()) return 0;
if (f >= iq1bn_max_value()) return 255;
iq1bn_scale_t s;
s.f = f;
#ifdef BITNET_IQ1BN_4x4
return ((((s.i >> 23) + 132) & 0xf) << 4) | ((s.i >> 19) & 0xf);
#else
return ((s.i >> 18) & 0x1f) | (((s.i >> 23) - 116) << 5);
#endif
}
static inline float iq1bn_fp8_to_float(uint8_t fp8) {
iq1bn_scale_t s;
#ifdef BITNET_IQ1BN_4x4
s.i = ((((fp8 >> 4) | 0xf0) - 132) << 23) | ((fp8 & 0x0f) << 19);
#else
s.i = (((fp8 >> 5) + 116) << 23) | ((fp8 & 0x1f) << 18);
#endif
return s.f;
}
#ifdef __cplusplus
}
#endif

1
iqk_mul_mat.cpp
View File

@@ -31,7 +31,6 @@
#include "ggml-impl.h"
#include "ggml-quants.h"
#include "iqk_mul_mat.h"
#include "iqk-quantize.h"
#define GGML_COMMON_IMPL_C
#include "ggml-common.h"