From 0a11f50da89d035ca37a87ae768ccc2b1a6fa330 Mon Sep 17 00:00:00 2001
From: Concedo <39025047+LostRuins@users.noreply.github.com>
Date: Tue, 18 Jul 2023 20:26:18 +0800
Subject: [PATCH] reenabled sched_yield, reduced sampler warning msg to once
 per session

---
 Makefile     | 4 ++--
 ggml.c       | 2 +-
 koboldcpp.py | 7 +++++--
 3 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/Makefile b/Makefile
index e580e6c6b..374ee0a74 100644
--- a/Makefile
+++ b/Makefile
@@ -144,7 +144,7 @@ ifdef LLAMA_CUBLAS
 	CUBLASLD_FLAGS = -lcublas -lculibos -lcudart -lcublasLt -lpthread -ldl -lrt -L/usr/local/cuda/lib64 -L/opt/cuda/lib64 -L$(CUDA_PATH)/targets/x86_64-linux/lib
 	CUBLAS_OBJS = ggml-cuda.o ggml_v2-cuda.o ggml_v2-cuda-legacy.o
 	NVCC      = nvcc
-	NVCCFLAGS = --forward-unknown-to-host-compiler
+	NVCCFLAGS = --forward-unknown-to-host-compiler -use_fast_math
 ifdef CUDA_DOCKER_ARCH
 	NVCCFLAGS += -Wno-deprecated-gpu-targets -arch=$(CUDA_DOCKER_ARCH)
 else
@@ -358,7 +358,7 @@ koboldcpp_openblas: ggml_openblas.o ggml_v2_openblas.o ggml_v1.o expose.o common
 	$(OPENBLAS_BUILD)
 koboldcpp_failsafe: ggml_failsafe.o ggml_v2_failsafe.o ggml_v1_failsafe.o expose.o common.o gpttype_adapter_failsafe.o k_quants_failsafe.o $(OBJS)
 	$(FAILSAFE_BUILD)
-koboldcpp_openblas_noavx2: ggml_openblas_noavx2.o ggml_v2_openblas_noavx2.o ggml_v1_failsafe.o expose.o common.o gpttype_adapter.o k_quants_noavx2.o $(OBJS)
+koboldcpp_openblas_noavx2: ggml_openblas_noavx2.o ggml_v2_openblas_noavx2.o ggml_v1_failsafe.o expose.o common.o gpttype_adapter_failsafe.o k_quants_noavx2.o $(OBJS)
 	$(OPENBLAS_NOAVX2_BUILD)
 koboldcpp_clblast: ggml_clblast.o ggml_v2_clblast.o ggml_v1.o expose.o common.o gpttype_adapter_clblast.o ggml-opencl.o ggml_v2-opencl.o ggml_v2-opencl-legacy.o k_quants.o $(OBJS)
 	$(CLBLAST_BUILD)
diff --git a/ggml.c b/ggml.c
index d6ba8ac97..a288b778d 100644
--- a/ggml.c
+++ b/ggml.c
@@ -16383,7 +16383,7 @@ static thread_ret_t ggml_graph_compute_thread(void * data) {
             // wait for other threads to finish
             const int last = node_n;
             do {
-                //sched_yield();
+                sched_yield();
                 node_n = atomic_load(&state->shared->node_n);
             } while (node_n == last);
         }
diff --git a/koboldcpp.py b/koboldcpp.py
index d269ae5b8..1c55fd302 100755
--- a/koboldcpp.py
+++ b/koboldcpp.py
@@ -242,8 +242,10 @@ def generate(prompt,max_length=20, max_context_length=512, temperature=0.8, top_
             for i, sampler in enumerate(sampler_order):
                 inputs.sampler_order[i] = sampler
             inputs.sampler_len = len(sampler_order)
-            if inputs.sampler_len>0 and (inputs.sampler_order[0]!=6 or inputs.sampler_order[inputs.sampler_len-1]!=5):
-                print("\n(Warning!!! Poor sampler_order detected! You will have reduced quality. Recommended values are [6,0,1,3,4,2,5])")
+            global showsamplerwarning
+            if showsamplerwarning and inputs.sampler_len>0 and (inputs.sampler_order[0]!=6 or inputs.sampler_order[inputs.sampler_len-1]!=5):
+                print("\n(Note: Sub-optimal sampler_order detected. You may have reduced quality. Recommended sampler values are [6,0,1,3,4,2,5]. This message will only show once per session.)")
+                showsamplerwarning = False
         except TypeError as e:
             print("ERROR: sampler_order must be a list of integers: " + str(e))
     inputs.seed = seed
@@ -277,6 +279,7 @@ modelbusy = False
 defaultport = 5001
 KcppVersion = "1.36"
 showdebug = True
+showsamplerwarning = True
 
 class ServerRequestHandler(http.server.SimpleHTTPRequestHandler):
     sys_version = ""