From: Richard <redacted>
Date: Wed, 12 Feb 2025 13:57:33 +0000 (+0000)
Subject: ggml : fix multi-threaded clamp_f32 (llama/11824)
X-Git-Tag: upstream/1.7.4+203~78
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=d597f83e1aa43d62556cc52b46ab6ffb819373e9;p=pkg%2Fggml%2Fsources%2Fwhisper.cpp

ggml : fix multi-threaded clamp_f32 (llama/11824)

* Bug fix for clamp_f32

When using tensors larger than 1d clamp operation does not work due to the restriction of returning if ith is not 0.

* Bug fix for clamp_f32

* Bug fix for clamp_f32
---

diff --git a/ggml/src/ggml-cpu/ggml-cpu.c b/ggml/src/ggml-cpu/ggml-cpu.c
index fdb430a4..fcbb5c23 100644
--- a/ggml/src/ggml-cpu/ggml-cpu.c
+++ b/ggml/src/ggml-cpu/ggml-cpu.c
@@ -9074,10 +9074,6 @@ static void ggml_compute_forward_clamp_f32(
 
     const struct ggml_tensor * src0 = dst->src[0];
 
-    if (params->ith != 0) {
-        return;
-    }
-
     float min;
     float max;
     memcpy(&min, (float *) dst->op_params + 0, sizeof(float));