From f98b1befdba40651406700fc06b35463e0f988bb Mon Sep 17 00:00:00 2001 From: usrlocalben Date: Tue, 19 Aug 2025 06:41:29 -0500 Subject: [PATCH] remove curious assertions (#705) This assertion can hit during prefill as MLA/KV tensors grow, e.g. Kimi K2 n_ctx >= 32768. --- ggml/src/ggml-cuda/cpy.cu | 3 --- 1 file changed, 3 deletions(-) diff --git a/ggml/src/ggml-cuda/cpy.cu b/ggml/src/ggml-cuda/cpy.cu index 7fb8d9db..04aa3e26 100644 --- a/ggml/src/ggml-cuda/cpy.cu +++ b/ggml/src/ggml-cuda/cpy.cu @@ -377,9 +377,6 @@ void ggml_cuda_cpy(ggml_backend_cuda_context & ctx, const ggml_tensor * src0, gg const int64_t ne = ggml_nelements(src0); GGML_ASSERT(ne == ggml_nelements(src1)); - GGML_ASSERT(ggml_nbytes(src0) <= INT_MAX); - GGML_ASSERT(ggml_nbytes(src1) <= INT_MAX); - const int64_t ne00 = src0->ne[0]; const int64_t ne01 = src0->ne[1]; const int64_t ne02 = src0->ne[2];