Running out of memory when allocating to second GPU

### What is the issue? No issues with any model that fits into a single 3090 but seems to run out of memory when trying to distribute to the second 3090. ``` INFO [wmain] starting c++ runner | tid="33768" timestamp=1729324300 INFO [wmain] build info | build=3670 commit="aad7f071" tid="33768" timestamp=1729324300 INFO [wmain] system info | n_threads=20 n_threads_batch=20 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | " tid="33768" timestamp=1729324300 total_threads=28 INFO [wmain] HTTP server listening | hostname="127.0.0.1" n_threads_http="27" port="56651" tid="33768" timestamp=1729324300 llama_model_loader: loaded meta data with 41 key-value pairs and 724 tensors from C:\Users\Joshua\.ollama\models\blobs\sha256-001c9aacecbdca348f7c7c6d2b1a4120d447bf023afcacb3b864df023f1e2be4 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = llama llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Llama 3.1 70B Instruct llama_model_loader: - kv 3: general.organization str = Meta Llama llama_model_loader: - kv 4: general.finetune str = Instruct llama_model_loader: - kv 5: general.basename str = Llama-3.1 llama_model_loader: - kv 6: general.size_label str = 70B llama_model_loader: - kv 7: general.license str = llama3.1 llama_model_loader: - kv 8: …