--- title: "Create Gpu" method: POST path: "/v1/admin/insert-gpu-specs" tags: ["Admin Management"] --- # Create Gpu `POST /v1/admin/insert-gpu-specs` ## Request - Base URL: `https://api.oneinfer.ai` (the document declares no server; this is the origin it was published from) - URL: `https://api.oneinfer.ai/v1/admin/insert-gpu-specs` - Auth: HTTP bearer Create a new GPU entry ## Request body - CreateGpuSpecsRequest - `gpu_name` string, required: GPU model name - `manufacturer` 'NVIDIA' | 'AMD' | 'Intel' | 'Google', required - `architechture` 'Blackwell' | 'Hopper' | 'Ada Lovelace' | 'Ampere' | 'Volta' | 'Turing' | 'CDNA3' | 'CDNA2' | 'Xe-HPG' | 'Custom', required - `clock_speeds` ClockSpeedsRequest: GPU clock speed specifications - `base_clock_mhz` integer, nullable: Base clock in MHz - `boost_clock_mhz` integer, nullable: Boost clock in MHz - `memory_clock_mhz` integer, nullable: Memory clock in MHz - `bus_interface` string, required: Bus interface type - `memory_specs` Memory, required - `size_gb` integer, nullable: Memory size in GB - `type` 'HBM3e' | 'HBM3' | 'HBM2e' | 'HBM2' | 'GDDR7' | 'GDDR6X' | 'GDDR6', required - `interface_bits` integer, nullable: Memory interface width in bits - `bandwidth_gbs` number, nullable: Memory bandwidth in GB/s - `ecc_support` boolean: ECC memory support - `chip_id` string, nullable: Reference to chip information from ChipInfoCollection - `compute_specs` GpuComputeSpecs - `cuda_cores` integer, nullable: Number of CUDA cores - `texture_mapping_units` integer, nullable: Number of texture mapping units - `render_output_units` integer, nullable: Number of render output units - `streaming_multiprocessors_count` integer, nullable, required: Number of streaming multiprocessors - `tensor_cores` integer, nullable: Number of tensor cores - `rt_cores` integer, nullable: Number of RT cores - `l1_cache_per_sm_kb` integer, nullable: L1 cache size in KB - `l2_cache_mb` integer, nullable: L2 cache size in MB - `technical_specs` TechnicalSpecsRequest: Technical specifications - `process_node_nm` integer, nullable: Manufacturing process node in nm - `transistor_count_billion` number, nullable: Transistor count in billions - `die_size_mm2` integer, nullable: Die size in mm² - `tdp_watts` integer, nullable: Thermal Design Power in watts - `pcie_version` string, nullable: PCIe version support - `nvlink_version` string, nullable: NVLink version - `interconnect_bandwidth_gbs` number, nullable: Interconnect bandwidth GB/s - `display_outputs` string[], nullable: Display output types - `dual_slot` boolean: Requires dual slot - `external_power_connectors` string[], nullable: External power connector types - `form_factor` 'SXM' | 'PCIe' | 'OAM' | 'Custom' - `gpu_performance` GpuPerformance - `pixel_rate` string, nullable: Pixel rate - `texture_rate` string, nullable: Texture rate - `ai_tops` number, nullable: AI performance in TOPS - `fp_64_tflops` number, nullable: FP64 performance in TFLOPS - `fp_32_tflops` number, nullable: FP32 performance in TFLOPS - `fp_16_tflops` number, nullable: FP16 performance in TFLOPS - `bf_16_tflops` number, nullable: BF16 performance in TFLOPS - `fp_8_tflops` number, nullable: FP8 performance in TFLOPS - `int_8_tops` number, nullable: INT8 performance in TOPS - `fp_4_tflops` number, nullable: FP4 performance in TFLOPS - `int_4_tops` number, nullable: INT4 performance in TOPS - `sparse_performance_multiplier` number, nullable: Sparse performance multiplier - `inference_performance` InferencePerformanceRequest: Inference performance for different model sizes - `model_7b_tokens_per_sec` number, nullable: 7B model tokens/second - `model_13b_tokens_per_sec` number, nullable: 13B model tokens/second - `model_30b_tokens_per_sec` number, nullable: 30B model tokens/second - `model_70b_tokens_per_sec` number, nullable: 70B model tokens/second - `model_235b_tokens_per_sec` number, nullable: 235B model tokens/second - `model_685b_tokens_per_sec` number, nullable: 685B model tokens/second - `optimal_batch_size` integer, nullable: Optimal batch size - `max_sequence_length` integer, nullable: Maximum sequence length - `memory_usage_7b_gb` number, nullable: Memory usage for 7B model in GB - `single_gpu_70b_capable` boolean: Can run 70B models on single GPU - `training_performance` TrainingPerformanceRequest: Training performance metrics relative to RTX 4090 baseline - `bert_large_multiplier` number, nullable: BERT Large training speed vs RTX 4090 - `gpt3_style_multiplier` number, nullable: GPT-3 style training speed vs RTX 4090 - `computer_vision_multiplier` number, nullable: Computer vision training speed vs RTX 4090 - `mixed_precision_multiplier` number, nullable: Mixed precision training speed vs RTX 4090 - `fp64_scientific_multiplier` number, nullable: FP64 scientific computing vs RTX 4090 - `overall_training_multiplier` number, nullable: Overall training performance vs RTX 4090 - `geekbench_scores` GeekbenchScoresRequest: Geekbench benchmark scores - `opencl_score` integer, nullable: OpenCL Geekbench score - `benchmark_results` BenchmarkResultsRequest: Standardized benchmark results - `mlperf_training_score` number, nullable: MLPerf training score - `mlperf_inference_score` number, nullable: MLPerf inference score - `llama2_7b_throughput` number, nullable: Llama2-7B tokens/second - `llama2_70b_throughput` number, nullable: Llama2-70B tokens/second - `stable_diffusion_its` number, nullable: Stable Diffusion iterations/second - `bert_large_samples_per_sec` number, nullable: BERT Large samples/second - `resnet50_images_per_sec` number, nullable: ResNet-50 images/second - `benchmark_date` string, date-time, nullable: When benchmarks were run - `benchmark_configuration` string, nullable: Benchmark configuration details - `multi_gpu_capabilities` MultiGPUCapabilitiesRequest: Multi-GPU scaling capabilities - `nvlink_support` boolean: NVLink support - `sli_support` boolean: SLI support - `max_multi_gpu_config` integer, nullable: Maximum multi-GPU configuration - `scaling_efficiency` number, nullable: Multi-GPU scaling efficiency - `mig_support` boolean: Multi-Instance GPU support - `max_mig_instances` integer, nullable: Maximum MIG instances - `software_support` SoftwareSupportRequest: Software ecosystem support - `cuda_support` boolean: CUDA support - `rocm_support` boolean: ROCm support - `oneapi_support` boolean: OneAPI support - `tensorrt_optimized` boolean: TensorRT optimization - `pytorch_support` boolean: PyTorch support - `tensorflow_support` boolean: TensorFlow support - `driver_maturity` string, nullable: Driver maturity level - `container_support` string[], nullable: Container runtime support ## Response `201` Successful Response - unknown ## Other responses - `422`: Validation Error --- [API](https://skmtc.dev/oneinfer/apis/oneinfer-apis.md) · [All operations](https://skmtc.dev/oneinfer/apis/oneinfer-apis/llms.txt) · [OpenAPI document](https://skmtc.dev/oneinfer/apis/oneinfer-apis/revisions/a49f22311a8f?raw)