The Register on MSN5dOpinion
Intel has officially missed the boat for AI in the datacenter
Intel's opportunities to capitalize on the AI boom may be shrinking in the datacenter, Chipzilla still has a shot at the ...
GitHub10d
TensorRT Model Optimizer Benchmark Reference
Config: H200, nvidia-modelopt v0.21.1, TensorRT-LLM v0.15, latency measured with trtllm-bench. Inference speedup are compared to the BF16 baseline. Speedup is normalized to the GPU count. Benchmark ...
GitHub28d
TensorRT Model Optimizer Benchmark Reference
The table below shows the MMLU loss in percentage compared to BF16 baseline. Config: H100, nvidia-modelopt v0.21.1, TenorR-LLM v0.15. Note that typically FP8 is the go-to choices for H100. 4-bit AWQ ...
DeepSeek research suggests Huawei's Ascend 910C delivers 60% of Nvidia H100 inference performance
Nvidia's GPUs remain the best solutions for AI training, but Huawei's own processors can be used for inference.
TweakTown13d
Chinese AI firm DeepSeek has 50,000 NVIDIA H100 AI GPUs says CEO, even with US restrictions
Chinese AI company DeepSeek says its DeepSeek R1 model is as good, or better than OpenAI's new o1 says CEO: powered by 50,000 ...
Digi Times2d
Huawei reports 2024 revenue, Ascend 910C hits 60% of Nvidia H100's AI performance
Huawei Chairman Howard Liang announced that 2024 revenue exceeded CNY860 billion (approx. US$118.6 billion) at the Guangdong ...