AMD's Ryzen AI 300 Series Outperforms Intel in LLM Performance Benchmarks

AMD Showcases Superior LLM Performance with Ryzen AI 300 Series

AMD has recently unveiled benchmark results demonstrating the superior performance of its Ryzen AI 300 series mobile processors, particularly the Ryzen AI 9 HX 375, in large language model (LLM) workloads. The tests, conducted using LM Studio, a popular desktop application for running LLMs locally, show significant advantages over Intel's competing Core Ultra 7 258V processor 1

1

2

2

.

Benchmark Results and Performance Gains

The Ryzen AI 9 HX 375 demonstrated up to 27% faster token generation speeds compared to Intel's Core Ultra 7 258V across various LLM models, including Meta Llama 3.2, Microsoft Phi 3.1, Google Gemma 2, and Mistral Nemo 2407 1

1

3

3

. Additionally, AMD's processor showed impressive improvements in latency, delivering up to 3.5x lower latency in certain models 2

2

.

Key performance highlights include:

• Up to 27% higher performance in consumer LLM applications in LM Studio
• 3.5x lower latency in the Mistral Nemo 2407 12b Instruct model
• 13.1% performance lead in Intel's AI Playground using the Mistral 7b Instruct v0.3 model
• 8.7% faster performance on Microsoft Phi 3.1 Mini Instruct test 1

  1

  2

  2

  4

  4

GPU Acceleration and Variable Graphics Memory

AMD's Ryzen AI 300 series processors leverage both CPU and GPU capabilities for AI tasks. When utilizing GPU acceleration through the Vulkan API, the Ryzen AI 9 HX 375's Radeon 890M iGPU demonstrated up to 23% faster performance than Intel's Arc 140V 1

1

3

3

.

Furthermore, AMD introduced Variable Graphics Memory (VGM) technology, which allows for memory reallocation in iGPU-oriented tasks. This feature, combined with GPU acceleration, resulted in a 60% higher performance boost compared to CPU-only operations 4

4

.

Comparison Considerations

It's worth noting that the comparison between AMD's flagship Ryzen AI 9 HX 375 and Intel's mid-range Core Ultra 7 258V may not represent a completely fair matchup. The Core Ultra 9 288V, Intel's top-tier offering, was not included in these benchmarks 3

3

5

5

.

Implications for Consumer AI Applications

These performance gains are particularly relevant as the demand for local LLM processing grows. AMD is actively working to make LLMs more accessible to users without extensive technical knowledge, highlighting the importance of user-friendly tools like LM Studio 1

1

2

2

.

As AI integration becomes more prevalent in everyday computing, the ability to run LLMs efficiently on personal devices could become increasingly important for consumers and businesses alike 4

4

.

Industry Competition and Future Outlook

The benchmark results underscore the intensifying competition in the AI hardware space, with both AMD and Intel vying for dominance in the rapidly evolving market for AI-capable processors. As the demand for AI performance in consumer devices continues to grow, we can expect further innovations and performance improvements from both companies 1

1

3

3

5

5

.