AMD showed how server processors affect AI computer performance during their recent technology presentation. Executive VP Forrest Norrod explained that data centers need balanced CPU and GPU combinations for best results. The company tested their EPYC 9575F processor against Intel Xeon 8592+ chips from two years ago. Both processors have 64 cores and similar power usage ratings. AMD paired each processor with their Instinct MI300X AI accelerator card for fair testing.
The EPYC processor delivered 6 percent better performance on average across different AI tasks. Performance gaps reached 17 percent when running larger AI models with more parameters. AMD focused their tests on inference workloads that process AI requests from users. The company used Llama 3.1 8B model for most benchmark comparisons. Results showed that choosing the right server processor makes a major difference for AI applications.
AMD continues gaining market share against Intel across the server processor business. Their EPYC chips have become popular choices for data center customers. The company credits balanced system design for their growing success. Server builders want processors that work well with AI accelerator cards. AMD expects their market position to keep improving as AI demand grows.
The EPYC processor delivered 6 percent better performance on average across different AI tasks. Performance gaps reached 17 percent when running larger AI models with more parameters. AMD focused their tests on inference workloads that process AI requests from users. The company used Llama 3.1 8B model for most benchmark comparisons. Results showed that choosing the right server processor makes a major difference for AI applications.
AMD continues gaining market share against Intel across the server processor business. Their EPYC chips have become popular choices for data center customers. The company credits balanced system design for their growing success. Server builders want processors that work well with AI accelerator cards. AMD expects their market position to keep improving as AI demand grows.
