Intel shows the performance of the 5th generation Intel Xeon Emeral Rapids
Intel has just released its latest internal projections and performance estimates for its upcoming 5th-generation Emerald Rapids and Granite Rapids server CPUs in the Xeon lineup. According to the results, the 64-core Xeons are expected to deliver up to 40% more performance than the 56-core Sapphire Rapids.
Intel talked about its Xeon CPUs during SC23. They claim to have shown real results from their upcoming 5th generation Emerald Rapids and forecasts for their next generation Xeon Granite Ridge CPUs. The results were compared to their own 4th generation Sapphire Rapids chips, while presenting the HPC metrics of the Xeon Max CPUs against AMD's 96-core EPYC Genoa models.
Intel boasts the power of the 5th generation of Intel Xeon CPUs
The 5th generation Intel Emerald Rapids CPUs will be backward compatible with the Sapphire Rapids CPUs of the Eagle Stream platform . They use the same LGA-4677 socket and primarily offer an optimized design with fewer cores and additional threads, larger cache reserves, and a focus on higher performance per watt.
The next Intel Xeon processors will use the LGA-4677 socket; learn about what a socket is.
They highlighted that the Xeon Emerald Rapids CPUs will offer optimized performance for workloads and energy- efficient computing . The addition of faster DDR5-5600 memory should result in greater bandwidth and increased transfer speeds. They will also support CXL 1.0 and 2.0 with up to 80 PCIe Gen 5.0 lanes.
Confirmed features of the 5th Gen Intel Xeon “Emerald Rapids” CPUs include:
- LLC cache up to 3 times larger.
- Higher memory speed.
- Increased number of CPU cores, with configurations of up to 64 cores.
- CXL Type 3 interface with higher bandwidth.
- Workload optimizations with Intel AE.
- Optimized feeding mode.
- Up to 17% increase in general purpose performance and power.
- Offloading CPU cores with Intel accelerator engines for greater energy efficiency
- Huge increase in performance in inference and training
- Intel AMX for integrated AI acceleration
- Immediate deployment with optimized software stacks
