While AMD’s Ryzen 7000 series, codenamed Raphael, and Intel’s 13th-generation Core series, codenamed Raptor Lake, have yet to go on sale, however, with more and more official information and leaked test results, there is no shortage of performance comparisons in various aspects on the Internet.
Recently, Twitter user @OneRaichu shared some details of the new generation processors of Intel and AMD, saying that the huge cache configuration of the Zen 4 architecture processor has greatly alleviated the dependence on memory. In addition, due to the impact of DDR5 memory delay, Gear 2’s DDR5-4800 is 15ns higher than Gear 1’s DDR4-3200 in latency. At the same time, OneRaichu also compared the IPC performance data of Raptor Cove, Golden Cove, and Zen 4 architecture cores. In addition to P-Core, Raptor Lake and Alder Lake E-Core based on Gracemont architecture were also added.
In the test conducted by SPECCPU2017 Rate 1, the frequency was fixed at 3.6GHz, DDR5-4800 memory was used, and the IPC performance under
DDR5-6000 memory was also compared.
Image: OneRaichu
When using DDR5-4800 memory, the Zen 4 architecture finally achieved the highest score in SPECint with a score of 6.66, even surpassing the Intel Raptor Cove architecture P-Core’s 6.63 points and the Golden Cove architecture’s score of 6.52 points. In the SPECfp test, Intel Raptor Cove, Golden Cove, and Zen 4 architecture scored 10.21 points, 9.91 points, and 9.99 points, respectively, and the overall performance was consistent.
Image: OneRaichu
If you switch to DDR5-6000 memory, which AMD calls the best memory frequency, the general situation is similar, and the IPC performance of the Raptor Cove and Zen 4 architecture is not much different. The biggest surprise comes from the Gracemont architecture used by the E-Core, the list is differentiated by GLC-12 (Alder Lake) and GLC-13 (Raptor Lake), it can be seen that after the L2 cache is increased, the Gracemont architecture is upgraded to a level comparable to Skylake’s IPC. Combined with the number of transistors that the E-Cores take up, it’s not hard to see why Intel is willing to give its processors a lot of E-Cores and a small amount of P-Cores.
