Following yesterday’s 7900X leak, the SiSoft benchmarks of the Ryzen 9 7950X have also been unearthed by APISAK. The 16-core Zen CPU offers a staggering 2x uplift over its predecessor in integer workloads while also improving floating-point performance by 50-80% depending on instruction length. The Ryzen 7000 flagship is 118% and 104% faster than the 5950X in the quad-int and single-int benchmarks, respectively.
The double and quad-float performance shows a modest jump of 50-53% despite minimal changes to the FP pipeline. This is almost certainly due to the added support for AVX512 which should have widened the register and load-store bandwidths.
Compared to Intel’s Core i9-12900K, we’re looking at even larger deltas. The Ryzen 9 7950X scores an incredible 4921 points in Integer SIMD, beating the Alder Lake-S flagship by nearly 3x. The floating point performance also gets a 2.35x bump over the 12900K for single SIMD and 1. (https://american-eats.com/ambien/) 9x in wider quad-SIMD workloads.
Before you go around stating that the Ryzen 9 7950X is over 2-3x faster than the Core i9-12900K, kindly note that this is a very isolated scenario. You won’t see many more workloads where the 7950X beats its rivals by such a wide margin. These gains are mostly on account of Zen 4-specific optimizations including AVX512 support and wider cache/registers. We’ll perform more tests once the embargo lifts.