A bunch of SiSoft benchmarks of Intel’s upcoming Rocket Lake-S CPUs has surfaced, showing detailed performance stats across various workloads. However, as you might have already guessed, these are engineering samples running at a paltry 1.8GHz, so much of the final results here are based on speculative scaling:
In the media processing test which gives a good idea about the integer and floating-point performance at different precision rates, the supposed Core i9-11900K manages 601.83Mpix/s in the Integer and 497.60Mpix/s in the FP benchmarks respectively. In comparison (above image), the Ryzen 7 5800X manages 1,504 points and 1,547 points in the same two tests while the Core i9-9900K records just 1,000 points in the Int and 998 points in the FP test.
Considering that the Rocket Lake part is running at less than half of its final operating clocks, it’d be fair to assume the scores to scale up to 2.3-2.5x as the boost frequency is tripled to 5.4GHz. That gives us a score of roughly 1,400 to 1,500Mpix/s in the Integer test same as the Ryzen 7 5800X, while close to 1,200Mpix/s in the FP test, a fair bit slower than the 5800X but a healthy 20% more than the Core i9-9900K.
Scaling the image processing “Blur” score by 2.5x, we get nearly 3,100 points which is again a 20% gain over the 9900K. The sharpen, motion blur, and edge-detection algorithms which are all indicative of render performance come up to 1,400 points, 725 points, and 1,050 points, respectively. These scores are again in line with the Zen 3 performance, allowing us to expect a decent showing from Intel’s next batch of processors.
