A peek at the AMD Ryzen 7 5800X3D’s performance with synthetic, non-gaming workloads has leaked out courtesy of Peru-based hardware site XanxoGaming. The upcoming 8C/16T chip with 3D V-Cache technology scored 1493/15060 (1T/nT) in Cinebench R23, 1639/10498 in Geekbench 5, and 617/6506 in CPU-Z. Here’s how those scores compare to what the AMD Ryzen 7 5800X and Intel Core i7-12900K are listed with on their respective benchmark databases.
|AMD Ryzen 7 5800X3D||AMD Ryzen 7 5800X||Intel Core i7-12900K|
|Cinebench R23||1493 / 15060||1594 / 15204||2003 / 27483|
|Geekbench 5||1639 / 10498||1671 / 10339||1992 / 17172|
|CPU-Z||617 / 6506||624 / 6328||831 / 11440|
XanxoGaming also tested the Ryzen 7 5800X3D in blender, the results of which can be compared against the Ryzen 7 5800X using Wccftech’s chart below. XanxoGaming says that it will be releasing gaming benchmarks for the CPU tomorrow.
AMD Ryzen 7 5800X3D Desktop CPU Benchmarks Leak Out, Synthetic Workloads Show Little To No Improvement (Wccftech)
What’s interesting is that the source decided to look at synthetic non-gaming workloads first for a chip that has the most notable gains promised around games. A range of benchmarks including Cinebench R23, Geekbench 5, CPU-z, and Blender was used.
In Cinebench R23, the CPU scored 1493 points in single-core and 15060 points in multi-core tests. Our AMD Ryzen 7 5800X is around 2% faster in multi-threaded and 5% faster in single-threaded performance on the same benchmark. Next up is Geekbench 5 where the chip scored 1639 points in single and 10498 points in multi-core tests. Here, the standard 5800X is 2% faster in single and 12% faster in multi-threaded performance. In CPU-z, the chip scores 617 points in single and 6505 points in multi-core tests. Unsurprisingly, the Ryzen 7 5800X beats the 3D part by 8% & 7% in the respective multi and single-core benchmarks.
In Blender, we can only compare the performance in the BMW scene since that’s the reference test we use in our own testing. The Ryzen 7 5800X3D ends up with a render time of 166 seconds while the standard chip completes the scene in 146 seconds. That’s 20 seconds shaved out without the extra 3D cache. An advantage of 14% for the non-3D part.