AMD Zen 4 EPYC Genoa Benchmark Leak Reveals A Massive L2 Cache Per Core Upgrade
by
Zak Killian
—
Monday, March 28, 2022, 04:46 PM EDT
If you, like this author, are focused more on the consumer side of things, you may not know that AMD's EPYC server CPUs actually use the same Core Complex Dice (CCDs) as its desktop Ryzen and Threadripper CPUs—EPYC just gets more of them, along with a beefier I/O chiplet (cIOD).
So saying, whatever CPU core innovations come to EYPC are likely to show up on the desktop side of things as well. That's why it's so exciting to read confirmed architectural details on AMD's next-generation server CPUs, codenamed "Genoa."
A fresh benchmark result over in the Geekbench browser, spotted by German site ComputerBase, lists an AMD Engineering Sample marked "100-000000479-13" running on a "Quartz" platform. If regular leaker ExecuFix is to be believed, this is a sample of the Zen 4-based Genoa processor. The particular processor in the benchmark listing possesses just 32 cores and runs at only 1.2 GHz, while the final EPYC Genoa release is expected to top out at fully 96 cores and likely triple or more of that clock rate.
That's not the interesting part of this entry, though. If we peer at the cache information for this CPU, we see that it has a whole 1MB of L2 cache for each individual core. Assuming that's accurate, that means Zen 4 will come with double the L2 cache of Zen 3.
That's surely a big contributor to the 1126 single-core score, which in a vacuum would be rather poor, but considering we're talking about an early sample of a server processor here is actually pretty amazing. That's over half the single-core performance of one of AMD's current Zen 3 chips despite this CPU having a base clock of only 1.2 GHz.
There are stark few details on Zen 4 so far, but we know AMD is focusing on improving its single-threaded performance after nailing efficiency with its refreshed Zen 3 mobile parts in the Ryzen 6000 series. One way to do that is by improving core occupancy, and the easiest way to do that is to simply increase the amount of data available closer to the CPUs.
AMD saw huge gains with this approach on Zen 2, Milan-X, and its recent Radeons, so continuing the trend makes perfect sense. We'll have to see whether the extra cache makes up for the lost clock rate when the Ryzen 7 5800X3D hits the market on April 20th.