A mysterious image was recently posted on Twitter by famed tech leaker HXL. It eventually turned out to be the L3 Cache test for AMD's upcoming R9 7900X.

It appears that HXL obtained a Zen4 chip accidentally. So, we now have some benchmarks for AMD's notoriously shrouded Ryzen 7000 CPUs.

Our R9 7900X achieved great results in the L3 Write, L3 Read, and L3 Copy tests using the AIDA64 testing tool, scoring 1494.8GB/s, 1445.7GB/s, and 1476.6GB/s, respectively.

With the new 7900X, even the latency has greatly decreased, reaching a new low of 10.1ns (Lower is better).




Of course, no one will object if I display numbers all day. I have therefore put together a few charts regarding relative performance to put these benchmarks into perspective.

A quick glance at this chart makes it obvious that the new R9 7900X is in the lead. The Alder Lake and Zen4 kings receive scores of 15.5 and 10.5 respectively. It won't be AMD vs. AMD, of course. If we use Intel's i9-12900K as a comparison, the new chip has a latency that is 34% faster than Intel's top products.


                                       L3 Cache Latency Zen4 vs Alder Lake

Yes, I am aware of the criteria for writing, reading, and copying. By achieving 1494.8GB/s, a 17 percent improvement over the previous record, the R9 7900X smashes all previous records. Moving on to the L3 Read tests, the Zen4 eats Alder Lake alive (figuratively speaking) by gaining an astounding advantage of more than 180%. The same pattern is seen for the L3 Copy speeds, where Alder Lake lags by 45%.

                                              L3 Cache Speed Zen4 vs Alder Lake

Therefore, if we use these standards strictly, Zen4 ought to be about 50% faster than Alder Lake. Not much; if you check at the R9 5950X in the same benchmarks, the results are almost equal. Additionally, your FPS and CPU-Z ST scores, where Intel reportedly surpassed the 1000-mark, are not truly determined by L3 Cache (many times).


The Zen4 CPUs from AMD are scheduled to go on sale on September 15th.