Leaked Ryzen 5 3600 Benchmark Claims Better Single-Threaded Performance Than the Core i9-9900K

(Image credit: AMD)

The benchmark and result database Passmark (also known as cpubenchmark.net) recently added the six-core, 12-thread Ryzen 5 3600 to its ranks and, if these results are accurate, the performance is nothing short of incredible. The Ryzen 5 3600 seemingly matches the performance of the Core i9-9900K (which has two more cores) in multi-threaded testing and actually beats the -9900K in single threaded tests (beating the 2700X by 36%), making the Ryzen 5 3600 the fastest single-threaded CPU, according to Passmark.

Photo (Image credit: Passmark)

However, it does have to be said; this sounds just a little too good to be true. The $200 CPU has lower clock speeds and fewer cores, making the test results dubious. Matching the -9900K, let alone beating it, seems too good to be true. Luckily, Passmark lets you look at the five most recent benchmarks for any given CPU, and at the time of writing, there are three we can examine (seemingly from the same test system).

The system reportedly uses a B450 Aorus M board, not an X570 board. According to the reported clock speed, the CPU doesn't seem to be overclocked either; all three tests show the same turbo of 4.21 GHz, and one result shows a "measured speed" of 3.37 GHz, and the other two 3.61 GHz. It doesn't seem like there was some sort of trick making this 3600 so fast, at least not something we can glean from Passmark's reported information.

Interestingly, the third benchmark for the 3600 uses a 16GB kit of 3200 MHz CL14 G Skill RAM, unlike the first two benchmarks which used a single stick of Crucial RAM at 2666 MHz CL16. The third benchmark reports a score of 7% faster than the two previous scores, which implies that Zen 2 and/or Passmark benefits heavily from having high-speed low-latency dual-channel RAM, something which previous iterations of Zen also benefit from.

Overall, this result seems legitimate, but AMD's lowest-end Zen 2 CPU beating the -9900K overall seems unrealistic at best, especially when AMD positions the beefier Ryzen 7 3800X against the Core i9-9900K. This benchmark does, however, prove that Zen 2 does some things far better than Coffee Lake, which bodes well for Zen 2's overall performance. July 7th is just eight days away now, so it won't be long before we know what the Ryzen 5 3600 can really do.

Matthew Connatser

Matthew Connatser is a freelancing writer for Tom's Hardware US. He writes articles about CPUs, GPUs, SSDs, and computers in general.

  • penn919
    It may not beat the 9900k, but I fully expect the 3rd Gen R5s to crush intel's i5s in everything.
    Reply
  • Math Geek
    yah the i5 seems like a better comparison. would be awesome if it beat the i9 but that seems a bit of a pipe dream overall.
    Reply
  • King Dranzer
    You guys missed the point. R5 3600(if the leaked benchmarks are to be considered accurate) is having single core core performance nearly as good as i9-9900K and we are not talking about multi core performance as it is obvious that $200 chip cannot be compared to $450+ chip.

    Knowing that out of the box R5 3600 base stats are no where as good as higher tier CPUs top up the line. Still when compared base performance to base performance(if that is what it is being tested in a legit manner) R5 3600 is very close to i9-9900K even in Cinebench R15(pic below). Ryzen R5 3600(196) vs Intel i9-9900K(204) crazy knowing that 3600 is lowest clocked CPU from Ryzen 7nm lineup.

    Reply
  • adnan_16a2
    If that is the truth, then I'll be damned for buying 9400F.
    Reply
  • mdd1963
    If it beat the 9900K in single core scores, i doubt very seriously it did so at it's standard Boost frequency...

    We expected a 18% boost in IPC, not 33% :)
    Reply
  • King Dranzer
    adnan_16a2 said:
    If that is the truth, then I'll be damned for buying 9400F.
    9400F will not be a competition for R5 3600. That is obvious.
    Reply
  • InvalidError
    Zen 2's cores have two more execution units per core than Coffee Lake's cores, which is bound to bump IPC by a considerable amount when the instruction mix allows the scheduler execute up to two more uOps per cycle. Intel's Sunny Cove is also ~25% wider, too bad it or Willow Cove (its successor) isn't happening for desktop in the foreseeable future, can't both make the core wider and maintain clock frequencies without a process shrink. (And Icelake couldn't simultaneously achieve both on 10nm either, will need 10nm++ or better to reclaim the clock regression from CFL on 14nm+++ to ICL on 10nm+.)
    Reply
  • adnan_16a2
    King Dranzer said:
    9400F will not be a competition for R5 3600. That is obvious.
    We can expect i7 9th generation to be its competitor
    Reply
  • bobalazs
    No such things as leaks. This stuff is intentional.
    Reply
  • cryoburner
    Considering there should be proper reviews available for these processors within the next week, there probably isn't much point in putting much thought into the legitimacy of these benchmarks. To me, it seems a bit unlikely that the 3600 will offer better single-threaded performance than a 9900K, at least without significant overclocking involved. I do suspect the Ryzen 3600 and 3600X will be strong i7 competitors though.

    mdd1963 said:
    We expected a 18% boost in IPC, not 33%
    It is worth noting that IPC can vary depending on the workload. AMD is citing 15% as an average, but the exact difference could be higher in some workloads and lower in others, depending on what operations are getting used.
    Reply