Overclocking Ryzen 3000

Experience with the MSI MEG X570 Godlike

Overclocking with the Ryzen 3000 series hasn't exactly titillated users since its launch with there not being much headroom available. The biggest issue with this boils down to all-core overclocks being limited by the worst core of the chip. and the new 7 nm processors running quite warm. To get the best experience and results with the Ryzen 3000, users with good quality AIOs and custom water cooling yields the best outcome, and even with voltages of 1.375 V, these chips can run a little too warm for comfort. 

The MSI Click BIOS 5 firmware is familiar to us and throughout the different chipsets including X399, X470, and Z390, the MSI MEG X570 Godlike uses the same design layout and setup. Within the OC section of the firmware, users can alter important settings such as the CPU ratio which can be changed in steps of 0.25 MHz, with relevant settings including CPU VCore, CPU Base Clock, and even FCLK frequency. Deeper into the firmware is settings to set the load line calibration (LLC) to a more aggressive or liberal profile, as well as all the Precision Boost Overdrive (PBO) settings. 

The three main settings that PBO works from include package power tracking (PPT), thermal design current (TDC), and the electrical design current (EDC). All three of these can be set within the firmware under the advanced section of the Precision Boost Overdrive menu, although MSI has included four different PBO profiles for users to select from. The higher the setting, the higher the thermal and power draw will be, and it may not necessarily matter if a sub-par cooler is installed; it's noted that Ryzen 3000 processors work better when they are cooler.

There are seven different MSI Game Boost profiles to select from and users can access them via the dial at the bottom of the board, or through the firmware itself. The Game Boost profiles within the firmware can be accessed through the dial in the top left-hand corner. These profiles range from 4.0 GHz through to 4.3 GHz, and we've tested each of them below.

Overclocking Methodology

Our standard overclocking methodology is as follows. We select the automatic overclock options and test for stability with POV-Ray and OCCT to simulate high-end workloads. These stability tests aim to catch any immediate causes for memory or CPU errors.

For manual overclocks, based on the information gathered from the previous testing, starts off at a nominal voltage and CPU multiplier, and the multiplier is increased until the stability tests are failed. The CPU voltage is increased gradually until the stability tests are passed, and the process repeated until the motherboard reduces the multiplier automatically (due to safety protocol) or the CPU temperature reaches a stupidly high level (105ºC+). Our testbed is not in a case, which should push overclocks higher with fresher (cooler) air.

Overclocking Results

As it currently stands, MSI's Game Boost overclocking profiles need quite a bit of refinement, something they have informed us that these will be rectified with a new firmware revision; we will test this when we receive the board back from MSI. The issue with these is CPU VCore, or in a nutshell, too much of it which cross the board into the unsafe territory of the Ryzen 3000 processors safety limits; both in CPU VCore and in temperature. When we activated PBO and MSI's preset profiles for this, it didn't make much of an impact over the stock settings, which seems to be a reoccurring issue so far across the range of models we have tested so far.

Manually overclocking the MSI MEG X570 Godlike proved very fruitful and we managed to hit what seems to be the limits of our Ryzen 7 3700X testbed CPU with an overclock of 4.3 GHz all-cores with a CPU VCore of 1.35 V, which seems to be the brick wall for our processor. Even with a CPU VCore of 1.50 V, we couldn't get 4.4 GHz to remain stable when any kind of semi-intensive load was placed on the processor, although we did manage to boot into Windows 10 without fanfare with these settings. The performance wasn't throttled in POV-Ray throughout each of the frequencies tested and once the temperature limit of 110ºC is exceeded, the system will safely shut down. From 3.6 to 4.2 GHz, we managed to do this with a set CPU VCore of 1.250 V which did fluctuate slightly under load under the auto LLC setting. It is safe to say that the MSI MEG X570 Godlike doesn't have a VDroop issue and even at 4.3 GHz, a set CPU VCore of 1.375 resulted in a maximum load voltage of 1.384 V which isn't too far off the mark. 

Gaming Performance Power Delivery Thermal Analysis
Comments Locked

116 Comments

View All Comments

  • inighthawki - Wednesday, August 28, 2019 - link

    Depends what you're doing. The 9900K is already, at stock, anywhere from 5-10% better single core perf than AMD, and also overclocks much higher. So if you're willing to accept the additional power draw and thermal output and your goal is raw single core performance, then absolutely.

    If you're working in highly multithreaded scenarios then the new 3900 and 3950 chips are clearly a much superior choice.

    Most people who are willing to spend this much money and OC their parts are also generally not too concerned with future proofing their systems. If you can afford (and are willing to spend) $700 on a motherboard, it stands to reason that money is not much of a concern and they probably don't care about having to buy a new motherboard for a new socket in a couple years.
  • StevoLincolnite - Wednesday, August 28, 2019 - link

    The 9900K is only an 8-core CPU though........ AMD can trounce it with it's higher core-count parts in multi-threaded scenarios.
  • yetanotherhuman - Thursday, August 29, 2019 - link

    He does say that, and you know he hasn't edited his post, because it's not possible :D
  • Qasar - Thursday, August 29, 2019 - link

    but why buy a 9900k when you can get the 3900X for the same price, use less power over all, pretty much the same IPC, and get 4 more cores on top of that ?? who cares if the 9900K can hit 5gz?? clock speed isnt everything, the Athlon 64 proved that way back when vs the P4. besides.. what if ryzen 3000 series did hit the same clocks as intel does, that would make intel's cpus look even worse
  • inighthawki - Thursday, August 29, 2019 - link

    Because the 9900K still edges out the 3900 (and presumably the 3950) in single threaded performance. Some people care about that more than having more cores.
  • AshlayW - Thursday, August 29, 2019 - link

    wow 5-10%, now tell me where that actually matters over the disadvantages of owning the 9900K vs Ryzen 3000.
  • Oliseo - Thursday, August 29, 2019 - link

    "wow 5-10%, now tell me where that actually matters over the disadvantages of owning the 9900K vs Ryzen 3000. "

    Compiling software for one. By it's very nature it's single threaded.

    Likewise DAWs also prefer high frequency high IPC cores rather than multiple lower specced lower frequencey cores.

    When you route your channels through an effects bus, that effects bus is allocated a core. It CANNOT be spread over multiple threads.

    Asking silly questions like that only demonstrate your own ignorance, where you highlight you don't actually know what it is you're talking about.

    It's okay though, we were all young once. I'd recommend in future putting aside your brand loyality though. It's not worth your effort, they aren't loyal to you in any shape or form.
  • peevee - Friday, August 30, 2019 - link

    "Compiling software for one. By it's very nature it's single threaded."

    Seriously? You obviously missed an option or two.
  • 29a - Friday, August 30, 2019 - link

    I'm not disagreeing with you but those last two paragraphs were unnecessary. You'll catch more flies with honey than vinegar.
  • Bus3rr0r - Friday, August 30, 2019 - link

    Maybe if you are writing single module programs in your mom's basement a singel core is fine.

    Most devleopment projects of even relatively small projects have hundreds with larger projects running into thousands of modules. It is not only common practise to compile modules in parallel butit woudl be ridiculous not to.

    The reality is there are very few applications that cannot take advantage of multi-core.

    You critise others for their ignorance while demonstrating your ignorance, remarkable.

Log in

Don't have an account? Sign up now