Oh ok thanks. So to find out if that overclock may work I should leave FCLK at default and push my ram on up? And then if its stable bump FCLK up to match?
Increase FCLK so that it is equal to MCLK (memory clock). Always keep it the same. Don't make it different, unless you're really pushing >4.2GHz ram and your FCLK isn't stable at 2100Mhz or something.
Also what are the chances of ram overheating causing a system crash in memtest86? I kinda want to push my ram to CAS 16 3733 at 1.45V to see if it could handle those speeds if voltage was high enough but if it crashed how would I know that it wasn't a temp problem?
RAM doesn't usually overheat itself. Assuming good case airflow, ram sticks don't actually generate a lot of heat. It's the components around it, mostly CPU, but some GPU, which can "cook" the sticks. 1.45V should be safe for daily, but don't push it any higher or else it may degrade. Temps should be okay, but checking that is what stability testing is for. Test in memtest86 all 11 tests for 4 passes. It's very rare for memtest86 crashes to be temp related, but possible. Even after it passes all memtest86 it may not be stable, since that only stresses and therefore heats the ram. So use prime 95 to stress CPU as well (AFTER using memtest86). There's a specific setting set that I use for testing.
I use version 286 since that supports Ryzen well but still has the appropriate FFT sizes. First off, disable AVX. Do this by going into local.txt and typing in a new line "CpuSupportsAVX=0". AVX results in higher unrealistic temps. You want to enable it for CPU oc testing but it's not necessary for mem oc testing.
Then go into p95. Select blend, then select custom. Set min FFT size to 512K, and max to 4096K. Set memory to use to approximately 75% of what is available. I generally just set it to 10000MB on my system with 16GB ram. Set time to run each FFT to 15 minutes.
Run for 4-8 hours. If all threads are still working by then, you are stable, or at least as close to stable as you can get. Generally, if the memory fails here but not memtest86, it's a temp issue.
This is, of course, very tedious. So I suggest just using memtest86 alone for minor tweaks and then once you've decided on the settings you want to run daily, then do the full suite of testing described above.