Pushing For Speed

With testing out of the way, I endeavored to see if these sticks have any headroom.


Given the Corsair Vengeance LPX comes with an external cooling device, I was curious about what kind of temperatures this kit would put out. After doing my tear-down shots, I installed one of my Omega Engineering SA1 Self Adhesive Thermocouple probes between an IC and the heat spreader on one of my sticks. I then booted up Memtest64 and had it run until the temperature stabilized. The maximum temperature recorded was 38°C (20°C ambient). I had originally intended to do a second run with the Vengeance Airflow device, but am confident it is not necessary unless case or ambient temperatures are very high.

For overclocking, I wanted to dig a little deeper. First, I followed the same procedure as I would normally with Intel to find the maximum frequency with XMP timings. It turns out that with my setup, the XMP rating is as good as it gets. Even cranking the memory voltage past 2 V did not net me an additional 100 MHz.


Next, I set the frequency to 3600 MHz and used DRAM Calculator for Ryzen to optimize the timings. If the "Fast" preset was not viable, start with the "safe" settings and try to at least get the primary timings as close to the "Fast" preset as possible. I then benched each setting with AIDA64 to showcase what kind of benefits you can expect from each. I also included the default (non-XMP) settings as a base reference point.


For timings, the "Fast" preset booted perfectly! Micron E-die is second only to Samsung B-die for tight timings, and CL14 was no problem for the Vengeance LPX.


Since the launch of Ryzen 3000, the go-to memory spec has been 3600 MHz with the tightest timings possible. While AMD postulates that fast enough memory will offset the mismatched Infinity Fabric ratio, the question has been how high you need to go to see the benefit. While the Vengeance LPX 5000 MHz loses out to that 3600 MHz spec (with a 2500 MHz IF) the loss is very slight, if not marginal.