More RAM typically doesn<|pad|>, can lead to higher memory latency.
More RAM typically doesn<|pad|>, can lead to higher memory latency.
You've increased your RAM capacity and confirmed the total is now 32 GB. The slight rise in idle memory latency could be normal, especially if the new memory isn't perfectly compatible or if there are minor compatibility issues. It's not uncommon for latency to shift slightly after upgrades, particularly if the new module doesn't match the exact specifications. The fact that you received a Hyundai die instead of a Samsung one might also play a role, as component quality can vary. Keep monitoring the performance to ensure stability.
It's normal for the memory controller to perform more work with double the memory sticks, possibly adding more ranks based on how the chips are laid out. The electrical signals become slightly noisier because of the increased proximity, and the DC-DC converter supplying power to the RAM might deliver less clean energy since it has to boost twice the voltage. This affects not only memory capacity but also latency, with other factors influencing the overall performance.
It’s not about the memory controller handling more load, you need to stop that. What’s happening actually? The internal sub timing has slowed down because you enabled auto mode. Reset your CMOS and check your subs using two sticks. Then reset the CMOS again and add more. Observe how the timings adjust.
The internal timing adjustments have become less precise, which is evident from the screenshot I took of my original 3200 MHz XMP sub-timings before adding more RAM. The timings I listed earlier are actually tighter than those generated automatically after installing the extra memory (the mlc latency is included in the photo). I therefore tested whether restoring the original sub-timings would improve performance. Interestingly, even with a 66 MHz overclock on the standard XMP frequency, my latency only slightly increased—about 0.4 nanoseconds, likely within acceptable limits. All testing was conducted in Linux; I then switched to Windows and used AIDA64 to check for changes. The results showed a minor improvement of roughly 0.5 nanoseconds after tightening the sub-timings, which is probably within error margins. In summary, unless you spot any flaws in my testing or have more insights into other variables affecting the latency rise from 2 to 4 RAM sticks, I’d lean toward attributing the small uptick to increased load on the memory controller. I’ll discuss this with @mariushm and consider whether overclocking might be the culprit.
RAM isn't completely trouble-free while your system is active. Data must be updated regularly to prevent loss and degradation. Adding more memory levels increases background tasks for the memory subsystem. It's accurate to say the memory controllers are handling more work.
Short version is that every bit stored in RAM have to be refreshed, or it will fade. And you don't want your 1's becoming 0's unless told so. This process is not instant. So there is a tiny bit of overhead involved. Some of the timings we change while tuning ram are the refresh timings. You will gain performance by extending the interval between refresh cycles. Go to far and you get data loss shown as errors in memtest f.ex.
Apologies for the mix-up. I didn't mention running Intel; my machine uses a 5600x with a b450 chipset. I relied on Intel's memory latency checker to obtain the idle memory latency reading. To be honest, if you intended to say you've never worked with software built by Intel, that's okay—usually it's not necessary to install software from a rival manufacturer unless you're using something like Ubuntu Linux and can't find reputable options elsewhere. That makes sense. However, what seems unclear is that the latencies I saw from my extra RAM sticks compared to those measured with manual timings before adding new RAM still matched according to both AIDA and Intel's MLC specs. In short, matching all the timings didn't reduce the latency enough to bring it back down to its original level. Do you think there are any hidden hardware adjustments that can't be changed via BIOS?