DDR4 RAM Latency

As a system builder, enthusiast, and gamer, we usually purchase for a new build is RAM (Random Access Memory). This RAM is used for fast temporary storage for your CPU (Central Processing Unit). RAM is similar to amount of desk’s top area. Persistence storage such as HDDs and SDDs are very slow to process data directly, plus they have a limited lifespan for reads and writes. These are like file cabinets. RAM is used between the CPU and the persistence storage, bringing faster application performance. We would want the fastest response RAM by looking at the lowest latency and including estimated cost per GB.

RAM is sold in terms of speed and CAS Latency. RAM speed is the built in clock cycles that refreshes the memory and is in terms of MHz or MT/s. CAS Latency is the delay time between memory controller asks for data and data is available on the RAMs output pins and usually is in nanoseconds.

Since RAM is sold with different speed and CAS Latency, the following formula will assist you on figuring out the relatively faster RAM as the 2 values determine the performance of the RAM.

Latency (ns) = Clock Cycle Time (ns) x CAS Latency (CL)

Latency (ns) = (1/Module Speed(MT/s) x 2) x CAS Latency (CL)

To find the higher performance RAM, look at the lowest CAS Latency and higher RAM speed to minimize the latency as much as possible. Shown below we calculated the latency. We’ve excluded some results due to not existing on the current market or only found very few results to provide an good average price per GB.

Officially, DDR4 memory limitation is set by JEDEC (https://www.jedec.org/) and the upper limit is at 3200 (MT/s) before needing to turn on XMP (Intel Extreme Memory Profile) in motherboard settings (BIOS/UEFI) to get the higher speeds on XMP Ready/Certified RAM. Some motherboards may not have this option to enable. Please refer to your motherboard specs/manual.

Module SpeedXMP Profile RequiredCAS LatencyLatency (ns)Est Price $/GB (6/2019)

The difference between the lowest and highest latency times are 7 nanoseconds apart. These differences might be noticeable, but very slightly.

As for our gamer audience, it is wiser to spend extra money on the GPU than on faster RAM. usually for gaming an individual computer only needs 8 to 16 GB depending on the game and near 10 ns in latency before getting higher prices.

3D Printer Calibration

Here’s what I use to calibrate our 3D printers. There are many guides out there for most of the methods and didn’t see the need to go into heavy details.

Calibration Tools used:

  • Pro-Grade Electronic Digital Caliper 82806
    • Know your accuracy tolerances, ours was +- .02 mm for lengths below 100mm with a resolution of .01 mm
  • Thin Wall Calibration https://www.thingiverse.com/thing:342198
  • Single sheet of paper

1.Initial Extruding Calibration

First thing we will need to calibrate on a new 3D printer is the filament extrude rate. It is very important to get this as accurate as possible, minor tweaking can be done with extrusion multipliers.

To figure this out by telling the printer to send 30 mm of filament through with the hot end disconnected and measuring with a caliper to see if the amount is accurate. Then update the stored value of Esteps/mm under Marlin Motion options by following this formula: Current Esteps/mm * 30mm / Actual amount of filament pushed

For example:

  • Current Esteps/mm: 93
  • Tell printer to extrude this amount of filament: 30mm
  • After measuring the amount of filament extruded: 29.33 mm
  • New Esteps/mm value = 93 * 30/29.33 = 95.124

2. Bed Leveling

Bed leveling is crucial for it will either allow your print complete properly or outright fail later on besides filament run-out or the disastrous thermal runaway. There are numerous ways of getting this done usually by sliding a piece of paper between the heated nozzle and heated bed, or something similar to a BLTouch or a conductive sensor.

The methods that are used often:

  1. 1st is done with piece of paper between heated nozzle and heated bed at normal printing temperatures for the printer. Since I print mainly in PLA, the temps would be at 200 C and bed at 70 C. Then manually level the middle, then the 4 corners where the paper is slightly catching between extruder and the bed
  2. Next, print Bed Leveling squares that looks like the 5 dot side on dice, 1 mm height and slightly adjust our leveling while it prints and wait a few lines are printed after adjusting and see if its flat and clean looking and doesn’t come off of the bed easily. An example of these squares on Thingiverse: https://www.thingiverse.com/thing:2789086

For Auto Bed leveling, a great bed visual is from OctoPrint with the plugin Bed Visualizer. It will highlight which corners need to be readjusted much better and readjust the dark red and blue slightly. Just keep an eye on the visual scale as it will re-adjust.

3. Wall thickness

Next up to calibrate our Wall thickness, please note that this will affect your bed leveling, extrusion multiplier, and/or Esteps/mm. Essentially with .2 mm nozzle, print a hollow cube with .2 mm thick walls with a layer height of 0.16 mm (80% the size of the nozzle), then measuring with a caliber. A good resource to use are the Thin Wall calibration files created by MEH4d at https://www.thingiverse.com/thing:342198

Once the .2 mm thick walls have been printed, measure the wall thickness with the caliber but exclude the bottom of the walls. The bottom will usually be a tad off due to 1st layer height/bed leveling is a tad off. If the thickness is off try adjusting the multiplier between .9 and 1.1, but keep in mind of your calibers limits. For example, if the measured result was .19, you could be very will be accurate.

If you had to adjust your multiplier to get best results outside of this range, place the multiplier back to 1 and adjust your ESteps/mm. Lower if the walls were too thick and the opposite if the walls were thin. After readjusting, you may need to readjust the bed level and layer heights.

Usually, a good max layer height for each of the nozzle sizes are at 80% as shown below:

  • 0.2 mm nozzle = 0.16 mm layer height
  • 0.3 mm nozzle = 0.24 mm layer height
  • 0.4 mm nozzle = 0.32 mm layer height
  • 0.5 mm nozzle = 0.40 mm layer height
  • 0.6 mm nozzle = 0.48 mm layer height

4. XYZ Calibration Cube

At this point, print a calibration cube and measure with the caliber. Usually, it’ll be within measured dimensions that are to be expected. Check how the cube looks by actually looking for infill, speed, and other issues and try to correct them in our slicer program. Slowing down the speed improves many issues at this point.

5. Stress Test

Now its time to stress test your printer. Good set of models to use are by ctrlV on Thingiverse especially https://www.thingiverse.com/thing:704409 and then part fitting a M4 nut in the slot. Also, look at how well the filament retraction works and increase or decrease the retraction length.

Nate15329's Blog