Important sequence for CL, TRCD, TRP and TRAS
Important sequence for CL, TRCD, TRP and TRAS
You're pushing the voltage beyond stock or close to it. Try around 1.450v perhaps. When you increase the voltage, the performance improves significantly. This works well with higher-tier memory options from Hynix and Micron. It really depends on the memory tier you choose. I also use active cooling for my RAM and don't depend on the CPU fan for temperature control. The setup is usually run on an open bench.
I noticed the fans are drawing air quite strongly, but there’s no active cooling. My RAM doesn’t have a temperature diode either, so I can’t tell if the 1.45V reading is safe. That’s why I’m hesitant to proceed. Maybe I’ll test with a probe later, but right now I won’t attempt it. My current RAM model is F4-3600C16D-32GVKC, running G.Skill Ripjaws V at 3600MHz DDR4-3600, stock voltage is 1.35V—if that’s relevant, that’s the standard for XMP profiles.
I just tested a few values to check basic functionality without rushing into stress. The sequence 16-16-16-26-48 didn’t work, and even booting failed. I’m planning to try the 3800 model but am already unsure about its reliability after previous low-confidence results with similar timings. The tRP max is capped at 27, not 28, and boots like 18-18-18-18-27-52 aren’t stable. So it seems unlikely any of these will be consistent. I’ll keep exploring formats if stability remains an issue, like 16-18-18-18 or 3600 vs 18-20-20-20, etc. Thanks for the guidance!
You're showing solid progress. Most folks don't match your dedication. I often emphasize refining and experimenting. Keep adjusting and checking. Even though I already know the optimal settings, I still tweak things to see what works best. Luckily, my memory is decent, which helps a lot. I've worked with Hynix modules and have Corsair 3000 LPX and LED sticks. They reach around 3466MHz at their peak 1.60v, but performance isn't great.
The jump from 3600 to 3800 clocks is noticeable, especially with timing differences of 16 vs 18. Latency stays similar, but the 3800 offers better read/write speeds due to higher clock speed. Keep in mind CPU speed also plays a role—test with consistent speeds for accurate results. Lower CPU frequencies mean more latency and slower operations.
It's tough against memory controllers, but you're definitely pushing your limits. Your enthusiasm is clear, and it shows. If you want, I can help visualize what you're seeing in your setup.
CL is the key metric to focus on during RAM OC configuration. Column A captures latency or CL, which represents the time gap between a memory controller's read request and when data becomes accessible. This value isn't fixed by clock rate alone. For DDR4, a CL of 14 at 3600 MHz translates to a first word speed of 7.784ns, while at 3866 MHz it drops to 7.252ns. You can perform simple calculations to estimate raw clock speeds relative to latency.
Here’s a revised version of your text with varied phrasing and structure:
For the calculation, I’m focusing on CL/performance ratio. In my current setup—CL 16, 3800—I calculated 16/1900 as a reference based on Luke’s video. This gives a solid estimate of latency. I’m also curious about other metrics like tRP and RAS. The key points I’ve gathered from online sources suggest prioritizing the sequence in which latencies appear: typically tRCD comes second, followed by tRP and then RAS. It seems important that CL is even, while the others don’t strictly need to be. There’s also a preference for closely spaced values, such as 16-16-16 instead of 15-17-18. This idea came from browsing recent discussions. I’m also considering how stable each component should be. My approach emphasizes consistency across numbers—like 16-16-16—not jumping between mismatched values. Regarding the Ryzen 5 3600, it delivers strong performance with 4.3 GHz per core at 1.25V, which feels like a solid choice in terms of efficiency. I opted to skip the 4.35GHz model because it would require higher voltage settings. I also chose not to use 4.35GHz due to potential power demands. Overall, I’m aiming for balanced performance across all areas.
My final configuration includes:
- CPU: F4-3600C16D-32GVKC
- Motherboard: CL16-19-19-39
- RAM: 1.35V XMP, DDR4-3600MHz
- BIOS: Hynix DJR
I tuned the CPU to consistently run at 4.3 GHz across all cores, maintaining a stable voltage. I adjusted the SOC voltage slightly and then settled on the lowest stable settings. For power efficiency, I plan to upgrade to a Corsair RM750 PSU and use a 650PS build for my dad’s system.
I’ve installed a basic CPU temperature sensor in the RAM fins since there’s no direct RAM temperature reading. At 1.40V (≈1.42V as reported), the cooler stays within safe limits, though actual chip temps are likely higher. I’m comfortable with the 1.42V max voltage setting. The large CPU cooler footprint and airflow from nearby fans help maintain stability.
Looking ahead, I’ll test three RAM heatsink configurations—16-16-16 versus 18-18-18—to see which offers the best balance of speed and reliability. Once I finalize the timings, I’ll focus on minimizing RAS latency and optimizing tRCD and tRP values.
I’m excited to share more details about my build, including the PC interior, cooling setup, and performance results.
I ran memtest for 20 hours across nearly 13 passes at speeds of 16-19-19-20-35 with a 3800 MHz and 1.37v setting (1.39 bios reported). I believe these configurations are optimal. Want tips on timing or importance order? Check out the link you shared.