I purchased a B-Die DDR4 for your i9-11900k and here are the details.
I purchased a B-Die DDR4 for your i9-11900k and here are the details.
Motherboard: ASUS ROG Strix Z590-E
CPU: Intel i9-11900k
RAM: G.Skill RipJaws V Series 32GB (4 x 8GB) – PC4-32000
GPU: AMD Radeon RT 6950 XT
Storage: 990 PRO 2TB
Source: https://pcpartpicker.com/user/paradoxuml...ved/syDkf7
For tuning, I’ll focus on secondary timings and adjustments. Here are some practical steps:
- **Timing basics:**
- Use XMP profiles for maximum performance.
- Secondary timings (slow/fast) help stabilize speeds.
- Common values: 36/28, 36/30, 36/32, 36/34.
- **Recommended changes:**
- Start with 36/28 for base stability.
- Try 36/30 for a balance between speed and stability.
- If temps are high, lower to 36/32 or 36/33.
- Adjust based on your CPU’s thermal behavior.
- **Tools in Windows:**
- Use MemTest86 or MemTestUtils to check stability.
- Consider MemTest.io for detailed timing analysis.
- If you want deeper control, install MemTestHelper and follow the OC guide.
- **Next steps:**
- Set up a small test environment.
- Monitor temperatures and voltages.
- Gradually tweak timings while observing performance.
Let me know if you need help interpreting results or setting up the tools!
I need to reduce the voltage first before aiming for a more normal level. ASUS MemTweakIt is widely available online, especially on HWBot forums: https://community.hwbot.org/topic/220910...0-apex-15/. For CPU overclocking, Turbo VCore is also helpful if you have an ASUS board. If you're only focusing on timing tweaks, prioritize tREFI and tRFC. These settings control memory refresh rates, impacting how much work your RAM can actually perform. Aim for low tRFC with high tREFI to keep memory availability near 250%, while keeping tREFi as low as possible. Good B die will help bring tRFC down to around 250, and a moderate setting might reach 300-350, depending on temperature. For maximum performance, set the values that allow your RAM temps to stay low—up to 65535 if conditions are ideal—but be cautious with high-power GPUs that can overheat memory. It’s wise to stress test with Furmark to ensure your RAM stays stable under load. These two parameters—tREFI and tRFC—have the biggest influence on memory speed, often accounting for half of the overall improvement from custom timings. Other settings remain useful mainly for stability during boot.
Edit: Make sure you’re using Gear 1. A 4266MT/s with 4x8GB is quite fast for Gear 1; unless your memory controller is top-tier, you’re likely in Gear 2. The extra latency from Gear 2 can be hard to beat with dual-rank memory. Running 3600 CL14 in Gear 1 may outperform 4266CL15 in Gear 2 overall.
Updated: March 7, 2024 – by RONOTHAN##
I increased the value from 1.480 after experiencing a WHEA crash while trying to start a 3dmark benchmark. I suspect a faulty chip. > To maximize performance, the reference temperature should be as high as possible. This explains why it never boots when lowering the temperature settings. I purchased these sinks; I’m planning to drill holes in them so I can attach my RAM waterblock. The Byski ram plates were damaged and no longer stay together. The new ones are much thicker metal, which helps dissipate heat more effectively. Right now, I’m just covering them with a waterblock tape until the carbon tipping drill bits arrive. Here’s the setup—still in development.
It seems like that's all right. Either your LLC configuration is unusual, or your RAM setup is a bit off. You didn't really need water-cooled RAM; a basic fan would have sufficed. However, if your RAM is water-cooled, you should be able to set tREFI to 65k and run the memory voltage up to 1.6V. I'm not sure if you noticed my earlier comments about memory gear modes—make sure you're in Gear 1 for optimal performance. Please confirm so we can send a notification.
I've always cooled my RAM simply because that's what I did, but recently learning about overclocking DDR4 dies and how temperature affects performance has made me feel more confident after all the doubts. Yeah, I've been using C1 settings. People mention gear or clock rates, and HWINFO shows those values at the bottom; should I adjust the command rate? I reduced trfc from 700 to 690, then tried 650 and 600, but stability issues forced me to step down more gradually. Is it impressive that Tras managed to drop from 36 to 28 without changing much else and then reached 4266? I don’t recall my CPU tweaks, but I lowered the LLC level to 2 and adjusted the offset voltage—seems like it’s rising in voltage but still stable. Desktop stayed asleep during testing. What about thermal management? I plan to re-seat the CPU block (velocity 2) and switch to MX4 instead of LM. I’ll do that around the same time the memory module is installed.
I still think it doesn't matter much because a 40mm fan with standard heat spreaders is sufficient to prevent overheating. You also need to consider other parts in your setup that might raise the water temperature and damage your RAM. What's your take? CR1 and Gear 1 are quite different. Gear mode refers to the clock speed ratio between memory and the CPU controller. Gear 1 offers a 1:1 ratio with much lower latency, whereas Gear 2 has a 2:1 ratio, meaning the controller runs at half speed for higher memory speeds but suffers from significant latency. You can technically beat the latency of Gear 2, but achieving that usually needs speeds above 4400MT/s, which is unlikely with dual-rank configurations. Stick with Gear 1 for better performance. Don't adjust command rates—modern CPUs often run at full speed regardless of mode because they count controller clocks. In Gear 2, the command rate appears doubled (2T), but this comes at the cost of higher latency. tRAS at 28 is typical for B-die chips; on AMD systems you can lower it to 25 or less, though some kits hit 3800CL14-13-13-21. Avoid tRAS at 690 unless you're certain your BIOS and hardware support it. Even the poorest B-die units usually stabilize in the 300s. If your memory is overheating (over 60°C), your BIOS might be faulty, or 4266 could be near its stability limit. That's a huge problem for a 14nm Intel chip. You should never exceed 1.5V for daily use to prevent CPU damage in months; ideally, keep it below 1.45V. For most ASUS boards, setting the LLC to Level 4 gives optimal results, and a static voltage is usually preferable. In your current configuration, sticking to stock settings is safer, as the chip should naturally run at 5.3 single-core/5.1GHz safely. The high voltage makes it hard to measure thermals accurately.
the bios suggests a minimum requirement for certain configurations, but it expects more power for higher speeds like 5.3ghz. running the ai overclock might push it beyond safe limits. the motherboard's built-in ai shouldn't be handling that unless temperatures are extreme. sticking to stock parts is safer, as the chip is designed for stable performance at lower voltages. the issue isn't thermal but rather voltage constraints.
I think it's unnecessary because a 40mm fan with standard heat spreaders should suffice to prevent overheating. You also need to consider other parts in your loop that might raise the water temperature and affect your RAM. You're right—water stays cooler than any component. When I play, my water temp hits around 42°C, and the RAM is already at 47°C. I'm gradually lowering the TDP, but why does it begin at 700 when 300 seems fine? And what about the next ones—520 and 320? (I don't remember their names well) Thanks.