This XMP profile isn't updating at 3600Mhz.
This XMP profile isn't updating at 3600Mhz.
the clock speed was 3733 with a few extra bits, needs around 1.55v for 3800. the settings 14 15 15 and 15 15 15 both worked previously (though not tested at 1.6v). based on recent stock updates, it seems acceptable as long as temperatures stay under control. these models were used by 8 sticks in two different builds, so if they match the SKU and don’t exceed cl14 3700 or cl15 4000, the issue likely lies with the imc. i’m planning a build soon and aiming for 3733 4x8, so I won’t try 4000 unless it’s dual rank. if they’re dual rank, it might be simpler to get it out of the box.
Wow, a lot of details here. I haven’t worked with VDDG IOD/CCD or VDDP before, but I’m willing to explore it! It’s not a big deal for me. I really appreciate diving into useful information. First, locate a bootable 1:1 ratio between MCLK and FCLK, then check the stability of the FCLK and fine-tune the settings afterward. I haven’t used Random-X or SSE OCCT yet. What configurations would work best for testing these programs? I usually stick to primary timings unless I’m using DRAN Calculator. I understand some timing parameters are linked and should probably be adjusted together, but I don’t have clear guidance on exact ranges. Which ones affect stability the most? I’m not expecting you to answer every single question, though you’re more than welcome. I know 1.5V DRAM is fine, but I’m unsure if I can maintain proper cooling without extra fans above the RAM. I’ll definitely try your suggestions and let you know if I need anything else! Thanks in advance for all the help!
They are single rank units—DDR4 doesn’t have 8GB dual-rank sticks. I meant two ranks per channel. I was trying to balance tighter timings with lower voltage, and I’m not sure if I can manage the temps at 1.55V without a RAM fan. I’ll give it a shot and let you know how it goes! Thanks for your feedback.
Hope you can get 3800MHz stable at least—otherwise it would be a letdown. Two 16GB sticks are usually simpler to tune than four 8GB ones, though you won’t like the empty RAM slots much (we all have quirks). 1.55V is already quite high; seeing 1.6V XMP before isn’t common. It might not be possible without a strong A+ cooler, I think. Who develops those?
They released yet another iteration featuring extremely relaxed clock speeds. I only recall it from the article I read a while ago.
No, there is an aida64 thread over at TPU where a bunch of us see what we can do with our systems. It’s pretty cool. All of those guys are awesome, including me haha
every stick launched at cl14 4000 1.59v (bios 1.575v) showed a 50-point improvement over b4. cpu-z was 200 points ahead of cl15 (confused). geekbench failed (it worked in February), latency dropped from 34ns to 37ns since Feb 2021. temps stayed around 49°C max, 20°C outside, 1.5v reached 47°C max. no big mem test in p95, worried about damaging ram in my zen build. tried trial version with voltage near 1.57, but i didn’t want a BSOD. 9w at 1.55v, 8w at 1.5v, 9w at 1.575v with some errors. temps varied 2-3°C, still wouldn’t run it nonstop