Hi everyone!

I’ve spent the last 5-6 days tuning and testing my Kingston 2x16gb Hynix M-die kit. Here is the current profile my PC:

  • CPU: AMD Ryzen 7 9850X3D (PBO Motherboard, CO per-core from -5 to -15, Scalar x1, +200 MHz, 5699-5709 MHz all core stable)
  • AIO: Arctic Liquid Freezer III Pro 360mm
  • Motherboard: Gigabyte X870 AORUS Elite WiFi7
  • BIOS: F9, AGESA ComboAM5PI 1.2.0.3g
  • Win: 11 Pro 25H2 (build 26200.8875, all last updates, new OS install 26.07.2026)
  • Memory: Kingston Fury Beast 32 GB (2x16gb, single-rank Hynix M-die)
  • Kit: KF560C30BBEK2-32 (2025, week 42)
  • GPU: Asus Prime 5080 OC (0.925mV @ 3000MHz +1500 mem clock, manual fan)

I have completed more than 40+ hours of testing in stress and games.

This included TM5 Extreme@ anta777, repeated y-cruncher FFTv4, N63 and VT3 runs, as well as several long gaming sessions. At this point, I can confidently call the profile fully stable =)

________________________

Today frequency: 6200 MHz

MCLK 3100 MHz, UCLK 3100 MHz, FCLK 2066.67 MHz (go 2133?)

Command Rate: 1T

GDM: Enabled

BGS: Enabled

BGS Alt: Disabled

Power Down: Disabled

Nitro: 1/2/1

Nitro Burst: 8X/8X

__________________
Primary timings

tCL: 28

tRCD: 38 (36 not stable after 2h TM5)

tRP: 38

tRAS: 56 (54,52?)

tRC: 96 (92,90?)

___________________________________
Secondary and tertiary timings

tRRDS: 8

tRRDL: 12

tFAW: 32

tWTRS: 4

tWTRL: 16

tWR: 48

tRTP: 12

tRDRDSCL: 6 (5 no difference, 4 not stable – error on 1:02h anta777 TM5)

tWRWRSCL: 6 (the same)

tRDWR: 16

tWRRD: 2

tRDRDSC: 1

tRDRDSD: 9 (1 is needed?)

tRDRDDD: 9 (1?)

tWRWRSC: 1

tWRWRSD: 15 (1?)

tWRWRDD: 15 (1?)

tCWL: 26 (BIOS auto, can't change)

tCKE: 0

tMOD: 44

tMODPDA: 32

tMRD: 44

tMRDPDA: 32

tPHYWRD: 6

tPHYWRL: 14

tPHYRDL: 36

tRDPRE: 4

tWRPRE: 4

__________________
Refresh timings

tRFC: 496

tRFC time: 160 ns (M-die wall =))

tREFI: 49151

tREFI time: 15855.16 ns

Refresh Mode: Normal

From what I have seen on Reddit, 16gb Hynix M-die generally does not scale much below approximately 160 ns tRFC, so I decided not to waste time fighting that wall for 0.1-0.2 ns =)

__________
Voltages

VSOC: 1.25 V (HWinfo reported 1.241 max)

CPU VDDIO MEM: 1.42 V

VDD: 1.52 V (set in BIOS) approximately 1.50-1.515 V reported ZenTimings and HWinfo

VDDQ: 1.45 V (set in BIOS) approx. 1.455–1.48 V reported

CLDO VDDP: 1.05 V (give more stable)

VDDG CCD: 1.00 V (give HUGE stable and VDDG IOD too)

VDDG IOD: 0.95 V

MEM VPP: 1.80 V

VDD MISC: 1.10 V

__________________

My DIMM temperatures reach approximately 50-52.4°C (5.475-5.625 W max in HWinfo) during heavy memory stress tests.

I will probably order (like Gigabyte Aorus Extreme wind blade mount for DDR5) or 3D-print a small RAM fan bracket, because I would feel more comfortable reducing the temperatures by another 5°C or so.

AIDA64 reports approximately 66-67 ns latency with Hyper-V enabled. FACEIT Anti-Cheat does not work on my system without the hypervisor, so I am not going to disable it. I am also not interested in enabling Latency Killer, booting Windows in Safe Mode or using other benchmark-only tricks just to produce prettier numbers like 59-61ns, its not real =)

Does anyone here have exactly the same Kingston kit and a similar 6200 CL28 profile? Have you managed to improve anything further while maintaining full stability?

Any suggestions or comparable M-die results would be appreciated.

I would especially be interested in results with FCLK 2133.