[RHBH]

A detailed gaming benchmark with RAM analysis (numbers of kits, SS or DS, etc...).
His conclusion is that 4 kits perform better than 2 sticks globally in games (for Ryzen 5000).
Of course, I assume that 4 kits is more difficult to overclock, especially in case of weak IMC:

Your conclusion is wrong.

Correction: 4 ranks is better than 2 ranks.

2 sticks with double ranks provides the same performance that 4 sticks with single ranks.

 

[Reica]

Your conclusion is wrong.

Correction: 4 ranks is better than 2 ranks.

2 sticks with double ranks provides the same performance that 4 sticks with single ranks.

Yeah, you seemingly want to match the number of ranks to the amount of channels your memory controller has.

The (odd?) way it works is like this:

• 2 single rank sticks = single rank
• 4 single rank sticks = dual rank
• 2 dual rank sticks = dual rank
• 4 dual rank sticks = quad rank

Ranks cannot be accessed simultaneously, but they can be addressed independently. This means the memory controller can write data to one rank and when that is done immediately receive read data from the other rank, instead of sending write data > having the RAM process it > then exporting read data back to the mem controller. This seems to give more bandwidth and throughput.

I think currently, unless you have a quad channel memory controller, running your RAM in a quad rank setup might be worsening your performance again. Could be wrong on this.

Edit: This is not just true for Zen 3. Ryzen 3000's and Intel CPU's also benefit from this. All as long as the application you're running is either CPU or Memory limited.

 

[Jesaul]

Ok. More fun with motherboard.
I've god 3rd set of memory - 2x8GB on qvl list. And it does not work in dual memory mode (0D error). I suspect that the person who told me that my 3600 is faulty may be true..

 

[RHBH]

Yeah, you seemingly want to match the number of ranks to the amount of channels your memory controller has.

The (odd?) way it works is like this:

• 2 single rank sticks = single rank
• 4 single rank sticks = dual rank
• 2 dual rank sticks = dual rank
• 4 dual rank sticks = quad rank

Ranks cannot be accessed simultaneously, but they can be addressed independently. This means the memory controller can write data to one rank and when that is done immediately receive read data from the other rank, instead of sending write data > having the RAM process it > then exporting read data back to the mem controller. This seems to give more bandwidth and throughput.

I think currently, unless you have a quad channel memory controller, running your RAM in a quad rank setup might be worsening your performance again. Could be wrong on this.

Edit: This is not just true for Zen 3. Ryzen 3000's and Intel CPU's also benefit from this. All as long as the application you're running is either CPU or Memory limited.

Tecnically having a quad rank setup (4 sticks with two ranks each) is the best performance you can get.

In practical terms you won't get the best performance from a quad rank setup because the amount of stress in the memory controller is much higher than having a dual rank setup.

So you most likely will lose performance due the lower clock / loosen timings required to stabilize all ranks, the benefit from a quad rank setup is not enough to compensate this.

If you can get 4x16GB stable at 4000MHz with a tight timing and 2000 FLCK, yeah that's the best possible performance, but you'll be luck if you can POST at more than 3600MHz with a quad rank setup.

That's the reason why AMD rates their memory controller as 3200MHz, because that's the higher clock they guarantee their memory controller will handle with whatever you throw at it.

 

[KingEngineRevUp]

Your conclusion is wrong.

Correction: 4 ranks is better than 2 ranks.

2 sticks with double ranks provides the same performance that 4 sticks with single ranks.

Yeah, you seemingly want to match the number of ranks to the amount of channels your memory controller has.

The (odd?) way it works is like this:

• 2 single rank sticks = single rank
• 4 single rank sticks = dual rank
• 2 dual rank sticks = dual rank
• 4 dual rank sticks = quad rank

Ranks cannot be accessed simultaneously, but they can be addressed independently. This means the memory controller can write data to one rank and when that is done immediately receive read data from the other rank, instead of sending write data > having the RAM process it > then exporting read data back to the mem controller. This seems to give more bandwidth and throughput.

I think currently, unless you have a quad channel memory controller, running your RAM in a quad rank setup might be worsening your performance again. Could be wrong on this.

Edit: This is not just true for Zen 3. Ryzen 3000's and Intel CPU's also benefit from this. All as long as the application you're running is either CPU or Memory limited.

So right now I am at 3800 MHz CL16 and 54ns of latency with dual channel single rank ram. Is there any benefit in me getting two more sticks for four sticks total? Specially if I game at 1440P?

 

[Jackalito]

Nice, good to know! What CL are You running on Your memory? And how many sticks do You have? What model? Are You running JEDEC, or D.O.C.P. memory settings? Or custom/manual?

My RAM is 16 GB. DDR4 G.SKILL TRIDENTZ F4-3600C15D-16GTZ (2x8GB) Single Rank, obviously.
No D.O.C.P., manually set as:

Note: VDIMM is 1.365V

Cheers!

 

[xeizo]

So right now I am at 3800 MHz CL16 and 54ns of latency with dual channel single rank ram. Is there any benefit in me getting two more sticks for four sticks total? Specially if I game at 1440P?

Latency will be worse as more ranks = more latency, but bandwidth will be higher. In games, bandwidth is more important for moar fps.

 

[KingEngineRevUp]

Latency will be worse as more ranks = more latency, but bandwidth will be higher. In games, bandwidth is more important for moar fps.

I watched the HU video and the single rank 3800MHz with 54.6ns seems to hold up against everything. Would you say there's hardly if any more benefit for me to getting 2 more sticks?

 

[xeizo]

I watched the HU video and the single rank 3800MHz with 54.6ns seems to hold up against everything. Would you say there's hardly if any more benefit for me to getting 2 more sticks? View attachment 2466881 View attachment 2466882 View attachment 2466883

The difference will be very small, it's never worth it unless you're hunting the last fps

 

[OblivionXT]

Wanted to share my personal 4-DIMM stability issues and "fixes" (Had to give up some minor functionality) These issues occur for me on virtually all BIOS versions, especially post 1302.

Current Build:
Crosshair VIII Hero Wi-Fi (BIOS 2702), Ryzen 3600, Team Group Xtreem 4500 - 4x8GB @ 3733 CL14 IF 1866

Issue #1: Randomly hangs up on POST with 1F or 01 Code, always passes POST once reset button is toggled.
Fix: Reset PROCODT and other termination resistances to Auto, RAM OC remained stable, no longer hangs up on POST

Issue #2: Frequent crashes, often within minutes of boot, no blue screen.
Fix: Reset C States to Auto from Enabled, these crashes seemed to occur when the CPU was unloading completed tasks and more cores were entering C6. C1 has had no issues thus far.

Issue #3: Occasional crashes intermittent and no pattern, no blue screen
Fix: Disabled all boost and PBO settings, set manual OC (4.2 Mhz @1.33v on my chip)

Hopefully I can get at least this level of stability once the 5900X arrives....next year, ha.

 

[RHBH]

So, let me clarify some things.

About the HU video

4 ranks running 3600CL16
Against
2 ranks running 3800CL16

Its obvious that higher frequency matters more than more ranks. But if you can run different number of ranks at the same frequency and timings more ranks will improve the performance.

In games going from two ranks to four ranks tends to show improvements around 2% to 4%, more improvements in memory bandwidth/CPU bound games.

Also there is a slightly mistake in HU video. Their 4000MHz are not in 1:1 as they were able to run 1900MHz FLCK. That's why you see 3800CL16 beating 4000CL16.

Regarding the "more ranks more latency myth".

It is true that more ranks will show you a higher latency in synthetic benchmarks such as AIDA64

These benchmarks will measure the DRAM latency by accessing all installed ranks, as explained before you cannot execute operations in more than one par at the same time. Thus the benchmark will measure more latency.

In practical terms more ranks will actually improve latency as your memory controller will be able to perform operations in one par of ranks while another par is refreshing. This is why more ranks show improvements in latency sensitivity applications such as games.

 

[dev1ance]

Have you guys with two sticks noticed that there's a problem using slots 1 and 3? I can do DDR4-3800 16-16-16 easily or DDR4-4000 on slots 2 and 4, but I wanted to try slots 1 and 3 but I'm getting the exact same problem that a lot of people complained about if I do (the problem is not booting above DDR4-3200 even though you can set IF higher, just memory cannot do above 3200 on slots 1 and 3). If I get another two sticks and put them in 1 and 3, will I be facing the same problem of not booting above DDR4-3200? Or is it not a problem if I'm running 4 sticks?

Following my previous post it seems that I am getting somewhere, yet it makes me feel wrong...
The 4x8Gb memory (F4-4000C15Q-32GVK) I got is unable to run anything above 3000Mhz with all 4 slots populated - immediate F9 error core

I checked the board with 2 DIMMs only :

1. With the A2 & B2 slots - 4000Mhz like a charm with the D.O.C.P. or manual. Tested all 4 sticks - 2 by 2 - any pair gets 4000 Mhz.
2. With the A1 & B1 slots - maximus is 3000Mhz - anything above is F9 as mentioned. - tested all 4 - any pair (or single) gets the same error above the 3000Mhz.

Reseated the CPU with the heatsink twice - same.
Checked with the BIOS v. 2402 beta and with the v. 2502 - same behavior.
However with the v. 2402 beta the MB allows 2000BCLK with the pair of the RAM sticks at A2 & B2 slots at 4000Mhz at 1:1 with stable Cinebench runs - not checked anything else yet. The recent BIOS v. 2502 is somehow limited with the BCLK - not getting the 2000 1:1 for me.

What is going on with this A1 B1 / A2 B2. Is it a faulty MB ? RMA ? Or is there any BIOS tweak I might try with the X570 to try to get the s slots to run above 3000Mhz ?
I have searched the thread and found different stories with the A1 B1 slots. Is there anyone here with this MB running fast 4 RAM sticks ?

Have you found a solution at all? I noticed something similar but DDR4-3200 is the limit on A1 and B1 for some reason

 

[OblivionXT]

Have you guys with two sticks noticed that there's a problem using slots 1 and 3? I can do DDR4-3800 16-16-16 easily or DDR4-4000 on slots 2 and 4, but I wanted to try slots 1 and 3 but I'm getting the exact same problem that a lot of people complained about if I do (the problem is not booting above DDR4-3200 even though you can set IF higher, just memory cannot do above 3200 on slots 1 and 3). If I get another two sticks and put them in 1 and 3, will I be facing the same problem of not booting above DDR4-3200? Or is it not a problem if I'm running 4 sticks?

This problem is specifically caused from only having sticks in the first and third slots. Without sticks in the second and fourth slots as well (four sticks total) you get signal reflection/interference.

 

[dev1ance]

This problem is specifically caused from only having sticks in the first and third slots. Without sticks in the second and fourth slots as well (four sticks total) you get signal reflection/interference.

Hmm, hopefully that's the case because IceB above is getting problems with 4 sticks populated and essentially being limited to A1 and B1 speeds...I have another kit incoming to run 4x8GB as well. Noticed it's working for you with Zen 2 though but the problem doesn't seem to be Zen 2 related as people were hitting higher speeds with their Zen 2 chips but Zen 3 was the culprit and limiting their speeds (not booting above 3200)

Is anyone else running 4 sticks just fine with Zen 3?

 

[lklem]

Hmm, hopefully that's the case because IceB above is getting problems with 4 sticks populated and essentially being limited to A1 and B1 speeds...I have another kit incoming to run 4x8GB as well. Noticed it's working for you with Zen 2 though but the problem doesn't seem to be Zen 2 related as people were hitting higher speeds with their Zen 2 chips but Zen 3 was the culprit and limiting their speeds (not booting above 3200)

Is anyone else running 4 sticks just fine with Zen 3?

Hmm, hopefully that's the case because IceB above is getting problems with 4 sticks populated and essentially being limited to A1 and B1 speeds...I have another kit incoming to run 4x8GB as well. Noticed it's working for you with Zen 2 though but the problem doesn't seem to be Zen 2 related as people were hitting higher speeds with their Zen 2 chips but Zen 3 was the culprit and limiting their speeds (not booting above 3200)

Is anyone else running 4 sticks just fine with Zen 3?

Ryzen 9 5900X, Asus C8HW, 4x8gb Gskill Trident Z 3200mhz CL14 @ 3600mhz CL14 (F4-3200C14D-16GTZR)

 

[lDevilDriverl]

Hmm, hopefully that's the case because IceB above is getting problems with 4 sticks populated and essentially being limited to A1 and B1 speeds...I have another kit incoming to run 4x8GB as well. Noticed it's working for you with Zen 2 though but the problem doesn't seem to be Zen 2 related as people were hitting higher speeds with their Zen 2 chips but Zen 3 was the culprit and limiting their speeds (not booting above 3200)

Is anyone else running 4 sticks just fine with Zen 3?

I'm running 4x8 bdie on 3800cl16 easily with c8hw and 5600x, same as on 3600x. On 2600x this kits was running on 3600cl14

 

[Syldon]

Looking for advice on tweaks. I think I set up as best as I can with the knowledge I have. I have ran Ryzen from CH6,7 and now 8 on 1800x, 2700x and now the 5950.

The stress tests included here do not include the 3090 in my profile. It arrives in about 1 hour's time. The GPU should have little to no effect on the stability runs.

Any advice is appreciated. Super impressed with what I am seeing so far.

Spoiler: Bios text dump

[2020/11/28 09:50:54]
Ai Overclock Tuner [Manual]
BCLK Frequency [100.0000]
Memory Frequency [DDR4-3800MHz]
FCLK Frequency [1900MHz]
Core Performance Boost [Auto]
CPU Core Ratio [Auto]
Core VID [Auto]
CCX0 Ratio [Auto]
CCX0 Ratio [Auto]
TPU [Keep Current Settings]
Performance Bias [Auto]
PBO Fmax Enhancer [Auto]
Precision Boost Overdrive [Auto]
Precision Boost Overdrive Scalar [Auto]
Max CPU Boost Clock Override [Auto]
Platform Thermal Throttle Limit [Auto]
DRAM CAS# Latency [16]
Trcdrd [16]
Trcdwr [16]
DRAM RAS# PRE Time [17]
DRAM RAS# ACT Time [34]
Trc [52]
TrrdS [7]
TrrdL [10]
Tfaw [28]
TwtrS [4]
TwtrL [12]
Twr [14]
Trcpage [Auto]
TrdrdScl [4]
TwrwrScl [4]
Trfc [333]
Trfc2 [Auto]
Trfc4 [Auto]
Tcwl [16]
Trtp [14]
Trdwr [14]
Twrrd [6]
TwrwrSc [Auto]
TwrwrSd [Auto]
TwrwrDd [Auto]
TrdrdSc [Auto]
TrdrdSd [Auto]
TrdrdDd [Auto]
Tcke [Auto]
ProcODT [53.3 ohm]
Cmd2T [1T]
Gear Down Mode [Disabled]
Power Down Enable [Disabled]
RttNom [RZQ/7]
RttWr [RZQ/3]
RttPark [RZQ/1]
MemAddrCmdSetup [Auto]
MemCsOdtSetup [Auto]
MemCkeSetup [Auto]
MemCadBusClkDrvStren [24.0 Ohm]
MemCadBusAddrCmdDrvStren [20.0 Ohm]
MemCadBusCsOdtDrvStren [20.0 Ohm]
MemCadBusCkeDrvStren [20.0 Ohm]
Mem Over Clock Fail Count [Auto]
Voltage Monitor [Die Sense]
CPU Load-line Calibration [Level 2]
CPU Current Capability [120%]
CPU VRM Switching Frequency [Auto]
VRM Spread Spectrum [Auto]
CPU Power Duty Control [T.Probe]
CPU Power Phase Control [Auto]
CPU Power Thermal Control [120]
VDDSOC Load-line Calibration [Level 1]
VDDSOC Switching Frequency [Auto]
VDDSOC Phase Control [Auto]
DRAM Current Capability [110%]
DRAM Power Phase Control [Extreme]
DRAM Switching Frequency [Auto]
CPU Core Current Telemetry [Auto]
CPU SOC Current Telemetry [Auto]
Force OC Mode Disable [Disabled]
SB Clock Spread Spectrum [Auto]
VTTDDR Voltage [Auto]
VPP_MEM Voltage [Auto]
DRAM CTRL REF Voltage on CHA [Auto]
DRAM CTRL REF Voltage on CHB [Auto]
VDDP Voltage [Auto]
1.8V Standby Voltage [Auto]
CPU 3.3v AUX [Auto]
1.2V SB Voltage [Auto]
DRAM R1 Tune [Auto]
DRAM R2 Tune [Auto]
DRAM R3 Tune [Auto]
DRAM R4 Tune [Auto]
PCIE Tune R1 [Auto]
PCIE Tune R2 [Auto]
PCIE Tune R3 [Auto]
PLL Tune R1 [Auto]
PLL reference voltage [Auto]
T Offset [Auto]
Sense MI Skew [Auto]
Sense MI Offset [Auto]
Promontory presence [Auto]
Clock Amplitude [Auto]
CPU Core Voltage [Auto]
CPU SOC Voltage [Auto]
DRAM Voltage [1.36000]
VDDG CCD Voltage Control [Auto]
VDDG IOD Voltage Control [Auto]
CLDO VDDP voltage [Auto]
1.00V SB Voltage [Auto]
1.8V PLL Voltage [Auto]
TPM Device Selection [Discrete TPM]
Erase fTPM NV for factory reset [Enabled]
PSS Support [Enabled]
PPC Adjustment [PState 2]
NX Mode [Enabled]
SVM Mode [Disabled]
SMT Mode [Auto]
Core Leveling Mode [Automatic mode]
CCD Control [Auto]
SATA Port Enable [Enabled]
SATA Mode [AHCI]
NVMe RAID mode [Disabled]
SMART Self Test [Enabled]
Hot Plug [Disabled]
Hot Plug [Disabled]
Hot Plug [Disabled]
Hot Plug [Disabled]
Hot Plug [Disabled]
Hot Plug [Disabled]
Hot Plug [Disabled]
Hot Plug [Disabled]
HD Audio Controller [Enabled]
PCIEX16_1 Bandwidth [Auto Mode]
PCIEX16_2 Bandwidth [Auto Mode]
When system is in working state [All On]
Q-Code LED Function [POST Code Only]
When system is in sleep, hibernate or soft off states [Aura Off]
Realtek 2.5G LAN Controller [Enabled]
Realtek PXE OPROM [Disabled]
Intel LAN Controller [Enabled]
Intel LAN OPROM [Disabled]
ASM1074 Controller [Enabled]
USB power delivery in Soft Off state (S5) [Enabled]
PCIEX16_1 Mode [Auto]
PCIEX16_2 Mode [Auto]
PCIEX1 Mode [Auto]
PCIEX16_3 Mode [Auto]
M.2_1 Link Mode [Auto]
M.2_2 Link Mode [Auto]
SB Link Mode [Auto]
ErP Ready [Disabled]
Restore AC Power Loss [Power Off]
Power On By PCI-E [Disabled]
Power On By RTC [Disabled]
Above 4G Decoding [Disabled]
SR-IOV Support [Disabled]
Legacy USB Support [Enabled]
XHCI Hand-off [Enabled]
KingstonDataTraveler 2.0PMAP [Auto]
USB Device Enable [Enabled]
U32G2_2 [Enabled]
U32G2_3 [Enabled]
U32G2_4 [Enabled]
U32G1_10 [Enabled]
U32G1_11 [Enabled]
USB12 [Enabled]
USB13 [Enabled]
U32G2_7 [Enabled]
U32G2_8 [Enabled]
U32G2_C9 [Enabled]
Network Stack [Disabled]
Device [SATA6G_8: Samsung SSD 850 EVO 2TB]
CPU Temperature [Monitor]
CPU Package Temperature [Monitor]
MotherBoard Temperature [Monitor]
VRM Temperature [Monitor]
T_Sensor Temperature [Monitor]
Water In T Sensor Temperature [Monitor]
Water Out T Sensor Temperature [Monitor]
CPU Fan Speed [Monitor]
CPU Optional Fan Speed [Monitor]
Chassis Fan 1 Speed [Monitor]
Chassis Fan 2 Speed [Monitor]
Chassis Fan 3 Speed [Monitor]
High Amp Fan Speed [Monitor]
W_PUMP+ Speed [Monitor]
AIO PUMP Speed [Monitor]
PCH Fan Speed [Monitor]
Flow Rate [Monitor]
CPU Core Voltage [Monitor]
3.3V Voltage [Monitor]
5V Voltage [Monitor]
12V Voltage [Monitor]
CPU Q-Fan Control [PWM Mode]
CPU Fan Step Up [0 sec]
CPU Fan Step Down [0 sec]
CPU Fan Speed Lower Limit [200 RPM]
CPU Fan Profile [Manual]
CPU Upper Temperature [68]
CPU Fan Max. Duty Cycle (%) [100]
CPU Middle Temperature [60]
CPU Fan Middle Duty Cycle (%) [100]
CPU Lower Temperature [30]
CPU Fan Min. Duty Cycle (%) [20]
Chassis Fan 1 Q-Fan Control [Auto]
Chassis Fan 1 Q-Fan Source [CPU]
Chassis Fan 1 Step Up [0 sec]
Chassis Fan 1 Step Down [0 sec]
Chassis Fan 1 Speed Low Limit [Ignore]
Chassis Fan 1 Profile [Silent]
Chassis Fan 2 Q-Fan Control [PWM Mode]
Chassis Fan 2 Q-Fan Source [CPU]
Chassis Fan 2 Step Up [0 sec]
Chassis Fan 2 Step Down [0 sec]
Chassis Fan 2 Speed Low Limit [Ignore]
Chassis Fan 2 Profile [Manual]
Chassis Fan 2 Upper Temperature [65]
Chassis Fan 2 Max. Duty Cycle (%) [100]
Chassis Fan 2 Middle Temperature [60]
Chassis Fan 2 Middle Duty Cycle (%) [75]
Chassis Fan 2 Lower Temperature [20]
Chassis Fan 2 Min. Duty Cycle (%) [40]
Chassis Fan 3 Q-Fan Control [Auto]
Chassis Fan 3 Q-Fan Source [CPU]
Chassis Fan 3 Step Up [0 sec]
Chassis Fan 3 Step Down [0 sec]
Chassis Fan 3 Speed Low Limit [200 RPM]
Chassis Fan 3 Profile [Standard]
High Amp Fan Q-Fan Control [Auto]
High Amp Fan Q-Fan Source [CPU]
High Amp Fan Step Up [0 sec]
High Amp Fan Step Down [0 sec]
High Amp Fan Speed Low Limit [200 RPM]
High Amp Fan Profile [Standard]
WATER PUMP+ Control [Disabled]
AIO PUMP Control [Disabled]
Fast Boot [Enabled]
Next Boot after AC Power Loss [Fast Boot]
Boot Logo Display [Disabled]
Bootup NumLock State [On]
POST Report [1 sec]
Wait For 'F1' If Error [Enabled]
Option ROM Messages [Force BIOS]
Interrupt 19 Capture [Disabled]
Setup Mode [Advanced Mode]
Launch CSM [Disabled]
OS Type [Other OS]
AMI Native NVMe Driver Support [Enabled]
Flexkey [Reset]
Setup Animator [Disabled]
Load from Profile [1]
Profile Name [3800 stable]
Save to Profile [1]
DIMM Slot Number [DIMM_A2]
Bus Interface [PCIEX16_1]
Download & Install ARMOURY CRATE app [Enabled]
CPU Frequency [0]
CPU Voltage [0]
CCD Control [Auto]
Core control [Auto]
SMT Control [Auto]
Overclock [Enabled ]
Memory Clock Speed [Auto]
Tcl [Auto]
Trcdrd [Auto]
Trcdwr [Auto]
Trp [Auto]
Tras [Auto]
Trc Ctrl [Auto]
TrrdS [Auto]
TrrdL [Auto]
Tfaw Ctrl [Auto]
TwtrS [Auto]
TwtrL [Auto]
Twr Ctrl [Auto]
Trcpage Ctrl [Auto]
TrdrdScL Ctrl [Auto]
TwrwrScL Ctrl [Auto]
Trfc Ctrl [Auto]
Trfc2 Ctrl [Auto]
Trfc4 Ctrl [Auto]
Tcwl [Auto]
Trtp [Auto]
Tcke [Auto]
Trdwr [Auto]
Twrrd [Auto]
TwrwrSc [Auto]
TwrwrSd [Auto]
TwrwrDd [Auto]
TrdrdSc [Auto]
TrdrdSd [Auto]
TrdrdDd [Auto]
ProcODT [Auto]
Power Down Enable [Auto]
Cmd2T [Auto]
Gear Down Mode [Auto]
CAD Bus Timing User Controls [Auto]
CAD Bus Drive Strength User Controls [Auto]
Data Bus Configuration User Controls [Auto]
Infinity Fabric Frequency and Dividers [Auto]
ECO Mode [Disable]
Precision Boost Overdrive [Auto]
LN2 Mode [Auto]
SoC Voltage [0]
SoC/Uncore OC Mode [Disabled]
VDDP Voltage Control [Auto]
VDDG Voltage Control [Auto]
NUMA nodes per socket [Auto]
Custom Pstate0 [Auto]
L1 Stream HW Prefetcher [Auto]
L2 Stream HW Prefetcher [Auto]
Core Watchdog Timer Enable [Auto]
SMEE [Auto]
Core Performance Boost [Auto]
Global C-state Control [Auto]
Power Supply Idle Control [Auto]
SEV ASID Count [Auto]
SEV-ES ASID Space Limit Control [Auto]
Streaming Stores Control [Auto]
Local APIC Mode [Auto]
ACPI _CST C1 Declaration [Auto]
MCA error thresh enable [Auto]
PPIN Opt-in [Auto]
Fast Short REP MOVSB [Enabled]
Enhanced REP MOVSB/STOSB [Enabled]
RdRand Speedup Disable [Enabled]
IBS hardware workaround [Auto]
DRAM scrub time [Auto]
Poison scrubber control [Auto]
Redirect scrubber control [Auto]
Redirect scrubber limit [Auto]
NUMA nodes per socket [Auto]
Memory interleaving [Auto]
Memory interleaving size [Auto]
1TB remap [Auto]
DRAM map inversion [Auto]
ACPI SRAT L3 Cache As NUMA Domain [Auto]
ACPI SLIT Distance Control [Auto]
ACPI SLIT remote relative distance [Auto]
GMI encryption control [Auto]
xGMI encryption control [Auto]
CAKE CRC perf bounds Control [Auto]
4-link xGMI max speed [Auto]
3-link xGMI max speed [Auto]
xGMI TXEQ Mode [Auto]
PcsCG control [Auto]
Disable DF to external downstream IP SyncFloodPropagation [Auto]
Disable DF sync flood propagation [Auto]
CC6 memory region encryption [Auto]
Memory Clear [Auto]
Overclock [Enabled]
Memory Clock Speed [Auto]
Tcl [Auto]
Trcdrd [Auto]
Trcdwr [Auto]
Trp [Auto]
Tras [Auto]
Trc Ctrl [Auto]
TrrdS [Auto]
TrrdL [Auto]
Tfaw Ctrl [Auto]
TwtrS [Auto]
TwtrL [Auto]
Twr Ctrl [Auto]
Trcpage Ctrl [Auto]
TrdrdScL Ctrl [Auto]
TwrwrScL Ctrl [Auto]
Trfc Ctrl [Auto]
Trfc2 Ctrl [Auto]
Trfc4 Ctrl [Auto]
Tcwl [Auto]
Trtp [Auto]
Tcke [Auto]
Trdwr [Auto]
Twrrd [Auto]
TwrwrSc [Auto]
TwrwrSd [Auto]
TwrwrDd [Auto]
TrdrdSc [Auto]
TrdrdSd [Auto]
TrdrdDd [Auto]
ProcODT [Auto]
Power Down Enable [Auto]
Disable Burst/Postponed Refresh [Auto]
DRAM Maximum Activate Count [Auto]
Cmd2T [Auto]
Gear Down Mode [Auto]
CAD Bus Timing User Controls [Auto]
CAD Bus Drive Strength User Controls [Auto]
Data Bus Configuration User Controls [Auto]
Data Poisoning [Auto]
DRAM Post Package Repair [Default]
RCD Parity [Auto]
DRAM Address Command Parity Retry [Auto]
Write CRC Enable [Auto]
DRAM Write CRC Enable and Retry Limit [Auto]
Disable Memory Error Injection [True]
DRAM ECC Symbol Size [Auto]
DRAM ECC Enable [Auto]
DRAM UECC Retry [Auto]
TSME [Auto]
Data Scramble [Auto]
DFE Read Training [Auto]
FFE Write Training [Auto]
PMU Pattern Bits Control [Auto]
MR6VrefDQ Control [Auto]
CPU Vref Training Seed Control [Auto]
Chipselect Interleaving [Auto]
BankGroupSwap [Auto]
BankGroupSwapAlt [Disabled]
Address Hash Bank [Auto]
Address Hash CS [Auto]
Address Hash Rm [Auto]
SPD Read Optimization [Enabled]
MBIST Enable [Disabled]
Pattern Select [PRBS]
Pattern Length [6]
Aggressor Channel [1 Aggressor Channel]
Aggressor Static Lane Control [Disabled]
Target Static Lane Control [Disabled]
Worst Case Margin Granularity [Per Chip Select]
Read Voltage Sweep Step Size [1]
Read Timing Sweep Step Size [1]
Write Voltage Sweep Step Size [1]
Write Timing Sweep Step Size [1]
IOMMU [Auto]
Precision Boost Overdrive [Auto]
Precision Boost Overdrive Scalar [Auto]
FCLK Frequency [Auto]
SOC OVERCLOCK VID [0]
UCLK DIV1 MODE [Auto]
VDDP Voltage Control [Auto]
VDDG Voltage Control [Auto]
SoC/Uncore OC Mode [Auto]
LN2 Mode [Auto]
ACS Enable [Auto]
PCIe ARI Support [Auto]
PCIe ARI Enumeration [Auto]
PCIe Ten Bit Tag Support [Auto]
cTDP Control [Auto]
EfficiencyModeEn [Auto]
Package Power Limit Control [Auto]
APBDIS [Auto]
DF Cstates [Auto]
CPPC [Auto]
CPPC Preferred Cores [Auto]
NBIO DPM Control [Auto]
Early Link Speed [Auto]
Presence Detect Select mode [Auto]
Preferred IO [Auto]
CV test [Auto]
Loopback Mode [Auto]
Data Link Feature Exchange [Disabled]

 

[CyrIng]

Those 500's lines of code may help with the DIMM(s) topology

As a low level program zencli can also peek into the SMU, FCH, PM, BIOS ...

CyrIng

 

[Deluxe1]

I've just installed a 5900x in my Crosshair VIII Hero and when playing games it won't boost at all and is locked to it's base clock of 3.7

If I run cinebench then it boosts just fine.

Is this just a bug?

 

[Syldon]

I've just installed a 5900x in my Crosshair VIII Hero and when playing games it won't boost at all and is locked to it's base clock of 3.7

If I run cinebench then it boosts just fine.

Is this just a bug?

Can you dump a copy of your bios settings ?