Mid-Life Crisis

[J7SC_Orion]

Good Progress ! With these Ryzen 5K, lots of permutations and combinations to find 'that setup'. 

 

FYI, per HardwareUnboxed vid, 4 sticks of Dual Rank > 2 sticks of Dual Rank, but not that drastic a difference.

 

Re. your Aida latency...there might still be some room (ignore the ZenTimings IF2000/DDR4 4000 at lower right - not fully stable yet). My daily setup latency w/ 4 sticks SR hovers between 54ns and 57ns 

 

 

 

 

 

 

Edited January 23, 2022 by J7SC_Orion

[ArchStanton]

5 hours ago, J7SC_Orion said:

Good Progress ! With these Ryzen 5K, lots of permutations and combinations to find 'that setup'. 

 

FYI, per HardwareUnboxed vid, 4 sticks of Dual Rank > 2 sticks of Dual Rank, but not that drastic a difference.

 

Re. your Aida latency...there might still be some room (ignore the ZenTimings IF2000/DDR4 4000 at lower right - not fully stable yet). My daily setup latency w/ 4 sticks SR hovers between 54ns and 57ns 

 

Thank you for the encouraging words.  Sub 60ns latency and an fclk of 1900 are hopefully in the cards for me.

[ArchStanton]

7hrs of prime95 and no errors.  Far from conclusive, but certainly encouraging.  I don't have much faith in some of the maximum temperature readings being reported by HWiNFO64 (or they are exceedingly fleeting), since if they were present for more than a flickering moment there should be a corresponding increase in recorded fan RPM.  I may let the test run again and log both the temps and RPMs just to have a better picture of the duration and frequency of the spikes.

 

[ArchStanton]

This particular piece of software has exceeded my expectations.  The screenshot was taken while prime95 was still active.  Hence the elevated temperatures on some of the components.  All of my fans are governed by the master "control curve" in the upper left of the picture.  It uses the maximum PWM duty cycle dictated by the 5 separate fans curves tied to individual temperature readings in the lower portion of the picture (I have each base curve set to 1-degree Celsius hysteresis with a 2 second response time to smooth changes in RPM out to levels I find acceptable).  Currently I do not have the option to create a base curve based on memory temperature as Fan Control v101 doesn't have the DRAM temperature readings in the list of possible sensors when creating a new curve.  I need to find or create a HWiNFO64 "plugin" for FCv101.  However, creating one is, for the moment, pretty far outside my comfort zone.  Regardless, those with more ambition and or time could really optimize the settings of their cooling system with FCv101.

[ArchStanton]

Reading comprehension is the most undervalued skill known to humanity...

 

 

[ArchStanton]

The good:  Another 8 hours of prime95 without issue.  I did save the .CSV log from HWiNFO64.  I'm pretty confident my VRM temperatures peaked at 40 Celsius rather than 95 indicated in the picture below since there was only 1 reading above 40 in 27,331 data points .  I also think it very possible that my mysterious wonkiness two nights ago was caused by the "Embedded Controller" issue highlighted in my post just prior to this one (rather than my RAM).  This controller's inability to play nice with HWiNFO64 at all times could explain some of the erroneous temperature readings I've seen in the past as well.

 

The bad:  Feeling like an idiot after wasting two nights because after reading that warning the first time I launched HWiNFO64 I completely ignored it from then on.

 

 

[ArchStanton]

Results with the most basic level of PBO turned on.

 

 

[ArchStanton]

Summary.

 

Significant jumps in multicore performance.  Slight degradation in single core performance.  Good for productivity and less than ideal for gaming.

 

[Sir Beregond]

How are you liking the Dark Hero board? That's what I have sitting in a box waiting for me to build it.

[ArchStanton]

1 hour ago, Sir Beregond said:

How are you liking the Dark Hero board? That's what I have sitting in a box waiting for me to build it.

In general, I like it very well.  I could really use a small novel explaining all the BIOS options, but I'm getting there one google search at a time .  I'm very glad there's an option to disable the MB's onboard ARGB effects, as there's a ridiculous amount of ASUS bloat required to control them, and they default to "rainbow puke".  It will be interesting to see what kind of fclk she can carry in the days to come .

[ArchStanton]

Began with some Core Optimizer tuning tonight.  I don't have it fully dialed in by a long shot, nor is stability truly confirmed, but it felt like progress none the less.

 

 

[Sir Beregond]

1 hour ago, ArchStanton said:

In general, I like it very well.  I could really use a small novel explaining all the BIOS options, but I'm getting there one google search at a time .  I'm very glad there's an option to disable the MB's onboard ARGB effects, as there's a ridiculous amount of ASUS bloat required to control them, and they default to "rainbow puke".  It will be interesting to see what kind of fclk she can carry in the days to come .

Oh that's good. I intend to turn off all RGB lights on the motherboard and my G.Skill RAM kit as well.

 

Looking forward to seeing what you get out of it. 

Edited January 24, 2022 by Sir Beregond

[ArchStanton]

Small update:

 

After a lot of additional reading, I swapped corecycler from the default test to 720K, and as others elsewhere had indicated, this exposed additional instabilities.  I reduced the CO offset (less negative) for about half the cores to regain stability.  Naturally, this resulted in regression for all benchmarks.

 

Currently, my plan of attack is to "tune" curve optimizer for the best stable (passes all benchmarks along with 24hrs of every permutation of corecycle I have the patience to throw at it) performance I can get from a single core.  Then, I will make use of the Dark Hero's "Dynamic Overclock Switching" feature to layer a per CCX static OC on top of the PBO/CO settings.  Once that is accomplished, I hope to start squeezing fclk and RAM.

 

In the meantime, I've been trying to improve my understanding of the PPT/TDC/EDC limits and their effects on performance (I believe I am correct in stating that these parameters are really only relevant for high thread count workloads).  The "auto" PBO limits are 140/95/140.  If I change the setting to "Motherboard", the Dark Hero's default limits are 395/255/200.  From what I've read, it seems quite common for the sustained amperage limit (TDC) to be set higher than the short-term amperage limit (EDC), but I've been unable to completely grasp why this should be the case. 

 

I made quite a few runs in CBR23 single last night with CPPC and CPPC Preferred Cores enabled.  Under those conditions, Windows always assigns the load to core 1.  So, I suppose it has the most aggressive factory V/F curve and will likely have a less negative CO offset when stable than my weaker cores.  I tried various combinations of Boost Override and Curve Offset on core 1 last night.  I do not consider the values in the table below to reflect the performance of a "stable" system as I did not run any other stress tests or benchmarks.

Edited January 26, 2022 by ArchStanton

[ArchStanton]

Based on last night's testing, I expect my peak stable single core performance to top out around 5000-5025Mhz.  I was seeing around 5050Mhz in effective clocks with no observed clock stretching, but I will likely need to go less negative on the offset for true stability, and this will bring the Mhz down some.

 

I tried some other settings not included in the table.  A Boost Override setting of +200 with a positive CO of 30 limited the actual boost to around 4650Mhz.  So, I have the impression that one wants the CO offset as negative as possible, and the BO set low enough to prevent clock stretching and maintain stability.

 

I've debated installing Ryzen Master to speed up the process but concerns over what might install alongside it have prevented me from doing so to date.

 

If anyone has suggestions/guidance regarding PPT/TDC/EDC, voltages for the static OC, best way to calculate Dynamic OC switch point, CPPC on vs CPPC off, or anything else, I'm all ears .

 

My impression is that the rabbit hole's width increases in proportion to its depth 

[ENTERPRISE]

Great information @ArchStantonI am a fellow owner of the 5950X is is a beast to get the most out of it. I actually ended going away from PBO and going for the PER CCX all core OC.  I did some playing with PBO and Curve tuner, but it was not for me, so went old school and abandoned focusing on single core performance and optimized for multithreaded. 

 

Keep up the good work

[J7SC_Orion]

The 5950X / Dark hero combo is already pretty much 'optimized' right out of the box. I've tried different PPT/TDC/EDC and ' + clock boost' options but I usually just get back to the stock bios settings for most apps, or the DynamicOC.

 

I would also be very careful with BCLK...in the olden days, that was a mainstay of my oc tuning (mostly for Intel back then, and I typically would start at 100MHz > 103.5MHz). But with the 5950X, either the nvme drive(s) disappear, or bench results drop dramatically because auto boost settings disable when doing smaller BCLK increments.   

[ArchStanton]

20 hours ago, J7SC_Orion said:

The 5950X / Dark hero combo is already pretty much 'optimized' right out of the box. I've tried different PPT/TDC/EDC and ' + clock boost' options but I usually just get back to the stock bios settings for most apps, or the DynamicOC.

 

I would also be very careful with BCLK...in the olden days, that was a mainstay of my oc tuning (mostly for Intel back then, and I typically would start at 100MHz > 103.5MHz). But with the 5950X, either the nvme drive(s) disappear, or bench results drop dramatically because auto boost settings disable when doing smaller BCLK increments.   

Thank you for the feedback .  As a journeyman PC "assembler" / novice PC OC'er, I don't intend to touch the BCLK at this time.  If last night's testing (detailed in a subsequent post to this thread) showed me anything, it showed me how truly daunting a task it's going to be to improve upon AMD/ASUS's "out of the box" optimizations.  It's relatively easy to "twist knobs" and improve a given benchmark.  Getting the system to pass a "comprehensive" set of stability tests afterwards is more challenging in the extreme.

 

This is all just for "shits and giggles".  I don't need the machine to perform at even half of its default capacity. If I wish to enjoy this "journey" long term, I'm going to have to train myself to ignore the desire to immediately "win".  

 

My current goal, beyond having fun and learning more about what's "behind the curtain" of the myriad of BIOS settings" is the best "stable" OC I can achieve.  For me, "stable" implies it passes not just benchmarks, but the worst-case scenarios tools like Corecycler can throw at it.  I am beginning to suspect that my gains, if any, are going to be minimal, but that is logical.  The default optimizations were formulated by people who do this for a living.  They aren't Olympian gods with mythical powers, and they may have left some crumbs under the table, but can a rat like me find them?  For myself, I think stamina is possibly the most important factor in determining how that question gets answered.  Thinking about it just now, this is just a "gaming computer" and I have decided, for the time being, that I want to play "Benchmarkz IV, Fool's Errand" .

[ArchStanton]

Before initiating last night's session, I opted to reinstall 64GB of RAM.  Not stopping right there setting the BIOS back to defaults and verifying the stability of both RAM and the system in general before proceeding one tweak at a time was a noobish naive numbskulled mistake.  Because now I have no way of knowing what particular change resulted in the never-ending series of errors testing exposed.  So, no charts with conditions and results today.  Tonight, its back to square one, after I tattoo "patience, grasshoppah" on my forehead .

Edited January 27, 2022 by ArchStanton

[Sir Beregond]

21 hours ago, J7SC_Orion said:

The 5950X / Dark hero combo is already pretty much 'optimized' right out of the box. I've tried different PPT/TDC/EDC and ' + clock boost' options but I usually just get back to the stock bios settings for most apps, or the DynamicOC.

 

I would also be very careful with BCLK...in the olden days, that was a mainstay of my oc tuning (mostly for Intel back then, and I typically would start at 100MHz > 103.5MHz). But with the 5950X, either the nvme drive(s) disappear, or bench results drop dramatically because auto boost settings disable when doing smaller BCLK increments.   

Even on Intel though, I've very rarely touched BCLK unless just trying to squeeze that last little bit out.

 

As for AMD, last AMD system I had was like an Athlon back in the day say 20 years ago. I am going to have to read up on how the Ryzen overclocking works. I'm a bit lost on that front.

[ArchStanton]

14 minutes ago, Sir Beregond said:

I am going to have to read up on how the Ryzen overclocking works. I'm a bit lost on that front.

Some light reading to get you started...

 

OFFICIAL 5900X and 5950X two chiplet Zen 3 CPUs Overclocking thread | Overclock.net

 

[J7SC_Orion]

15 minutes ago, ArchStanton said:

Some light reading to get you started...

 

OFFICIAL 5900X and 5950X two chiplet Zen 3 CPUs Overclocking thread | Overclock.net

 

 

I also find this OCN thread very helpful re. the Asus CH8 series...Asus ROG also posts beta bios there - for us guinea pigs  > here

[ArchStanton]

1 hour ago, ArchStanton said:

Some light reading to get you started...

Here are a few more.

 

Asus Crosshair VIII DARK Hero.... Official Thread!? | Overclock.net

 

CoreCycler - tool for testing Curve Optimizer settings | Overclock.net

 

[Official] AMD Ryzen DDR4 24/7 Memory Stability Thread | Overclock.net

 

I'm still in the process of reading through them.  I've never actually reached the end of any of them, and I've reread portions of them many many times...

[ArchStanton]

However, I see you are aware of at least one of those threads already...

[Sir Beregond]

1 hour ago, ArchStanton said:

However, I see you are aware of at least one of those threads already...

Yep. I did manage to snag a Dark Hero, but still on the 5600X for now.

 

Going to look at snagging a 5900X around the time Zen 4 comes out I think. I don't have a want/reason to go 5950X since I primarily will just be gaming, but would like to have some fun with OC too.

Edited January 27, 2022 by Sir Beregond

[ArchStanton]

Update.

 

 

CoreCycler seems to be the consensus tool for verifying Curve Optimizer settings on the Ryzen 5000 platform.  As such, I have made frequent use of it to date.  However, CoreCycler appears to indicate I have "out of the box" instabilities.

 

I have previously checked the RAM installed in the system (2 DIMM's only, 32GB total) using 10 cycles of Testmem5 ([email protected]) and OCCT (6 hours) with DOCP both enabled and disabled.  I have never had either program indicate a memory error.

 

If I clear CMOS and start from scratch, I get the results detailed below.

 

CMOS reset settings.txt