ClockTuner for Ryzen Simplifies "Zen 2" Overclocking, Squeezes Out Double-digit Percent Performance

[Agent_D]

Or maybe you're the one that's clueless and doesn't know how to use it? It worked perfectly fine for me with CJR modules. I'm up 200MHz with lower latency than my modules were sold as and that's with four sticks. So please keep your opinions to yourself.

Its his experience with the tool; I have the same experience with it on the CJR and DJR modules in that I cannot get them to work no matter which settings are chosen, but I have also used it on the samsung bdie and had no issues with it. There may be some very poorly binned CJR/DJR that just doesn't work with the settings the tool recommends. No reason to get snarky about others' experience.

 

[thesmokingman]

What does this have to with the software you're trashing? It doesn't even support your modules
And yes, you clearly have no idea what you're talking about.

Get the F over yourself. Dram calc doesn't always work and the values get whacked from version to version. Not everyone has this dream experience and your odd fanboying is really weird.

 

[TheLostSwede]

Sorry, but incompetence doesn't make it a bad tool. Keep in mind what it is, it's a third party tool to get you started, it's not guaranteed to work and it's not from AMD.
Seriously, what are you people expecting? It works perfectly fine on my CJR modules, so I'm sorry to say, that it's most likely down to incompetence if you can't get it working and maybe if you can't do something as simple as changing a handful of setting in the UEFI, then maybe you shouldn't tinker around in there at all.

But yeah, it's all me, while it's apparently ok to attack the creator of this tool and call his software out as trash, makes perfect sense

 

[Max(IT)]

Sorry, but incompetence doesn't make it a bad tool. Keep in mind what it is, it's a third party tool to get you started, it's not guaranteed to work and it's not from AMD.
Seriously, what are you people expecting? It works perfectly fine on my CJR modules, so I'm sorry to say, that it's most likely down to incompetence if you can't get it working and maybe if you can't do something as simple as changing a handful of setting in the UEFI, then maybe you shouldn't tinker around in there at all.

But yeah, it's all me, while it's apparently ok to attack the creator of this tool and call his software out as trash, makes perfect sense

You seem to be very happy about DRAM calculator, and that’s fine.
And you are right about there is no point attacking 1usmus about his wonderful tools, which are useful and great.
But the DRAM calculator is not working for everyone, and that has nothing to do with incompetence. Sometimes DRAM modules are just too bad to go any faster than XMP values.
In my case (Hynix DJR on G.Skill 3600CL16 sticks) the suggested values worked but the performance gain were negligible at best, so I’d stick with defaults XMP.

Cant wait to try this. I am basically settled with my settings but more fine tuning is better.

Linus did say however that as per 1usmus, this works better with a cpu with more ccx.

I’m intrigued about the LLC level 4 (Asus) suggested setting.
During my fine tuning of the Ryzen 9 3900X I was under the impression that LLC Level 4 on my Asus ROG Strix B550-F was providing a sort of benefit for my results.
I asked for advice here but someone told me “leave it on AUTO because you are overvolting without a reason”. Since I’ve just recently jumped ship from Intel (where I was quite expert with those settings) I didn’t argue about that.
Now it is interesting to see 1usmus to suggest the same setting I was using with my slightly undervolted 3900X.

 

[nangu]

what about single core performance ? After you tuned the CPU for all core, I mean..

Single core is worse, but only measurable in benchmarking tools. While gaming, my fastest core at stock + PBO sustain 4.3 / 4.35 Ghz and jumping all over the place due to the boost algorithm, and very temperature dependant.

My CCX overclock is 4.35/4.35/4.25/4.25 @1.3v (1.26v after droop), so for gaming is the same as PBO, but I gain performance in multithreaded applications such as rendering, encoding, etc., at slightly more power draw, about 10 watts at 100% load, and performance is always the same regardless the ambient temp. Stock + PBO I loss performance in summer because the raised ambient temp and how the boost algorithm depends on that.

In a normal system, that you use everyday with Windows bloatware and a lot of background apps, I never saw my CPU sustain that 4.6Ghz single core in apps or games. The best I got was 4.4Ghz under certain scenarios, and around 4.5Ghz in benchmarks such as CB, AIDA or CPU-z, so the theoretical single core boost advertised by AMD has zero meaning in everyday usage in my opinion. It's only a peak, instantaneous value, sustained is much lower in a normal system with a lot of apps loaded. My all core sustained clocks @ stock are 4.05 / 4.2Ghz load dependant, so in my particular system, dialing a CCX overclock as I did is a no brainer.

For others with much better binned CPUs the situation would be different, of course.

 

[TheoneandonlyMrK]

I’m curious enough to try tuning my 3800X. It would be one way to find out if it’s a gold bin.

It is what it is, they're are higher clocked stock chip's so I wouldn't worry about gold bins.

I'll try it too but doubt it is beating my best 4.5 all core run or my usual Oc tactics of defaults ,and memory OC.

Regardless Ty 1usmus I hope you prove me wrong and it's nice to have some tools to mess with, Ryzen master could be better, I like it ,but it could, per core clock's for example.

 

[nangu]

It is what it is, they're are higher clocked stock chip's so I wouldn't worry about gold bins.

I'll try it too but doubt it is beating my best 4.5 all core run or my usual Oc tactics of defaults ,and memory Of.

Regardless Ty 1usmus I hope you prove me wrong and it's nice to have some tools to mess with, Ryzen master could be better, I like it ,but it could, per core clock's for example.

Agree, per core clock and AVX all core max clock and voltage would be great IMO.

 

[Max(IT)]

Single core is worse

as usual...

but only measurable in benchmarking tools.

most probably true.

While gaming, my fastest core at stock + PBO sustain 4.3 / 4.35 Ghz and jumping all over the place due to the boost algorithm, and very temperature dependant.

Well, gaming it’s not a single core duty...
Not many game engine can populate all the cores of the 3900X, but most modern games can spread the workload over many cores, so single core performance aren’t relevant even for gaming.
at stock on my 3900X using Doom Eternal I’m seeing 4.2/4.4 GHz on many cores. Same for AC Odyssey.

My CCX overclock is 4.35/4.35/4.25/4.25 @1.3v (1.26v after droop), so for gaming is the same as PBO, but I gain performance in multithreaded applications such as rendering, encoding, etc., at slightly more power draw, about 10 watts at 100% load, and performance is always the same regardless the ambient temp. Stock + PBO I loss performance in summer because the raised ambient temp and how the boost algorithm depends on that.

it could be just me, but I’m scared by fixed voltage on a cpu, so overclock in a Ryzen isn’t my kind of business...

In a normal system, that you use everyday with Windows bloatware and a lot of background apps, I never saw my CPU sustain that 4.6Ghz single core in apps or games. The best I got was 4.4Ghz under certain scenarios, and around 4.5Ghz in benchmarks such as CB, AIDA or CPU-z, so the theoretical single core boost advertised by AMD has zero meaning in everyday usage in my opinion. It's only a peak, instantaneous value, sustained is much lower in a normal system with a lot of apps loaded. My all core sustained clocks @ stock are 4.05 / 4.2Ghz load dependant, so in my particular system, dialing a CCX overclock as I did is a no brainer.

On my PC during everyday usage the CPU is hitting 4541 MHz on many cores, with 4616 MHz (the best I’ve ever seen) on the two best cores.

For others with much better binned CPUs the situation would be different, of course.

Yep. Every cpu is a different story. Mine is extremely temperature dependent.

 

[ThrashZone]

Well i hope this turns out better than DRAM Calc as that may work fine for B-Die but offers completly unusable numbers for RAM using Hynix chips (DJR, CJR, JJR)

Hi,
Hynix is useless memory anyway even on intel micron is hit and miss.

 

[nangu]

Well, gaming it’s not a single core duty...
Not many game engine can populate all the cores of the 3900X, but most modern games can spread the workload over many cores, so single core performance aren’t relevant even for gaming.
at stock on my 3900X using Doom Eternal I’m seeing 4.2/4.4 GHz on many cores. Same for AC Odyssey.

That's correct. Even a heavily single core game (like IL-2 series) can't sustain high core clocks because there are a lot of things going on in Windows itself, so these high clock figures are almost impossible to reach outside benchmarks.

it could be just me, but I’m scared by fixed voltage on a cpu, so overclock in a Ryzen isn’t my kind of business...

I'm on a Gigabyte board. If you use offsets for core voltage, the CPU downvolts, downclock and put cores to sleep as usual, it's not like the fixed voltages as we seen in previous generations.

On my PC during everyday usage the CPU is hitting 4541 MHz on many cores, with 4616 MHz (the best I’ve ever seen) on the two best cores.

Yes, mine too, but these are clock spikes which last nanoseconds and when there are almost no load on the core, so not performance impacting in any meaningfull way. Which it's important here is sustained (or effective) clock speeds from start to finish for a given task. That effective clock speed is much lower than that spikes. On mu CPU, it's around 4.35 Ghz at stock + PBO.

 

[Max(IT)]

still no news about the 1usmus tool ?
He wrote about a beta, on Twitter, but never released one.

 

[chevy350]

Stated in Sept. no specific date
From OP "1usmus expects to release CTR 1.0 in September 2020."

 

[Max(IT)]

Stated in Sept. no specific date
From OP "1usmus expects to release CTR 1.0 in September 2020."

Yep. I was hoping about a beta release...

 

[Max(IT)]

according to last tweets from yesterday, he is improving the tool and there will be a late September beta release

 

[biffzinker]

https://twitter.com/i/web/status/1310526652758274051

 

[phill]

Looking forward to it

 

[Taraquin]

This will be interesting!

 

[SeventhReign]

the linustechtips video on it didn't show much benefit really, they were not able to replicate the 10% gain. at end of day Ryzen overclocks itself well enough, just leave everything default in BIOS. if you want fps gains, then you want to get good bdie ram and overclock it with dram calculator made by the same guy here 1usmus

but since all of us here do enjoy tweaking from time to time, I have to admit I will probably give this a shot at some point if it ends up supporting ryzen 4800x, seems neat anyway

This is a total lie. They may not have been able to get 10% but they DID see significant improvements to both performance and voltage. If you are going to spout nonsense, try not to do it about a video with several million views.

 

[Space Lynx]

This is a total lie. They may not have been able to get 10% but they DID see significant improvements to both performance and voltage. If you are going to spout nonsense, try not to do it about a video with several million views.

sure thing. enjoy your 3% gains. and voltage depends on mobo and BIOS used in said mobo, can't just take results from two setups and say its definitively better at voltage control prediction.

 

[biffzinker]

Looking forward to it

It's here if you want to try. https://www.guru3d.com/articles-pages/clocktuner-for-ryzen-ctr-guide-download,1.html

Download Link: https://www.guru3d.com/files-details/clocktuner-for-ryzen-download.html

I downloaded everything just need to hop over to the bios and change a couple of settings.

It’s moved the clockspeed on the 3800X from 4,175 MHz all cores to 4,300 with the same voltage from the beginning at 1.25. Nvm it was sleep mode.

Energy Efficiency so far has went from 3.35 to 3.48

Looks likes moved on to testing the individual CCX1 & CCX 2. And it’s done.

 

[Space Lynx]

It's here if you want to try. https://www.guru3d.com/articles-pages/clocktuner-for-ryzen-ctr-guide-download,1.html

Download Link: https://www.guru3d.com/files-details/clocktuner-for-ryzen-download.html

I downloaded everything just need to hop over to the bios and change a couple of settings.

It’s moved the clockspeed on the 3800X from 4,175 MHz all cores to 4,300 with the same voltage from the beginning at 1.25. Nvm it was sleep mode.

Energy Efficiency so far has went from 3.35 to 3.48

Looks likes moved on to testing the individual CCX1 & CCX 2. And it’s done.
View attachment 170201

what do you mean by "nvm it was sleep mode"?

 

[biffzinker]

what do you mean by "nvm it was sleep mode"?

Windows power management kicked in, did the shut off and suspended to RAM.

Anyways I missed a step for the benchmark that shows before and after change.

The set OC with CTR in Cinebench, score is up, and power usage is down.

https://twitter.com/i/web/status/1310992125828501507

https://twitter.com/i/web/status/1311017352486584320

 

[Nordic]

I have results. It says I have a silver sample while I thought I would have had a trash tier bronze sample. that is good to know. Efficiency is improved but performance is nearly the same. I am not even reaching the default performance for a 3900x but that could be because I have my memory set to 2133mhz for maximum stability.

Spoiler: Diagnostic Results

***ClockTuner for Ryzen 1.0r by 1usmus***
```AMD Ryzen 9 3900X 12-Core Processor (870F10)
ASUSTeK COMPUTER INC. TUF GAMING X570-PLUS
BIOS ver. 2607 SMU ver. 46.63.00
DRAM speed 2133 MHz
09/29/2020 18:37:09


AVX light mode
Cycle time: 30000 ms
Reference frequency: 4050MHz
Reference voltage: 1175 mV
Voltage step: 6 mV


Manual overclocking mode enabled
Sets overclocking parameters...
CCX1  Quality 148  Frequency 4050 MHz  Voltage 1175 mV
CCX2  Quality 130  Frequency 4050 MHz  Voltage 1175 mV
CCX3  Quality 148  Frequency 4050 MHz  Voltage 1175 mV
CCX4  Quality 130  Frequency 4050 MHz  Voltage 1175 mV

Step# 1
Diagnostic VID voltage: 1175 mV
Stress test started!  09/29/2020 18:37:19
CPU Vdroop: 1 %  CPU Temperature: 44.9°
Stress test stopped!  09/29/2020 18:37:53

Step# 2
Diagnostic VID voltage: 1169 mV
Stress test started!  09/29/2020 18:37:54
CPU Vdroop: 1.6 %  CPU Temperature: 46.3°
Stress test stopped!  09/29/2020 18:38:28

Step# 3
Diagnostic VID voltage: 1163 mV
Stress test started!  09/29/2020 18:38:28
CPU Vdroop: 1.6 %  CPU Temperature: 46.6°
Stress test stopped!  09/29/2020 18:39:02

Step# 4
Diagnostic VID voltage: 1157 mV
Stress test started!  09/29/2020 18:39:02
CPU Vdroop: 1.6 %  CPU Temperature: 47.4°
Stress test stopped!  09/29/2020 18:39:36

Step# 5
Diagnostic VID voltage: 1151 mV
Stress test started!  09/29/2020 18:39:37
CPU Vdroop: 1.7 %  CPU Temperature: 47.2°
SOC SVI2 1.013
Thread# 25 fall down, usage 87.9%
Stress test stopped!  09/29/2020 18:40:04

Step# 6
Diagnostic VID voltage: 1157 mV


Diagnostic results:
Energy efficient: 3.5
AMD Ryzen 9 3900X 12-Core Processor - Silver sample
Recomended values for Overclocking:
Reference frequency: 4250 MHz
Reference voltage: 1250 mV
Recomended values for Undervoolt:
Reference frequency: 4050 MHz
Reference voltage: 1150 mV```

 

[Mussels]

Running this i just keep seeing 'error on CCX1' and ever decreasing clocks down to 2GHz at 1.25v?
Have i likely got something screwing with it, or am i missing how its meant to work?

 

[Calmmo]

Looks neat, gave it a try just to see what it is, how it works. Just confirmed what i knew, that my ccd1 is pretty solid, ccd2 trash