• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Ryzen clock tuner 2.0

Soldato
OP
Joined
29 May 2005
Posts
4,899
This new version is really conservative.

for the same voltage it has lowered my OC by 200-250MHz.

I used to run 4.35/4.375/4.25/4.25 on 1.25V. Now it says 4.1/4.2/4.0/4.05 on the same CCX and my system was completely stable on the original OC.

this is with AGESA 1.0.0.8 with 3900X.
 
Last edited:
Soldato
Joined
12 Sep 2003
Posts
10,072
Location
Newcastle, UK
Having some better luck actually with reducing PPT, TDC and EDC in the BIOS. I was hitting over 85º in Cinebench and that was stopping CTR. At "stock" for the 5800X these are:
  • PPT: 142
  • TDC: 95
  • EDC: 140
After some reading online at what others were trying, I reduced these to:
  • PPT: 125
  • TDC: 90
  • EDC: 120
This knocked my Cinebench temps down and I only went down from a score of 5960 with "stock" PPT, TDC and EDC to 5921.

...
Phoenix ready!
Cinebench R20 started
Cinebench R20 finished with result: 5921
Voltage: 1.32 V PPT: 125 W Temperature: 71.3°
...

This helped me achieve a final P1 Profile of 4.450GHz @1.256v and a P2 Profile of 4.525Ghz @1.331v. Temps during the stress testing were between 55ºC and 63ºC.
 
Associate
Joined
7 Dec 2002
Posts
1,833
Location
Frauenfeld, Switzerland
Having some better luck actually with reducing PPT, TDC and EDC in the BIOS. I was hitting over 85º in Cinebench and that was stopping CTR. At "stock" for the 5800X these are:
  • PPT: 142
  • TDC: 95
  • EDC: 140
After some reading online at what others were trying, I reduced these to:
  • PPT: 125
  • TDC: 90
  • EDC: 120
This knocked my Cinebench temps down and I only went down from a score of 5960 with "stock" PPT, TDC and EDC to 5921.



This helped me achieve a final P1 Profile of 4.450GHz @1.256v and a P2 Profile of 4.525Ghz @1.331v. Temps during the stress testing were between 55ºC and 63ºC.
My default values for the 5800X were even higher: 175 PPT, 190 EDC and 115 TDC. I really find CTR confusing: yesterday my chip was rated silver sample and could hit 6013 on Cinebench R20, now it is bronze and only manages 5800 tuned and 5600 stock. With no BIOS change in between
 
Last edited:
Soldato
OP
Joined
29 May 2005
Posts
4,899
Just finished my tests on my 3900x

AGESA 1.0.0.8 has higher OC by 50Mhz after tuning compared with AGESA 1.2.0.0

however my manual CCX tuning yielded better result than the CTR. Took ages but got 4.3/4.3/4.25/4.25 on 1.25V where CTR was saying 4.1/4.3/4.0/3.9 on 1.25V

CB R20 with auto PBO - 7100
CB R20 with CTR tuning - 7240
CB R20 with manual CCX tuning - 7440
 
Associate
Joined
7 Dec 2002
Posts
1,833
Location
Frauenfeld, Switzerland
Also it seems from the guru 3d guide written by 1usmus that the PPT, EDC and TDC values are rather determined by the motherboard not just the processor.

I finally got some good, stable diagnostic and tuning runs using PPT 150, EDC 150 and TDC 100 on my X570 TOMAHAWK. I will post them when I get back home. Basically boosted from ~5680 stock to 6045 whilst reducing temperatures, voltage and power. I am pretty sure my original stock 6013 on the old BIOS was either unstable or the 1 2.0.0 BIOS is just much more conservative as stated by several others. Either way it ran at much higher voltages than I am now getting.
 
Associate
Joined
7 Dec 2002
Posts
1,833
Location
Frauenfeld, Switzerland
So my "stable" CTR settings of P1 4500mhz at 1.256V and P2 4575mhz at 1.331v crashed on 3D Mark Timespy. I dialed down the values to 4475mhz and 4550mhz at the same voltages and it passed again but with almost the same score as with just PBO enabled at the default values (which according to Ryzen Master is the same as Default: PPT 142, TDC 95, EDC 140). From the CPU frequency graph it runs in P0 mode for most of the test then settles at the P2 value before dropping to the P1 value for the CPU test. I tested these values in the Shadow of the Tomb Raider Benchmark on Ultra and in Heaven and found no benefit to 1% lows or average fps over disabling CTR. Whilst I can see the benefits in Cinebench in terms of voltage and temperatures, it does not seem to have a noticeable impact in games. For now I am disabling CTR and running at stock PBO values (upping the PBO PPT in Ryzen master did nothing either, probably from thermal throttling).

I will have another go once CTR v2.1 is released to adjust the P0 values to run lower voltages all the time but for now I feel like I wasted 2 days of tuning
 
Last edited:
Associate
Joined
7 Apr 2017
Posts
1,762
From all the results it seems better to just use curve optimiser and bump the clocks a minimum of 150mhz if you can keep it cool enough?
Pretty sure most people's were seeing around 4.6ghz in all core workloads and still hitting 5ghz single core?

Still no mention of any real world application of CTR outside of benchmarks yet? I was impressed with it on my 3900x, until I tried to play games or covert blurays
 
Associate
Joined
10 Aug 2009
Posts
695
Location
Malvern
I have this working just fine on my 3700x, just not sure it is worth it, with diagnose my cb20 score was over 5100, now its down 300 points ,single threaded performance is lower too, when I use hybrid mode, no wonder it s more efficient and keeps temps lower (about 4c for me) I'm also getting lower performance!
Going back to Pbo and auto ac, maybe someone can convince me otherwise or later versions improve performamce
 
Soldato
OP
Joined
29 May 2005
Posts
4,899
I have this working just fine on my 3700x, just not sure it is worth it, with diagnose my cb20 score was over 5100, now its down 300 points ,single threaded performance is lower too, when I use hybrid mode, no wonder it s more efficient and keeps temps lower (about 4c for me) I'm also getting lower performance!
Going back to Pbo and auto ac, maybe someone can convince me otherwise or later versions improve performamce
Under PBO single core of my 3900x will boost to 4505MHz but on sustained all core loads it wouldn’t go past 4.1GHz

with the tuning I am doing way more than that on all core.
 
Soldato
Joined
21 Jan 2016
Posts
2,915
From all the results it seems better to just use curve optimiser and bump the clocks a minimum of 150mhz if you can keep it cool enough?
Pretty sure most people's were seeing around 4.6ghz in all core workloads and still hitting 5ghz single core?

Still no mention of any real world application of CTR outside of benchmarks yet? I was impressed with it on my 3900x, until I tried to play games or covert blurays

To me it's more interesting when it gets the automatic curve optimiser as that will potentially be a real time saver.
 
Associate
Joined
21 Jan 2010
Posts
554
Can anyone confirm if this works with 2600x chips? I saw the supported chips, but only showed 3000 series and above.
My x470 board has a beta bios for the new agesa. If this tuner proved useful I'd be tempted to flash it
 
Associate
Joined
10 Jan 2003
Posts
885
Location
staines
***ClockTuner for Ryzen 2.0 RC3 by 1usmus***
AMD Ryzen 7 5800X 8-Core Processor
ASRock X570 Taichi
BIOS ver. P4.00 SMU ver. 56.43.00
TABLE ver. 3672324
DRAM speed 3600 MHz
02/10/2021 13:18:54


Phoenix ready!



AVX Light mode
Cycle time: 360000 ms
Reference frequency: 4375 MHz
CCX delta: 25 MHz
Reference voltage: 1125 mV
Target voltage: 1125 mV


Step# 9
14:12:10: WHEA-counter: 7
14:12:10: CCX1 (127): 4475 MHz, 1125 mV OC+
14:12:11: Stress test #1 started...
14:12:17: CPU Vdroop: 2.7 % temperature: 67.3°
14:13:50: Stress test stopped.
14:13:51: Stress test #2 started...
14:13:57: CPU Vdroop: 2.6 % temperature: 66.6°
14:15:29: Stress test stopped.
14:15:31: Stress test #3 started...
14:15:36: CPU Vdroop: 2.2 % temperature: 60.8°
14:17:09: Stress test stopped.
14:17:10: Stress test #4 started...
14:17:16: CPU Vdroop: 2.2 % temperature: 58.5°
14:18:49: Stress test stopped.

Step# 10
14:18:49: WHEA-counter: 7
14:18:49: CCX1 (127): 4475 MHz, 1119 mV OC+
14:18:50: Stress test #1 started...
14:18:56: CPU Vdroop: 2.2 % temperature: 58.7°
14:20:28: Stress test stopped.
14:20:30: Stress test #2 started...
14:20:36: CPU Vdroop: 2.2 % temperature: 57.3°
14:22:07: Stress test stopped.
14:22:09: Stress test #3 started...
14:22:15: CPU Vdroop: 2.2 % temperature: 57.1°
14:23:47: Stress test stopped.
14:23:49: Stress test #4 started...
14:23:54: CPU Vdroop: 2.2 % temperature: 57.5°
14:25:27: Stress test stopped.

So is this doing well now?
 
Associate
Joined
7 Dec 2002
Posts
1,833
Location
Frauenfeld, Switzerland
***ClockTuner for Ryzen 2.0 RC3 by 1usmus***
AMD Ryzen 7 5800X 8-Core Processor
ASRock X570 Taichi
BIOS ver. P4.00 SMU ver. 56.43.00
TABLE ver. 3672324
DRAM speed 3600 MHz
02/10/2021 13:18:54


Phoenix ready!



AVX Light mode
Cycle time: 360000 ms
Reference frequency: 4375 MHz
CCX delta: 25 MHz
Reference voltage: 1125 mV
Target voltage: 1125 mV


Step# 9
14:12:10: WHEA-counter: 7
14:12:10: CCX1 (127): 4475 MHz, 1125 mV OC+
14:12:11: Stress test #1 started...
14:12:17: CPU Vdroop: 2.7 % temperature: 67.3°
14:13:50: Stress test stopped.
14:13:51: Stress test #2 started...
14:13:57: CPU Vdroop: 2.6 % temperature: 66.6°
14:15:29: Stress test stopped.
14:15:31: Stress test #3 started...
14:15:36: CPU Vdroop: 2.2 % temperature: 60.8°
14:17:09: Stress test stopped.
14:17:10: Stress test #4 started...
14:17:16: CPU Vdroop: 2.2 % temperature: 58.5°
14:18:49: Stress test stopped.

Step# 10
14:18:49: WHEA-counter: 7
14:18:49: CCX1 (127): 4475 MHz, 1119 mV OC+
14:18:50: Stress test #1 started...
14:18:56: CPU Vdroop: 2.2 % temperature: 58.7°
14:20:28: Stress test stopped.
14:20:30: Stress test #2 started...
14:20:36: CPU Vdroop: 2.2 % temperature: 57.3°
14:22:07: Stress test stopped.
14:22:09: Stress test #3 started...
14:22:15: CPU Vdroop: 2.2 % temperature: 57.1°
14:23:47: Stress test stopped.
14:23:49: Stress test #4 started...
14:23:54: CPU Vdroop: 2.2 % temperature: 57.5°
14:25:27: Stress test stopped.

So is this doing well now?
So at 1125mv you are testing the undervolting frequency and voltage suggestion it seems? It is currently checking every frequency between the initial frequency (4375mhz) and the max frequency (whatever is input or default) until it crashes, at the voltage you have chosen and at 6mv above and below. If stable it goes up by 25mhz until then tries again at 6mV below to see if still stable. Once it crashes it will use Phoenix mode to revert to a frequency 25 or 50mhz below the crash and at the voltage 6mv above then you can choose to save these values as P1 or P2.

When you originally ran the diagnostic it suggested P1, P2 and undervolt starting speeds and voltages. Normally you just do P1 then tune for optimisation, save as P1 profile followed by the same for P2. The undervolt option allows you to replace either P1 or P2 (or more commonly replace P2 with P1 and P1 with the undervolt settings). This lets you run at lower voltage and temperatures but at lower performance. You will need to compare the Cinebench scores for each profile and the Hybrid mode combined P1 and P2 profile to decide which you will keep. I recommend also running other benchmarks with CTR active in the different modes (P1 alone, P2 alone, hybrid P1 + P2 and finally replacing P1 or P2 with the undervolt values) applied to see if you get any benefit. Personally I found almost none so have disabled the program for now.

Your saved P2 profile will be the most aggressive for 1/2 to 3/4 core loads such as gaming and P1 is less aggressive for running on all cores like rendering. As stated earlier mostly in benchmarks and light CPU load games (<25% utilisation) it will run in P0 mode which is not adjustable and is basically PBO.

From the values you posted it is in the middle of adjusting the undervolt values but I found mine would often crash so badly it would corrupt thr phoenix mode data and eventually I had to input more conservative (higher voltage and lower max speed) values to start the tuning.
 
Last edited:
Associate
Joined
10 Jan 2003
Posts
885
Location
staines
So at 1125mv you are testing the undervolting frequency and voltage suggestion it seems? It is currently checking every frequency between the initial frequency (4375mhz) and the max frequency (whatever is input or default) until it crashes, at the voltage you have chosen and at 6mv above and below. If stable it goes up by 25mhz until then tries again at 6mV below to see if still stable. Once it crashes it will use Phoenix mode to revert to a frequency 25 or 50mhz below the crash and at the voltage 6mv above then you can choose to save these values as P1 or P2.

When you originally ran the diagnostic it suggested P1, P2 and undervolt starting speeds and voltages. Normally you just do P1 then tune for optimisation, save as P1 profile followed by the same for P2. The undervolt option allows you to replace either P1 or P2 (or more commonly replace P2 with P1 and P1 with the undervolt settings). This lets you run at lower voltage and temperatures but at lower performance. You will need to compare the Cinebench scores for each profile and the Hybrid mode combined P1 and P2 profile to decide which you will keep. I recommend also running other benchmarks with CTR active in the different modes (P1 alone, P2 alone, hybrid P1 + P2 and finally replacing P1 or P2 with the undervolt values) applied to see if you get any benefit. Personally I found almost none so have disabled the program for now.

Your saved P2 profile will be the most aggressive for 1/2 to 3/4 core loads such as gaming and P1 is less aggressive for running on all cores like rendering. As stated earlier mostly in benchmarks and light CPU load games (<25% utilisation) it will run in P0 mode which is not adjustable and is basically PBO.

From the values you posted it is in the middle of adjusting the undervolt values but I found mine would often crash so badly it would corrupt thr phoenix mode data and eventually I had to input more conservative (higher voltage and lower max speed) values to start the tuning.

All i did was run the tune when i booted it up.. i just ran the diagnosics now to see what it says

***ClockTuner for Ryzen 2.0 RC3 by 1usmus***
AMD Ryzen 7 5800X 8-Core Processor
ASRock X570 Taichi
BIOS ver. P4.00 SMU ver. 56.43.00
TABLE ver. 3672324
DRAM speed 3600 MHz
02/10/2021 15:16:22


DIAGNOSTIC RESULTS
AMD Ryzen 7 5800X 8-Core Processor
Max temperature: 76.4°
Energy efficient: 4.14
Your CPU is GOLDEN SAMPLE
Recomended values for overclocking (P1 profile):
Reference voltage: 1250 mV
Reference frequency: 4700 MHz
Recomended values for overclocking (P2 profile):
Reference voltage: 1325 mV
Reference frequency: 4775 MHz
Recomended values for undervolting:
Reference voltage: 1125 mV
Reference frequency: 4450 MHz

Phoenix deactivated!

so what should i imput where?
 
Associate
Joined
7 Dec 2002
Posts
1,833
Location
Frauenfeld, Switzerland
Did it finish the tuning for the 1125mv that you had running before? If so it should have given a summary after a reboot with a new cinebench run and a before and after graph on the benchmark tab. If it crashed and didn't recover then it will not have done that. If it finished ok then save those values (mhz and mv) for later from the results summary in the dialogue box.

Before starting run some real world benchmarks to see where you are starting from. Also read the CTR guide from Igor's lab and Guru 3d which are earlier in this thread.

I would click the option under Tuning for run in tray, then put the recommended P1 frequency and voltage (1250mv and 4700mhz for you) in the tune page on default settings then run Tune. Once it crashes and restarts after 45 minutes or so it should reopen the program (might need to double click on the tray icon if it doesn't start by itself after 2 minutes). Then save a screenshot of the benchmark page to see default and tuned score voltage and power draw. Now save the final values from the dialogue box of the Tuning tab into the Tuning profiles tab under P1 (fill then apply then save).

Now do the same with the P2 suggested values (1325mv and 4475mhz), tune and compare the benchmark scores. Save under P2 profile. Then enable Hybrid mode with P1 and P2 saved and run some real benchmarks to see how it runs compared to stock settings.
 
Last edited:
Back
Top Bottom