Home
Artists
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register
Partychan Telegram ThePornDude
Home Artists Posts Import Register

CTR 2.1 RC6 (ver.18) - experimental build (GAME MODE) (Patreon)

Published:
2021-06-02 17:24:24
Edited:
2021-06-02 17:28:33
Imported:
2021-08

Downloads

Content

I suggest checking out the new mode for gamers. There is no profile switching, but there is a dynamic frequency change.  The profile is turned on automatically if the video core load is more than 75%. In this mode there is a frequency hold, i.e. the processor will fight with lags and frequecy droop. The voltage profile is lower by 100mV than AMD, it allows to make the system cooler and also get extra performance. To make the profile work, it is enough to do the diagnostics without the OB test. If you want more, do an OB test and switch the game mode to Custom. In this case it will use OB. For comparison, I have published the result of 3D Mark Time Spy (click on the picture). Higher frequency. Stabilized frequency and cooler processor.

Unfortunately I can not guarantee stability 100% of the time, so if you are having stability problems my suggestion is to lower the OB by 10 for all profiles. I also welcome discussion in the discord channel (to join the community, fill out your Patreon account and you will automatically be connected to Discord).
_______

CTR requirements:

  • Stable DRAM.
  • Disabled Curve Optimizer (in BIOS).
  • Manual CPU LLC (Load Line Calibration). ASUS - 3, MSI - 4, Asrock - 2, Gigabyte - High.
  • CPU Voltage - Auto (in BIOS).
  • Windows 10 build 2004 or newer.

_______

How to make a diagnosis & profile creation (VIDEO)
How to activate profiles + quick performance test for PX and P1 profiles  (VIDEO) 

The program is in a 7-ZIP archive.  Download the archiver
.NET Framework 4.8.  Download
_______

What's new?

* New diagnostic. Better accuracy, bug fixes.
* Ability to test OB for PX HIGH / MID / LOW (automatic blind search).
* Game mode (available after diagnosis, the OB test for it is optional).
* EXTRA BOOST is included by default (the rules are more lenient and apply only to P1 and P2 profiles).
* Max temperature limit - 95 degree.
* Other minor improvements and optimizations.
_______

Thank you for your support!
_______

Special thanks to @Riwwelorsch, @Ann Martiné, @ACE, @arcanazen, @DayDreamer, @Gadfly, @rizlah , @&&Devin, @Bloax , @RdWing and @wanderinarcheologist  for help with development and testing RC6.

Files

Comments

Anonymous

Can I update to this new version since my previous version ? Or I need to delete old version and download this one ?

1usmus
>>61095258

This build does not expand with auto-updates. You need to install it manually. The previous build must be deleted.

Anonymous

sorry but what does OB mean?

Anonymous
>>61104843

His guide videos explain what you need to know, such as OB, etc. Worth reading all the documentation on Guru3D before doing anything/using CTR: https://www.guru3d.com/files-details/clocktuner-for-ryzen-download.html

Anonymous

Honestly Yuri, I think you are rushing and adding too many features in a very short timeframe and I think you need search the stability for every user first before rush more new features. Well in my case im still in RC6 Ver.13 and I never got a default stable PX diagnostic a single time, both AVX and MIX always black screen crash with default diagnostic values and I need lower the PX high by around 150-200 with my 5950x and x570-E (from 5000 to 4800 in MIX and 4950 to 4800 in AVX). Also im getting much lower default values in P1 and P2 diagnostic in RC6 compared to RC5 (P2 lowered from 4400 to 4100 and P1 4100 to 3900). And an offtopic suggestion, I think we could get a better advantage if you update your DRAM calculator and add it to CTR.

Anonymous

Can this game mode be used also with Zen 2? Or it is just for Zen 3?

1usmus
>>61106743

That's not exactly true. The current update is designed to reduce the load on the PX profile. This will provide additional stability. Regarding profiles. The base frequency has no effect on the final frequency. The frequency is controlled only by the OB. Even if it is 3400 MHz in the profiles - everything will work the same way. You should also understand that it is impossible to guess the behavior of all systems. Some systems will always require manual intervention by the user. This is absolutely natural for overclocking programs. Regarding your situation. Have you tried using OB 1 ?

1usmus
>>61108754

Zen 2 unfortunately does not support CO and does not support dynamic mode.

Anonymous

Incredible! Good job Yuri! https://i.imgur.com/g8g9Bhl.png

Tim H

I noticed you went back to LLC settings -- is Auto no longer the recommended setting?

Anonymous

This version is really good, first time in some time it is better than my pbo+doc optimized in games, game mode is really good. As a recommendation, maybe add that 75% gpu limit in settings, some games (e.g. ESO) on my 3090 is using around 65% GPU and game mode is not engaged.

Anonymous

What up everyone ! Thank you Yuri. Fantastic Version. This worked just as well as V. 13, except i notice that it is more aggressive on the core speeds. This may just be my own isolated experience, and it seems like user preference to have a more aggressive or V.13, which seems more laid back in my experience. BOTH are great! While gaming, V.18 holds my cores at 4900 Mhz... crazy ! I love it :) While gaming with V. 13, my cores regularly stayed around 4600-4700 MHZ... More aggressive, and I notice a little more heat with V.18, but understandable do to it holding higher core speeds. Most importantly, this could be personal user experience, because i added 50 Mhz higher frequencies to P1 and P2 after V. 18 diagnostic completed. I wanted to match the frequencies exactly to V. 13

1usmus
>>61137271

Thanks buddy, I'll make a threshold adjustment. It'll probably be 50%.

1usmus
>>61133387

In some situations Auto gives more performance, but in any case manual LLC makes the system more stable. Be careful, after changing the LLC you need a new diagnostics.

1usmus
>>61142976

The game profile is controlled by the PX LOW voltage. You will be able to lower it slightly (if necessary).

Anonymous

When using diagnostic 5800X should be on 16 threads or 8?I always use 8

Anonymous

I mean always in gaming however in CTR diagnostics I have SMT on , then disabled after new profile , testing etc….

Anonymous

Have you removed the tune functionality for Ryzen 5950x? The button is grey out and no linger accessable

DeeJay

Couple of observations for this build (5950X): - P1 and P2 calculated are very low (like 4k and 4.2k - to be fair, also at slightly lower voltages) vs. previous builds 4.4 and 4.6k, and these profiles are used under heavier loads, so overall a worse performance (tested both Auto and Level 3 LLC, as well as SOC override 1.15V vs Auto, on my Asus board, no real difference) - when calculating OB for Mid/Low, got an application exception and CTR exited - Results did work again in RC6v13, but no longer work in v18 (I think results don't come up when there's a WHEA/Reboot and CTR could not save the settings/results in time before it happens) - Gaming mode generally a great idea (my system under gaming could definitely use less heat thru undervolting), just don't yet see how with the other issues it will help

Anonymous
>>61300030

I have same issues with my 5950x, lower P1-P2 base frequencies compared to previous builds, using LLC3

Anonymous

Newest updates seems to working nicely, getting a even flat'er CPU mhz line in 3dmark timespy: https://imgur.com/a/aJmEhTf Pretty much the only feedback i have is that the auto "diagnostic" seems to be very conservative.. It give my CCD0 something like 40 OB and my CCD1 from -5 to 3 OB.. Have to adjust these numbers way up.. Numbers in screenshot above i can run i cinebench without problems on the different thread-counts. (4-8-24 for "PX-mid", "PX-low" and "P2" The are also some programs that really dont match with CTR it seems.. I can never run Aida FP64-ray trace for example, even with default (slow) settings diagnostic gives me +LLC4. Geekbench 5 is one other of these programs that dont work well.. (can run GB3 and GB4 depending on settings) If i were to guesstimate the reason, CTR cant switch the profiles fast enough / the app is blocking CTR from working (?) --> If i enabled static P1 profile here is no problem running aida 64 ray trace benchmark. Its the programs that seems to change the load in very short/quick bursts that seem to be the problem. (blocking mutex and/or dont use cppc preferred?)

1usmus
>>61366857

For some users the values are conservative and for some users the values are critically unstable. At this time I do not know the reason why in some cases SMU calculates and offers unstable values. I am preparing a tutorial to correct this problem. 4 years ago AMD introduced the first Ryzen processors. Until then software developer did not use CPPC as Intel processors always have the same boost for all cores. The task was always manually attached to certain cores. This is not optimal for current AMD processors. In a future project I will have to make an emulation of Intel so that all the old test programs show good results. By the way - the current game profile is an emulation of the mode without CPPC.

Anonymous

Hello, first a big thank you to Yuri :) I currently use a 5900x and I have to say what CTR gets out of it is awesome. I reach the 5 GHz mark. Unfortunately I have the problem that the whole thing is not quite stable. I can run about 20x Time Spy and everything is fine but on the 21 run I get WEHA errors. I suspect as well as domdtxdissar that is due to the profile switch? Thanks for your hard work.

OpenGrape

This message appears during diagnosis. What should I do? CMD_REJECTED_PREREQUISITE 93