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

HYDRA 1.1H2 PRO - RELEASE (Patreon)

Published:
2022-06-13 16:47:12
Edited:
2022-06-13 16:47:33
Imported:

Downloads

Content

I have prepared another quick update for you!
The main focus was on diagnostic stability and solving infinite loop problems. Also at your request, you can search for the ideal voltages with the target EDC and TDC.

INFO ABOUT PREVIOUS VERSION
VIRUSTOTAL REPORT


HYDRA 1.1H2 PRO (1109) - what's new?

* The ideal volt search can now be configured by additional parameters such as EDC and TDC (4 parameters in total).
* A number of stability improvements for diagnostics and "Phoenix".
* If there are no CPPC tags, HYDRA will not start and thus will not damage the configuration files.
* User can safely remove "inpoutx64.dll" if some anti-cheat scolds HYDRA.
* Some improvement for "CUSTOM" profiles.
* Added hints and made corrections to some descriptions.

Let's move on :)

Files

Comments

Anonymous

My objective is to get the better CO values to them set it on my bios.. not interested for now on hybrid OC or other features :) (but i would lot to see my ram (micron 3600 2x 32gb) supported on DRAM calculator)

Anonymous
>>85784342

CO values I think will be luck of the draw I really don't know but the objective is to decrease the voltage used for PBO2 while getting gains without the PBO2 sucking up1.4+ volts all the time, Personally I run my PBO2 with PPT 230W EDC 150A TDC 200A use this as a base then just let Hydra do the testing https://www.reddit.com/r/overclocking/comments/qikcbd/zen_3_pbo_curve_optimiser_tweaking_guide/

Anonymous

I am wondering about this too @marcexeu. I´ve set my own PPT / EDC / TDC values but each time the Diagnostic starts, it switches to the standard values (and also says that in the log). Why is that and how we can we avoid it?

Anonymous

I have no idea, hope @Yuri will be able to help us :)

Anonymous

Additionally I´ve noticed that the CO offsets come out different with every run, with sometimes huge variations. I´ve already increased the test times to 120 seconds (from 30) and added FMA3 tests as well, but still, there are rather big CO offset variations between each run for the same cores - ending up in an unstable system. How to avoid this? As it´s pretty useless this way to get any stable or meaningful CO offset results if they vary *that* much between each run even with the heavily increased testing times.

1usmus
>>85784342

The calculator is not required to find the ideal values for you, it is a compass tool, nothing more. All values are chosen manually by the user.

1usmus
>>85814971

Diagnostics must only be performed on standard EDC/TDC/PPT/THM values. HYDRA corrects all values by itself if the user forgets or purposely tries to corrupt the diagnostic results.

1usmus
>>85836529

This means that you have unstable DRAM overclocking. One of the requirements for diagnostics is a stable system because Prime is very sensitive to DRAM errors. There will always be an error of +-2 units. Every time system starts up, it makes a series of boost-calibrations. You will never achieve identical values.

Anonymous

Hey, Yuri, not sure if you will see this or even if you will feel the same but in some release can we get something done about the diagnostic power limits and safety limits I will explain. A lot of users including myself always hit a power limit or thermal limit during MT diagnostic stability testing I say for me it would 98% of the time the test will come to a halt and the diagnostic will stop when in reality the CPU does not behave like this it would actually throttle, so to amend this headache my suggestion would be to switch up how the diagnostic is handling the power and thermal limits to instead continue and alert the user that the CPU is throttling, retain the read-out as to which parameter flag it tripped on but simply continue anyway under a throttled state if that is at all possible. People are getting this Diag stopping at both low and high power limits, now I have mentioned before how tjmax for the 5950X and 5900X is 90c I did see this got changed from 95c which is good because the CPU will throttle anyway beyond that 90c limit. Regarding Power limits I run pretty much as high as you would want to go for a 5900X EDC: 200A TDC 150A-170A PPT of 200W-240W Beyond this the value is just more heat without many returns, ofc you could just set this too unreasonable like 500W as it will never actually get that power draw but this is only valid for users with good AIO or custom loop water cooling. I always get diagnostic STOPPED because some value was reached for me it is usually the PPT or Thermal limit I find if I set PPT to 260W then the next to fail is thermal I hit 91.5c almost instantly on any MT Prime Load and you would know to expect that. I also am aware I could very much tell the test to use 95c but I don't know if that is a good idea as the that is beyond tjmax according to AMD spec sheet for those CPU's, then I have to battle with restarting a long process just to increase it by 5c and get around this temporary unrealistic load to keep the safety limits happy even tho it probably is not hurting the CPU as it is going to throttle anyway. I use an EK 360 AIO so plenty of headroom for high power limits and the board is a MSI Tomohawk Max X570 which for its price point has stupidly overkill VRMs not even a 5950X OC is capable of pushing the VRM too far. I know it will throttle and it would just a hack of sorts to get around the diagnostic fail but a better solution would be to simply let the CPU throttle back and print that out, a real dirty solution would be to print what it tripped on then PAUSE diagnostic testing allow the user to adjust the safety or pwr limits on the fly then resume, as the load would be temporary we could continue and finish the testing this way without starting over.

Anonymous

I think I found a bug... Setting thermal limits higher than 90c under PBO2 Tab results in THM/LIMIT showing xx/9 Effectively this creates a situation where my 5900X has THM limit of 9c and I verified this in Task Manager the CPU freq was under 1ghz. Setting THM Limit back to 90c resets the THM/LIMIT back to xx/90 and all is well.

Anonymous
>>85836529

Thanks Yuri, but I doubt that the RAM is the issue. It´s been tested for a whole week before using Hydra (with standard PBO2 settings) with Karhu RAM Test, TestMem5 and AIDA64, and Prime95 without a single error in these 7 days. I am getting a variation of up to +-3 units, so that might be normal as you say - didn´t know that. Another question: 12 of my cores 16 on my 5950X are hitting -30 without issues, also on repeated tests. I guess they could go even lower than -30 as there is most likely even more headroom on these cores. As we can´t go lower than -30 (can we??), would using a global negative voltage offset in the BIOS help with this? So that the cores get less voltage in general so that the CO offsets will be lower (e.g. -20 instead of -30) too which should resul in overall less voltage for these cores to max them out further? Thanks.

Anonymous
>>85869115

Same problem exists here. I´d appreciate if this could be fixed and let us allow to enter at least up to 94°C. Thanks.

Anonymous
>>85864182

It came to my attention that in 1.1H2 and likely 1.1H that the power limits are on a standard for diagnostic which for 5900X and higher is never going to pass the MT testing phase as even with AIO these CPU's will get to 90c and they already would use more than 95A TDC in any OC situation.

Anonymous

After several runs I can now confirm that Hydra is finding almost exactly the same CO values over and over again - impressive. Very good software, great speed improvements for my 5950X. Thanks again for creating it Yuri!

Anonymous
>>85938877

Btw Yuri, is there any way to go lower than -30 for the CO values? Because I have about 8 of my 16 cores that are at -30 and I am sure they could work at even lower values. But -30 seems to be the limit?

Anonymous

For the Hybrid CO - it starts at -55, and takes bloody ages.... Current CO values CORE | CO C01 161 C02 -55 C03 -55 C04 -55 C05 -55 C06 -55 C07 -55 C08 -55 C09 -55 C10 -55 C11 -55 C12 -55 C13 -55 C14 -55 C15 -55 C16 -55

Anonymous

2hrs 45mins to check 1 core - that can't be the design?

Anonymous

Do I just remove "inpoutx64.dll" directly from the profile if Hydra gets scolded by Anti-cheats? I removed from the profile but still get detected by Anti-Cheats

Anonymous
>>85938877

I don't think so as these are the same values you would set in using a negative offset in the BIOS and the bios at least on my X570 board will only allow up to -30 I cannot set the value any lower, I think this is a ZEN3 limit of -30.

1usmus
>>85836529

It is HYBRID OC that exists for cases where cores can do more than -30. There are no limits in HYBRID OC. Double diagnostic will find both AMD CO and optional CO for HYBRID OC. Negative volt offset will not change the situation, because the curve is calculated with respect to the working volt.

1usmus
>>85869115

Bug confirmed. Thanks! I have some concerns about the maximum allowable 95 degrees. But if you insist, the limit will be changed.

1usmus
>>86000902

If you have completed a successful AMD CO search, then -55 is zero for AMD CO -30. Each processor has its own "zero value". For some processors it can be -80 and this is normal. In your case apparently AMD CO diagnostics were not required. I have made some changes in a future patch that will optimize the time a bit for similar cases (- 30-50% ).

1usmus
>>86030340

You can remove this dll from the program folder. It is safe to do so.

1usmus
>>86000953

It's a curious situation. I will try to find a way to replicate it.

1usmus
>>85938877

It is HYBRID OC that exists for cases where cores can do more than -30. There are no limits in HYBRID OC. Double diagnostic will find both AMD CO and optional CO for HYBRID OC.

1usmus
>>85864182

You are confusing the protective values of HYDRA and AMD PBO2. HYDRA protection values are in the SETTINGS section, and they are what can stop your diagnostics. AMD PBO2 except for AMD CO search does not take any part, no matter what settings you set, it will not affect the result in any way. Once again. In order not to stop your diagnostics, tell HYDRA in SETTINGS that you want to use your specific settings, whatever you want.

1usmus
>>86000953

UPD: I have made some changes in a future patch that will optimize the time a bit for similar cases (- 30-50% )

Anonymous
>>85864182

Gotcha, I see what is going on now thanks. The settings can be set in the Profile Creation during Diagnostic to prevent Hydra from stopping during testing. I go into Settings and if I must set an unusually higher than the wanted limit for safety just so the testing can finish smoothly then set it back to a safer daily limit after. thanks for clearing that up. I think by default now the safety limits were raised as of 1.1H3

Anonymous
>>85869115

Oh me too, the AMD spec sheet says tjmax for a 5900X is 90c but if you use 90c safety limit the test always stops on the MC tests on processors with 10+ cores especially if you are going for a moderate OC. I said what I did based on many diagnostic attempts with a 5900X, I raised the limit to 95c in the SETTINGS > Safety Limits temporarily just so it can get through the CO compare and MC diagnostics then afterwards set it back to 90c, Which should be ok because the AMD hard limits in the CPU will force it to throttle back above tjmax the higher limit, in this case, would be temporary I still recommend users keep an eye on the testing in this situation I found with a 360mm EK AIO my CPU would hit 92c max during the MT and Compare tests, this is not daily driving at those temps it's just that prime95 is a heavy load on all the cores and the diagnostics are really heating the CPU up, for a short burst it should be fairly ok. For other CPUS the 5600 and some of the lower SKU's the tjmax is actually 95c https://www.amd.com/en/products/cpu/amd-ryzen-5-5600x specs *Max. Operating Temperature (Tjmax) 95°C* That was another reason I said may as well allow us to use 95c as not all of the SKU's are limited to 90c before throttling due to Tjmax temp reached.