Asus Strix GTX970 4GB Massive problem, need help. [SOLVED]

[xvi]

Disabling cores would be the easy way to go (and probably best way to troubleshoot quickly).

For undervolting, it's been a while since I've played with any of this, but I'd nab AMD Overdrive and move the CPU voltage offset down until your PC crashes from instability. Set it slightly higher than where it crashed and you should be good.

Okay, I saw that feature. But if it is the temps going up and down like crazy which is causing the CPU to throttle...Should I try by changing the cooler again to the old one (later today) and see what happens?

I'd say no. I don't think it's your CPU's temperatures, I think it's something supporting the CPU (like the VRMs). Undervolting your CPU will reduce the load on your VRMs, cooling them down.

Really, I'm just guessing it's the VRMs since a 8-core AMD proc on that board seems like it's pushing it.

Found a thread on Tom's Hardware Forums of someone with the same issue.

Edit 2: Maybe see if you can place a fan over the CPU socket? I've got a watercooler on both my 8350 and my 3930k and VRM temps are nuts because there's not as much airflow around there.

 

[Foxiol]

Disabling cores would be the easy way to go (and probably best way to troubleshoot quickly).

For undervolting, it's been a while since I've played with any of this, but I'd nab AMD Overdrive and move the CPU voltage offset down until your PC crashes from instability. Set it slightly higher than where it crashed and you should be good.


I'd say no. I don't think it's your CPU's temperatures, I think it's something supporting the CPU (like the VRMs). Undervolting your CPU will reduce the load on your VRMs, cooling them down.

Really, I'm just guessing it's the VRMs since a 8-core AMD proc on that board seems like it's pushing it.

Found a thread on Tom's Hardware Forums of someone with the same issue.

Edit 2: Maybe see if you can place a fan over the CPU socket? I've got a watercooler on both my 8350 and my 3930k and VRM temps are nuts because there's not as much airflow around there.

Aha...I knew this was a not so good motherboard from the beginning. It gave me this problem once I put a new and powerful GPU in it...so it is now using the CPU "as expected" in this case. I had a Gigabyte GTX560Ti OC 1GB with it before and never had problems with it but of course the CPU was more than good enough for the GPU.

I was thinking in getting a new motherboard like the Asus Sabertooth R2.0 but I need more conclusions and you guys sure are helping with it.


PS: Going to reboot and undervolt my CPU a bit.

Edit:

PS2: Undervolted it to 1.250V and it runs at least. The lower voltage is 1.175V. Going to do the tests.

 

[Foxiol]

Guys...first of all hats off to you all who wanted or tried to help me with my issue...I'm literally crying from happiness right now...a 30 year old man. If you wanted to make a good thing for this year you can put a note on me because you did!!

Undervolted my CPU to 1.250V as suggested by @okidna and see this...take a look at this!!!!

Spoiler

If I could hug you (no homo at all ) @okidna I will do it and very hard my man!!

The game ran perfect without a single drop, it had the regular weird drops in frames in replays (while playing 180/200FPS and in replays it drops to 120FPS...with VSync while playing 60FPS and in replays 30FPS...console port of course what else) but the game ran smooth as butter!!

I literally can't believe that it was just this thing. So now I know that this motherboard is not so good with temperatures...and to think that I wanted to do an Overclock to my CPU...I can imagine the fire already. haha

So guys...should I keep voltages there and it will be okay...or do I really need to buy a new motherboard or save and wait for a new gen (maybe upcoming Zen and AM4 motherboards which are coming next year...)...What do you think?


And again...I wish you all that helped me here in this thread and that guy on private (he also gave me good advices), all the best for this upcoming 2016, you literally made not just my night...my whole year...I spend 600€ in components to make my PC a little bit more up to date and really...times are hard enough to have to spend so much money in things that one maybe don't really need at all. (I'm talking about that new PSU that now is going to get a new use for sure but still...it was like throwing 150€ out of the window for just a PSU)

I won't close this thread now since I need to do more tests but I'm going to put it as "SOLVED" since you all can see that was the heat on my VRM causing the CPU to throttle. Also hope this help other people if they have the same issue as me.


Thank you all and happy new year! @okidna @xvi @newtekie1 @DarthBaggins @flmatter @mstenholm and the guy in private that maybe don't want to be called in here who knows.

You made me a really happy guy...best ending of a year possible!! Let's go and game again!!!

--------------------------------------------------------------------------------------------------------


Quick Edit: My problem is not resolved at 100%. Did a Prime95 test and because of the "low voltage" 3 cores stopped working and after a while my PC crashed/froze literally.

So now I'll have to mess with voltages to not heat the VRM and at the same time keep all my cores fed.

I'm trully thinking in getting that Asus Sabertooth and forget about motherboard problems since now I know where my problem is.

I'll do some tests and update this post later or tomorrow.

My Prime Test with cores failures before the crash at 1.2500V:

Spoiler

 

[Kanan]

Glad it got solved man. But ignore Prime, it's not a game, you need game stable/program stable, Prime isn't important, really. Also, I would keep that system and only upgrade with new CPU etc.

 

[okidna]

Glad you got it sorted.
If you really insist on Prime stability try to increase the voltage bit by bit, @xvi already gave a great suggestion by changing the voltage via AMD Overdrive and test it and see if it's stable or not.
Judging by your CPU-Z screenshot your stock voltage is around 1.35v, so I suggest you try to set it between 1.275v-1.3v.

 

[Foxiol]

Glad you got it sorted.
If you really insist on Prime stability try to increase the voltage bit by bit, @xvi already gave a great suggestion by changing the voltage via AMD Overdrive and test it and see if it's stable or not.
Judging by your CPU-Z screenshot your stock voltage is around 1.35v, so I suggest you try to set it between 1.275v-1.3v.

Yeah I put it at 1.275V but I see all my cores working and up to some point my PC crashes/freeze. (in Prime95) I see my VRM temp rise a bit (maybe because I did tests for the whole day) over 60ºC, if it gets lower than 60's the games and all run great.

So I'll probably leave it there or upgrade the motherboard. Need to download a lot of games (The Witcher 3...it will take like 10 hours...xD) and see what happens...hope none of them stress the CPU hard enough to make my PC crash.

At least I know that is not my CPU, my GPU or my RAM or whatever...fiushhhh!!


Again guys, thank you all!

 

[xvi]

Awesome! Like I say, you might be able to increase the airflow over the VRMs to keep them cool as well (or some combination of undervolting and airflow). At least we know what's going on.

But ignore Prime

+1. Programs that are built to stress components can push them further than you'll typically see real-world. I'd keep it in the back of your mind, but I wouldn't worry about it too much. I would bump voltage up a little. Prime95 will generate a LOT of heat, so real-world should have less impact.

It gave me this problem once I put a new and powerful GPU in it

Sounds about right. Old GPU, the CPU is probably less active since it's waiting for the GPU more often. The higher frame rates from the new GPU probably lets the CPU really stretch its legs which in turn put more load on the VRMs.

Really love all the testing and documentation you put in your first post too, by the way. It's really helpful seeing everything you've done up to the point when you posted.

Edit: If you have the budget for it, a new motherboard would probably be the best solution.

 

[Foxiol]

Awesome! Like I say, you might be able to increase the airflow over the VRMs to keep them cool as well (or some combination of undervolting and airflow). At least we know what's going on.

+1. Programs that are built to stress components can push them further than you'll typically see real-world. I'd keep it in the back of your mind, but I wouldn't worry about it too much. I would bump voltage up a little. Prime95 will generate a LOT of heat, so real-world should have less impact.

Sounds about right. Old GPU, the CPU is probably less active since it's waiting for the GPU more often. The higher frame rates from the new GPU probably lets the CPU really stretch its legs which in turn put more load on the VRMs.

Really love all the testing and documentation you put in your first post too, by the way. It's really helpful seeing everything you've done up to the point when you posted.

Edit: If you have the budget for it, a new motherboard would probably be the best solution.

Woah man...what a ride! haha

Yeah I like to put all the details so good people like you can give a proper advice or see what is going on with my problem. More people should do this and trust me...I saw some posts in where people have no idea what are they talking about and thank god this one brought the right people. Two pages and problem solved. A miracle!!

But yeah, I should be looking at a new motherboard since I don't want to have my head thinking in having a crash while playing or stuff like that. And the better part is that I don't need to go to the store back again and ask for a new GPU. I really like this model and as far as I could tell it runs everything maxed or almost maxed with great frame rates. (not counting those slwodowns tho...haha).

Oh well time to sleep with a smile on my face.

I'll update the post after some days with more games tested and such. And hope this help others of course.

You all did a great job...thank you!!! Good night, and good morning to others.

 

[xorbe]

Control Panel -> Power -> set min CPU state to 100% -> report back

Sounds to me that the CPU tries to power down a bit, then goes well crap, and speeds back up, and repeats.

 

[azraelumaro]

Hi Foxiol! I'm really glad your trouble is solved, but have you tried to fix the value of your cpu multiplier? When this setting is on auto, cpu saves some power running at 7x sometimes. Also can disable hpc mode, cpu spread spectrum and turbo mode (note 1, this improves oc potential. Note 2, don´t know if your mobo allows it, I have a sabertooth 990fx rev 2.0, same cpu, that I love too much ). It could help without undervolting, since cpu runs smooth and stable at 4.0 GHz. It seems your problem was caused by this kind of fluctuation.
I'm new on forums like this, but have wide experience (a 10 year hobby and sometimes cash source ). Feel free to tell me if this was useful, or if undervolting was better.
You guys are awesome, I've learnt so much reading you.

 

[Foxiol]

Hi Foxiol! I'm really glad your trouble is solved, but have you tried to fix the value of your cpu multiplier? When this setting is on auto, cpu saves some power running at 7x sometimes. Also can disable hpc mode, cpu spread spectrum and turbo mode (note 1, this improves oc potential. Note 2, don´t know if your mobo allows it, I have a sabertooth 990fx rev 2.0, same cpu, that I love too much ). It could help without undervolting, since cpu runs smooth and stable at 4.0 GHz. It seems your problem was caused by this kind of fluctuation.
I'm new on forums like this, but have wide experience (a 10 year hobby and sometimes cash source ). Feel free to tell me if this was useful, or if undervolting was better.
You guys are awesome, I've learnt so much reading you.

Hi there, well I "undervolted" it and it is working just fine since then and at 4.0Ghz in manual mode of course. I first set the voltaje to 1.275V and it worked fine but in demanding games and such sometimes my PC crashed (I guess the power was not enough for the CPU to work and had a freeze, quite obvious), but then I raised it a bit up to 1.285 or 1.295 (can't remember exactly which value) and since then I had no more problems whatsoever.

Compared some benchmarks with the performance I'm getting with my FX8350/Asus Strix GTX970 combo and is working as expected and sometimes even better than in benchmarks with even "better" CPU like Core i7's and the likes. So you bet I'm quite happy with this.

Also I wanted to change this motherboard but after thinking about it and having in mind the upcoming AM4 motherboards...I better wait for them to come and then make the right upgrade. After all, this is working as of now, let's see what happens during the summer since temperatures are going to be a little bit higher and hope to not have this issue again. Probably not because is more of an internal issue with power energy consumption more than the heat itself over the VRM's. Let's cross fingers...if that happens, upgrade the motherboard and problem again solved.

Thank you for the recommendations and I'm glad that this can help some more people.

Have a good day.

 

[xorbe]

but then I raised it a bit up to 1.285 or 1.295

This is the approx range I settled on also, before temp and watts start to get out of hand with a typical CM Hyper 212 hsf setup. It won't service p95/ibt (99c / drops to 3.8 ghz) but I don't use p95/ibt either.

 

[azraelumaro]

Great!!! This because I'm planning to upgrade my old 750 Ti to an Strix 970 and googled some opinions of 8350/strix 970 owners. Strix has always took my attention due to its power efficiency (single 8 pin ~210 watts under full load vs. 224 or 233 from msi, gigabyte and evga). Seem there are no big problems with that combo.
And same situation, waiting for AM4 and Polaris, but want to run one of the acclaimed gtx 970 before upgrade.

 

[Vellinious]

You got 8 wire power supply. Look at any of the other GTX 970s - they have 6+8. I got 3 MSI and one ASUS. Same problem when I fold. Live with it or get a MSI.

This situation has absolutely zero to do with power delivery, power limits, or power connectors. The steep drop in clocks are directly correlated with the perf cap "util".

/smh

 

[Foxiol]

@xorbe Yeah, I checked back again and it is set to 1.295V. And to think that its default voltaje is 1.360! I bet once I upgrade my motherboard I could let it stock (the voltaje) and do some OC up to 4.5Ghz with the same voltaje. Yet I'm not in the need of doing that and with the stock boost to 4.2Ghz is enough.

@azraelumaro This GPU is also really cold (have in mind this is winter...well, it was) and my case is has a very good airflow. After playing The Witcher 3 or even Arma 3 for hours the GPU max temp was 68ºC. Lower temps than my old and retired Gigabyte GTX560Ti OC which went up to 73/76ºC.
It is a good looking card and well crafted in my opinion.

My advice is to wait for Polaris and/or Pascal if you want to upgrade. Those GPU's will be better than current ones of course.

Well...gotta game some. Bye.