Trying to overclock 2990WX

[r1v0]

So, I will post here my month long back and fourth with MSI support, you can see just how bad their support is and maybe in the end you would be even able to help me.
As a claimer, this post will be VERY long, so I do not blame you if you don't go through it
Here we go:
ME:

Quote

Hello, so I am trying to overclock my TR2. Right now I am stuck. I turned on the AMD's Overdrive option and it works, but only just, because the processor is only pulling 250W and not a Watt more. As soon as I force the voltage up, and system starts pulling more than 250W, the whole setup locks up and crashes (have to switch off psu to shut it down). I have a lot of thermal headroom. so my question is how do I adjust the power envelope to pull more like 400W from the PSU. My PSU is 750W, just for info.

MSI:
 

Quote

1. What is the frequency of the cpu under the default settings? did you overclock the CPU in BIOS? Have you tried to open the GAME BOOST option in the BIOS? Or can you change the CPU Ratio manually for overclock?
2. About the voltage, How much voltage do you set? Did you set the voltage in bios?
3. About the power, system starts pulling more than 250W---how do you check the power?  how do I adjust the power envelope to pull more like 400W from the PSU.--We are sorry that we don't undertand your mean. In theory, it doesn't need to adjust anythings to use the cpu and psu. 

Please clear CMOS referring to the manual, then enter BIOS, press F6 to load optimized defaults, press F10 to save and reset to have a try.

https://www.msi.com/Motherboard/support/X399-GAMING-PRO-CARBON-AC#down-manual

If problem still exists, please download and flash the latest MB BIOS v1C from below link for a try,if the problem still persists,please tell us. Thanks.

ftp://jerryuser:2018gaming@ftp2.msi.com.tw/BIOS/BIOS0829/E7B09AMS.zip

ME:

Quote

1. I am trying to overclock it with the precision boost overdrive system that AMD has, I have tryed to OC it manually, crashes after 3.4Ghz or so every time, and yes, i have given it more volts, default clock for 2990wx is 3.0.
2. I tryed giving enough volts, but now it is the amd's system that is managing the volts etc.
3. Again, AMD has a nice program called Ryzen master, where you can see and change stuff and further in this letter I am explaining how I managed to change the power envelope in bios. You as a manufacturer should have known about that setting where you can adjust PPT.

ME:

Quote

Here is where I am so far, I found from Bios, that if I put Precision Boost Overdrive to manual, I can change the PPT, TDC and EDC manually. Default figures are 250w, 215A and 300A. If I cange the first two, cpu overclocks to around 3.2Ghz and basically is limited by the third figure. Third one I can only change up to like 340A and then it is stable at 3.5 or so, if I go higher, the whole thing crashes, even the power button does not work. I have to cut the psu. Temps at the same time, still below 60.
another issue that I have now is, when it overclocks to 3.5, the clock starts to fluctuate after a minute of stresstesting, it goes from 3500Mhz to 600 every second and just keeps jumping back and forth, and it is 100% not thermal throttling, temps are waaay down for that. 
So that is where I am stuck right now, how to get rid of this fluctuation and how to get the mobo to give more amps to the cpu without crashing the whole thing?

ME:

Quote

Some updates, I think I know why the computer is shutting down and the clock is jumping, it is basically throttling. But the issue is, that AIDA64 and hwmonitor and those programs are reporting cpu temps as high as 90-95 degrees while stress testing, while ryzen master and bios has other temps for the cpu (around 60 degrees). At the same time the cpu water block is not too hot to touch and the air blowing through radiator is cool. Most probably the shutdowns happen because when I start stress testin that temp is climbing too fast and as a precaution it is an emergency shutdown, but in reality I think that temp is a bogus one.

MSI:

Quote

1. If you press F6 to load optimized defaults, only manually adjust the CPU ratio,  press F10 to save, can it overclock the CPU successfully? Will it occur crash?

2. If you press F6 to load optimized defaults, only manually set the CPU voltage,  press F10 to save, Will it occur crash?

3. About the temp issue, please download and install the MSI Command Center to check the temperatures.

https://www.msi.com/Motherboard/support/X399-GAMING-PRO-CARBON-AC#down-utility

4.We suggest you change the plan power to the saver power for a try.

ME:

Quote

If I overclock manually, I can push it very little, maybe to 3.5ghz and that is it. Then the whole thing crashes, extra voltage just crashes it faster.
I made couple os screenshots, one is idle and the other one under load, all while overclocked to 3.6 with amd overdrive option. You can see, that only ryzen and your program can see the real cpu temp, other programs see some other temp and when that temperature hits 95 degrees, it starts throttling. I am using enermax 360 aio cooling on it, so temps should not be an issue

MSI:
 

Quote

Do you mean that OS will crash if you only manually adjust the cpu ratio to 35 in bios? Don't change other settings in bios. Thanks.

ME:

Quote

No, multiplyer only is fine until a limit, but of course after increasing voltage and when that “fake cpu temp” gets too high, it will shut down. 
Helpme find out what temp it is and how to change some settings that it would not throttle with that temp hitting 95.

MSI:

Quote

Please describle the issue clearly. Do you mean that OC function can work fine now? Please tell us the test step clearly. We are so sorry thta we don't understand what your current problem is. Is the your issue the overclock or the cpu temerature? Thanks.

ME:

Quote

Okay, I will describe it once again. When I overclock (it does not matter, whether manually or automatically by the cpu technology), the temps naturally go up, obviously. So couple of post ago I sent you images, where you can clearly see 3 applications side by side, 2 of which say that cpu temp is 60 under load and one that says cpu temp is 90. That is where things start getting bad, because something starts throttling the cpu clock, because it sees the temp being 90+, when the temperature in actuality is around 60, shown by MSI control centre. At idle, MSI control centre says my cpu temp is 30 degrees, but other hardware monitoring apps say that it is 60+ degrees, what clearly is nonsense.If I overclock it more aggressively, then cpu temp rises more rapidly on both, MSI control centre and on the other apps aswell, then the system cannot react quickly enough to throttle it down and after the temp goes over 95 degrees (not the temp in MSI control centre), system locks up and shuts down.

And do you really look surprised when you adjust cpu ratio only and the system crashes? Obviously it crashes if you do not add voltage to the CPU. It would be perfect if I could push the cpu from 30 multiplyer to 40 without any voltage change and the thing would run stable. That is not how overclocking works.

MSI:

Quote

1. Do you mean that the issue only happened under the load? 
2. Please enter to the bios, press F6 to load optimized defaults, then make your settings, press F10 to save and reset. And before you press "Yes", please press F12 to save the BIOS screenshot to a USB stick and send the screenshot to us. It can show us the detailed settings. You can refer to the attached screenshot. Thanks!

ME:

Quote

1. Yes, of course only under load, that is when temperatures rise.
2. that screenshot will come a bit later

ME:

Quote

So here is the screenshot. I have a question, do you have any idea what Precision Boost Overdrive is? You keep pushing me to overclock the system manually, but I actually like the AMD's solution better, it does the work great and I do not have to figure out the voltages etc myself. That is why I keep on using it. I am clocking the ram to 2933, to get me more stability, ram is rated to 3200. everything else I am changing to get me better overclocking.

MSI:

Quote

Precision Boost Overdrive:  Disabled: stock board values for standard IRM.  Enabled: Loads board limits for EDC, TDC and PPT.  Manual: manual setting of PPT, TDC and EDC

-- but I actually like the AMD's solution better, it does the work great 

Please send us the pictures about how to set the overclock CPU in the AMD software for our reference. We will test it according to your settings.Thanks.

ME:

Quote

If you put precision boost overdrive to manual from bios and start upping the PPT, TDC and EDC, then the CPU does all the overclocking for you. Default values are as follows: PPT 250W, TDC 215A and EDC 300A. In theory you can max them all out to some really high numbers and then the processor would overclock itself until the thermal limitation. My settings were as follows: PPT 500W, TDC 400A and EDC 349A. My overclock was limited by EDC, as soon as I went over 349, the computer started crashing. It should not be like that.
To answer completely, once you do everything from BIOS, you do not have to set anything from software. Here is how the program should look like after you set the values, I have not touched anything inside the software. (note the discrepancy between the temperature on AIDA64 cpu temp and Ryzenmaster CPU temp)

MSI:

Quote

We tested it with the 1950X CPU, the temp is same in the AMD Ryzen master and the AIDA64. Please refer to the attached picture. It seems related to the CPU.

About the OC issue, because your settings are too much, so we only set the PPT 500W, TDC 400A and EDC 349A, we don't change other irrelevant settings. But the CPU frequency doesn't change. And we do the load test by system stability test. The system works fine. Would you please tell me more information?

ME:

Quote

Guys, cmon, you DESIGN motherboards, you are the ones that implement these features into your motherboards etc. I know that I most probably am talking to a customer service rep and not to a engineer. But really? You cannot type "precision boost overdrive" into google search and read literally the first link that comes up and the FIRST sentence says "Precision Boost Overdrive (PBO) is a powerful new feature of the 2nd Gen AMD Ryzen™ Threadripper™ CPUs."

And there you are, trying stuff on first gen threadripper, amazed that stuff does not work...
speechless

MSI:

Quote

Thanks for your information. We will try our best to find the 2990WX CPU for test. Sorry for the inconvenience caused you.

ME:

Quote

You are saying, that, MSI, a multi BILLION dollar company, does not have a Threadripper 2 CPU on their hands? 
You live and you learn, amazing.

MSI:

Quote

We will check the issue as soon as possible. We will reply to you once we get the result. Please wait patiently. Thanks!

MSI

Quote

Sorry for my late reply.

We have set the same options in the bios(attached picture1), then enter to OS, use the AIDA64 to stress CPU. The system didn't crash (attached picture2). We tested many times. Test time is around 5 minute or 10 minutes. And the power of the power supply we test is 500W.  There are many factors to cause crashed under load, such as the ram、system、CPU...

About the temp issue, as the AIDA64 is third-party software. It maybe didn't update the version to correct the temp for 2990X CPU. We suggest you check the temp of CPU by the command center and amd ryzen master. Thanks.

ME:

Quote

well, this config does not crash it, but now try and put the EDC to like 400 or even 500, it does not matter. I can already see from your picture, that you have the same exact issue that I am having, where AIDA is showing different cpu temp than what ryzen master is showing. The 2990WX should easily clock to 3.8-4.0Ghz. 
You can even max out PPT, TDC and EDC, it should not matter, then the cpu will clock itself to max limit.

and be sure that the CPU actually overclocks, because on your picture it shows that it is running at 0.7Ghz... that is not overclocking

MSI:

Quote

Did you set other settings? such as the cpu voltage. We noticed your cpu voltage is 1.3, but from the bios screenshot, it doesn't show you set it. Does it need to set other option to overclock? We would appreciate if you can tell us all settings to get your issue. In addition, we have set the EDC to 500, but system still doesn't crash. Please see the attached picture.

We aren't a bios engineer. please forgive us for being unprofessional. But we are still tried our best to solve your issue. Thanks.

ME:

Quote

I understand.
No, Precision Boost overdrive actually changes the cpu voltage by itself, if you have it on, it should overclock the cpu until its thermal limitation automatically, it will adjust the voltage and multiplyer by itself. No input from the user. What overclock do you get? 

You can try and overclock your system manually too, see if that changes anything. For me, that aida64 reported cpu temp goes over 95 degrees and then the system shuts down.

On your system it is even more strange because your cpu does not even run with the stock speed of 3.0 ghz. That is already an issue what you should look into, you can clearly see that your voltage is not even 1v and aida is reporting 95 degrees on the cpu.
You can try and reset everything and just look if your cpu can run 3.0 without any overclock on it, if not, then there is your issue to solve.

MSI:

Quote

the Please don't check the temp of the CPU by the AIDA64. It maybe didn't update the version to correct the temp for 2990X CPU. We suggest you check the temp of CPU by the command center and amd ryzen master. 

If we keep default settings under bios, AMS Ryzen Master will show that the peak speed is 4.0Ghz. If we set same settings with yours, the peak speed is around 3.2Ghz under idle. We cannot copy your issue, we are afraid we don't know what causes it. Sorry for the inconvenience caused you! 

ME:

Quote

And it runs at 4.0ghz under loadtest? Show me.

MSI:

Quote

If we keep default settings under bios, AMS Ryzen Master will show that the peak speed is 4.0Ghz under idle.  In the previous letter, we test it many times. As we didn't copy your issue, we didn't know what caused it. Thanks and sorry for the inconvenience caused you.

ME:

Quote

Guys, UNDER IDLE, i can get it to show 4,0ghz as well, obviously it is under load when it crashes. I hope you understand that under idle does not mean anything. Do not try to shake this issue off. Those settings that I sent you were the last stable settings for me, and those settings did not work for you either. Show me speeds over 3.6ghz under load and I belive you it is me, not your board. Do not try to shake me off, belive me, I will not stop until it is resolved.

You did replicate my issue, I saw it from pictures.

MSI:

Quote

Are the CPU_PWR 1 and CPU_PWR 2 connected to the 8 pin power supply? If you manually set the Precision Boost Overdrive option, please use above 800w power supply, and also, both CPU 12V-in connect (CPU-PWR1/CPU_PWR_2) should be connected properly.

ME:

Quote

Yes, everything is connected and I am giving enough power to the cpu. Now can you show me how your pc looks if you put it under load with 4ghz to the cpu. I truly have a feeling that your motherboard is not able to overclock the cpu that far.

MSI:

Quote

Now, the current issue is that: 
1. If you did the settings in the bios,(bios screenshot you sent us), will the CPU frequency drop under load? Or is it always stable to 3600? If the CPU frequency will drop, what is the frequency after the reduction?
2. Do you think the CPU frequency will increase to 4.0 under load after you did the settings in the bios（bios screenshot you sent us）？

Global C-state Control:[auto]->[Disable]
Spread Spectrum:[Auto]->[Disabled]
Relaxed EDC throtting:[auto]->[Enabled]
AMD Cool'n' Quiet:[Enabled]->[Disable]
Precision Boost Overdrive:[auto]->[Manual]
PPT Limit:[auto]->[500W]
TDC Limit:[auto]->[450A]
EDC Limt:[auto]->[349A]

If you only set the above options, don't change other options. Will the CPU frequency drop under load? Or is it always stable to 3600? If the CPU frequency will drop, what is the frequency after the reduction? Thanks.

ME:

Quote

So, first of all, since I have access to another identical motherboard, I tried swapping the cpu to the other PC, no changes. Then I swapped the PSU to a 1000W one, not an issue with a PSU, everything remains the same.
Second, the most stable I have had it is at 3.6... but when I leave the stresstest running for a couple of minutes, the clock and voltage of the cpu starts fluctuating... There is no certain value thet it drops down to, but it usually is around 0.9Ghz and 0.7V... then the fluctuation just keeps going every couple of seconds... it goes to 100%, drops to 30%, back to 100% etc... It is something thermal related, because initially it does not do that... If I leave the case open from both sides, it even runs without these drops sometimes... but bear in mind, CPU temp does NOT go over 60-65 degrees no matter if the case is closed or open.

Then I tried manually overclocking... did not get it stable even at 3.6... voltage was 1.375 and under stress it crashed (ram running 2133). The only way it is "kind of stable" is when running with precision boost overdrive, then I can run the ram at 2933 even (tried 3200, crashed, ram is rated at 3200, cl15).
But all this overclocking is with a HUGE spare on the temperature department. I still have about 20 degrees to spare, the clock should be higher.
There is no problem when I am trying to overclock TR1 with this board, but TR2 seems to have huge issues. I cannot even return this board either, too much time has passed, should have gotten an ASUS board, that is confirmed to run TR2 with 4.0GHz with ease. I really am disappointed guys. Please release a BIOS update or something, or just tell me, show me, how your board is doing 3.8 or 4.0, what settings etc.

MSI:

Quote

---If I leave the case open from both sides, it even runs without these drops sometimes..
Do you mean that cpu frequency will be stable at 3.6, no drop?

--Then I tried manually overclocking... did not get it stable even at 3.6... voltage was 1.375 and under stress it crashed
What do you mean that manually overclocking? Does it mean that you adjust the cpu ratio to 36 and adjust the cpu voltage to 1.375 manually? Or Does it mean that you still use the precision boost overdrive for manually overclock?

Would you please test the issue when you keep the memory frequency default settings? Only set the below settings. The overclock for rams also will affect the cpu frequency. So please firstly keep the issue simple. Only change the below settings to test. Thanks.

Global C-state Control:[auto]->[Disable]
Spread Spectrum:[Auto]->[Disabled]
Relaxed EDC throtting:[auto]->[Enabled]
AMD Cool'n' Quiet:[Enabled]->[Disable]
Precision Boost Overdrive:[auto]->[Manual]
PPT Limit:[auto]->[500W]
TDC Limit:[auto]->[450A]
EDC Limt:[auto]->[349A]

ME:

Quote

1) Yes, sometimes the CPU frequency will be stable at 3.6, no drops (not always) again, largely depending on the temps.

2) Yes, that means that I adjusted the ratio and voltage myself, no overdrive. But overdrive seems to run the minor overclock that I have, more stable.

3) I am always testing it with default memory frequency, after I have my setting somewhat stable, I try to turn up memory frequency. All new tests I do with memory at 2133Mhz.

MSI:

Quote

Thank you for your information. There are many factors to cause the system to crash\unstable. Include overclock、temp、CPU itself.....We have consulted with related department, If you set the voltage to 1.375v, the system crashed under load, it maybe caused by temp rise quickly. We suggest you reduce the cpu voltage to try, such as1.2v. About the PBO option, if you set the PPT to 500W, It is possible to occur crash under load for X399 GAMING PRO CARBON. We suggest you reduce the PPT\TDC\EDC value to try. And make sure your cooling system can keep the CPU temperature below 70° under load. If you want to set the PPT to 500W, the MEG X399 CREATION is better than the X399 GAMING PRO CARBON. Anyway, we are sorry to say that overclocking is not guaranteed. Very sorry for the inconvenience caused you! 

ME:

Quote

Temperature wise, I have a 360mm radiatior cooling the CPU, to this day I have not seen temperatures more than 60 degrees.

About PBO situation, have you been following ANYTHING that I have written to you? You are suggesting me to reduce the PPT/TDC/EDC values. That is the reason, my EDC is 349A, I cannot go over that, even though I should be able.

Again, my cooling is VERY good.

Do not try to sell me another motherboard, when i just bought a 400$ motherboard from you. What you should do, is make this motherboard work.

AND THAT IS WHERE WE ARE....

 

 

 

 

[Ashleyyyy]

i read the whole thing, and wow. after all that they try to convince you to buy a more expensive board? wow... 

[r1v0]

they are just unbelivable, now they are just trying to get rid of me and responding to everything with:

Quote

Very sorry for the inconvenience caused you! 

That is what you get with some Asian under-payed guys on the other end...

[Origami Cactus]

IF you bought it on amazon just return it and get an ASUS board.

[Streetguru]

32 minutes ago, r1v0 said:

 

The gaming pro carbon if it has limits, is probably there because it doesn't support overclocking a 32 core CPU much if any lest the VRM blow up...

need a TLDR on that post though

 

There's a reason the MEG board exists.
 

 

[r1v0]

1 minute ago, Origami Cactus said:

IF you bought it on amazon just return it and get an ASUS board.

nope, bought it from a local shop, and it has been over a month now dealing with this crap... 

[r1v0]

3 minutes ago, Streetguru said:

 

 

hmm... got me thinking.... maybe I should try and cool the VRM separately? What if I just put a dedicated fan blowing right on it?

[Streetguru]

5 minutes ago, r1v0 said:

hmm... got me thinking.... maybe I should try and cool the VRM separately? What if I just put a dedicated fan blowing right on it?

It's mostly just a physical limit of those VRMs, they were built for 16 cores, not 32 cores overclocked.
 

This is what the MEG board can support

 

[r1v0]

3 minutes ago, Streetguru said:

 

 

 

Already getting more support with 30 minutes from a forum than I have gotten from MSI within a month. Mind blown...