New AMD build freezing at random times

averylazysloth · Nov 14, 2018

Hello everyone,

I am new to this forum and I don't know where to put it.
I have recently build a new computer (I am new to this). Everything seems to work fine besides that my computer freezes at random times (while playing SteamVR, gaming, or doing nothing).
The freeze requires me to hard reset my computer because nothing is responding.

Specs:

Mobo: Asus ROG STRIX X470-F Gaming
CPU: AMD Ryzen 7 2700X
Memory: G.Skill 16 GB DDR4-3000 (2 x 8 GB)
PSU: Thermaltake SMART PRO RGB 750W
GPU: GIGABYTE AMD Radeon RX Vega 64 8 GB
Storage: Samsung 860 EVO, 500 GB SSD and Seagate BarraCude, 2 TB
OS: Windows 10 pro 64-bit

I've been monitoring the temperatures inside the case and it always seems to be stable. I've also included some benchmark results.

(Should I raise the frequency to 2933 Mhz?)

What i've done so far to try and fix this problem:
- Downloaded latest Windows 10 updates.
- Updated BIOS version to 4024
- Updated drivers (GPU, LAN, ...)

Can someone please help quide me through this problem? I really want my PC to be stable.
Thank you in advance and have a nice day.

Jason Pieters

kanewolf · Nov 14, 2018

Your statement of "Everything seems to work fine besides that my computer freezes at random times" means everything does NOT work fine. If it is locking up with stock settings you need to do basic stability troubleshooting.

One item you didn't list in your parts list above is the power supply...

averylazysloth · Nov 14, 2018

kanewolf :

I am using the Smart Pro RGB 750W Bronze Power Supply. I have no idea where to start with this problem. I've done memtest and got no errors. stress tests also seem to run alright.

kanewolf · Nov 14, 2018

How long did you let MEMTEST run? Overnight?

What cooling are you using for your CPU? Air? AIO?

Do you have the latest BIOS?

averylazysloth · Nov 14, 2018

kanewolf :

Going to let let memtest run for a couple hours. I only did it for a short period of time earlier.

I am using the cooler master masterliquid ml240l rgb aoi.

I did update to the latest BIOS.

Redneck5439 · Nov 14, 2018

You have to set your RAM to the G.Skill specifications of your kit. You list you have 3000Mhz RAM yet your running it at ~2130. Ryzen loves fast ram, and looking at your timings I would bet your running CL14 Samsung B die ram with horribly loose timings. That alone can cause Ryzen to not perform well at all. Saying your not running stable go into Bios and set the Ai Overclock Tuner to D.O.C.P. Standard (or will list your ram). It should auto detect your RAM and set for the proper clock speed and timings for your kit. If for some reason it doesn't you'll have to manually set the clock speed and timings yourself. You should also set the SOC voltage for no more than 1.1V, you probably won't even need that much but leaving it at auto puts too much voltage through the controller.

Have you done any overclocking? Have you set a negative offset or manually set the Vcore of the processor? I never "overclocked" my rig but by setting a negative offset (as on Auto my core voltage was exceeding 1.5V which is absurd), setting the SOC to 1.1V (so its not being over volted), setting my RAM to 3600Mhz with stock timings (just setting it to the kit specs) and enabling Precision Boost Overdrive with a scalar of 10X I have all core boosts of 4.175 - 4.225Ghz and four cores that boost up to 4.3 - 4.350Ghz. My Cinebench R15 scores are all above 1900 topping off at 1920. Considering that I haven't done any real "overclocking" that is a very nice boost provided by Precision Boost Overdrive.

One last thing, Memtest is a good tool and an old standby for stability testing, however I have found that its not all that reliable with testing Ryzen. I've had my RAM overclocked to 3800Mhz and it passed Memtest running all night, but it would cause any game I would run to crash to desktop within mere minutes of game play and I had some other odd stability issues. I found I had to raise the voltage of my RAM to ~1.475V to get it "stable" in games (even though according to Memtest it was fine). I later just set my RAM to run at the stock kit settings (for me 3600Mhz @ 1.35V) as I have found that really above 3466Mhz you don't get much more performance. Don't trust that just because you pass Memtest your don't have memory issues on Ryzen. Its always best to run the kit at its specified speed, timings, and voltage.

averylazysloth · Nov 15, 2018

I've let memtest run overnight and it produced no errors. I've also set the ai overclocker tuner to DOCP and the VDDCR SOC to 1.1V. My bios seemed to autodetect the 3000Mhz and the associated 1.35V for the memory. If I open Ryzen master it shows me that my rig is still using 1067Mhz memory.

Redneck5439 · Nov 15, 2018

averylazysloth :

Like I said, Memtest cannot be 100% trusted for Ryzen. Setting for DOCP is good and the SOC won't be overvolted now, which is also good. You may want to re-check bios to make sure your RAM is set for 3000Mhz and check to see if your RAM timings are set to the proper specifications of your RAM kit. If your memory is set in bios to 3000Mhz you may be getting a false reading in Ryzen Master. I have found that Ryzen Master gives me false readings on my core voltage. Download and launch a program called CPU-Z. Its totally free and will give you a lot of information about your system. Check in the memory tab and see what it is listing for your DRAM Frequency. You should be getting 1500Mhz or very close to it and it will also list your timings so you can just double check that as well. Ideally you'll be able to get your RAM to 3200Mhz, but 3000 isn't bad and we are first trying for 100% system stability. If your RAM is now running at the kit specifications you can go ahead and test out your system and see if you are still having issues. If you are then we have to start looking elsewhere such as some other setting you changed in bios or possibly having a closer look at your power supply unit.

averylazysloth · Nov 15, 2018

Ryzen Master confirms that the ram frequency is 3000Mhz now, I think DOCP is running correctly. The freezing still occurs though. I haven't changed anything in bios besides CODP and optimized fan control. What else can I do?

Redneck5439 · Nov 15, 2018

averylazysloth :

What are your thermals like, how hot is your processor getting? Are you on stock cooling? Do you have adequate cooling in your case? These are the next questions to address. I suggest downloading a program called HWmonitor as you can see the temp of your processor, and it will also give you several temp readings on your motherboard. You have to ensure that something isn't being overheated for some reason. HWmonitor will automatically save the highest recorded temp and voltages for every sensor it can read so after running benchmarks you can see how high everything spiked. Have a look at what the highest temp reading on your motherboard components are and what the highest temp reading for your processor is while under load/ right before you get a system freeze.

You should also have a look at the hardware itself. Is this a totally new build or are some components bought used, or are being reused? If everything is new I doubt it is a hardware failure, unless you were unlucky enough to get a bad component. Are you running the latest drivers for your Vega 64? Have you overclocked or modded your Vega 64? At this point I don't think its the RAM and if you haven't touched any CPU options (ie changed multipliers, LLC or something) I doubt its that unless your hitting high thermal loads and its throttling. That basically leaves the PSU and GPU as most likely culprits.

One thing I just thought of that could be an issue and would be a rookie mistake is what slots did you put your RAM in? For the Strix Asus recommends installing Dimm (when using 2 sticks) in the B2 and A2 slots (page 1-5 of your motherboard manual). Make sure you have your RAM in the proper slots. You probably do, but it never hurts just to make sure.

averylazysloth · Nov 16, 2018

I am using liquid watercooling for the processor (cooler master masterliquid ml240l), 2 fans in the front, 2 on the bottom and 1 in the back creating an airflow from the front to the back.

I ran HWMonitor with prime95 (maximum temperature test) for about 20 minutes. (it didn't freeze)
I am also running it again for a longer period of time.
These were the recorded max temperatures:
MOBO
TMPIN0: 82°C
TMPIN1: 29°C
TMPIN2: 66°C
CPU: 83°C
GPU: 33°C
DRIVES: 33°C

It is a totally new build so every component is bought new. I am running AMD Radeon v18.11.1 and I didn't change any CPU options.

I checked my motherboard manual and the RAM is in the right slots.

Redneck5439 · Nov 16, 2018

averylazysloth :

In my opinion your temperatures are on the high side for a R7 2700X under liquid cooling. I'm hitting the same temps after about 30 minutes of Prime 95 testing and I'm on the stock Prism cooler. I usually have Ryzen Master, task manager and HWmonitor running in the background. Temp reading I trust HWmonitor and clock speed readings I trust Ryzen Master and task manager. During torture testing I get a constant all core boost of 4.2 - 4.225Ghz all core. What is your all core clock speed coming in at? If your boosting higher than mine it could explain the temp difference, but you should be at least 10C cooler than me with liquid cooling saying I'm on the stock Prism. I do have more case fans (5 140mm intake fans and 2 140mm exhaust fans creating positive pressure) which could explain my cooler motherboard readings, but that shouldn't effect the processor temps by that much, not compared to liquid cooling. Keep in mind your still not technically "over" the thermal limit- AMD specs Ryzen at a thermal limit of 95C, but usually temps over 80C with liquid cooling are the product of overclocking, or of course running the stock cooler.

Redneck5439 · Nov 16, 2018

Don't overdo the Prime95 too much. I know a lot of guys talking about running it all day and night- the famed 24hour "burn", but most modern processors are going to error out within a half hour if there is a problem so the 24 hour burn is in my opinion outdated and not necessary. If you can pass 30 min of torture testing, which will push your computer far harder than any other task you'll ever do, then you are "stable" as far as Prime95 testing goes. Now just because your stable with Prime95 doesn't mean your 100% rock solid either. I've actually had past builds that could pass 24 hour torture testing but would crash randomly during games or while running a light duty task. The very best stability testing there is... Just use your system.

You may want to crank your case fans up a bit, have them run faster and create more air flow, would help with motherboard temps. You may also want to check your thermal paste, reapplying with a better quality paste could drop your CPU temp. Do a 30 min run with Prime95 and see how your doing temp wise and what your all core clock speed is boosting to. If you pass 30 min your "stable" and we can focus on tweaking some bios settings. At the very least you can get a negative offset on your Vcore, lower core voltage and get some lower thermals.

Redneck5439 · Nov 16, 2018

Updating my previous post. I just ran 40 min of Prime95 and my temps on the stock cooler are much lower than yours on liquid cooling. After 20min my max temp was 68C and after 40min my max temp was 70C. The motherboard readings were tmpin0 70C, tmpin1 23C, and tmpin2 was 51C. The motherboard temps could easily be I have more case cooling than you do, but on liquid cooling you should have much lower temps than I do with the stock Prism cooler. You are running far too hot, and that could easily be your issue.

Before going any further we need to know what your average all core boost clock speeds are. I need a low and a high. For example my rig runs Prime 95 between 4.175 as a low and 4.225Ghz (with a max CPU Vcore of 1.385V also important information we need to know) as a high on all cores during Prime95. If you rig is running higher clocks or massively higher Vcore then that could explain the higher temps, if not than there is another issue that needs to be addressed.

averylazysloth · Nov 16, 2018

I re-applied some Arctic silver thermal paste. I ran Prime95 for 30 minutes (no freezing).
Max temperatures:
TMPIN0: 84°C
TMPIN1: 29°C
TMPIN2: 66°C
CPU: 85°C
GPU: 34°C

The speed I got from the CPU was between 3900 and 4000Ghz with a max CPU Vcore of 1.531V.
I have a Sharkoon nightshark RGB case and I can't really afford a new one at the moment.
I used the optimize fan control option in bios to tweak my fan speed. I'll up their speed now.

Redneck5439 · Nov 16, 2018

averylazysloth :

Its good that you aren't having system freezes anymore, that is real progess

However you have to get the heat under control. You should have much better thermals, especially under liquid cooling.

Your Vcore is much, much too high and could be resulting in the higher temps. Running at 3.9 - 4.0Ghz these Ryzen processors can easily do that on ~1.3 - 1.36V. These processors have an absolute max of 1.55V on the Vcore and its not recommended to run it above 1.5V. I know on AUTO for some reason they do use absurdly high Vcore. I would suggest setting a negative offset in bios of I would say -0.1125 would be a safe starting point. That should lower your Vcore to max out just below 1.4V under full load. That should keep your system 100% stable while lowing temp and will overall help your processor as it is being over volted.

With liquid cooling you should be able to run Precision Boost Overdrive, should be hitting 4.2 - 4.225 all core boosts and should be doing all that at below 70C. There are some real heat related issues going on. How do you have your fans mounted on your water block? Are they pushing air though the block or drawing air away from the block? They should be set up to push air through the block and straight out the top of your case. Lowering the Vcore will help but there are other heat related issues going on here.

I'd have to find some reviews on your case. It's a pretty case to be sure, but for future reference as a rule of thumb the more glass the less air flow. I have a case with a solid tempered glass side, but the top and front are all open mesh with only the dust filters interfering with the air flow. With a tempered glass front the intakes have to try to draw the air through the side vents and that is a lot of restriction. Like I said, just for future reference.

averylazysloth · Nov 16, 2018

I applied the offset of -0.1125 and the max CPU Vcore never went past 1.431V. My 2 front fans are fushing air through the block of my watercooler.
The problem isn't fixed though, my computer just crashed again while it was running a game. I'll be getting a better case in the future when I have more money but i'll have to do with this one.

Redneck5439 · Nov 16, 2018

averylazysloth :

Do you have top vents on your case? It doesn't look like it from what I could find, but if you do moving your waterblock to the top of the case with fans blowing air through the block and straight out the top of the case would be the best setup. If you can do that then move your bottom fans to the front panel as intake fans.

If you can't do that, then just to be clear the way your waterblock should be set up is your waterblock should be installed first right up against the panel then the cooling fans are attached to the waterblock pushing air through the block and out (in this case) the front of your case. Basically then the front fans act as exhaust fans. If that is the only way to set it up (ie no top exhaust) then I think for your particular case you may have better luck keeping the waterblock set up as described, keeping the bottom fans set to intake and then flip your rear exhaust fan to make it an intake as well. In some low airflow cases (especially ones lacking a top exhaust) this seems to help. Make sure you have your fans facing in the right direction for exhaust and intake, its an easy rookie mistake to make.

I also think you should try one more thing with your bios. Take some photos of your bios screen for RAM settings, SOC 1.1V and the negative offset, so you remember those. Now select optimized defaults save the changes and restart. Before the system goes to windows mash the del key and set your RAM back to its proper settings, set the SOC to 1.1V and set the negative offset. Save and restart and see how your computer runs now. By doing that if you changed something without realizing it will be set back to default. So the only changes now are to RAM (clock and timings), SOC at 1.1V and negative offset. If your still having issues after that then its either do to the excess heat that better cooling will have to fix or its a defect in hardware itself (which is very rare in my opinion).

Redneck5439 · Nov 16, 2018

I was just looking over your case. Did you leave the RGB case fans in the front of the case? If you left them there did you flip them so they are exhaust. I'm trying to get an idea of how you have it set up. The AIO came with fans that should be set up to push air through the AIO. So the waterblock is mounted to the front panel of your case, the fans that came with it are installed next directly to the block pushing air through the block and out the front panel. If you left your front case fans in place and never flipped them then you have a big problem as all those fans are cancelling themselves out. It has to be set to what is called a push pull, configuration. The AIO fans are pushing air through the block and the front case fans are pulling the heated air and exhausting it out the front. All the air is flowing in one direction that way. If those front RGB fans are still there then they need to be flipped so they are exhaust and not intake fans.

averylazysloth · Nov 17, 2018

Redneck5439 :

This is my current setup. I left the installed RGB fans on the front and I mounted the AIO block onto them. I mounted the cooler master fans on the bottom of the case (as the sharkoon fans are a bit prettier). So the best thing I can do is to flip the front fans so they pull air through the block out of the front?

Redneck5439 · Nov 17, 2018

You will have to get some more case fans. Did you use the fans that actually came with the waterblock on the bottom of the case as intakes? The correct way to cool your waterblock is for the fans to be on the other side of it pushing air though the block. If your front fans are set up as intakes blowing air through the block then you are blowing hot air from your waterblock directly into your case heating your components, not cooling them.

Redneck5439 · Nov 17, 2018

The best thing you can do is flip the front fans to be exhaust, mount two fans on the GPU side of your waterblock pushing air through the block (air flow is going through the block and through the front of the case exhausting all the hot air from the block to straight out the front of your case). This is a push/ pull configuration and will be exhausting the hot air from your waterblock directly out the front of the case instead of blowing that hot air into your case. The way you have it set up now is like putting a heater in your system as it is constantly blowing hot air on your components. You are literally cooking your components.

Best set up is to use the bottom fans to mount on the GPU side of your waterblock pushing air though it to set up the push/ pull for your waterblock. Flip the rear exhaust fan to an intake fan and get two (or 3) more 120mm or two 140mm fans (if they will fit) to install as new intake fans on the bottom of your case. In this configuration you will have 3 intake fans (the rear fan and two bottom fans) and four total exhaust fans which are also doubling as your waterblock cooling fans in the front of your system.

averylazysloth · Dec 6, 2018

I've brought my PC to a trusted repair shop and they found out my power supply unit was faulty. So now I got the corsair rmx 850W instead of the thermaltake RGB 750W and it seems to be working fine now.
Thank you for your help though!

New AMD build freezing at random times

Titan

Titan

Honorable

Honorable

Honorable

Honorable

Honorable

Honorable

Honorable

Honorable

Honorable

Honorable

Honorable

Share this page