This will be a long one so you can start pouring your coke [or insert favourite beverage here] and popping your popcorn [i'll be nice enough to let you choose the food of your choice too].
First and foremost, here are the specs of my build: https://ca.pcpartpicker.com/list/pcJbM8
Note: take as a premise that i've been researching all matters of PC building and overclocking on and off since 2014 (when I started contemplating the idea of building my own PC) and more intensively since the last 2 months. Even though it doesn't exclude the fact that I don't have absolute knowledge, I like to be thorough in everything I do so I consider asking for help here as a last resort after doing my own research and looking at forums as I understand it can be difficult to figure out what is going on from a distance. TL;DR: i'm a novice--yes--but i've done my homework (maybe not all of it, but what I could) and I know what i'm doing.
While I initially had certain issues that I imagine at least a certain portion of first-time PC builders like myself encounter (atleast from what I assume and what i've observed on forums concerning the matter), I was able to overcome them and get to the point where I am now (which means getting windows, other programs, and games--at least partially, lol--to run)
Issues i've been having:
My first concern was with temperatures, which seem to always have hovered around 40-45 in the UEFI no matter the version (those temps aren't alarming but still warm for idle. I figure this is due to the offset applied to X Ryzens either by Asus or AMD themselves, and i've seen people get higher temps in their BIOS than when idling on Windows, which I also observed myself by looking at temps on HWMonitor while idling on desktop later on. Although i'm still not 100% certain what temps to look at with all the confusion on the Internet, the package and TMPIN temps all seem to stay below 40 and package sometimes goes down to the 20s for a second on idle and i've observed an around mid-60s maximum during a Cinebench test).
So after doing some research and stressing out about this for a while, I finally decided to install Windows 10 and convinced myself my system was sophisticated enough to protect itself and shut itself down in the event of excessive temps. I initally experienced display signal-loss/ black-screen issues (which I describe later on below) when I first tried to install Windows but this was due to a too-drastic undervolting (-0.13125 offset, to be exact, in fear that my motherboard was overvolting my CPU with voltages of 1.4v and up, which it goes up to from its 1.375 default setting) on my part, which I corrected by recovering optimized defaults. Rest assured I proceeded to reinstall Windows after to avoid any chance of corruption due to these issues I experienced during my first installation.
After the Windows installation (plus updates) and then installing drivers one by one with a restart in between each (see the issue I had with Nvidia driver 385.41 at the start below, which is not an issue anymore for some reason) was done came my growing concern with voltage, which my Asus ROG Strix B350-F sets at 1.375 (a voltage that i've heard Ryzen can handle but a bit on the high side and close to the threshold at which you start to observe degredation after a while, according to me and what i've seen on the interwebs); that is, as of BIOS 0809. Unfortunately, I don't remember which UEFI version my board came with as I almost flashed right away after testing to see if it POSTed.
The thing is when I leave my UEFI settings on optimized defaults, it sets my Vcore--or my VDDCR, as they call it--on 1.375 auto but I can see it hover around 1.4-1.45 in my UEFI hardware monitor (screenshots of my UEFI screen opon request), which I can also see momentarily spike up to in HWMonitor on Windows. I know it is normal for CPUs to ramp up and down due to cores waking up and falling to sleep/different states but unless it is normal for it to do that, it means my MB is clocking my CPU up to 4Ghz on at least 2 cores (according to AMDs XFR/Precision Boost Technology) and also while idling on desktop with only HWMonitor open, which it shouldn't and is a bit excessive, in my opinion.
When I set VDDCR to 1.2875 using a -0.0875 offset from the standard 1.375, it can go down to around 0.5v when completely idle and nothing is open, but as soon as I open a program, I rarely see it go down to even 0.8v (i'm not sure if this is a problem) and often goes higher than 1.3v (I've seen it go up 1.363 when running games or a Cinebench test). Maybe i'm not looking at the right voltage since there are many voltages (and temps) to look at in HWMonitor. Also, if I set a CPU core ratio of 38 and let it set a voltage on its own, it sets it to around 1.5v so that means I have to downvolt my CPU from default to attain 3.8Ghz without frying it.
- Since updating from CCleaner 5.33 to 5.34, Windows Defender detected a backdoor threat in the 5.33 installer .exe and somewhere in a ryzen-related directory (although I should have recorded the location and failed to do so due to the stress of having a threat on my newly-built PC on which I never did anything or went anywhere near where there was a chance I could have gotten a virus, it was probably a chipset driver related file). Also, pretty much right before I entered the OS I had flashed from UEFI v0809 to 0811, so basically I can't isolate what happened. When I got into Windows, I decided to perform a Cinebench test to see if my scores were still the same since I felt like they had lowered at some point (due to background programs, I had assumed) but they seemed to still to be on par with what I got before the Cinebench freezing issue started to appear.
- My computer sometimes fails to POST on restart when clocked to 3.8Ghz on 1.287v (either via restarting through windows or pressing the restart button on my case)
- The backlit writing on my Gtx 1070 doesn't light up evenly. Also, i've started experiencing coil whine (i'm pretty sure that is what i'm hearing) since a long session of BF1, although this doesn't seem to affect performance
I know saying this makes it look like my GPU is the culprit but what makes more suspicious about my CPU is the micro-stuttering in games and the recent freezing in Cinebench, which looks similar to when I tried installing Windows 10 while undervolting my CPU.
My Cinebench scores with corresponding UEFI settings are as follows:
*DOCP=Asus' term for XMP Profile
**CPB=Core Performance Boost
Completely stock (anywhere around 1.35v-1.45v with DOCP and CPB On): the highest I ever got was 1243
Stock with DOCP and CPB off (around 1.22v @3.6Ghz): 1150-1170
1.287v @3.8Ghz with DOCP on and CPB off: the highest I ever got was 1278
Game Performance:
All game cutscenes seem to have this kind of micro-stutter effect that makes objects twitch and look laggy in general (especially GTA V and BF1)
Nonetheless, I get pretty good FPS in the games i've tried.
Space Engineers: 90-120 on average but it dips to below 60 when I crash a ship or when it loads terrain on a planet
GTA V: 70-110, except for 55-60 fps lag spikes sometimes when changing camera angle rapidly, in intense events such as colliding with surfaces, while driving in grassy mountains, and in menus
BF1: 80-120, stutter in intense moments during campaign, starts after 3 matches in multiplayer (sometimes items in the spawn screen flash), and 144 in menus
Chivalry - Medieval Warfare: hovers around the 120 fps extended cap I set in the config file as I didn't like the choppiness of 60 fps since i've gotten used to the 144Hz of my screen but it seems to also micro-stutter
Day of Infamy: especially when I run or am in certain areas (right in front of the left bunker on the american side in Dog Red, for example). I can clearly see stuttering in this one because I can actually see it freeze and i've moved from where I was before when it "unfreezes" (lasts like a fraction of a second)
Minecraft: FPS drops to below 60 after a while. I've tried from playing with in-game settings, to allocated ram (up to 8gb) in launch instructions, and using Optifine. To no avail, but it seems to lessen the effect if I lower chunk render distance so I guess it's an issue related to chunk render distance
What i've tried as of now:
- Undervolting since I deemed voltage set too high on auto (1.375). At the very start I went as low as a -0.13125 offset and it seemed to still POST and get to the UEFI but it started losing display signal and black-screening, which interrupted my attempts to instal Windows. The solution was a hard-reboot via holding the power button. Worry not as I since then realized I was by far undervolting my CPU and did a clean install of Windows in case these interruptions could have caused OS corruption during installation or any kind of corruption in general.
- Several clean installations of Windows 10 Pro x64, cleaning the drive with Window's Diskpart utility during the installation before installing the OS and keeping my WD 1tb storage drive unplugged during the process
- Several Nvidia drivers (the newest 385.41 version seemed to stutter for a second upon startup and once had my GPU pegged on 100% clock speed when not running games for no apparent reason so I resorted back to the 384.94 default windows driver and I've had no issues that I can directly relate to this particular version since then)
- Temporary fix for stutter in BF1 (only one i've tried): high performance mode (It is always set on Ryzen Balanced Power Plan since I downloaded the AMD chipset driver it was packaged with)
UEFI settings that seemed to work and be stable on games (minus the micro-stuttering) and Cinebench before I started having problems with it:
- DOCP according to SPD determined by Team Group
- Many different Load Line Calibration settings (from auto to high) for CPU and SOC. Doesn't seem to help stability, only to slightly increase voltage and lower my Cinebench scores the higher I go
- CPU core ratio 38(x100) @ VDDCR 1.287v using a negative offset to the 1.375v auto setting my motherboard sets on its own. It used to work on 1.287v @3.8Ghz before I started experiencing freezing in Cinebench.
(I even tried AIDA64 for a couple minutes even though I don't feel like that kind of stability/stress testing would reflect my real-world usage of this computer and I didn't want to risk damaging my new CPU even more, plus different errors/problems can arise when performing different tasks)
In no particular order, here are possible causes for issues i'm having:
- Inappropriate UEFI settings (too low voltage?)
- Software corruption (Windows or installed programs)
- I Dropped the SSD on my table while taking it out from its packaging, only from a couple inches off the table at most though, not even 5 (Unlikely to cause issues. I asked a guy I know that works with and builds servers and he said apart from the read/write limit SSDs have (in other words, their lifetime), they either work or don't. Now, I know I can't necessarily take that for granted, but I find it logical to think so considering our collective technical know-how)
- Possible bent CPU pins (unlikely, but it is still a possibility since I reinserted the CPU to make sure it was properly seated in the socket, being as careful as possible of course. But, I don't see why it would give me the issues i'm encountering. I assume the PC just wouldn't turn on or would fail to POST). I can check for bent pins if it is deemed necessary.
- Corrupted installation media (I installed Windows 10 Pro x64 from a bootable-USB I made using Microsoft's media creation tool. Even though I did this a couple of times along with a couple Win 10 clean installations, I assume this couldn't really be the issue since the tool reformats the USB every time)
- UEFI versions (Nothing seems to be out of the ordinary except for Cinebench and the games I tried--especially not in the UEFI, except for the temp and voltage "issues" I mentioned I was having at first--but I was wondering if certain parts of a UEFI version could remain even when you flash to a newer one because the issues with Cinebench started when I flashed to 0811 and even remained when I reflashed to 0809. This is unless this is not all related to a software issue caused by a windows update made in the background or the CCleaner 5.33 backdoor virus I mentioned earlier)
I ordered my parts on (August 12) and my computer was up and running on (August 26). I wanted to pinpoint the culprit(s) of the issues i've been having before the end of my return window, but unfortunately, all this testing plus me being busy with school and life in general has made it impossible for me to do so. I am more than a month in but I can still rely on manufacturer warranty so I was wondering if you guys could help me find out what the problematic variable is (i'm thinking a defective CPU, but then again i'm wondering if it can show symptoms of being degraded like it seems to be or if simply would not work. This is an answer i'm having trouble finding on the Internet)--if applicable--so I can proceed to RMA if necessary. Also, I don't know if he will be able to help me or not, but I will possibly be getting my PC checked out by somewhat of a professional in the matter so I'll update you guys on that if necessary.
While I know I haven't tested every possible variable in every possible instance (which is impossible to do anyway), i've done my best to troubleshoot everything as thoroughly as I could and I feel like something isn't normal here or at least as it should be. Also, I know it's not always an easy plug 'n' play experience with PC-building but I think with the amount of time and money (especially with money totalling around $2k CAN with the screen and keyboard) i've spent, it should be at least a bit moreso than it is now. It would be fun to be able to use my PC without worry of damaging something or that some problem is lurking right around the corner every time I try something new so I am ready to do what it takes to be able to use it like i've been wanting to for a bit more than a month now.
I have a couple more things I have up my sleeve, such as setting everything back to auto/optimized defaults (although reluctantly) as if something happens (such as burning my CPU out due to excessive voltage) it is not my fault since logically ASUS set those defaults in accordance to AMDs recommendations and I am still under warranty. I will also once again do a clean install of Windows just for good measure.
First and foremost, here are the specs of my build: https://ca.pcpartpicker.com/list/pcJbM8
Note: take as a premise that i've been researching all matters of PC building and overclocking on and off since 2014 (when I started contemplating the idea of building my own PC) and more intensively since the last 2 months. Even though it doesn't exclude the fact that I don't have absolute knowledge, I like to be thorough in everything I do so I consider asking for help here as a last resort after doing my own research and looking at forums as I understand it can be difficult to figure out what is going on from a distance. TL;DR: i'm a novice--yes--but i've done my homework (maybe not all of it, but what I could) and I know what i'm doing.
While I initially had certain issues that I imagine at least a certain portion of first-time PC builders like myself encounter (atleast from what I assume and what i've observed on forums concerning the matter), I was able to overcome them and get to the point where I am now (which means getting windows, other programs, and games--at least partially, lol--to run)
Issues i've been having:
My first concern was with temperatures, which seem to always have hovered around 40-45 in the UEFI no matter the version (those temps aren't alarming but still warm for idle. I figure this is due to the offset applied to X Ryzens either by Asus or AMD themselves, and i've seen people get higher temps in their BIOS than when idling on Windows, which I also observed myself by looking at temps on HWMonitor while idling on desktop later on. Although i'm still not 100% certain what temps to look at with all the confusion on the Internet, the package and TMPIN temps all seem to stay below 40 and package sometimes goes down to the 20s for a second on idle and i've observed an around mid-60s maximum during a Cinebench test).
So after doing some research and stressing out about this for a while, I finally decided to install Windows 10 and convinced myself my system was sophisticated enough to protect itself and shut itself down in the event of excessive temps. I initally experienced display signal-loss/ black-screen issues (which I describe later on below) when I first tried to install Windows but this was due to a too-drastic undervolting (-0.13125 offset, to be exact, in fear that my motherboard was overvolting my CPU with voltages of 1.4v and up, which it goes up to from its 1.375 default setting) on my part, which I corrected by recovering optimized defaults. Rest assured I proceeded to reinstall Windows after to avoid any chance of corruption due to these issues I experienced during my first installation.
After the Windows installation (plus updates) and then installing drivers one by one with a restart in between each (see the issue I had with Nvidia driver 385.41 at the start below, which is not an issue anymore for some reason) was done came my growing concern with voltage, which my Asus ROG Strix B350-F sets at 1.375 (a voltage that i've heard Ryzen can handle but a bit on the high side and close to the threshold at which you start to observe degredation after a while, according to me and what i've seen on the interwebs); that is, as of BIOS 0809. Unfortunately, I don't remember which UEFI version my board came with as I almost flashed right away after testing to see if it POSTed.
The thing is when I leave my UEFI settings on optimized defaults, it sets my Vcore--or my VDDCR, as they call it--on 1.375 auto but I can see it hover around 1.4-1.45 in my UEFI hardware monitor (screenshots of my UEFI screen opon request), which I can also see momentarily spike up to in HWMonitor on Windows. I know it is normal for CPUs to ramp up and down due to cores waking up and falling to sleep/different states but unless it is normal for it to do that, it means my MB is clocking my CPU up to 4Ghz on at least 2 cores (according to AMDs XFR/Precision Boost Technology) and also while idling on desktop with only HWMonitor open, which it shouldn't and is a bit excessive, in my opinion.
When I set VDDCR to 1.2875 using a -0.0875 offset from the standard 1.375, it can go down to around 0.5v when completely idle and nothing is open, but as soon as I open a program, I rarely see it go down to even 0.8v (i'm not sure if this is a problem) and often goes higher than 1.3v (I've seen it go up 1.363 when running games or a Cinebench test). Maybe i'm not looking at the right voltage since there are many voltages (and temps) to look at in HWMonitor. Also, if I set a CPU core ratio of 38 and let it set a voltage on its own, it sets it to around 1.5v so that means I have to downvolt my CPU from default to attain 3.8Ghz without frying it.
- Since updating from CCleaner 5.33 to 5.34, Windows Defender detected a backdoor threat in the 5.33 installer .exe and somewhere in a ryzen-related directory (although I should have recorded the location and failed to do so due to the stress of having a threat on my newly-built PC on which I never did anything or went anywhere near where there was a chance I could have gotten a virus, it was probably a chipset driver related file). Also, pretty much right before I entered the OS I had flashed from UEFI v0809 to 0811, so basically I can't isolate what happened. When I got into Windows, I decided to perform a Cinebench test to see if my scores were still the same since I felt like they had lowered at some point (due to background programs, I had assumed) but they seemed to still to be on par with what I got before the Cinebench freezing issue started to appear.
- My computer sometimes fails to POST on restart when clocked to 3.8Ghz on 1.287v (either via restarting through windows or pressing the restart button on my case)
- The backlit writing on my Gtx 1070 doesn't light up evenly. Also, i've started experiencing coil whine (i'm pretty sure that is what i'm hearing) since a long session of BF1, although this doesn't seem to affect performance
I know saying this makes it look like my GPU is the culprit but what makes more suspicious about my CPU is the micro-stuttering in games and the recent freezing in Cinebench, which looks similar to when I tried installing Windows 10 while undervolting my CPU.
My Cinebench scores with corresponding UEFI settings are as follows:
*DOCP=Asus' term for XMP Profile
**CPB=Core Performance Boost
Completely stock (anywhere around 1.35v-1.45v with DOCP and CPB On): the highest I ever got was 1243
Stock with DOCP and CPB off (around 1.22v @3.6Ghz): 1150-1170
1.287v @3.8Ghz with DOCP on and CPB off: the highest I ever got was 1278
Game Performance:
All game cutscenes seem to have this kind of micro-stutter effect that makes objects twitch and look laggy in general (especially GTA V and BF1)
Nonetheless, I get pretty good FPS in the games i've tried.
Space Engineers: 90-120 on average but it dips to below 60 when I crash a ship or when it loads terrain on a planet
GTA V: 70-110, except for 55-60 fps lag spikes sometimes when changing camera angle rapidly, in intense events such as colliding with surfaces, while driving in grassy mountains, and in menus
BF1: 80-120, stutter in intense moments during campaign, starts after 3 matches in multiplayer (sometimes items in the spawn screen flash), and 144 in menus
Chivalry - Medieval Warfare: hovers around the 120 fps extended cap I set in the config file as I didn't like the choppiness of 60 fps since i've gotten used to the 144Hz of my screen but it seems to also micro-stutter
Day of Infamy: especially when I run or am in certain areas (right in front of the left bunker on the american side in Dog Red, for example). I can clearly see stuttering in this one because I can actually see it freeze and i've moved from where I was before when it "unfreezes" (lasts like a fraction of a second)
Minecraft: FPS drops to below 60 after a while. I've tried from playing with in-game settings, to allocated ram (up to 8gb) in launch instructions, and using Optifine. To no avail, but it seems to lessen the effect if I lower chunk render distance so I guess it's an issue related to chunk render distance
What i've tried as of now:
- Undervolting since I deemed voltage set too high on auto (1.375). At the very start I went as low as a -0.13125 offset and it seemed to still POST and get to the UEFI but it started losing display signal and black-screening, which interrupted my attempts to instal Windows. The solution was a hard-reboot via holding the power button. Worry not as I since then realized I was by far undervolting my CPU and did a clean install of Windows in case these interruptions could have caused OS corruption during installation or any kind of corruption in general.
- Several clean installations of Windows 10 Pro x64, cleaning the drive with Window's Diskpart utility during the installation before installing the OS and keeping my WD 1tb storage drive unplugged during the process
- Several Nvidia drivers (the newest 385.41 version seemed to stutter for a second upon startup and once had my GPU pegged on 100% clock speed when not running games for no apparent reason so I resorted back to the 384.94 default windows driver and I've had no issues that I can directly relate to this particular version since then)
- Temporary fix for stutter in BF1 (only one i've tried): high performance mode (It is always set on Ryzen Balanced Power Plan since I downloaded the AMD chipset driver it was packaged with)
UEFI settings that seemed to work and be stable on games (minus the micro-stuttering) and Cinebench before I started having problems with it:
- DOCP according to SPD determined by Team Group
- Many different Load Line Calibration settings (from auto to high) for CPU and SOC. Doesn't seem to help stability, only to slightly increase voltage and lower my Cinebench scores the higher I go
- CPU core ratio 38(x100) @ VDDCR 1.287v using a negative offset to the 1.375v auto setting my motherboard sets on its own. It used to work on 1.287v @3.8Ghz before I started experiencing freezing in Cinebench.
(I even tried AIDA64 for a couple minutes even though I don't feel like that kind of stability/stress testing would reflect my real-world usage of this computer and I didn't want to risk damaging my new CPU even more, plus different errors/problems can arise when performing different tasks)
In no particular order, here are possible causes for issues i'm having:
- Inappropriate UEFI settings (too low voltage?)
- Software corruption (Windows or installed programs)
- I Dropped the SSD on my table while taking it out from its packaging, only from a couple inches off the table at most though, not even 5 (Unlikely to cause issues. I asked a guy I know that works with and builds servers and he said apart from the read/write limit SSDs have (in other words, their lifetime), they either work or don't. Now, I know I can't necessarily take that for granted, but I find it logical to think so considering our collective technical know-how)
- Possible bent CPU pins (unlikely, but it is still a possibility since I reinserted the CPU to make sure it was properly seated in the socket, being as careful as possible of course. But, I don't see why it would give me the issues i'm encountering. I assume the PC just wouldn't turn on or would fail to POST). I can check for bent pins if it is deemed necessary.
- Corrupted installation media (I installed Windows 10 Pro x64 from a bootable-USB I made using Microsoft's media creation tool. Even though I did this a couple of times along with a couple Win 10 clean installations, I assume this couldn't really be the issue since the tool reformats the USB every time)
- UEFI versions (Nothing seems to be out of the ordinary except for Cinebench and the games I tried--especially not in the UEFI, except for the temp and voltage "issues" I mentioned I was having at first--but I was wondering if certain parts of a UEFI version could remain even when you flash to a newer one because the issues with Cinebench started when I flashed to 0811 and even remained when I reflashed to 0809. This is unless this is not all related to a software issue caused by a windows update made in the background or the CCleaner 5.33 backdoor virus I mentioned earlier)
I ordered my parts on (August 12) and my computer was up and running on (August 26). I wanted to pinpoint the culprit(s) of the issues i've been having before the end of my return window, but unfortunately, all this testing plus me being busy with school and life in general has made it impossible for me to do so. I am more than a month in but I can still rely on manufacturer warranty so I was wondering if you guys could help me find out what the problematic variable is (i'm thinking a defective CPU, but then again i'm wondering if it can show symptoms of being degraded like it seems to be or if simply would not work. This is an answer i'm having trouble finding on the Internet)--if applicable--so I can proceed to RMA if necessary. Also, I don't know if he will be able to help me or not, but I will possibly be getting my PC checked out by somewhat of a professional in the matter so I'll update you guys on that if necessary.
While I know I haven't tested every possible variable in every possible instance (which is impossible to do anyway), i've done my best to troubleshoot everything as thoroughly as I could and I feel like something isn't normal here or at least as it should be. Also, I know it's not always an easy plug 'n' play experience with PC-building but I think with the amount of time and money (especially with money totalling around $2k CAN with the screen and keyboard) i've spent, it should be at least a bit moreso than it is now. It would be fun to be able to use my PC without worry of damaging something or that some problem is lurking right around the corner every time I try something new so I am ready to do what it takes to be able to use it like i've been wanting to for a bit more than a month now.
I have a couple more things I have up my sleeve, such as setting everything back to auto/optimized defaults (although reluctantly) as if something happens (such as burning my CPU out due to excessive voltage) it is not my fault since logically ASUS set those defaults in accordance to AMDs recommendations and I am still under warranty. I will also once again do a clean install of Windows just for good measure.