PC Upgrade Issues
-
A few days ago I picked up a Ryzen 9 5900X for a good price and decided to throw it in my system. However, I immediately started running into issues. I was experiencing an immediate shutdown, like it was hitting some kind of power limits; My first thought was the PSU, which was a Thermaltake Smart 700w (notoriously unreliable unit), so I swapped it out with a Corsair RM1000X that a friend had in his old build.
This didn't fix the issue. I put my memory into a friend's system and ran Memtest86 overnight and my memory shows no issues whatsoever. It seems to power off when one of 3 things are happening: Booting windows (only sometimes), Doing something that puts a heavy load on the system (loading into GTA V), or, bizarrely, running a memory or power test in OCCT.
I don't think it's GPU related, but I also don't have a spare one on hand to verify that theory, and none of my friends have a spare one to loan to me for a day. I'm pretty much just down to either the motherboard being faulty/not up to the job, or maybe there's something I haven't thought of. Anyone have any thoughts?
In Lucas we (don't) trust. All things Saab, old British cars, and general shitboxes.
-
@black-villain What type of motherboard are you running, and did you update the bios?
-
@cb Yes, Bios is on the latest version from Asrock. It's... one of their lower end B450 boards I got for free when I built my system a couple of years ago (B450M-HDV R4.0), so it wouldn't surprise me if it just can't handle it. However, the chip is officially listed as supported on their site for the board, so I figured it would at least be enough to run at stock speeds for a while until I got a new board.
In Lucas we (don't) trust. All things Saab, old British cars, and general shitboxes.
-
@black-villain So, you replaced the CPU, things started to get extra crispy, and you fired the parts cannon to try and fix the issue? Have you tried putting the previous CPU back in?
"I drank what?" -- Socrates
-
@black-villain said in PC Upgrade Issues:
@cb Yes, Bios is on the latest version from Asrock. It's... one of their lower end B450 boards I got for free when I built my system a couple of years ago (B450M-HDV R4.0), so it wouldn't surprise me if it just can't handle it. However, the chip is officially listed as supported on their site for the board, so I figured it would at least be enough to run at stock speeds for a while until I got a new board.
There's the issue, probably.
-
@aremmes Yup, first thing I did. It runs fine at stock speeds with the old processor. Kinda why I thought my old PSU just wasn't up to the job. Now that that has been ruled out, yeah it's probably just the board.
-
@cb Well, looks like that board upgrade is coming sooner rather than later lol. Might pick up an Asus WS Pro X570 Ace this weekend
In Lucas we (don't) trust. All things Saab, old British cars, and general shitboxes.
-
@black-villain Yeah does sound power related, but honestly both supplies you tried should be good enough. If the board says its supported, it should work, but again who knows. I assume you've tried resetting the BIOS to default settings? I'm not familiar with OCCT, does it run within Windows? If not can you run memtest86 on your system from a liveUSB? Does a Linux liveUSB distro conk out if you stress it? How are the thermals, anything getting excessively hot? Are you able to boot the system without a GPU, and does it stay up that way (ping it from a laptop or something)?
-
First idea: try a B550 motherboard
Second idea: try RGB ram -
@facw Yeah, UEFI/Bios has been reset multiple times, no change. OCCT does run within windows, I had HWInfo running a log while I was testing stuff to see if it could catch something before it crashed. I did run Memtest86 from a liveUSB and it crashes on my system after about 20 seconds. Tested my memory in a friends system and it passes. My linux drive seems to work better than windows, but I can still get it to crash if I push it.
Thermals seem to be fine, I even underclocked the CPU to ~2.3ghz to see if that improved things, and although it did, I was able to crash it when putting it under a synthetic load.
Haven't tried booting without a GPU, but I may try that and see if it changes things.
In Lucas we (don't) trust. All things Saab, old British cars, and general shitboxes.
-
@black-villain Stupid question, have you verified all the power connections on the mobo are fully seated?
and the ones on the PSU itself?Some kind of sentient shark, usually doesn't bite
-
@speedy964 Yup, everything is seated on both the PSU and board end. Where/how it shuts down also has not changed when switching between the two PSU's
-
@black-villain on this note, I have a MSI b450 board that in reviews by people who know what they’re talking about they mentioned the mosfet arrangement was great, but not for the highest end Ryzen 9’s and the mosfets would just be driven too hard.
What’s your cooling arrangement and is there sufficient airflow to the mosfets? (Do they even have sinks on them or just open to the air?)
‘22 Ridgeline
‘78 mini
Automation: 1.5 wins, 12 podium, 6 middle, 5 bottoms -
@black-villain unlikely to get far without a GPU as it wont have onboard graphics
are you using the stock cooler? if so it could potentially be doing a thermal shut down, particularly if the thermal paste application wasn't great
-
@grindintosecond said in PC Upgrade Issues:
What’s your cooling arrangement and is there sufficient airflow to the mosfets? (Do they even have sinks on them or just open to the air?)
Wraith spire downdraft, so it's definitely getting airflow onto the board.
@samv8 said in PC Upgrade Issues:
are you using the stock cooler? if so it could potentially be doing a thermal shut down, particularly if the thermal paste application wasn't great
Using my old Wraith Spire (until my Dark Rock Pro 4 arrives tomorrow), which I know isn't adequate enough for the 5900x, but should at least get it past windows booting I would think. Or at the very least it would downclock itself and throttle before initiating a thermal shutdown. I've been watching the temps in HWinfo while using the system when it's on and it never gets near hot enough to where it should shutdown.
In Lucas we (don't) trust. All things Saab, old British cars, and general shitboxes.
-
@black-villain Can you monitor mobo temps with HWinfo? As @Grindintosecond said, the mobo mosfets might be getting hot, even with the air flow. Worth a check at least
'89 Civic Wagon RT4WD, '01 Miata Special Edition, '00 Honda Insight 5Speed
-
@civicwagonengineer Unfortunately no, HWinfo can't read my Mobo temp for some reason, it just reads as 106C regardless of load or which CPU is in or whatever, so I'm pretty sure it's just using a default value since it can't read it
In Lucas we (don't) trust. All things Saab, old British cars, and general shitboxes.
-
This post is deleted! -
@black-villain looks like it's a known issue with HWinfo and Zen3. I would still say mobo is the issue since it's basically the one thing you haven't swapped out, but tough to say for sure.
'89 Civic Wagon RT4WD, '01 Miata Special Edition, '00 Honda Insight 5Speed
-
@black-villain on that mobo I now run a top down fan from a 3700x on a 1600x (a 2nd gen unicorn version). That style helps a lot.
-
@civicwagonengineer said in PC Upgrade Issues:
@black-villain looks like it's a known issue with HWinfo and Zen3. I would still say mobo is the issue since it's basically the one thing you haven't swapped out, but tough to say for sure.
Yeah, I have no idea. The system seems to be working fine with my old CPU in. I've tested the memory on a different system and it passed, the PSU is a known-good RM1000X (overkill enough to where it shouldn't hit any limit), I swapped out the gpu with an old basic display adapter and it's still happening with the new cpu... The only thing that has not been swapped out or verified is the motherboard.
In Lucas we (don't) trust. All things Saab, old British cars, and general shitboxes.
-
@black-villain Since the zen launch there has been mobo issues, so it makes sense. I wouldn't even be surprised if the mobo is actually good, just not fully compatible with that new of a CPU
'89 Civic Wagon RT4WD, '01 Miata Special Edition, '00 Honda Insight 5Speed
-
@civicwagonengineer Looks like I'm doing a mobo upgrade a little sooner than expected lol. Thinking about an Asus Pro WS X570 Ace. Oh well, got a fantastic deal on the CPU so I'm still out ahead
-
You could install the AsRock softwares to have the temperature infos on the mosfets or maybe AIDA64 does support your motherboard.
The memory on Ryzen is very important, i would definitely check the RAM voltage and push it a little if it's low (DDR4 default spec is 1.2v but most sticks have an XMP profile at 1.35v and might not run well if ran at 1.2v) and do the same for the VCore SOC (which is managing the memory).
Another possible issue can be poor voltage management, for example on my Gigabyte X370 Gaming 5 (it's a model known to be crippled with bugs), if i set a specific CPU voltage, it usually won't result in having the right voltage and it won't be kept if i go into sleep mode and wake the computer, resulting in the computer freezing frequently with the overclocking still in place so i have to set a relative value to the default to end up with something good (and definitely check while doing a load test if it's still within the right bounds).
-
@jb-boin said in PC Upgrade Issues:
You could install the AsRock softwares to have the temperature infos on the mosfets or maybe AIDA64 does support your motherboard.
As far as I know, this specific board doesn't actually have a temp sensor for the VRMs or mosfets or anything. It reads the same values at HWInfo, regardless of if I have them covered up with a piece of paper, fan blowing directly on them, or whatever. It pretty much just reads 38C/100F at all times, even when using Asrock's software
The memory on Ryzen is very important, i would definitely check the RAM voltage and push it a little if it's low (DDR4 default spec is 1.2v but most sticks have an XMP profile at 1.35v and might not run well if ran at 1.2v) and do the same for the VCore SOC (which is managing the memory).
Yeah, it's sitting at the default 1.2v right now. The thing is, I'm running the same ram settings on my old CPU, have been for about 2 years, and haven't had any issues from it. Would switching to 5000 series really make it want more voltage?
Another possible issue can be poor voltage management, for example on my Gigabyte X370 Gaming 5 (it's a model known to be crippled with bugs), if i set a specific CPU voltage, it usually won't result in having the right voltage and it won't be kept if i go into sleep mode and wake the computer, resulting in the computer freezing frequently with the overclocking still in place so i have to set a relative value to the default to end up with something good (and definitely check while doing a load test if it's still within the right bounds).
I tried messing with the voltages yesterday and eventually gave up after a couple of hours. Only way I could keep the system decently stable was to underclock it to about 2.3Ghz and then it seemed fine for a while. Tbh my current board is known to be a piece of junk, and is listed as not recommended for anything over Ryzen x600X from some community/forum posts I've found.
In Lucas we (don't) trust. All things Saab, old British cars, and general shitboxes.
