Seeking help with weird slowness issue, Ryzen 7 3700X

M

mvmiller12

[H]ard|Gawd
Joined
Aug 7, 2011
Messages
1,523
The setup: Crosshair VII Hero WiFi on BIOS 0503, Ryzen 7 3700X CPU, Radeon 6800XT GPU, 16G of Corsair DDR4-3600 RAM, 512G m.2 Samsung 970 Pro SSD, 4TB HDD (just data storage), 850W Seasonic Focus Gold PSU, Asus 1080P IPS Monitor using DVi cable and DVi-to-Displayport adapter, Windows 11 Pro. The video card and CPU are custom loop liquid cooled with 2x240 rads with everything in serial, flowrate is plenty good. CPU temps idle at 35c in Windows, 42c in BIOS. GPU is similar

The problem: The computer sometimes out of the blue becomes extremely slow. Sometimes it works fine, sometimes for no known reason, it just goes to shit. As in, laggy, jumpy and unresponsive. EVEN THE BIOS -- As in I can literally watch the screen refresh when switching pages in the BIOS. Power cycling and unplugging the system for a few minutes does not effect this when it decides to be an ass. BIOS has been cleared to no effect. BIOS has been updated to 0503 using BIOS Flashback, to no effect. This happens at stock from CMOS settings.

Other than pulling the system entirely apart and testing each bit individually (I have multiple AM4 systems so I have parts, but I'd rather not tear one of them down unless I have to), I am unsure what else to try or what I should be looking at.

Anyone here want to pipe in on this?
 
DVOM into an unused molex connector to monitor psu voltages and maybe HW info to monitor vrms and cpu/gpu throttling. Maybe sub a cpu in there if nothing appears out of order. Just sounds like some sort of throttling occurring.
 
Failing hard drive can cause this. Would check the 4tb hdd first. Secondly, I would check USB devices connected to the PC. :)
 
No WHEA errors in event viewer?? Soc voltages etc all within spec??
 
The thing is, it does this in THE BIOS SCREEN. Like I can literally watch the screen redraw from top to bottom when I switch pages in the BIOS. Frequencies, Voltages and temps all read as normal.
 
Cmos battery dead? Had this only once in many years of fiddling with pc's and it made it crawl through initialization of every onboard device. Worth checking anyway.
 
Verify SSD firmware is latest version. Unplug HDD (to rule out a SATA/cable issue). Unplug all USB devices except keyboard and mouse. Try different keyboard. Try different mouse. Try unplugging keyboard or mouse when lag starts. Unplug network cable when lag starts. Try different monitor. Try different video cable. Try without DVI-to-DP adapter. IMO the lag is due to an I/O issue, be it excessive interrupts or lag while processing the interrupts.
 
mvmiller12 said:
The thing is, it does this in THE BIOS SCREEN. Like I can literally watch the screen redraw from top to bottom when I switch pages in the BIOS. Frequencies, Voltages and temps all read as normal.
Click to expand...
Which is why anything I/O related is suspect, such as SATA cabling, networks, USB devices, etc. BIOS coding isn't usually too sophisticated as far as I/O is concerned, and if you're seeing an I/O related hang, it's quite likely to show up at the BIOS level.

Have you tried swapping in a different GPU temporarily? might be worth a shot, and it's not as annoying a swap as say RAM or PSU.
 
mvmiller12 said:
The thing is, it does this in THE BIOS SCREEN. Like I can literally watch the screen redraw from top to bottom when I switch pages in the BIOS. Frequencies, Voltages and temps all read as normal.
Click to expand...

Like I said, I went through the same process with an older system, I want to say it was a 4790K. I tried all sorts of BIOS stuff, updates, etc. In the end I stripped it out of the case and ran just the board, CPU, one stick of RAM with the igpu on top of a box to isolate things one at a time. It had an m.2 boot drive that worked perfectly and then when I plugged in the SATA drive it just stopped posting by that point, drive totally failed.
 
Update: I have pulled the components out for bench testing - this means no SATA HDD and an entirely different Corsair 750W power supply. At first, it worked great but then it started acting up again - I noticed the mainboard was a bit warped around the CPU socket, so I loosened the water block's death grip a bit and the problem got worse. This may or may not be related as I moved the board around a few times on the bench. I may actually have more than one problem component here.
 
mvmiller12 said:
Update: I have pulled the components out for bench testing - this means no SATA HDD and an entirely different Corsair 750W power supply. At first, it worked great but then it started acting up again - I noticed the mainboard was a bit warped around the CPU socket, so I loosened the water block's death grip a bit and the problem got worse. This may or may not be related as I moved the board around a few times on the bench. I may actually have more than one problem component here.
Click to expand...
pull off the aio and chip, then check the pins and contacts. make sure they are clean and pins are straight, and then reseat everything.
 
Update: I believe I have discovered the source of my trouble.

My son's 6800XT (and the Vega 64 before it) does not fit in his case when mounted traditionally. So we opted for a vertical mount (which his case has the slots for) and got a PCIe 3.0 x16 slot ribbon cable to accommodate the card. I did my initial testing just before bed last night after pulling the whole rig apart, so I didn't spend very long on it. I initially tested without the slot cable and the system worked fine, and then I put it in... and it still worked fine -- all the way up until it didn't. It worked fine for a few minutes and then suddenly the slowness reappeared and eventually got so bad that the computer would not even POST anymore. I came back this afternoon after work to the same NO POST situation and the qcodes on the mainboard display were all not listed in the Asus manual - but it was consistently hanging with the white LED light on. Sometimes a BIOS reset would bring it back for a minute before it failed again with the terrible slowness or no POST condition. So I removed the cable again and it booted right up like it didn't have a care in the world. And it kept booting up just fine. Re-added the cable one last time, making sure it is definitely seated properly. Problems immediately came back.

Um.... I think that PCIe 3.0 x16 slot cable might be bad :) I should have suspected this was the case when his Vega 64 started doing this same behavior (well, the slowness anyway - the Vega NEVER refused to POST). He had suspected that the card was failing because it is old and had been in a few builds before he ended up with it, coupled with the fact that he wanted an upgrade anyway had led him to replacing it with the 6800XT. Diagnostics have been fine so far since plugging the card directly into the mainboard so I'm going to reassemble the computer tomorrow and order a new ribbon cable and call this one solved.

I have NEVER in 30+ years of being a PC enthusiast and industry professional experienced (and, mind you, I have seen some weird shit) this sort of failure before. TIL indeed.
 
Hopefully the new cable will fix it electronics get weird when the impedance/inductance of circuit is changed particularly at higher frequencies.
 
mvmiller12 said:
I have NEVER in 30+ years of being a PC enthusiast and industry professional experienced (and, mind you, I have seen some weird shit) this sort of failure before. TIL indeed.
Click to expand...

Yeah this is why we suspected the SATA drive. Something filling the bus with interrupts and bad data causes this kind of crap, PCI-E riser cables are quite prone to failure especially the cheap unshielded ones. I didn't realize you even had one or that would have been my first guess, I've seen a number of those go bad in ITX cases that flip the GPU onto the back.

At least you got it figured out now.
 
His build in that case initially used a reference Radeon RX480, and if you recall, those cards are really short. When he got the Vega we had to get the PCIe riser cable and switch to vertical mount because the last <1/4" of the card was physically blocked by the pump and reservoir. The 6800XT is a hair longer even than that. I ordered a Phanteks PCIe 4.0 rated one as a replacement. It has the wires separated into narrow bundles which should allow a little greater flexibility and they appear to be well insulated. Also, since his Crosshair VII Hero WiFi board tops out at PCIe 3.0, the cable is well within spec. Frankly, that 6800XT is dramatic overkill for his 1080p60 monitor but he got a good price on it ($440 shipped from a seller here on the [H]) and is very happy with his IPS monitor. Shit, the Vega 64 was nearly overkill for 1080p60 :)
 
You must log in or register to reply here.
Back
Top