Suggestions for my recent problems with my 5950X

[Mpegger]

...and it still didn't capture w/e was causing the 'spike'. It reads the UPS... the UPS 'caught' it, but no devices are out of line power-wise. I wonder if you left Hwinfo running for a lot longer, that it would catch one of those spikes? Like the GPU says 500/600W.... or something? To me, you're missing a couple of hundred W somewhere to make that come true... I don't know. I believe you're seeing 600W or so (500W between CPU and GPU and 100W for the system) actual system use. But even adding inefficiencies, fans/HDDs/pump etc., just doesn't add up (and it could all be in my head) to 1.1KW.

I agree, which is why I added in the Kill-A-Watt to get a 2nd reading. I'd also add in a clamp power meter, but I don't currently have a spare power cable I'm willing to cut up so I can clamp the hot wire and see with a third measuring device (I plan on making a cable of my own for this once I get the RaspverryPi monitor in).

As I said above, the Kill-A-Watt is always within ±20'ish Watt's of the UPS. UPS shows 600W, Kill-A-Watt shows 600W. UPS shows 900W, Kill-A-Watt shows 900W. I'm just as boggled by this as again, I agree with you that this system shouldn't use more then 650-700W from the PSU (740-800W from the wall), but both meters lying?

Although, I do have to say that the MB taking 200W+ is realistic to me. 5W on a X570 chipset? When it requires a dedicated fan of its own because it can easily pass 80C on a chip of that size without one, there is no way its just using 5W.

 

[freeagent]

What happens when you don't use the UPS?

 

[EarthDog]

Although, I do have to say that the MB taking 200W+ is realistic to me. 5W on a X570 chipset? When it requires a dedicated fan of its own because it can easily pass 80C on a chip of that size without one, there is no way its just using 5W.

Look it up. I dont think it was over 10W. X470 was around 3W IIRC. Intel z690 chipsets are 6W.

Intel® Z690 Chipset - Product Specifications | Intel

Intel® Z690 Chipset quick reference with specifications, features, and technologies.

www.intel.com

I had a 5W mining ASIC that burned my fingers without a heatsink... it doesn't work like you're thinking. A yellow flame on a lighter is the same temperature as yellow flames in a huge bonfire... which has more energy and more difficult to cool?

Worth noting, x570 didn't really need a fan in the first place. All the 2nd revisions don't have any and the heatsinks aren't really any beefier. It'll take a bit more than a 2" around 1/4" thick piece of metal with some ridges on it to cool 200W IC... nor can a cpu be cooled in the same manner... something that actually outputs 200W.

EDIT: For clamping meters, don't you put the cord between the clamps? I didn't think you needed to skin it to get W readings off them.

 

[freeagent]

There is no way your board alone is drawing 200w. FWIW running F@H on my CPU and GPU, both overclocked, can do 640w on my meter. Do you know how hard it is to try and cool 200+ modern watts? It is intense. At least 7nm.. Not like trying to cool 200 old school watts. I would try without the UPS first, if that doesn't help, I would take a closer look at whatever tune you put on the system.. specifically memory. And a closer look at your VDDG's and SOC.

 

[Johan45]

Although, I do have to say that the MB taking 200W+ is realistic to me. 5W on a X570 chipset? When it requires a dedicated fan of its own because it can easily pass 80C on a chip of that size without one, there is no way its just using 5W.

Here's some actual X570 chipset testing https://hothardware.com/news/amd-x570-chipset-voracious-power-consumption-compared-x470





As for the 5950X unless you were under extreme cooling that CPU would max at 200 W or less using P95.
Here's our review of the 6900XT https://www.overclockers.com/msi-radeon-rx-6900-xt-gaming-x-trio-review/
Max system is under 600W using a 10900K which is thirstier than the 5950X.

 

[EarthDog]

Btw, the Kill-A-Watt is connected directly to the PC, after the UPS.

What happens when you don't use the UPS?

This. ^^

Like we said previously, take the UPS out of it and just put the PC on the KaW...let's see if the system w/o the UPS still spikes. I know it updates at once /s but, you shouldn't see huge spikes to 900+ W with a system that's actually pulling 550-600W max.

For my 12900K/3080Ti system (more wattage on paper than 6900XT/5950X), I happily run on a 850W unit. Last I checked, the system pulled around 500-550W at the wall when gaming (PUBG). It ran on a 750W unit, but I found a 'cheap' 850W Titanium upgrade so I jumped on it.

 

[freeagent]

We aren't prone to brownouts here, but they do happen from time to time. If you don't need the UPS, don't use it. It doesn't seem to be able to keep up with the action anyway, so set it free.. I do use a line conditioner on my main rig, but my second rig is just plugged into a nice power bar. Works great. Good luck!



Edit:

I like that ED, a "cheap" 850 Ti unit.. I have been on the lookout, I did notice steep PSU price drops all around..

 

[EarthDog]

IIRC, I got it on sale for ~$150? The normal price is $249 or something?

Seasonic PRIME GX-750, 750W 80+ Gold, Full Modular, Low Noise, Premium Japanese Capacitor, 12 Year Warranty, Nvidia RTX 30/40 Super, AMD GPU Compatible, SSR-750GD2 - Newegg.com

Buy Seasonic PRIME GX-750, 750W 80+ Gold, Full Modular, Low Noise, Premium Japanese Capacitor, 12 Year Warranty, Nvidia RTX 30/40 Super, AMD GPU Compatible, SSR-750GD2 with fast shipping and top-rated customer service. Once you know, you Newegg!

www.newegg.com

 

[freeagent]

That's a 300 dollar PSU up here in Canukistan. Newegg.ca is showing 100 bucks shipping on it which at that point might as grab the KW unit with 7.99 shipping lol. It should be free to 8 bucks to ship normally. I do see they have a GX850 for 135 But I have a brand new EVGA 750 Super Nova driving my 5600X rig, I was forced to buy it with my 3070 Ti last summer. Or else I would scoop it right now. I almost should anyways, just because 750 is a bit thin at the top end with this rig. I am just a few 3.5 HDD's away from the top lol

 

[JLK03F150]

I'd also add in a clamp power meter, but I don't currently have a spare power cable I'm willing to cut up so I can clamp the hot wire and see with a third measuring device (I plan on making a cable of my own for this once I get the RaspverryPi monitor in).

I have a splitter for taking amp readings from a cord without the need of ruining a cord. It has a 10x multi to get the reading into a higher, more accurate range of the meter. IMO anyone that owns a clamp-on amp meter should own one of these too.

 

[Mpegger]

So, I finally got around to wire up my own plug for a clamp meter. I power down and disconnect the PC, throw in my wire so I can use my clamp meter, get a tripod to film the meter annnnddddd...

Now suddenly nothing is showing more then 900'ish Watts being pulled from the wall.

I tried throwing Cinebench all threads alongside MSFS 2020, and it was a steady 6.8xA on the clamp meter, 816W from the wall, 718W from the PSU the majority of the time. I tried Furmark along with the CPU coin miner, Rust, VMs running in the background, all combinations of loading my CPU and GPU as much as possible, and was only able to get a handful of spikes in the low 900Watt from the wall (which is only around 820Watts from the PSU). Between previous testing and this round of testing the only major thing that changed was that I powered down the PC to include the plug for the clamp meter, so I have no idea why I'm no longer seeing 1kW+ spikes on anything.

Still, 718W steady still is more then expected from this system, and those 820W spikes this round of testing are still pretty high imo for a near 8 year old PSU.

I'm going to be spending this week slapping on the water block for the GPU and redoing the water loop, as well as switching out the PSU for the Corsair RM1000x I now have, so no more testing until after all that is done. And of course, with a change of PSU, it wouldn't be the same results, or at least I wouldn't expect the same results, but we'll see.

 

[EarthDog]

....seems like the UPS isn't guessing output very well, lol.

718W is a lot more than expected out of that system. Are you measuring the UPS cord or PC only?

Really weird how you're pulling more wattage than say... my system which uses more W........ on paper. A factory overclocked 3080ti and 12900k... lol

 

[Mpegger]

....seems like the UPS isn't guessing output very well, lol.

UPS, Kill-A-Watt and the clamp meter. Again, the Kill-A-Watt already read around ±20W of the UPS, but since the UPS samples and updates 4x faster then the Kill-A-Watt, it could show those transient spikes (and dips) that the Kill-A-Watt wasn't showing. Just for some reason now everything seems to stay at a steady average and isn't jumping all over the place like it was before. I know what I saw, and it's been like that for weeks (months now), but like always when one tries to actually show the issue, that's when things seemingly go back to "normal".

 

[EarthDog]

Or.... you never saw what you thought you saw....

(kidding course!)

I edited a bit up there, note.

Have you taken the ups out of the equation to measure... the kaw, or your meter? I know a couple of us were wondering what readings were like that for a while now.

 

[Mpegger]

The readings are the same regardless if the UPS is inline or not. So no, the meter on the UPS is not reading wrong if that's what everyone is thinking, it's just not accurate to the single/double digit because of how its been programmed as I described in a previous post, but it's still somewhere around 3% tolerance, much like my clamp meter.

 

[Mpegger]

Well now, here is something interesting.

I have not seen the UPS display high wattage readings till the other day when I was playing a game that hits the GPU hard, and I alt-tabbed to my browser. At one point when I alt-tabbed, the UPS suddenly started beeping, which normally means there is some kind of problem (I thought a brown out as that's what happens in this neighborhood irregularly), but when I looked at the display, the wattage was jumping all over the place, from 0 watts, to 999 watts, and everything inbetween. When I alt-tabbed back to the game, the wattage stopped jumping all over the place and went back to a "normal" reading in the current gaming state (around 650ish watts). Alt-tab back to my browser, al the wattage goes wild again. Alt-tab back to the game, wattage reading goes back to normal.

This isn't the first time I alt-tab from a game to my desktop. In fact, in this particular game I do so regularly. But this is the first time this has happened at all.

The UPS is a old unit, and I do have it plugged into the computer via its USB cable for monitoring and shutdown, but this only started happening about 2 days ago, so I haven't a clue what the issue may be, other then me needing to reboot my system (44 days, 20 hours+ since last reboot ).

 

[Mpegger]

For anyone still watching this thread, I think the PSU might be going bad, though nothing seems out of the norm with it. Voltages measured and reported are within the norm being supplied to the system, but for some reason the wattage being pulled from the outlet just keeps slowly getting higher and higher with weird spikes all over the place. A very brief look inside through the vents and I don't see any bad capacitors, nothing looking burned or cooked, nor any such smells to indicate there is overheating going on. Using a different psu and the wattage being pulled form the outlet are lower and not jumping all over the place, even while gaming. I'll need to open up the PSU and take a closer look + somehow test it underload.

 

[EarthDog]

Try using a more consistent load and see how it behaves. You're getting a vague idea, but gaming isn't a steady load so that can blur things a bit.