Question Why does TDP and PPT differ, on consumer CPUs? And what role does Core Performance Boost and Turbo Clocks have on TDP and wattage?

[VirtualLarry]

Serious question. I've got a 65W-rated TDP Ryzen R5 1600, in a rig, on a 65W-rated AMD stock heatsink. It's blue-screening, crashing, and the CPU temps just keep going up and up.

I updated HWMonitor, and it's showing a "Package Power" for the entire chip, at 82W or so. No wonder it's constantly overheating and crashing. 82W TDP CPU > 65W TDP Heatsink.

The worst part is, this is AFTER limiting the number of PrimeGrid threads, down from 12 to 9. That's right, I'm not even running the CPU at a full thread load.

Edit: Yes, I know that the obvious answer, is to "get a better heatsink", and that the "stock heatsink" for the 1600 was the 95W TDP model. Which, at the time, was stated that AMD wanted to give users the ability to OC on the stock heatsink. Now I know that was a lie, it's because AMD CPUs (at least, the 1600), are NOT able to stay within their stated rated specs.

Edit: A slight update, very important, actually. My original premise for this thread, was that I *thought* I was using a 65W TDP-rated AMD stock Wraith Stealth cooler with my Ryzen R5 1600 CPU, and it was crashing, at "stock BIOS" settings, which includes "Core Performance Boost" on "Auto", which defaults to enabled, to allow "Turbo Clocks" (the 1600 has an ACT of 3.4Ghz). I was initially placing the blame on AMD for the fact that HWMonitor reported the "Package Power" as something like 82W, which I thought was overcoming the 65W-rated heatsink. As it turned out, I actually was using a 95W Wraith Stealth (copper-cored) in the first place. Yet, it was still crashing due to overheating of the CPU. Part of this was due to the heat load of dual GPUs mining, and part of it was due to using a case that had NO vents on top, no fan mounts, no rad mounts, nothing but a solid steel top, and only a single 120mm exhaust out the rear, combined with the fact that my PCs are in desk cubbies. They are open to the front, and have dual 120mm intakes and vented fronts, but that still wasn't enough to prevent the CPUs from slowly creeping up in temp, passing 95C, and crashing/restarting.

Thus far, I have split the two GPUs up, one per PC (same case, same type cubby, same EVGA 650W G1+ 80Plus Gold PSUs), and disabled CPB on both of them (one has a 3600 CPU, one has a 1600 CPU), and then also in Wattman, set the Power Limit for the RX 5600XT (which was a refurb, both of them) to -20%. Thus far, overnight, they seem to have stabilized at under 90C on the CPU, and haven't crashed.

 

[Thunder 57]

I just chacked and you are right. Howver, I still think you are way too close to the PSU limit, and WAY over the cooling requirments. Not to mention the video card modded bios, the motherboard setting that may be giving too much power/voltage and the reconditioned products are in question too. There are so many vartiables here as to the real power draw and stability, I still question you blaming it on the CPU. I have 18 boxes (2 of which are Xeon, and 6 EPYC) and I have never had the kind of issues you are talking about.

I wouldn't be so sure.

What is Power Supply Efficiency and Why is it Important?

What is Power Supply Efficiency, 80 Plus, Bronze, Silver, and Gold and why are they important specs when building your next PC?

www.velocitymicro.com

80 Plus is a certification designed to denote that a power supply is at least 80% efficient at 20%, 50%, and 100% loads. This means a 500W 80 Plus rated PSU would draw a maximum of 625W at 100% load.

Hmm, random forum post from 2008, or an article on 80+ from last year? I bet there would be a class action lawsuit if they advertised a rate they could never supply. Damn lawyers.

EDIT

Ha!, just realized I was proving @VirtualLarry right.

 

[Velgen]

One thing I am wondering about for that CPU Temp Limit is that applying that 20C offset Ryzen first gen had to temps onto it or is that after the 20C offset? No idea how that board reads out on that version if it still has the 20C temp offset or not. Know my ASRock X370 Taichis read the CPU temp with the 20C offset even in BIOS. So if that is 80C + the offset it wouldn't thermal throttle until 100C, but if that value was taking into account the 20C offset it should be throttling at 60C as read out from HWMonitor or any other CPU readout tool because they threw out the offset number I believe. Potentially anyway not 100% sure on that because the temp readouts on my X370 boards are all a little weird and don't feel like I can particularly trust what it's outputting.

Edit: Could also be the real temp value for that CPU Temp Limit, but who knows that whole offset on first gen Ryzen makes setting temp limits or fan curves a pain in the butt.

 

[VirtualLarry]

Good point on the temp limit offset. Insightful.

I guess I may have made another boo-boo. I never went into "Smart Fan 5" in BIOS, and manually set all of the fans to "full speed" rather than "normal", and based on CPU temp. So I did that, after flashing BIOS F31 (bridge), and F50a (AGESA ComboPI V1 1.0.0.4 B).

Disabled Core Performance Boost, but did NOT set cTDP and CPU Temp Limit this time.

 

[Thunder 57]

Well, I interrupted the mining, restarted, went into the BIOS looking for a Vsoc fixed voltage option, but when I got into the BIOS, there was a message that my BIOS settings had been reset. So I set XMP and existed. Which, of course, removed my setting for disabling Core Performance Boost, as well as my cTDP 45 and CPU Temp Limit 80. I rebooted into Windows.

And just now, the PC restarted.

Let's recap. Ryzen R5 1600 CPU, Wraith Spire 95W copper-cored heatsink, properly pasted and torqued down, with a refurb Gigabyte RX 5600XT with BIOS F3 (current newest, added support for 14Gbit/sec memory, and boosted TDP to 180W), and a 75W PCI-E slot-powered Gigabyte GTX 1650 4GB D5 ITX card in the lower slot.

Mining on both GPUs, and PrimeGrid on the CPU, for less than 10 minutes.

BOOM! Hard restart.

Not looking good.

I don't have BOTH RX 5600XT cards in there, just one of them. Sigh.

Edit: Ok, on the 1600 rig, I went back into BIOS, disabled Core Performance Boost, set cTDP to 45, and then set CPU Temp Limit to 80. Rebooted into Windows. Went into Wattman, set Power Limit on the RX 5600XT to -20%. Exited.

Letting it mine and crunch PrimeGrid.

Edit: The more that I think about this, the more that it seems like possible the mobo IS overvolting the CPU, somehow. Maybe if I flash to a newer BIOS? They're up to like F50 or so, I think, maybe higher now. I thought that the PinnaclePI 1.0.0.6 was best for Ryzen 1 CPUs though.

You know I like you, but you are really trying to find someone else to blame for your poor decisions. I would never run a PSU that close to a limit. Power spikes are real. It could very well be the PSU. I had one for about 10 years and eventually just playing Battlefield 1 would cause it to spontaneous reboot at times. Took me a bit to track it down. Got another solid PSU and the problem is gone.

And that PSU was under low load for most of its life. Not running 80% or so all the time. Imagine revving an engine 80% to it's redline. Think it'll last as long as normal driving? Sorry, I had to, since we've been doing cars in this thread . It sounds like your mistake was getting a case that was totally inadequate for what you were going to end up doing with it

That was just like setting a timebomb that would go off at some point. Apparently that times is now. Learn from it and move on. Trying to find someone to blame (AMD lies about it's TDP!) isn't the answer. Under normal circumstances, if cool enough it will boost a bit higher and use a bit more power until it has to throttle down. I'd say that's a good thing. We've been doing it for years. Now if you go overboard with it like Intel, that's another topic. Even then, it's really the BIOS for those Intel chips. Just like it is with AMD.

 

[Makaveli]

Thread title is kinda click baity.

lol why is this thread 6 pages.

Get a better cooler

/thread

 

[Thunder 57]

And I was serious about the UPS thing. I have a cheapo one for my monitor. Takes some 40W off my main UPS so I get extra runtime should I need it. You have a 650W PSU + 100W TV all in one UPS rated for 810W rated UPS. That's too much. At best you would have a runtime of a few minutes if you're burning even 600W total. Maybe that's all you want. If so, then that's OK I guess.

I have an estimated runtime of 70-75 minutes. Realistically it's more like an hour or maybe even less since the batteries have aged some since I last really tried. It's outputting 11-15% doing basic browsing. Youtube brought it up to nearly 20%.

 

[Shivansps]

A few things i wanted to point out:

1) For the record: I did not created this thread.

2) TDP numbers does not really means what it used to mean in the past, the new turbo tech is designed to clock past the TDP as long as the cooler allows it.

3) Temperatures should not be a issue unless it gets past 95°C

4) Wraith Stealth is good in two aspects: Durability and low noise, cooling performance is not bad, but is not its strong point, this is why AMD went with the Wraith Spire on the 65W 3400G and the more expensive 3600 still has the Stealth.

The AMD "square" cooler (the one that was used since forever and still comes with the Athlons) may outperform the Wraith stealth in cooling performance, but durability is awful.

 

[DisarmedDespot]

They aren't more than a couple of years old. The PSUs have a 10-year warranty, and the APC UPS has a 3-year warranty. All of that equipment is still under warranty.

Warranty generally doesn't cover batteries as they're expendable. I had a 500 watt UPS that couldn't handled a 200 watt load after a few years. Try running it at your normal load, then trigger a failover to battery power either by testing it or unplugging it from the wall. If it shuts off, the batteries need replaced. They're probably sealed lead-acid so probably not expensive, just heavy.

 

[Schmide]

Old hardware, esp when taxed 247, dies.

Trends of a single point are pointless

 

[maddogmcgee]

This thread is crazy. If it was me I would test every single piece, one at a time, in a different PC and see if they all work under load. Mobo could be dying, PSU could be dying, cards could be defective, who knows?

When younger, my wife had a HP laptop she would put on the bed so the heat exhausts were blocked. When it would turn off, she would turn it back on again. Eventually the GPU broke. Should I have called Nvidia and demanded they improve the thermal throttling on the GPU?

 

[Thunder 57]

Warranty generally doesn't cover batteries as they're expendable. I had a 500 watt UPS that couldn't handled a 200 watt load after a few years. Try running it at your normal load, then trigger a failover to battery power either by testing it or unplugging it from the wall. If it shuts off, the batteries need replaced. They're probably sealed lead-acid so probably not expensive, just heavy.

You can usually find longer lasting batteries too. 7Ah batteries are very common. I replaced mine with 9Ah batteries on my old one. My main one actually came with 8.5Ah batteries so I left them alone. The 1.5-2 extra Amp hours helps a good bit. Just saw a pair of 9Ah batteries can be had for just under $40 shipped. They also only have a warranty for one year, because of obvious reasons.

 

[Kenmitch]

Larry's threads sometimes make me dizzy.

 

[thesmokingman]

Seriously....???

 

[ksosx86]

Serious question. I've got a 65W-rated TDP Ryzen R5 1600, in a rig, on a 65W-rated AMD stock heatsink. It's blue-screening, crashing, and the CPU temps just keep going up and up.

I updated HWMonitor, and it's showing a "Package Power" for the entire chip, at 82W or so. No wonder it's constantly overheating and crashing. 82W TDP CPU > 65W TDP Heatsink.

The worst part is, this is AFTER limiting the number of PrimeGrid threads, down from 12 to 9. That's right, I'm not even running the CPU at a full thread load.

Edit: Yes, I know that the obvious answer, is to "get a better heatsink", and that the "stock heatsink" for the 1600 was the 95W TDP model. Which, at the time, was stated that AMD wanted to give users the ability to OC on the stock heatsink. Now I know that was a lie, it's because AMD CPUs (at least, the 1600), are NOT able to stay within their stated rated specs.

Has anyone here yet mentioned or considered activating ECO mode - most UEFI flavors support it you'd find it under AMD's advanced settings along or around the same section mentioning precision boost overdrive. In cases you'll need to enable PBO then you'll see as an option as opposed to PBO "advanced enabled disabled etc" "ECO mode" anyways I've given enough information to point someone who can make a peanut butter and jelly in the right direction.

Regarding ECO & the 3900 (A different CPU I know but it's a good in-depth analysis)

EDIT: Now back to work on the night shift at chipzilla while posting solutions about AMD chips That's me rolling eyes at me. My coworkers would probably be more inquisitive about RDR2 on Stadia that I play during breaks on my Lenovo. Had to jailbreak an iPhone + Mullvad + "tetherme" = circumvents T-Mobile's hotspot caps & surprisingly it is actually playable.

Anyhoo... try Eco mode, my friend

 

[Leeea]

gut instinct:
Having read all of this, I think those refurb GPU's are poison.

If you can, send them back and get some of your money back. It is not worth the fight.

If you cannot get your money back, you can always study them with a magnifying glass looking for defective parts. A lot of board components leave visible indicators of failures. capacitors bulge or leak, vrm's leave burnt marks, diodes sometimes leave a brown stain, etc.

There is nothing stopping those cards from pulling more then 230 watts. There are ratings, and there is what happens when things degrade and start to fail.


Who did you buy them from? Anyone reputable should let you return them. Otherwise you might have just been ripped off. It is not unheard of for a seller to buy a pallet of "Untested Customer Returns" wholesale, purchase a verified high rated ebay/paypal account, and then ebay the entire pallet for a quick buck.

 

[ksosx86]

Maybe I got some "miner trash" GPUs that someone was overclocking and generally messing with, and they "burned them out", and Gigabyte gave them a once-over, and re-flashed the BIOS, and sent them back out again.

If you're still within the Refurb's Warranty RMA them .. one at a time if need be if you need a card (having trouble digesting this thread so cannot tell if you have two cards in the same rig for crossfire and or SLI or not but whatever - RMA them)

• Take one card out of your rig
• Run all settings at stock (BIOS defaults - preferably the latest unless there's a generational conflict with your Ryzen CPU)
• Get a USB drive and flash it as a live WinPE environment and or Ubuntu/Debian etc place some known good apps on it
• I'd recommend a known good distro for WinPE (Himen's etc GandalfPE etc) they come with AIDA and such pre loaded
• Using all those default settings still (and one card) run in PE and use AIDA (stress the system) lookup how to run the AIDA system stress test if you don't know how you can stress the GPU the CPU CACHE HDD/SSD RAM etc all at the same time with that one app.
• Try to recreate your issues with Thermals
• The first step I'd try if it is so bad that you still hit 95C right away is to remount the AMD PRISM cooler and for testing purposes use a CPU_PWR header for its fan not "CPU" this way it will run at 100%
• Make sure it's mounted nice and firm 95C is bizarre and unusual even for a case with bad airflow
• I write in bold because I tend to think people ask for help then ignore the relevant parts of instructions lol - not to be a dick with the bold.
• Umm what else yeah apply ECO mode
• Try to recreate the 95C scenario again.
• At this point I dunno ...I figure you'll probably be in touch'

 

[leoneazzurro]

I have a 3700X in a laptop case, running at full power envelope (65WTDP-88W PPT), and I never saw temperatures above 85°C. 95°C is screaming about bad thermal contact/malfunctioning of the fan speed control.

 

[Magic Carpet]

I’d try undervolting / underclocking that thing to see if it makes a difference. If you can’t, test the cpu in a different mobo.

 

[coercitiv]

Edit: Could also be the real temp value for that CPU Temp Limit, but who knows that whole offset on first gen Ryzen makes setting temp limits or fan curves a pain in the butt.

The offset is most likely variable based on SKU. From what I understand some of the higher SKUs had a 20C offset, while my 1600X had a 10C offset based on HWInfo readings.

At this moment @VirtualLarry is making it unnecessarily hard to configure a proper throttle mechanism on his "budget" AM4 board that has no voltage controls whatsoever and a mesed up XMP implementation which he insists on keeping turned on: he's changing temp limits and cTDP settings at the same time, while reporting little or anything about how these settings are affecting HWINfo readings. (temps, voltages and most importantly clocks)

Look at his mess:

Well, I interrupted the mining, restarted, went into the BIOS looking for a Vsoc fixed voltage option, but when I got into the BIOS, there was a message that my BIOS settings had been reset. So I set XMP and existed. Which, of course, removed my setting for disabling Core Performance Boost, as well as my cTDP 45 and CPU Temp Limit 80. I rebooted into Windows.

That is not how one goes about solving stability issues, at all.

• Turn the damn XMP off. In fact, reset the BIOS to make sure every hidden voltage/power parameter is back to stock.
• Start by using "CPU Temp Limit" at Auto and observe system behavior. After you have a value for stock config, proceed with incremental lower values on that parameter alone to check whether firmware actually enforces this configurable limit.
• Once you have proper understanding of the "CPU Temp Limit", leave in on Auto and start working with cTDP parameter the same way - change only this parameter and observe temps, voltages, clocks.
• After you're done with independent testing - if they both work - chose a combination that suits your needs and test stability.
• If the system is stable, enable XMP and watch it go insane.

My intuition says you're very likely to achieve stable thermals for the CPU by disabling XMP and reverting to stock settings. You may even be able to get memory speeds back up if you write down main memory timings and then manually configure the overclock (change mem clocks to 3000, main timings and voltage to 1.35V). With stock Vsoc you should still be able to hit 3000.

Even if everything goes as planned above, this won't necessarily fix your thermal and/or instability issues while running the dual GPU setup.

 

[Atari2600]

Serious question. I've got a 65W-rated TDP Ryzen R5 1600, in a rig, on a 65W-rated AMD stock heatsink. It's blue-screening, crashing, and the CPU temps just keep going up and up.

I updated HWMonitor, and it's showing a "Package Power" for the entire chip, at 82W or so. No wonder it's constantly overheating and crashing. 82W TDP CPU > 65W TDP Heatsink.

The worst part is, this is AFTER limiting the number of PrimeGrid threads, down from 12 to 9. That's right, I'm not even running the CPU at a full thread load.

Edit: Yes, I know that the obvious answer, is to "get a better heatsink", and that the "stock heatsink" for the 1600 was the 95W TDP model. Which, at the time, was stated that AMD wanted to give users the ability to OC on the stock heatsink. Now I know that was a lie, it's because AMD CPUs (at least, the 1600), are NOT able to stay within their stated rated specs.

You've taken a supermini, fired a towbar on it and expect it to tow a 3 tonne mini digger.

Its amazing that someone who has built so many pcs has so little appreciation of the nuances of suitable configuration and operation.

 

[VirtualLarry]

gut instinct:
Having read all of this, I think those refurb GPU's are poison.

If you can, send them back and get some of your money back. It is not worth the fight.

If you cannot get your money back, you can always study them with a magnifying glass looking for defective parts. A lot of board components leave visible indicators of failures. capacitors bulge or leak, vrm's leave burnt marks, diodes sometimes leave a brown stain, etc.

There is nothing stopping those cards from pulling more then 230 watts. There are ratings, and there is what happens when things degrade and start to fail.


Who did you buy them from? Anyone reputable should let you return them. Otherwise you might have just been ripped off. It is not unheard of for a seller to buy a pallet of "Untested Customer Returns" wholesale, purchase a verified high rated ebay/paypal account, and then ebay the entire pallet for a quick buck.

Hmm,. I bought them from Newegg, as a Gigabyte factory refurb. I don't know if I'm still within my 30-day return window.

Anyways, I don't know if I have the PBO section in the BIOS, these Gigabyte BIOSes are WAY stripped-down as far as options go, as compared to my Asus and ASRock boards. I'll look for "ECO mode", although I would think setting the cTDP to 45 would be the same.

And that paragraph that I wrote, about setting XMP, and having my cTDP and CPU Temp Limit settings reset... I didn't say setting XMP did that, I was referring to the previous sentence, in which I mentioned the BIOS said that my settings were re-set, and then I went in and set XMP again, but because the BIOS reset my settings, NOT because I set XMP, the cTDP and CPU Temp Limit settings were no longer in force.

Anyways, the 1600 rig, in which I disabled Core Performance Boost (and updated to F50), seems to have been stable while I slept.

The rig with the 3600 and Gammax 400, which hoverered around 93C current and max when I went to sleep, was frozen when I woke up. However, black-screen, the blue power LED was on, but I COULD NOT hold the ATX soft-off power button down on the case to force power-off, I had to pull it out of the cubby and use the switch on the back of the PSU to reset it.

I then went into the BIOS, it was also version F50 for the B450 AORUS PRO WIFI (AGESA ComboPI V1 1.0.0.4 B). I disabled Core Performance Boost in that BIOS, and started Windows 10 again, and made sure that the AMD Wattman settings for the RX 5600XT were set to -20% Power Limit as well. I also observed that the Frequency of the cores didn't exceed 3600Mhz, because CPB was disabled, so it wasn't boosting to 4.0Ghz all-core anymore. But temps were hovering around 67C. I'll check it again in a bit. (I may not have been mining, or mining long enough to heat up the case at that point.)

 

[ksosx86]

Hmm,. I bought them from Newegg, as a Gigabyte factory refurb. I don't know if I'm still within my 30-day return window.

You get 90 days total if you're out of that window.
EDIT: Looking into BIOS specific settings to relay for the B450 AORUS PRO WIFI ...
That 1600 rig that resets and loses your BIOS settings, is that a constant issue, intermittent or was that just an isolated incident?

Update 1: I noticed your B450 AORUS PRO WIFI is dual bios, perhaps your other board is too and you can switch to the other BIOS? (you'll likely have to update it if you haven't yet) The only other time I can recall settings dropping was CMOS batteries going but... that'd be unusual... Christ... one thing after another xD

 

[VirtualLarry]

That 1600 rig that resets and loses your BIOS settings, is that a constant issue, intermittent or was that just an isolated incident?

It was the first time in a long time that it happened to me. Maybe the CMOS battery is going, or maybe something triggered it (bad PSU?).

 

[ksosx86]

Yep my thoughts exactly, if it persists try switching to your secondary BIOS if it's a motherboard with dual if you still can't resolve .. CMOS I'd guess... either way you shouldn't have to do this sort of thing. Hopefully, you have Warranty worse case.

 

[DrMrLordX]

n a low-wattage GTX 1650 4GB D5 card into the secondary slot, just because I can, for extra mining.

Okay, I'm not sure why multiquote mangled this, but I was trying to acknowledge that you have a proper HSF for that CPU. And yes Summit Ridge can and will exceed its TDP. If you really care about power usage, you should tweak voltage by hand and aim for static clocks instead of relying on thermal throttling and the . .. not all that great boost algo. Seriously. Setting cTDP isn't a terrible idea. But you can dial in exactly the power usage you want just by adjusting voltage. And that will help you cope with board overvoltage as well. Due to the operating conditions extant, I would think that a static undervolt is in order, and I highly recommend you start @ 1.2v with a conservative clockspeed (maybe 3000 MHz) and see how temps look during intended operating conditions. Then try upping clocks until you can't get stability. You should know the drill by now.