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.

 

[VirtualLarry]

Can someone speak to my theory, that there may be something amiss ("shorting", for lack of a better term), in one or both of the GPU's power-delivery section, and possibly, they are drawing more +12V power (say, from the PCI-E slot), and thus, lowering the input voltage to the CPU's VRMs, thus causing them to transfer more current to the CPU, thus causing the CPU to heat up more? Could this even be a thing? Or am I grasping at straws here (I'm no EE), and the reality is more like AMD specifies PPT to the CPU socket as higher than TDP, and I'm trying to use a 65W TDP cooler on it, and thus, the equation fails to keep temps under control. To say nothing about my potentially-overvolting Gigabyte mobo BIOS.

 

[blckgrffn]

Larry, this is me giving you the side eye. (╯°□°）╯

You are seriously heatsoaking your case, running a tiny tin CPU cooler, running PrimeGrid and complaining about stability without tuning the CPU by hand first?!?

ALSO. Aren't you mining? Shouldn't you be reserving a core or two to ensure your GPUs are running 100%? I don't know how much that is a thing, but even if you were doing GPU compute for saving the world you'd want the GPUs to be fed as a much higher priority than loading your CPU due to the amount of work they can accomplish, relatively speaking. Even a 1% performance regression on the GPUs is too much.

Based on your ludicrously detailed posts on other topics I find this situation shocking! Get a box fan, take the side off, do something!

 

[alexruiz]

Have you run a Ryzen R5 1600 on wraith stealth? For more mundane, "normal" tasks, like web browsing, sure, it works fine.

But try running PrimeGrid, 12 threads of pure AVX2 torture for the CPU. The Package Power is over 80W, and termps on a 65W TDP heatsink, will just go up, and up, and up.

I have built dozens of Ryzen 5 1600s, and I have almost always used the stealth.
Stress tested and such.
Yes, the package power shuts up to over 80W, and yes, the temps get high and the fan gets loud, but they DO NOT crash.
The thermal throttle algorithm is smart enough to dial it down. Of course, I always make sure that my system air flow is good.
I let the UEFI pick the boosting strategy.

 

[Markfw]

Here are 2. $25 (affordable) and a little more but WOW the cooling capacity. If I wanted good air, the second would be one choice for me.

Are you a human?

www.newegg.com

Are you a human?

www.newegg.com

 

[Leeea]

Get real. I never said that the case temps were 95C. If they were, I would be getting some bacon to cook on my case. CPU core temps of 95C (silicon hotspots), are necessarily going to be a lot higher than the actual air temp inside the case, which is again going to be higher than ambient temps in the room.

Once it is burnt, from to much power draw, or to much heat, your non-CPU components do not heal themselves. Sometimes if you drop the load you will be ok, but normally it is replacement time. Get a multimeter and start probing sort of thing. There are components on the pcb boards you cannot replace, and there are parts you can. Maybe you will get lucky and it will be an easy cap swap.

Odds are you have moved beyond having a cooling problem or a power problem to real problems.

 

[Hitman928]

Can someone speak to my theory, that there may be something amiss ("shorting", for lack of a better term), in one or both of the GPU's power-delivery section, and possibly, they are drawing more +12V power (say, from the PCI-E slot), and thus, lowering the input voltage to the CPU's VRMs, thus causing them to transfer more current to the CPU, thus causing the CPU to heat up more? Could this even be a thing?

No

the reality is more like AMD specifies PPT to the CPU socket as higher than TDP, and I'm trying to use a 65W TDP cooler on it, and thus, the equation fails to keep temps under control. To say nothing about my potentially-overvolting Gigabyte mobo BIOS.

Again, your CPU should be reducing frequency if thermal limits are reached. If it's not doing this, something is wrong in your setup. If it is doing this and still crashing, then something is wrong in your system. Again, I would look at the far more likely sources which are PSU, GPU, MB, and even memory.

 

[Zstream]

I don't feel like reading all this dribble, but I've run more AMD 6 core chips than most people have fingers and toes. I haven't had a chip crash due to thermals in a really really long time, and that was pre-bulldozer. The chips, especially ryzen 1600, 1700's, and 3600's will lower the voltage and decrease chip speed.

Something is really messed up in the bios, or some setting you have changed.

 

[Hitman928]

As @Leeea mentioned, by putting your system into an environment with far too inferior ventilation for it's power dissipation needs, it's very possible that other components in the system (memory, PSU, etc.) or even on the motherboard have experienced physical damage to the point that they will no longer be fully stable even when put into an open air environment. CPUs/GPUs have had protections in place for a long time to guard against this, but other components, not as much. You really need to consider this possibility and stop fixating on the CPU overheating angle.

 

[UsandThem]

Can someone speak to my theory, that there may be something amiss ("shorting", for lack of a better term), in one or both of the GPU's power-delivery section, and possibly, they are drawing more +12V power (say, from the PCI-E slot), and thus, lowering the input voltage to the CPU's VRMs, thus causing them to transfer more current to the CPU, thus causing the CPU to heat up more? Could this even be a thing? Or am I grasping at straws here (I'm no EE), and the reality is more like AMD specifies PPT to the CPU socket as higher than TDP, and I'm trying to use a 65W TDP cooler on it, and thus, the equation fails to keep temps under control. To say nothing about my potentially-overvolting Gigabyte mobo BIOS.

Once it is burnt, from to much power draw, or to much heat, your non-CPU components do not heal themselves. Sometimes if you drop the load you will be ok, but normally it is replacement time. Get a multimeter and start probing sort of thing. There are components on the board you cannot replace, and there are parts you can. Maybe you will get lucky.

Odds are you have moved beyond having a cooling problem or a power problem to real problems.

I think we have moved from tech discussion into an episode of the X Files...... or The Twilight Zone.

When troubleshooting, you start with the simple stuff first. In this case, you are running two GPUs and a CPU with a application that pushes their use to 100% over very long periods. You have one 120mm exhaust case fan, along with one stock cooler that is not designed for what you are doing.

Open the case up, blast it with a box fan, and eliminate heat being the issue. Outside of dirty power, heat is electronics biggest enemy.

 

[VirtualLarry]

You are seriously heatsoaking your case, running a tiny tin CPU cooler, running PrimeGrid

And mining on the GPUs. I currently only have 9 threads out of 12 running PrimeGrid (80% of total CPUs in BOINC). Mining takes very little CPU time, people build mining rigs all of the time with a Celeron G3900 Skylake dual-core, with a mobo with 8 or 13 PCI-E x1 risers, and drive them all just fine on that little CPU.

I have some 240mm AIO WC kits that I intended to upgrade with, I just need some new cases with top ventilation. I mean, my PCs used to "run hot", but not, "crash every six hours HOT". They were at least stable.

Also, one more data-point, when the power went out here after a storm, my battery-backup (APC consumer Pure-Sine 1350VA / 810W) on the rig with the two RX 5600XT cards in it, shut off immediately, and the UPS started to whine, not the normal beeping when the power was cut off. It showed an "F01" error code, which is "battery overload". I then had further issues with it, with the PC. What I don't get, though, is the PC has 2x 180W TDP GPUs, 1x 65W TDP CPU, some RAM, chipset, an SSD, etc. I have an EVGA 650W 80Plus Gold G1+ modular PSU in both of the secondary rigs. But the crashing/shutoff happened to both of them, with both of these GPUs in them. (Are they cursed or something? Or is something shorted?) So it's not the PSU.

I switched the power plug from "Battery" to "Surge Only", and I still got crashes (on the Ryzen 3600 rig with the Gammax 400 cooler). So I pulled out the rig with the 1600 and stock Wraith Stealth cooler (yeah, I knew it was kind of underpowered, but it could CPU-mine OK). Put the 2x RX 5600XT (did I mention yet, they were factory refurbs, that had to flash the BIOS on them, they came with the 12Gbit/sec BIOS with 150W TDP, I put the 14Gbit/sec BIOS with the 180W TDP onto them).

Anyways, the CPU + 2x GPUs + system load, shouldn't be too much for my 650W Gold PSU, and in turn, that shouldn't be too much (at the wall) for my 810W-rated battery backup.

An identical model battery backup, still kept my other (main) rig running, with an RX 5700 reference card, and a GTX 1660 ti Gaming X running and mining for a few minutes (400W power draw on the meter), until I manually shut it down, without issue.

Edit: I mean, maybe... maybe the problem is BOTH EVGA PSUs? And that the high CPU temp is a red herring, and in fact, AMD's thermal-throttling is... gasp... working, and that's why it was pinned at 95C for a Package Temp, but the EVGA PSUs are unstable under load, having living for a year or two at the bottom of the cubby, and not getting proper cooling, or just being a pile of ****? Do the G1+ models have a bad rep? ( @aigomorla , you might know )

Edit: I mean, it was doing it also in the rig with the 3600 and the Gammaxx 400, which is roughly equivalent in performance, AFAIK, to the Hyper 212 EVO cooler. It's a 4-heatpipe tower with a 120mm fan. And we know that the 3600's thermal-throttling does work, more-or-less. And if temps were CONSTANTLY increasing inside the case, then why were the GPU temps as reported by Wattman for both GPUs, roughly 67C (not at all concerning, in my mind). If case temps were rising due to lack of case cooling capacity, one would expect the non-blower GPUs to start increasing in temp too, wouldn't you?

The cubby is open in the front, and the case has dual 120mm intake fans in front.

 

[coercitiv]

and to think I was just getting ready to ask if you're running custom BIOS on those cards

I put the 14Gbit/sec BIOS with the 180W TDP onto them

 

[blckgrffn]

And mining on the GPUs. I currently only have 9 threads out of 12 running PrimeGrid (80% of total CPUs in BOINC). Mining takes very little CPU time, people build mining rigs all of the time with a Celeron G3900 Skylake dual-core, with a mobo with 8 or 13 PCI-E x1 risers, and drive them all just fine on that little CPU.

I have some 240mm AIO WC kits that I intended to upgrade with, I just need some new cases with top ventilation. I mean, my PCs used to "run hot", but not, "crash every six hours HOT". They were at least stable.

Also, one more data-point, when the power went out here after a storm, my battery-backup (APC consumer Pure-Sine 1350VA / 810W) on the rig with the two RX 5600XT cards in it, shut off immediately, and the UPS started to whine, not the normal beeping when the power was cut off. It showed an "F01" error code, which is "battery overload". I then had further issues with it, with the PC. What I don't get, though, is the PC has 2x 180W TDP GPUs, 1x 65W TDP CPU, some RAM, chipset, an SSD, etc. I have an EVGA 650W 80Plus Gold G1+ modular PSU in both of the secondary rigs. But the crashing/shutoff happened to both of them, with both of these GPUs in them. (Are they cursed or something? Or is something shorted?) So it's not the PSU.

I switched the power plug from "Battery" to "Surge Only", and I still got crashes (on the Ryzen 3600 rig with the Gammax 400 cooler). So I pulled out the rig with the 1600 and stock Wraith Stealth cooler (yeah, I knew it was kind of underpowered, but it could CPU-mine OK). Put the 2x RX 5600XT (did I mention yet, they were factory refurbs, that had to flash the BIOS on them, they came with the 12Gbit/sec BIOS with 150W TDP, I put the 14Gbit/sec BIOS with the 180W TDP onto them).

Anyways, the CPU + 2x GPUs + system load, shouldn't be too much for my 650W Gold PSU, and in turn, that shouldn't be too much (at the wall) for my 810W-rated battery backup.

An identical model battery backup, still kept my other (main) rig running, with an RX 5700 reference card, and a GTX 1660 ti Gaming X running and mining for a few minutes (400W power draw on the meter), until I manually shut it down, without issue.

These extra details are exactly what I am talking about. It's like all trees in here, see the forest.

Heat soak kills your stuff. Sounds like your PSU or your motherboard is no longer good.

Crazy power drops like what you experienced may have caused damage too.

This damage can show itself in many ways, but is exacerbated in my experience with really edge case usage patterns that stress hardware.

To your specific use case: I would assume that the 810W rating the UPS is upper bounds, while I would assume instantaneous draw on the PSU could be quite high. Remember, they are rated to pull even more from the wall - their rating is what they can supply to components after conversion.

I would start validating the components one at a time. For my part, I no longer trust most of that PC you are having issues with. Maybe the storage is still good, and ram can be pretty well validated in another rig.

 

[Velgen]

So question not 100% sure from your post but it sounds like you've tested those GPUs in another separate system and have had issues with them have you tried flashing them to the 150W TDP bios? Not sure if they got refurbed due to them being unable to run the 180W bios or if it was that power failure issue that caused it, but better to rule things out. As for the CPU setup run it by itself for a while to ensure it's all stable then try adding a different set of GPUs. Although I am hesitant to recommend that until you've got a case that can actually handle that level of heat. Really just need to test everything 1 by 1 to ensure nothing has been damaged by heat or that power failure so best to reduce it down to the bare minimum.

Edit: Another thought probably best to test both PCI slots before trying another set of GPUs in that 1600 rig

 

[VirtualLarry]

Once it is burnt, from to much power draw, or to much heat, your non-CPU components do not heal themselves. Sometimes if you drop the load you will be ok, but normally it is replacement time. Get a multimeter and start probing sort of thing. There are components on the pcb boards you cannot replace, and there are parts you can. Maybe you will get lucky and it will be an easy cap swap.

Odds are you have moved beyond having a cooling problem or a power problem to real problems.

Well, that's kind of what I'm wondering, about the refurbished GPUs, that once I added them to (either) secondary rig, I was having insane heating and crashing problems.

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.

 

[blckgrffn]

So question not 100% sure from your post but it sounds like you've tested those GPUs in another separate system and have had issues with them have you tried flashing them to the 150W TDP bios? Not sure if they got refurbed due to them being unable to run the 180W bios or if it was that power failure issue that caused it, but better to rule things out. As for the CPU setup run it by itself for a while to ensure it's all stable then trying adding a different set of GPUs. Although I am hesitant to recommend that until you've got a case that can actually handle that level of heat. Really just need to test everything 1 by 1 to ensure nothing has been damaged by heat or that power failure so best to reduce it down to the bare minimum.

LURKER DETECTED. I expect you to post again in roughly two years

 

[VirtualLarry]

and to think I was just getting ready to ask if you're running custom BIOS on those cards

Nope, stock BIOS. Gigabyte's own web site "makes you" (if you want them running at the "proper" 14Gbit/sec spec than an RX 5600XT is supposed to run at) flash the BIOS on the card yourself.

Radeon™ RX 5600 XT GAMING OC 6G Key Features | Graphics Card - GIGABYTE Global

Discover AORUS premium graphics cards, ft. WINDFORCE cooling, RGB lighting, PCB protection, and VR friendly features for the best gaming and VR experience!

www.gigabyte.com

Click Support, click BIOS, it shows the upgrade BIOSes that they want you to install.

 

[Leeea]

Well, that's kind of what I'm wondering, about the refurbished GPUs, that once I added them to (either) secondary rig, I was having insane heating and crashing problems.

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.

The problem with refurbished products is you know at least one other person had a problem with them.

As for Gigabyte, they probably stuck it in a tester, it came back good, and then resold it. It is unlikely they would have attempted to repair, re-flash, or anything. It is typically pass or fail.

If you have a different video card lying around test the system and see what happens. If your problems go away you likely located the issue. Send defectives back to the vendor for a refund.

( If it is a dodgy vendor they may have purchased Gigabytes rejected scrap pallet and resold it. See https://www.directliquidation.com/how-it-works/merchandise-conditions-explained/ and pallets labeled "Tested Not Working", "Tested Not Working (Salvage)", "Tested Not Working (SCRAP)", "Untested Customer Returns" )

 

[Velgen]

I mean technically yes it's stock, but that was a BIOS update on release with the 5600XT so if they are early revision cards it could potentially be possible they were refurbished due to not being able to run the higher clocked BIOS. It is possible some weaker GPUs passed the testing for the first revision BIOS that may not of been able to pass the BIOS AMD pushed through to better compete on launch.

@blckgrffn I am indeed a lurker been lurking since before this account was made just felt like posting Today you may be right about how long it is till I post again lol

 

[PrinceXizor]

This is very simple from a physics standpoint not a computer standpoint.
1. What is thermal throttling on a CPU?
-A reduction in electrical specifications to reduce the thermal output of the processor itself.

2. What causes a CPU to enter thermal throttling?
-Simplified - a high temperature at a specific sensing location

3. Does that temperature automatically lower when throttling occurs?
-Very Often - the specific areas to be protected by thermal throttling are generally the hottest by far and thus thermal energy flows fast out of it once its no longer heating a lot.

4. Is this always the case?
Absolutely Not - every system has a maximum thermal dissipation rate. It's an extremely complicated mixture of thermal transfer characteristics.

It can be somewhat simplified in considering the air inside the case as a homogenous whole. There is a maximum rate that thermal energy can be transferred from the air inside the case to the air outside the case (even with a completely open case this becomes the thermal gradient between the air near components and air further away. A lack of air movement significantly degrades this gradient. This is an actual known occurence in HVAC when dealing with the thermal envelope of a house. A thin layer of stagnant air just outside the home is a key thermal characteristic to be accounted for). As you generate additional heat from alternative sources (dual GPUs) that heats the inside air which reduces the thermal transfer rate from the cooler to the air which reduces the transfer rate from the core to the fins, and so on and so forth.

5. Will this shutdown your CPU?
Usually no. Process manufacturers are smarter than that and thermal throttling algorithms DO take these extreme measures into mind.

6. What gives?
The CPU and particularly the RAM relies on other things that are adversely affected by high heat that don't have as robust "throttling" protections as modern CPUs. Power FETs on both the motherboard and particularly in the PSU come to mind. Electrolytic caps (as mentioned previously and are used highly in regulating the various electrical signals used in microprocessors) are very succeptible to heat issues. I've seen many strange power anomalies in a commercial/industrial environment that were caused by failing or overtemp FETs and caps.

7. Side note - any "rating" always has a set of defined or assumed parameters that accompany the rating. A dive watch for example can advertise 200m water resistance. That doesn't mean that you can submerge the watch indefinitely and expect it to never take in water. That is not the assumed usage when testing the 200m water resistance. TDP and similar specs are the same.

 

[VirtualLarry]

Oh yeah, the mobo was a refurb, in the 1600 rig. (The Gigabyte X370 Gaming ATX board.) The rig with the 3600, the AORUS PRO WIFI ATX was a brand-new board when I put it in.

But both were mining/crunching whatever, with dual GPUs before (GTX 1660 ti), without issue / crashes / UPS beeping. It's when I installed this blasted pair of RX 5600XT refurbs that everything started to go severely downhill.\

 

[Jimzz]

Here are 2. $25 (affordable) and a little more but WOW the cooling capacity. If I wanted good air, the second would be one choice for me.

Are you a human?

www.newegg.com

Are you a human?

www.newegg.com

Heck you can get good coolers cheaper now. I use one of these 6 pipe Chinese coolers on my 2700x. Less than $20 if you look around and sometimes in the $15ish range. Super quiet and cool.

Amazon.com: Shumo 6 Heat Pipe 4 Wire Single Fan Without Light Cpu Fan Cpu Heatsink For Intel 775/1150/1155/1156/1366 For Adm All: Electronics

Buy Shumo 6 Heat Pipe 4 Wire Single Fan Without Light Cpu Fan Cpu Heatsink For Intel 775/1150/1155/1156/1366 For Adm All: CPU Cooling Fans - Amazon.com ✓ FREE DELIVERY possible on eligible purchases

smile.amazon.com

Amazon.com: Silent CPU Fan, 800-2000 revs Dual Tower Temperature Control 6 Heat Pipe CPU Cooler for PC Computer(Blue Light) : Electronics

Buy Silent CPU Fan, 800-2000 revs Dual Tower Temperature Control 6 Heat Pipe CPU Cooler for PC Computer(Blue Light): CPU Cooling Fans - Amazon.com ✓ FREE DELIVERY possible on eligible purchases

smile.amazon.com

 

[VirtualLarry]

have you tried flashing them to the 150W TDP bios? Not sure if they got refurbed due to them being unable to run the 180W bios or if it was that power failure issue that caused it, but better to rule things out.

That's a really good point that I was just contemplating in the bathroom before I read your post. Maybe, these refurbed cards are unstable with the 180W TDP / 14Gbit/sec BIOS, and Gigabyte "refurbed" them, by flashing them to the 150W / 12Gbit/sec BIOS. Hmm. You might be on to something there.

But that would mean that I basically wasn't getting the product that I thought that I was paying for.

 

[amrnuke]

Oh yeah, the mobo was a refurb, in the 1600 rig. (The Gigabyte X370 Gaming ATX board.) The rig with the 3600, the AORUS PRO WIFI ATX was a brand-new board when I put it in.

But both were mining/crunching whatever, with dual GPUs before (GTX 1660 ti), without issue / crashes / UPS beeping. It's when I installed this blasted pair of RX 5600XT refurbs that everything started to go severely downhill.\

But you think it's the 1600 that's the problem?

 

[VirtualLarry]

@blckgrffn I am indeed a lurker been lurking since before this account was made just felt like posting Today you may be right about how long it is till I post again lol

Please don't let my "insane (at times) ranting" drive you away. This is a (mostly) good site. And I'm (usually) a bit more "on the ball".

 

[Hans Gruber]

So question not 100% sure from your post but it sounds like you've tested those GPUs in another separate system and have had issues with them have you tried flashing them to the 150W TDP bios? Not sure if they got refurbed due to them being unable to run the 180W bios or if it was that power failure issue that caused it, but better to rule things out. As for the CPU setup run it by itself for a while to ensure it's all stable then try adding a different set of GPUs. Although I am hesitant to recommend that until you've got a case that can actually handle that level of heat. Really just need to test everything 1 by 1 to ensure nothing has been damaged by heat or that power failure so best to reduce it down to the bare minimum.

Edit: Another thought probably best to test both PCI slots before trying another set of GPUs in that 1600 rig

He's been a member since 2013 and is now up to 5 posts.