• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

5950x and PBO overheated something or busted custom loop?

Anyone have any recommendations on the right equipment/parts to use for temperature monitoring that can be picked up by HWiNFO64?

I want to attach multiple thermal sensors to a used Master SLI/ac (the motherboard that started this thread) that I was able to pick up dirt cheap.
Various locations will be on or around the PCB, cpu socket, and VRM.
 
Adding my two cents in case that it helps someone down the road; I've read somewhere that the x470 wasn't designed with a 5950x in mind (obviously), thus it wasn't a good match in terms of power delivery and the mobo struggles more than it should to handle it. I had a similar setup (Aorus x470 + 5950x). The CPU temp would constantly go beyond 90 degrees, the mobo was heating up a lot and I started smelling stuff (even lightly). 2 months in, the computer started behaving erratically, not booting up randomly, crashing, and I ended up having to replace the motherboard. I switched it for a x570 (X570S Aorus Elite AX, no southbridge fan, woohoo!) and sure enough, no more problems, with 5 to 10C degrees less on the CPU for the same usage. For a $100, it's worth keeping it in mind.
 
Just out of curiosity. Do people really have problems colling down there 5950X with PBO?

Compared to Zen 4 and intels 2 latest gen. Zen 3 seems to be fairly easy to cool. I have no trouble keeping my 5600X or 5950X cooled down. Even on hot sommer days with only aircooling.

Maybe i am misunderstanding something as i have not read this thread throw.

I can share my temp with aircooling is people should desire that. Let me know and what test you desire to see temp with. Exsample Cinibench 2024 or prime 95.

My cooling on my 5950X is a Noctua NH-D15 chromax black with two Noctua IPPC 3000 RPM 120/140 MM fans and thermal grizzly kryonaut extreme paste. So yeah hig-end air cooling, but no exotic water cooling either.
 
Dahita said:
Adding my two cents in case that it helps someone down the road; I've read somewhere that the x470 wasn't designed with a 5950x in mind (obviously), thus it wasn't a good match in terms of power delivery and the mobo struggles more than it should to handle it. I had a similar setup (Aorus x470 + 5950x). The CPU temp would constantly go beyond 90 degrees, the mobo was heating up a lot and I started smelling stuff (even lightly). 2 months in, the computer started behaving erratically, not booting up randomly, crashing, and I ended up having to replace the motherboard. I switched it for a x570 (X570S Aorus Elite AX, no southbridge fan, woohoo!) and sure enough, no more problems, with 5 to 10C degrees less on the CPU for the same usage. For a $100, it's worth keeping it in mind.
Click to expand...
For me the Master SLI/ac actually had no problem with the 5950x (and 3950x prior) until I went off the rails with PBO. My CPU temps were always ok and VRM temps on high load were not catastrophic when using the monoblock.
 
Tomgang said:
Do people really have problems colling down there 5950X with PBO?
Click to expand...
Mine run cool too.. but to answer your question.. yes, many people have a hard time cooling modern CPUs.
 
Tomgang said:
Just out of curiosity. Do people really have problems colling down there 5950X with PBO?

Compared to Zen 4 and intels 2 latest gen. Zen 3 seems to be fairly easy to cool. I have no trouble keeping my 5600X or 5950X cooled down. Even on hot sommer days with only aircooling.

Maybe i am misunderstanding something as i have not read this thread throw.
Click to expand...
If you like a lot of teardown and build pictures it's worth a look through the thread. Here are some bookmarks you might find helpful.
Tomgang said:
I can share my temp with aircooling is people should desire that. Let me know and what test you desire to see temp with. Exsample Cinibench 2024 or prime 95.

My cooling on my 5950X is a Noctua NH-D15 chromax black with two Noctua IPPC 3000 RPM 120/140 MM fans and thermal grizzly kryonaut extreme paste. So yeah hig-end air cooling, but no exotic water cooling either.
Click to expand...
I didn't think the 5950X was a problem to cool at all. Barely an inconvenience.
 
Last edited:
freeagent said:
Mine run cool too.. but to answer your question.. yes, many people have a hard time cooling modern CPUs.
Click to expand...
When it comes to Zen 4 and intels 12 and 13 gen i am well aware of it. These chips redlines all the time and has much higher TDP ratings than Zen 3. So off cause these chips runs hot and needs very good cooling.

Zen 3 has lower ratings and does not reach clock speeds of nearly 6 GHz. exsample 5950X is rated 105 watt stock while realty is 141 watt and 7950X is rated for 170 watt while realty its like 230 watt and intels I9 chips can sip around 300 watts. So i am in no douts new chips are hard to cool. AMDand Intel let the chips redlines as stated before when they can. First throttle when max temp or power limits are reach. So if your cooling is not up to the task, you will end up with at hot runnings thermal throtling cpu all the time.

A Computer Guy said:
If you like a lot of teardown and build pictures it's worth a look through the thread. Here are some bookmarks you might find helpful.

I didn't think the 5950X was a problem to cool at all. Barely and inconvenience.
Click to expand...
Nope neither here. 5950X is not a problem to cool. also why i am still on aircooling for my 5950X still. I was hesitant to go air with 5950X at first, but deside to try and i do not regret it.
 
On of the reasons is max power limit on AM4 is what ~170W while on AM5 it's much higher.
 
Tomgang said:
When it comes to Zen 4 and intels 12 and 13 gen i am well aware of it. These chips redlines all the time and has much higher TDP ratings than Zen 3. So off cause these chips runs hot and needs very good cooling.

Zen 3 has lower ratings and does not reach clock speeds of nearly 6 GHz. exsample 5950X is rated 105 watt stock while realty is 141 watt and 7950X is rated for 170 watt while realty its like 230 watt and intels I9 chips can sip around 300 watts. So i am in no douts new chips are hard to cool. AMDand Intel let the chips redlines as stated before when they can. First throttle when max temp or power limits are reach. So if your cooling is not up to the task, you will end up with at hot runnings thermal throtling cpu all the time.


Nope neither here. 5950X is not a problem to cool. also why i am still on aircooling for my 5950X still. I was hesitant to go air with 5950X at first, but deside to try and i do not regret it.
Click to expand...
Stock Zen 3 is super easy to cool. I run my 5900X at 245/165/190 and she rips to 5150. Granted it’s not 5500+. If you were curious, she will do the full 245w PPT with just PBO.
 
A Computer Guy said:
Anyone have any recommendations on the right equipment/parts to use for temperature monitoring that can be picked up by HWiNFO64?

I want to attach multiple thermal sensors to a used Master SLI/ac (the motherboard that started this thread) that I was able to pick up dirt cheap.
Various locations will be on or around the PCB, cpu socket, and VRM.
Click to expand...
Corsairs commander pro works really well, you can set fan speeds static that it remembers and has 4x temp probes
1694334154147.png

The new core XT variant only has 2 temp probes, but its easier to get.

1694334174259.png



They work fine in HWinfo as long as you don't have iCue installed.

R0H1T said:
On of the reasons is max power limit on AM4 is what ~170W while on AM5 it's much higher.
Click to expand...
142W

It's only all core OC's that go past that limit, unless PBO has been disabled or altered - some boards do this by default, which is a problem.

A Computer Guy said:
I didn't think the 5950X was a problem to cool at all. Barely an inconvenience.
Click to expand...
Dual CCX means it's very easy to cool vs a 5800x, it has double the surface area

The single CCX chips are generally better for gaming, but pushed to their limits heat density is an issue - which cannot be compared to how intel chips behave since it's all about the IHS limiting heat transfer, and not something a bigger cooler can help with.
 
Mussels said:
Corsairs commander pro works really well, you can set fan speeds static that it remembers and has 4x temp probes
View attachment 312908
The new core XT variant only has 2 temp probes, but its easier to get.

View attachment 312909


They work fine in HWinfo as long as you don't have iCue installed.
Click to expand...
I was kind of hoping there was something with like 8 to 16 sensors on it so I don't have to build my own. (as fun as that sounds)
 
A Computer Guy said:
I was kind of hoping there was something with like 8 to 16 sensors on it so I don't have to build my own. (as fun as that sounds)
Click to expand...
Nothing exists with that many sensors and it gets a bit pointless - you cant put sensors IN the components so you're dealing with lower temps than they are internally as well as items with a heatsink will all equalise to the same temperature in the long run anyway
 
Mussels said:
Nothing exists with that many sensors and it gets a bit pointless - you cant put sensors IN the components so you're dealing with lower temps than they are internally as well as items with a heatsink will all equalise to the same temperature in the long run anyway
Click to expand...
I was thinking I can potentially detect when/where the hotspots on the front and back of pcb are occurring and hopefully fast enough to throttle, stop, or shutdown the experiment before real damage kicks in unlike before I had no warning until it actually started to burn.
 
A Computer Guy said:
I was thinking I can potentially detect when/where the hotspots on the front and back of pcb are occurring and hopefully fast enough to throttle, stop, or shutdown the experiment before real damage kicks in unlike before I had no warning until it actually started to burn.
Click to expand...
By the point things are that hot, the odds on the PC working correctly enough to trigger a shutdown are basically zero. You'd need to wire up these sensors to an external controller with a cutoff to the mains power and hope that the heat isn't in a place you can't measure.
 
I only monitor one temp in my setup and that's water temp. Everything else is determined with water temp in mind. In other words pump/fans monitor water temp.
 
Mussels said:
By the point things are that hot, the odds on the PC working correctly enough to trigger a shutdown are basically zero. You'd need to wire up these sensors to an external controller with a cutoff to the mains power and hope that the heat isn't in a place you can't measure.
Click to expand...
I was thinking I don't need to totally uncap the system like I did before but nudge it up enough to see the pattern of temperature based on the damaged areas I already documented in pictures.
Of course this could just be a fools errand perhaps all was fine up into the split second (assuming) thermal runaway took over.
 
thesmokingman said:
I only monitor one temp in my setup and that's water temp. Everything else is determined with water temp in mind. In other words pump/fans monitor water temp.
Click to expand...
That's helpful for fan speeds and not much else. Certainly not something you could use for any sort of safety measure.

A Computer Guy said:
I was thinking I don't need to totally uncap the system like I did before but nudge it up enough to see the pattern of temperature based on the damaged areas I already documented in pictures.
Of course this could just be a fools errand perhaps all was fine up into the split second (assuming) thermal runaway took over.
Click to expand...
I did this with the 2500k test system I just ran, past certain wattages/voltages the VRM efficiency went to total and utter shite.
Screenshot 2023-09-10 162427.png



Looking into the board now to find out the maximum amperage the VRM's can put out, so you can get a safe limit to begin your testing with

Measure your power draw at the EPS connectors or at the wall vs what software is saying, once you get the drastic differences like i ran into here you'll know you're in the danger zone for the boards power delivery.

4.5GHz had 27% 'wasted' power as heat, but maxing it out at 1.6v for 5.3GHz kicked that to 97% - and even with a heatsink and direct airflow they were pushing 90c. Without that fan, they'd have likely gone kaboom like yours did.


The moment you run a static OC, all safeties are disabled on the board - only the CPU's thermal throttle is active.
PBO with maxed out values is the same, those values *are* the safeties.


Buildzoids video on the Asrock X470 Master SLI/ac

They claimed 12 phase, 45A each (540A total/peak) - an extremely high end value we know can't be the full story
(Edits to come)


Low side fets are
Nikos pk618ba

Datasheet here
PK618BA Datasheet(PDF) - NIKO SEMICONDUCTOR CO., LTD. (alldatasheet.com)

1694404364045.png


So at 70C, they're 12A each
12x12 = 144A, which is barely enough within PBO limits. You'll need active cooling on the VRM's (heatsink and fans) to even reach 15Ax12 (180A)

I cant recall the value but we found 5950x values of over 200A with PBO overclocking in one of these threads
 
Last edited:
Mussels said:
That's helpful for fan speeds and not much else. Certainly not something you could use for any sort of safety measure.


I did this with the 2500k test system I just ran, past certain wattages/voltages the VRM efficiency went to total and utter shite.View attachment 313056


Looking into the board now to find out the maximum amperage the VRM's can put out, so you can get a safe limit to begin your testing with

Measure your power draw at the EPS connectors or at the wall vs what software is saying, once you get the drastic differences like i ran into here you'll know you're in the danger zone for the boards power delivery.

4.5GHz had 27% 'wasted' power as heat, but maxing it out at 1.6v for 5.3GHz kicked that to 97% - and even with a heatsink and direct airflow they were pushing 90c. Without that fan, they'd have likely gone kaboom like yours did.


The moment you run a static OC, all safeties are disabled on the board - only the CPU's thermal throttle is active.
PBO with maxed out values is the same, those values *are* the safeties.


Buildzoids video on the Asrock X470 Master SLI/ac

They claimed 12 phase, 45A each (540A total/peak) - an extremely high end value we know can't be the full story
(Edits to come)
Click to expand...
Don't forget I was using that monoblock. I think the VRM didn't blow out sooner because of that monoblock but instead the PCB traces to the CPU took the brunt of the damage and that blew up instead, not just once but twice. My thought is with no OCP on board the monoblock enabled a potentially more hazardous situation when over stressing the VRM and it became like a delayed fuse bomb.
 
Last edited:
Mussels said:
That's helpful for fan speeds and not much else. Certainly not something you could use for any sort of safety measure.
Click to expand...

:rolleyes:
 
A Computer Guy said:
Don't forget I was using that monoblock. I think the VRM didn't blow out sooner because of that monoblock but instead the PCB traces to the CPU took the brunt of the damage and that blew up instead, not just once but twice. My thought is with no OCP on board the monoblock enabled a potentially more hazardous situation when over stressing the VRM and it became like a delayed fuse bomb.
Click to expand...
edited in the information, the VRM's cant provide past 15A each even if they're at 25c

The PWM controller/wiring they used doesnt use a proper doubler setup either - which is bad and I'll need to figure out how to explain that problem better
Buildzoid explains its more like a 4 phase with 3 of each component stuck side by side, since the PWM controller isn't aware they exist it can't alternate the timing correctly to take advantage of the extra components, no cooldown time, no better spread of the heat.

Oh god was that their thinking - 3x15A = 45A?
Was it designed for a 45A phase, and they slapped three capable of 15A in parallel expecting it to be the same?

The controller is designed for 3+1, so it seems like that actually IS what they've done. Madness.

Speaking of 3+1, that means only 3/4 of these are on the CPU and that value may be even lower than expected, 15A x8 for the CPU (120A)?
Without probing the parts it's hard to know how many go to each phase - BZ does this in the video, but not to find the answer to that question
Stock is 142w/90A/140A... so this board can't even sustain that?
I can't find the 3000 series stock PBO values listed anywhere, in case they were lower and thats what it was designed for.

thesmokingman said:
:rolleyes:
Click to expand...
helpful.
He asked for advice on a particular need, not for how you deal with an entirely unrelated topic.
 
Last edited:
Mussels said:
helpful.
He asked for advice on a particular need, not for how you deal with an entirely unrelated topic.
Click to expand...

iirc computer guy has long since gone to a far better board though, no reason he can't join the rest of the custom loop master race and just run fans off water temp alone. Really no active airflow needed under any normal PC use circumstances for upper mid-range and above boards.
 
tabascosauz said:
iirc computer guy has long since gone to a far better board though, no reason he can't join the rest of the custom loop master race and just run fans off water temp alone. Really no active airflow needed under any normal PC use circumstances for upper mid-range and above boards.
Click to expand...
Just to clarify on my x570 I do run my fans off of water temp but keep pump at fixed % to manage noise because of the distroplate.

Mussels said:
edited in the information, the VRM's cant provide past 15A each even if they're at 25c

The PWM controller/wiring they used doesnt use a proper doubler setup either - which is bad and I'll need to figure out how to explain that problem better
Buildzoid explains its more like a 4 phase with 3 of each component stuck side by side, since the PWM controller isn't aware they exist it can't alternate the timing correctly to take advantage of the extra components, no cooldown time, no better spread of the heat.

Oh god was that their thinking - 3x15A = 45A?
Was it designed for a 45A phase, and they slapped three capable of 15A in parallel expecting it to be the same?

The controller is designed for 3+1, so it seems like that actually IS what they've done. Madness.

Speaking of 3+1, that means only 3/4 of these are on the CPU and that value may be even lower than expected, 15A x8 for the CPU (120A)?
Without probing the parts it's hard to know how many go to each phase - BZ does this in the video, but not to find the answer to that question
Stock is 142w/90A/140A... so this board can't even sustain that?
I can't find the 3000 series stock PBO values listed anywhere, in case they were lower and thats what it was designed for.


helpful.
He asked for advice on a particular need, not for how you deal with an entirely unrelated topic.
Click to expand...
Keep in mind it had no problem handling 3950x or 5950x (at stock) with the monoblock for quite a long time prior to the incident.

It's been a long time but when the 5950x was stock I recall running all core workloads at 60c to 65c (for like an hour) and the vrm was often also around 60c :kookoo: and that's when I only had the EK 360 PE rad at maybe 60% fan speed. Now that I just wrote that I see now the numbers aligned like an evil omen :twitch:.
 
Last edited:
to sum this thread up Way to much current on a so-so motherboard = melted stuffs
 
OneMoar said:
to sum this thread up Way to much current on a so-so motherboard = melted stuffs
Click to expand...
Plus bonus features for additional entertainment value. I took the liberty of indexing them in this post for everyone's amusement.
www.techpowerup.com

5950x and PBO overheated something or busted custom loop?

Anyone have any recommendations on the right equipment/parts to use for temperature monitoring that can be picked up by HWiNFO64? I want to attach multiple thermal sensors to a used Master SLI/ac (the motherboard that started this thread) that I was able to pick up dirt cheap. Various...
www.techpowerup.com www.techpowerup.com
 
A Computer Guy said:
Just to clarify on my x570 I do run my fans off of water temp but keep pump at fixed % to manage noise because of the distroplate.


Keep in mind it had no problem handling 3950x or 5950x (at stock) with the monoblock for quite a long time prior to the incident.

It's been a long time but when the 5950x was stock I recall running all core workloads at 60c to 65c (for like an hour) and the vrm was often also around 60c :kookoo: and that's when I only had the EK 360 PE rad at maybe 60% fan speed. Now that I just wrote that I see now the numbers aligned like an evil omen :twitch:.
Click to expand...
I was gonna say a D5 is rather quiet so that distro must create a bit of noise.

I've got a 3900x in a x370 board... it's like hanging on for dear life that 3900x is bone stock. Anymore and that board will probably pop.
 
You must log in or register to reply here.
Back
Top