AMD Ryzen 1600X & TMPIN4?

[Sirholg]

Does anyone know what TMPIN4 is? I can't make sense of it.

I'm doing some basic clocking on my 1600X, and that sensor keeps approaching 90c after 10min of Prime95 Small-FFT test.
I think it's a MB sensor, but it rises with the OC setting. Currently Running 3.9Ghz with +1.3875V stable.

HWInfo is during load.

 

[Blaylock]

I'm guessing it's your VRM's. I would try using Aida64 for your stress test. Prime95 has been know to stress unrealistically with the addition of AVX2.

If you'd like to know for sure if it's your VRM sensor hit the Heatsink with a long shot of compressed air with it under stress. The cold air should cause that sensor to drop quite a few degrees.

EDIT: Just checked the last 1600X build I had and your voltage is right on par with mine. I was at 1.40v @ 4.0Ghz.

 

[trents]

Your CPU vcore is getting up there, exceeding 1.4. If I were you I would check those temps with Ryzen Master. Seems like I've read on this forum that other temp monitoring software gives some inaccurate results.

 

[Thentilian]

Does anyone know what TMPIN4 is? I can't make sense of it.

I'm doing some basic clocking on my 1600X, and that sensor keeps approaching 90c after 10min of Prime95 Small-FFT test.
I think it's a MB sensor, but it rises with the OC setting. Currently Running 3.9Ghz with +1.3875V stable.

HWInfo is during load.

What cooler are you using for this? I know that if I don't use a water cooler with my 1600 with that much voltage, (1.381 specifically) the included stock cooler would cause failure during load and hit around 90C on the core. Granted I was only doing this for a cooler review, I knew it was more than likely going to cause instability and fail. But with water cooling, I only hit about 70C at the peak in a room that was about 25C ambient temp when I did the testing. Was a hot day that one...

I'm guessing it's your VRM's. I would try using Aida64 for your stress test. Prime95 has been know to stress unrealistically with the addition of AVX2.

If you'd like to know for sure if it's your VRM sensor hit the Heatsink with a long shot of compressed air with it under stress. The cold air should cause that sensor to drop quite a few degrees.

EDIT: Just checked the last 1600X build I had and your voltage is right on par with mine. I was at 1.40v @ 4.0Ghz.

I also think this is your VRM temp sensor as well, unless you are using some kind of air cooler, I would expect those temps.

Your CPU vcore is getting up there, exceeding 1.4. If I were you I would check those temps with Ryzen Master. Seems like I've read on this forum that other temp monitoring software gives some inaccurate results.

I'd be wary of trusting the software telling you what voltage you are running. I'm curious as to what settings you are using specifically? As I am using offset to hit my voltage with LLC 3 on an Asus Prime X370-Pro motherboard. I have been tinkering with this for a couple of months now, and I have a pretty solid handle on OCing with this board and what to expect from the CPU and SOC while overclocking.

I saw some very high readings in voltage from HWINFO as well. One sensor is completely off. Says I hit upwars of 1.46 or 1.47 volts peak, which I'm not. I ran IBT on a custom setting using 12GB of RAM, and used my multi-meter to measure voltage output for the SOC and CPU at the back of the socket. I had 1.37 to 1.38 voltage during the entire run of IBT, even after it finished, I didn't see any transient peak, but that could of happened faster than my multi-meter could detect. But this is one reason I'm using LLC 3, and not 5. 5 is too extreme and buildzoid himself says to never use extreme LLC settings, it's just too much.

SOC voltage was a solid 1.0 to 1.1 volt at the back of the socket as well. I bumped it up for stability even though my RAM is only clocked to 2400 currently. 2933 is possible, but I have to run 1.38v on the RAM and 1.08 SOC to get it there, and I just don't see a difference in speed when gaming. Probably because my GPU is already fed well enough. Say if I had a GTX 1070 or 1080 rather than my R9 390, I'd probably see some FPS gains.

Anyhow, went off target a bit, but that's just my experience with my board, so listing some more specs of your board, and what kind of cooler you are using, could narrow down what that temp sensor is. It's probably different for every motherboard. I don't appear to even have a VRM sensor, as I can't find any on HWINFO that looks to be the right temps and I haven't really looked into it, since I have a fan blowing air down over the VRM with my AIO set in a pull configuration rather than push.

 

[Sirholg]

Anyhow, went off target a bit, but that's just my experience with my board, so listing some more specs of your board, and what kind of cooler you are using, could narrow down what that temp sensor is. It's probably different for every motherboard. I don't appear to even have a VRM sensor, as I can't find any on HWINFO that looks to be the right temps and I haven't really looked into it, since I have a fan blowing air down over the VRM with my AIO set in a pull configuration rather than push.

Thanks for all the replies.

Turns out you're correct, it is the VRM sensor according to Gigabyte software I installed.

I'm using a Corsair H60 with only pull. Functions as exhaust aswell. Have three 120mm Corsair SP120 x1 (Bottom HDD Rack) and AF120x2 to blow cold air over the system.
Temps on CPU not OC'ed never exeeds 55c at full load (Prime95 Small-FFT). Other components gets good temps aswell, but the VRM seems to be high when not using air cooled CPU.

Never thought of that as a drawback with AIO. Thanks, that makes sense.

As i have good CPU temps with the H60, even when clocked at 3.9-4.0Ghz - I don't feel like having to install Push\Pull on it. Maybe mod something to have my extra SP120 blow air on the VRM Controller.

As far as Cooling is concerned, I'm out of FAN connectors on my MB (5). I want the ability to control them in BIOS as i do now. A PWM splitter, would that work without causing problems with power supply to the connected FAN pin?

Specs:

• Gigabyte GA-AX370 Gaming K3 MB
• Gigabyte Geforce GTX 1060 OC Windforce 3GB
• AMD Ryzen 5 1600X CPU
• Corsair Vengeance LPX DDR4 2666MHz 16GB
• Corsair RM650x 650W PSU
• Corsair H60 Water cooling
• Kingston SSDNow UV400 240GB 2.5" SSD
• 2x Seagate Firecuda 1TB SSHD
• Cooler Master MasterBox 5
• Windows 10 64bit
• 1x Corsair SP120 Led fans & 2x Corsair AF120 Led fans

 

[Blaylock]

Yes adding spot fans directly on the VRM's will help. You can split your headers but make sure you dont draw more than each mobo header can provide (usually 1A).

Here I added 2x60mm's on an old AMD FX build known for VRM temp issues.

 

[Sirholg]

Yes adding spot fans directly on the VRM's will help. You can split your headers but make sure you dont draw more than each mobo header can provide (usually 1A).

Nice. How did you make those stick onto the heatsink? Zipties? Any advice?

 

[Blaylock]

I've seen people use zip ties, double sided tape, or in the pic above I found some Niteize cable ties with an adhesive pad on one side. It worked perfect. Looks like I used a couple black bread ties to hold the fans together too.

 

[Thentilian]

Thanks for all the replies.

Turns out you're correct, it is the VRM sensor according to Gigabyte software I installed.

I'm using a Corsair H60 with only pull. Functions as exhaust aswell. Have three 120mm Corsair SP120 x1 (Bottom HDD Rack) and AF120x2 to blow cold air over the system.
Temps on CPU not OC'ed never exeeds 55c at full load (Prime95 Small-FFT). Other components gets good temps aswell, but the VRM seems to be high when not using air cooled CPU.

Never thought of that as a drawback with AIO. Thanks, that makes sense.

As i have good CPU temps with the H60, even when clocked at 3.9-4.0Ghz - I don't feel like having to install Push\Pull on it. Maybe mod something to have my extra SP120 blow air on the VRM Controller.

As far as Cooling is concerned, I'm out of FAN connectors on my MB (5). I want the ability to control them in BIOS as i do now. A PWM splitter, would that work without causing problems with power supply to the connected FAN pin?

Specs:

• Gigabyte GA-AX370 Gaming K3 MB
• Gigabyte Geforce GTX 1060 OC Windforce 3GB
• AMD Ryzen 5 1600X CPU
• Corsair Vengeance LPX DDR4 2666MHz 16GB
• Corsair RM650x 650W PSU
• Corsair H60 Water cooling
• Kingston SSDNow UV400 240GB 2.5" SSD
• 2x Seagate Firecuda 1TB SSHD
• Cooler Master MasterBox 5
• Windows 10 64bit
• 1x Corsair SP120 Led fans & 2x Corsair AF120 Led fans

Nice. How did you make those stick onto the heatsink? Zipties? Any advice?

I took a look at your case specs, I was going to suggest that you put a fan in the top of the case as an intake fan, but your case doesn't have any options for fans at the top of the case. Best bet would be to do what Blaylock did with his motherboard if you are that worried about VRM temps. 88C might seem really hot, but VRMs are usually good for over 100C before there is an issue with them. Over an extended time doing a long video edit or render (if you do such things) that's when the VRM could reach much higher temps with an extended load.

Also, looking at your readouts it looks like your CPU temps are hitting 64C rather than 55C. The temp node 0 reading is more than likely your CPU die temp. My board has two temp readings for the CPU. One is CPU Tdie, which is the one I pay attention to, which is hitting 64.3C right now with P95 running small FFT's in the background, and my other temp reading is CPU which is 57C. I'm using a Corsair H150i Pro which is a 360mm rad, your temps seem way too low for a single rad cooler and the voltage you are running.

I suspect your voltage readout is faulty and you are probably running less volts under load then your software says. Given the volts you posted, you should be running slightly warmer than me to begin with, since my voltage is slightly lower, but I'm running hotter temps with a much bigger cooler. If the software is off anywhere close to how mine reads, it's probably more like 1.35 or 1.36 volts under load that you are seeing. My 1600 (non-x) takes more voltage than others to hit 3.9Ghz, but with your 1600X it looks to be a bit better chip than mine. I can't even get 4Ghz to run on my machine, even with 1.45 volts. It'll boot, but crash when doing any kind of stress test.

EDIT: I took a shot of my fans in the top of the case for reference.

 

[Blaylock]

There could be a couple factors for the package temps to be inaccurate. If memory serves me correctly some of the older BIOS's for Zen displayed incorrect readings due to the way the sensors work (max T ∆ iirc). Newer BIOS's have this issue corrected. Since OP hasn't listed his BIOS it's not possible to confirm ATM. Also, running multiple monitoring software can lead to inaccurate readings.

 

[Sirholg]

Also, looking at your readouts it looks like your CPU temps are hitting 64C rather than 55C. The temp node 0 reading is more than likely your CPU die temp. My board has two temp readings for the CPU. One is CPU Tdie, which is the one I pay attention to, which is hitting 64.3C right now with P95 running small FFT's in the background, and my other temp reading is CPU which is 57C. I'm using a Corsair H150i Pro which is a 360mm rad, your temps seem way too low for a single rad cooler and the voltage you are running.

I don't know. Those temps was under no overclock. When OC'ed I've seen it go to 60-65c, but not higher (At 4Ghz). HWMonitor, gigabyte Xtreme Engine, AIDA64 and more all report the same temps. My motherboard likes to pretend it's cooler (In Gigabyte's SIV APP) for some reason. Also, during normal operating states this Ryzen CPU will agressively clock itself between 3.7Ghz (Multiplier set 36) and 4.1Ghz, this is due to it detecting headroom through AI and core boost. The voltage will adjust itself as it pleases, all the way up to 1.5V sometimes to get that 4.1Ghz turbo. This is normal for a Ryzen 1600X to do (If allowed in BIOS, and ON by default). Using the Ryzen power plan also. This voltage peak might seem high, but it's the CPU itself that discharges that by force to reach 4.1Ghz Precicion boosts.

It's XFR that hammers that voltage up like that, and I don't think I'll keep it on.

The CPU tDIE node is reporting more or less accurate, I've stopped looking at that other CPU surface node - it's way off and tells me it's running at 15c or so.

 

[trents]

Nice. How did you make those stick onto the heatsink? Zipties? Any advice?

I've used double sided adhesive strips with good results but take the paper label off the fan first as the adhesive under the label will let go at some point.

 

[Sirholg]

I was at 4Ghz OC for a while now, and tried 4.1 with voltage on auto but then had to reset CMOS. 4.1 works, atleast with Ryzen Master. I won't dare to up the voltage more without some more cooling on the VRM controller as it will at 4Ghz + go up into the 100c area. Sticking with a 3.9 @ 1.392V for now. Temps at this clock and higher is very good on the CPU. Current 3.9Ghz has a max load temp of 65c and idle down to 30c, but the VRM temp is shooting up with more power and I fear it's not good on the board.

Current VRM temp with 3.9Ghz OC is maxed out at 97c, would you consider this "safe"? Temperature achieved using Prime95 Small-FFT run over atleast 20min.

 

[Blaylock]

So, just out of curiosity here, what will you be using this rig for?

Here's my thought, if it's only gaming and surfing on a single core mostly you'll be better off leaving it at stock as the turbo will auto OC to 4.1 on single core.

If you're doing video encoding or gaming on multi cores then the OC would make sense.

 

[Sirholg]

So, just out of curiosity here, what will you be using this rig for?

Here's my thought, if it's only gaming and surfing on a single core mostly you'll be better off leaving it at stock as the turbo will auto OC to 4.1 on single core.

If you're doing video encoding or gaming on multi cores then the OC would make sense.

Yeah, I've been thinking that myself. I'm not going for much video enc or MT games, and I play at 1080P. I just got bitten with the OC virus after having good results with my GPU OC. Stable at +120 core and +780Mhz (1.6Ghz) GPU RAM speed.
Loving to see higher bench results and the feeling I get when achieving #2 for my system type in 3DMark's Time Spy.... hehe.

 

[Thentilian]

Yeah, I've been thinking that myself. I'm not going for much video enc or MT games, and I play at 1080P. I just got bitten with the OC virus after having good results with my GPU OC. Stable at +120 core and +780Mhz (1.6Ghz) GPU RAM speed.
Loving to see higher bench results and the feeling I get when achieving #2 for my system type in 3DMark's Time Spy.... hehe.

Yeah, I forgot about the XFR on the "X" chips. I game, so having mine at 3.9Ghz is best for me. And I'm on a 1600 so I don't get the XFR boost anyways. I don't think you need to worry about your VRM temps that much. Most are rated for 115 to 125C. 97 is hot, but if you get some air flow over them, probably drop them 10 to 15C and you'll have nothing to worry about.

The XFR doing it's thing, will cause more VRM heat, if it's really peaking 1.5 volts, which I kind of doubt, as AMD recommends a max of 1.45 volts. That software reads the voltages wrong. Only way to know is to check it with a multi-meter at the back of the socket. If you use offset, which I think is the only way you can OC on that board, I would do that, rather than setting a 1.39 volts. It'll override C-states and run that voltage all the time, which would lead to more heat in the VRM.

So in the end, I guess it depends on if you want the XFR to function, or back off of it, and go for a 3.9 or 4.0 OC with offset voltage. Either way finding a way to put some air over the VRM is a good idea.

 

[Sirholg]

So in the end, I guess it depends on if you want the XFR to function, or back off of it, and go for a 3.9 or 4.0 OC with offset voltage. Either way finding a way to put some air over the VRM is a good idea.

Yep. I did some tests with XFR disabled. I do get some extra juice in heavy games or heavy apps that suddenly draw x amout of power with XFR enabled, which I guess is the purpose of XFR.
I monitored temps under both configurations and they are more or less the same. VRM goes to a maximum of 80c after 20min Small-FFT etc and stays there. CPU however stays at a constant 3.7Ghz compared to 3.6Ghz with this setting off.
That is because XFR also mesures the heat and cooling solution used and then clocks up 100Mhz by default, without any gains in temp (maybe 2c). + adding a boost of 4.1-4.2Ghz when it wants to (one or two cores at a time).

So, Precicion Boost\XFR gives me an automatic OC of +100Mhz constant and boosts if needed on one or two cores to 4.1-4.2Ghz at good temps.
Voltage may be faulty read by monitors, but it does atleast jump up\down very fast as the CPU AI tells it to. Nothing to worry about, and normal.

After trying OC'ing and testing that performance against XFR with my board, I reached a conclusion. I'll stick with XFR on, as temps are well within range and the performance is very close to overclocked @ 4.Ghz.
If I get that tiny bit extra from OC, I'll rather upgrade the MB with more VRM load dividers and better cooling later.

 

[Blaylock]

Yep. I did some tests with XFR disabled. I do get some extra juice in heavy games or heavy apps that suddenly draw x amout of power with XFR enabled, which I guess is the purpose of XFR.
I monitored temps under both configurations and they are more or less the same. VRM goes to a maximum of 80c after 20min Small-FFT etc and stays there. CPU however stays at a constant 3.7Ghz compared to 3.6Ghz with this setting off.
That is because XFR also mesures the heat and cooling solution used and then clocks up 100Mhz by default, without any gains in temp (maybe 2c). + adding a boost of 4.1-4.2Ghz when it wants to (one or two cores at a time).

So, Precicion Boost\XFR gives me an automatic OC of +100Mhz constant and boosts if needed on one or two cores to 4.1-4.2Ghz at good temps.
Voltage may be faulty read by monitors, but it does atleast jump up\down very fast as the CPU AI tells it to. Nothing to worry about, and normal.

After trying OC'ing and testing that performance against XFR with my board, I reached a conclusion. I'll stick with XFR on, as temps are well within range and the performance is very close to overclocked @ 4.Ghz.
If I get that tiny bit extra from OC, I'll rather upgrade the MB with more VRM load dividers and better cooling later.

This is the exact same conclusion I came up with for the DismalDIRGE build I did. I did leave a 4.0Ghz overclock profile in the BIOS for Dismal for the few times he does video encoding, but for 24/7 running XFR was eaiser on the processor (IMO) than running OC'd with higher volts and temps with minimal real gains.

 

[Thentilian]

Yep. I did some tests with XFR disabled. I do get some extra juice in heavy games or heavy apps that suddenly draw x amout of power with XFR enabled, which I guess is the purpose of XFR.
I monitored temps under both configurations and they are more or less the same. VRM goes to a maximum of 80c after 20min Small-FFT etc and stays there. CPU however stays at a constant 3.7Ghz compared to 3.6Ghz with this setting off.
That is because XFR also mesures the heat and cooling solution used and then clocks up 100Mhz by default, without any gains in temp (maybe 2c). + adding a boost of 4.1-4.2Ghz when it wants to (one or two cores at a time).

So, Precicion Boost\XFR gives me an automatic OC of +100Mhz constant and boosts if needed on one or two cores to 4.1-4.2Ghz at good temps.
Voltage may be faulty read by monitors, but it does atleast jump up\down very fast as the CPU AI tells it to. Nothing to worry about, and normal.

After trying OC'ing and testing that performance against XFR with my board, I reached a conclusion. I'll stick with XFR on, as temps are well within range and the performance is very close to overclocked @ 4.Ghz.
If I get that tiny bit extra from OC, I'll rather upgrade the MB with more VRM load dividers and better cooling later.

Glad you got it sorted then. 80 to 82 on the VRM is not bad. Cooler is always better. Keeping everything cooler helps to extend the life of components, after all. I think if I had a 1600X, I probably would of done the same, and left it stock with XFR. Since I have a 1600 though, it's stock (3.2) and boost clocks (3.6) are not that great, so an everyday 3.9 OC is going to be better in this instance.

I managed to keep my MSI 870-G45 alive with a 3.9Ghz 24/7 OC for many years with my Phenom II 965. In fact, after I upgraded to Ryzen, I gave that board, CPU, and RAM to my mom. That MSI board was notorious for having a bad VRM, had no sinks on it, so I kept air on it all the time, and it's still rocking. I put the clock down to 3.5 with stock voltage though, as my mom doesn't need an OC. I tried to squeeze every last ounce of performance, even had the NB clocked to 2600, it's now back to 2000 as well.

 

[Sirholg]

Thank you all for your input, it helped me reach a conclusion and learn quite a few things in the proccess. Researching myself and listening to your replies has truly given me more insight in OC and system stability!