AMD EPYC 7401P running temperatures

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

anomaly

Active Member
Jan 8, 2018
235
48
28
Any owners of AMD EPYC 7401P based systems that could give some information about the running temperatures they are experiencing with different workloads?

I'm considering building an EPYC system but see some challenges with making it run cool and quiet, as the TDP probably requires a lot of thermal capability in the chassis (which rules out short depth ones, I guess).
 

MiniKnight

Well-Known Member
Mar 30, 2012
3,072
973
113
NYC
P is single socket. Just get Threadripper quiet solutions. There are many out there. "Running temperatures" depend on a lot of factors in servers right? Heatsinks, ambient, chassis fans, active cooling on the heatsink, workloads. Still, you're either using a barebones that comes with a heatsink or you're building a desktop type system where you'd just use a TR cooling solution.
 

anomaly

Active Member
Jan 8, 2018
235
48
28
While I appreciate you taking time to respond, you are not adding value to the thread per se. It is obviously implied that when somebody asks for running temperatures, a response describing the setup will be descriptive of the system itself. My question was verbalized clearly to incite people to elaborate such details, including the workloads associated with it.

Don't be offended, I just think this is a typical occurrence in every online forum there is, and the signal to noise ratio gets skewed in the process. An actually useful response would include practical non vague information, like what TR cooling options you use (if you have an EPYC system) in rack mounted equipment. If any custom ducts were used, etc.
 

Tha_14

Server Newbie
Mar 9, 2017
72
10
8
While I appreciate you taking time to respond, you are not adding value to the thread per se. It is obviously implied that when somebody asks for running temperatures, a response describing the setup will be descriptive of the system itself. My question was verbalized clearly to incite people to elaborate such details, including the workloads associated with it.

Don't be offended, I just think this is a typical occurrence in every online forum there is, and the signal to noise ratio gets skewed in the process. An actually useful response would include practical non vague information, like what TR cooling options you use (if you have an EPYC system) in rack mounted equipment. If any custom ducts were used, etc.
Hello
 

MiniKnight

Well-Known Member
Mar 30, 2012
3,072
973
113
NYC
You never specified "rackmount" system in your OP. It was extremely vague. I read your OP and thought you were trying to do a short depth ATX build or something like that.

If you are going rackmount, just buy a barebones or completed system. They're reasonable these days and the TR coolers will be too big. If something goes wrong, you have a vendor support to fall back on.

In a rackmount, you want chassis fans cooling a passive heatsink. So you need a shroud that fits the chassis and the heatsink. EPYC in rackmount has been solved with dozens of models. Temperatures in these cases are set via fan speed and BIOS settings so I'm lost as to how this fits with the ask in the OP.

I appreciate you like people to only think your way. For your question, it seems the answer is just buy a rackmount server. We have no issues cooling any of the brands we have used so far including HPE, Supermicro and Dell. If you aren't going to just use a barebones, then you're in a small minority of people. EPYC isn't a big market share yet and those who have the systems are most likely to buy at least barebones that are assembled by their resellers or complete systems from the factory. Both where cooling isn't an issue.
 

anomaly

Active Member
Jan 8, 2018
235
48
28
The problem here is not that I do not appreciate you thinking "differently", the problem was that you did not think at all, or much, when you responded with a message that doesn't provide numbers, or any factual data meaningful to me or other users. You are spewing sales chatter for all anyone could care, "buy a ready made barebones". This is STH, last time I checked not everyone wants a meter-deep rack cabinet. Or wants to buy an off the shelf barebones system. Or hey, maybe they don't suit their needs and require something different.

Let's revisit my original message though:

"Any owners of AMD EPYC 7401P based systems that could give some information about the running temperatures they are experiencing with different workloads?"

So, when are you actually going to share numbers, if you insist on participating in the thread? If reading comprehension is not the problem here, it does seem pretty clear that was the point.

If it is still vague, I can chew all that brain processing for you if it's still too much vagueness to handle:
  • Temperature readings along load / workload information:
    • Idle / 50% / 100% utilization (raw)
    • What kind of duty the system is performing.
  • Characteristics of the chassis:
    • Airflow/thermal capability: CFM/fan setup.
    • Depth
    • Rackmount/tower
    • etc...
Do you need also optimization flags of the toolchain in the OS to be specified to narrow down the 0.0004% influence on thermal? o_O
 
Last edited:

EffrafaxOfWug

Radioactive Member
Feb 12, 2015
1,394
511
113
So, I'm being negative or hurting some snowflakey feelings because I insisted on having useful input? LOL!
Perhaps you can try starting with a useful question next time? You say you're contemplating a custom build with unspecified requirements, and then want to know what other people running a completely different build are experiencing. As MiniKnight quite rightfully pointed out, there's dozens of factors that'll affect whatever running temperatures you see, so he's trying to politely nudge you towards giving some more information on the requirements and restrictions you have before you, and yet you seem to take offence at him asking this. Because he doesn't want to say the servers he runs in an Antarctic research station idle at -42°C, only to have you complain about inaccuracy when you install them inside a blast furnace. Context matters.

You haven't hurt anyone's feelings, because no-one really cares about people acting like bratty teenagers on the internet; you just started being rude for no adequately justifiable reason. If you were going to pour scorn on anyone who replied without the numbers you wanted for whatever hypothetical scenario you've got going on in your head, you should have warned people in your first post that you weren't interested in why's and wherefore's.

(Not an Epyc owner, so I'm not remotely qualified to even look at your posts TBH)
 

chilipepperz

Active Member
Mar 17, 2016
212
64
28
54
(Not an Epyc owner, so I'm not remotely qualified to even look at your posts TBH)
As an EPYC owner, @MiniKnight added the right advice to a question formulated as such.

Being rude is why people like me are just not going to answer your thread with actual numbers and not answer other threads you post on.

You are asking for a free consulting service which is fine. You are then being rude to people giving honest input for free. That's a great way to alienate yourself from an entire community by being rude to a longtime member with thousands of posts and hundreds of likes in that community.
 

101

Member
Apr 30, 2018
35
12
8
TDP is marketing BS and has little to do with actual power dissipation. Don't let high numbers scare you, or low numbers fool you.