3800X build bad performance - what am I doing wrong?

[A Computer Guy]

MSI x570 Godlike - don’t flame me for overkill reasons

Just gonna share my latest idle temps. I do think everything is just too high in general. Very funny though is that it won't change with the side glass of the case off. So when I don't have a case/airflow issue, why are they so "+high at idle. I'm very curious if it will change with the new board or stay the same:

• [AMBIENCE TEMP: ca. 25°C]
• CPU: 50°C (CPU Core V: 1.406V)
• CPU Package: 62°C
• GPU: 60°C - it is a MSI 2070S. The card has a zero fan profile so it is normal that it's a bit high with no fans, but there are ppl with the same card reporting ca. 15°C lower idle temps
• Motherboard: 48°C
• Chipset: 70°C
• DIMM's: ca. 44°C
• M.2: 60°C (remember in the thread that once it was down to a normal 35°C? it is just fishy
• Upper case temp: 35°C (temp probe)

I think I saw in a prior post you had iCue running. Turn off iCue and Steam and your idle temps and idle voltages should drop. If I recall properly those programs had some stupid high timer resolution set and that basically was ensuring your CPU would end up boosting as frequently as possible. ( I use commander pro hardware mode so I can basically turn off iCue. )

For your comparison I have a 3800x as well. Here are my temps at about ambient 28c (external to the case using a cooking thermometer) on a custom loop.

https://imgur.com/a/wkacVze

 

[ecopsorn]

@tabascosauz @A Computer Guy
Thanks, I will see how it behaves when the mobo arrives.
regarding the case fans, I did even put a tiny feather in front of every fan to confirm that the direction of airflow is correct. Even when I put them at max speed 1600rpm, temps only get down a little bit.

Here are some more data. I cold booted after the computer was off the whole night.
Before I did it, I made sure to have no background processes running, no steam, no icue, no razer, no nothing

values in °C

	Cold Boot (5min. after start of PC)	+20min.	+60min.	5min top fan speeds 1650rpm
Chip	54	63	64.3	62.5
Dims average	28.8	38	39-41	39
Drive	39	52	54	51
Case	28	27.5-39 (second number below GPU)	28-41	28-40
GPU	34	49	44	49
CPU	47	43	40	40
CPU Package	50	49	54	47
Fans	715in x3 / 685out x3	715in x3 / 685out x3	715in x3 / 685out x3	715in x3 / 685out x3

 

[heky]

@tabascosauz @A Computer Guy
Thanks, I will see how it behaves when the mobo arrives.
regarding the case fans, I did even put a tiny feather in front of every fan to confirm that the direction of airflow is correct. Even when I put them at max speed 1600rpm, temps only get down a little bit.

Here are some more data. I cold booted after the computer was off the whole night.
Before I did it, I made sure to have no background processes running, no steam, no icue, no razer, no nothing

values in °C

	Cold Boot (5min. after start of PC)	+20min.	+60min.	5min top fan speeds 1650rpm
Chip	54	63	64.3	62.5
Dims average	28.8	38	39-41	39
Drive	39	52	54	51
Case	28	27.5-39 (second number below GPU)	28-41	28-40
GPU	34	49	44	49
CPU	47	43	40	40
CPU Package	50	49	54	47
Fans	715in x3 / 685out x3	715in x3 / 685out x3	715in x3 / 685out x3	715in x3 / 685out x3

These are normal temperatures! What is your issue again?

 

[TheLostSwede]

These are normal temperatures! What is your issue again?

At idle? No, not really.

 

[heky]

At idle? No, not really.

Really, which one in particular is not normal? He is aircooling in a case that is also not great at airflow...

 

[TheLostSwede]

Really, which one in particular is not normal? He is aircooling in a case that is also not great at airflow...

Where did you get that from?
By the looks of his pictures and choice of fans, it shouldn't be that airflow.

3800X build bad performance - what am I doing wrong?

So for the less informed of us, can you tell me why the board affects memory when the memory controller is in the CPU? What is contained in the board besides the slot and traces that would affect memory compatibility? I have never bought qvl in my life. Also, never had a problem. And no, not...

www.techpowerup.com

Regardless, the CPU shouldn't be over 60 degrees C on idle, even with an air cooler.
Even the stock cooler should keep it around 50 degrees C. Example in the link.

Is 40 - 50c on idle the normal temp for the 3800x?

As I am exploring the "Internets", high temps for the 3800x seem to be a common theme. So is it normal for this cpu to be running on these temps? and what of the longevity? This is my newly built rig: Thermaltake V200 Mid-Tower case with 3 120mm fans in the front and one 120mm Rear Fan...

community.amd.com

 

[heky]

Where did you get that from?
By the looks of his pictures and choice of fans, it shouldn't be that airflow.

3800X build bad performance - what am I doing wrong?

So for the less informed of us, can you tell me why the board affects memory when the memory controller is in the CPU? What is contained in the board besides the slot and traces that would affect memory compatibility? I have never bought qvl in my life. Also, never had a problem. And no, not...

www.techpowerup.com

Regardless, the CPU shouldn't be over 60 degrees C on idle, even with an air cooler.
Even the stock cooler should keep it around 50 degrees C. Example in the link.

Is 40 - 50c on idle the normal temp for the 3800x?

As I am exploring the "Internets", high temps for the 3800x seem to be a common theme. So is it normal for this cpu to be running on these temps? and what of the longevity? This is my newly built rig: Thermaltake V200 Mid-Tower case with 3 120mm fans in the front and one 120mm Rear Fan...

community.amd.com

His case has a glass front so the air has to get sucked trough a really tight gap. The airflow in that case is really bad. Period. Also he is running the fans under 1000rpm, which makes matters worse. (Running fans at higher RPM actually made the temps better)

And where did you get the cpu over 60° on idle from? His last post says something else, check again. Its actually spot on with the article you linked as being the normal temps. So wtf?

 

[TheLostSwede]

His case has a glass front so the air has to get sucked trough a really tight gap. The airflow in that case is really bad. Period. Also he is running the fans under 1000rpm, which makes matters worse. (Running fans at higher RPM actually made the temps better)

And where did you get the cpu over 60° on idle from? His last post says something else, check again. Its actually spot on with the article you linked as being the normal temps. So wtf?

Sorry, you're right, I misread his table
I was looking at the chip temperature, not the CPU.

 

[heky]

Sorry, you're right, I misread his table
I was looking at the chip temperature, not the CPU.

Yeah, kinda figured that must be the case, because i read it that way at first too, lol.

But seriously, his temps are not that bad if you consider that he is also only running the middle fan on the noctua...

 

[TheLostSwede]

Yeah, kinda figured that must be the case, because i read it that way at first too, lol.

But seriously, his temps are not that bad if you consider that he is also only running the middle fan on the noctua...

Yeah, it was much worse before. I guess the OP is just concerned about every little detail at this point, as he's encountered a bunch of odd issues all at once. The temperatures seem to be down about 4-5 degrees on idle now and much more in line where it should be.

 

[ecopsorn]

Yeah, it was much worse before. I guess the OP is just concerned about every little detail at this point, as he's encountered a bunch of odd issues all at once. The temperatures seem to be down about 4-5 degrees on idle now and much more in line where it should be.

yes, I'm just still curious if the SSD temp really has to do with the fact that it is located under the GPU where fans are inactive during idle (won't be the case with the new board), and all the other temps just seem a bit high but to be honest, with all the issues I had, temps were just one thing that I was considering to be a possible root cause.

Once I get my system stable and the performance in line where it should be, I'm sure I won't even check temps anymore
All the temps during load are perfectly fine.

His case has a glass front so the air has to get sucked trough a really tight gap. The airflow in that case is really bad. Period. Also he is running the fans under 1000rpm, which makes matters worse. (Running fans at higher RPM actually made the temps better)

And where did you get the cpu over 60° on idle from? His last post says something else, check again. Its actually spot on with the article you linked as being the normal temps. So wtf?

Actually not a bad point, I will try to remove the glass front to see if that changes anything.

 

[INSTG8R]

I just want to chime in. If you have AIDA64 with a license (I do to, need renew it next month) you don’t need HWInfo as well AIDA can give you all the same info and doesn’t “spike” like HWInfo does.Not too mention running multiple monitoring software polling for readIngs at the same time won’t help anything. I’d look in AIDA to see if it”s seeing your apparently missing RAM voltage. They are amazing at fixing stuff like that quickly if they can. It was showing double my VRAM speed for my 5700XT., asked for some logs, gave me a beta to test, it was fixed and added to the last update..

 

[TheLostSwede]

yes, I'm just still curious if the SSD temp really has to do with the fact that it is located under the GPU where fans are inactive during idle (won't be the case with the new board), and all the other temps just seem a bit high but to be honest, with all the issues I had, temps were just one thing that I was considering to be a possible root cause.

Obviously the SSD is going to be a few degrees hotter under the graphics card, even more so if the fans on the card stop at idle.
Of the two NVMe drives I have, the WD Black one is hotter, both at idle and in use, even though it's above the graphics card. However, it's a spot in my system that doesn't get a ton of cooling. That said, it's usually below 40 degrees C at idle and hits 60+ when heavily loaded and 50-60 when in normal use. That said, I don't see any performance abnormalities running five loops of CrystalDiskMark, unlike what you did earlier on so...
However, from your last set of benchmarks it seems more in line with what it should be.

 

[Waldorf]

@heky
just because fans are running below a certain rpm doesn't automatically mean bad flow.
as long as you dont know how many cfm it produces, rpm is only an indicator.

@ecopsorn
whats your min cpu perf level in the power profile? change it to 20 or 30%, if its higher.
yes, it is supposed to be 99%, but it only makes the cpu much more jumpy on boost clocks (and v/temps),
for virtually no increase in "snappiness" of win.
its not like a pro sprinter runs around the house at max speed to get things done quicker,
same way you dont need to run 8 cores at full boost for 5s just because you opened a browser.
even when gaming, i never seen the need for more than about 30% for min.

install aida (trial) and turn of any other monitoring sw, just to check if thats causing it.
to see if you might need a bit higher min load once rig is running, install afterburner (turn off any power related monitoring in settings) ,
play your fav games, and after closing game, check the cpu load (%) and look at the lowest (peak) on the graph.
adding 5-10% to that and set as min cpu per. this makes sure your not increasing v/t at low loads, while getting enough min output for gaming.

 

[heky]

@heky
just because fans are running below a certain rpm doesn't automatically mean bad flow.
as long as you dont know how many cfm it produces, rpm is only an indicator.

LOL...nothing more to say to that. You can have magical unicorns producing a gazillion CFM behind that glass panel it will do jack! Also, low RPM just can not produce high CFM, basic physics. (At least with the same FAN size)

 

[A Computer Guy]

yes, I'm just still curious if the SSD temp really has to do with the fact that it is located under the GPU where fans are inactive during idle (won't be the case with the new board), and all the other temps just seem a bit high but to be honest, with all the issues I had, temps were just one thing that I was considering to be a possible root cause.

Once I get my system stable and the performance in line where it should be, I'm sure I won't even check temps anymore
All the temps during load are perfectly fine.


Actually not a bad point, I will try to remove the glass front to see if that changes anything.

Yea your NVMe temps looked a bit high but that might be due to different factors including internal case temperature or junk heatsink. Otherwise your CPU temps seem ok to me.

If I recall correctly typically the first NVMe slot under the CPU is the preferred one to use first as those get the PCIe lanes from the CPU instead of from the chipset. Check your motherboard manual to be sure. As you stated your SSD is under your GPU so I imagine it's not getting the coolest air that it can. I kinda miss having side intake fans.

Gamers Nexus has a case review...
https://www.gamersnexus.net/hwreviews/2689-corsair-570x-270r-case-review?showall=1

From the list you indicated you're using Corsair QL120's. I wonder if those fans might not be good enough to pull very well through your case dust filter and limited glass clearance.
Try removing the glass and dust filter to see what different that makes in your airflow. You might consider getting fans that can deal with a higher static pressure for your intake on that case.

 

[Waldorf]

@heky
so what your saying is, there cant be any difference in cfm, based on the amount of blades and/or shape?
then how is it, that a basic 120 mm arctic cooling fan produces the same cfm (74) at 1100 rpm LESS,
than lets say a corsair ML 120?
so i just reduced rpm by 40%, without any loss in cfm/airflow/cooling.

but yeah, lol.

 

[heky]

@heky
so what your saying is, there cant be any difference in cfm, based on the amount of blades and/or shape?
then how is it, that a basic 120 mm arctic cooling fan produces the same cfm (74) at 1100 rpm LESS,
than lets say a corsair ML 120?
so i just reduced rpm by 40%, without any loss in cfm/airflow/cooling.

but yeah, lol.

LOL again...dude you must be smoking some really good shit...

 

[ecopsorn]

Vcore it feeds the chip gives me between 0.025-0.05V of extra voltage to play with, so I always set -0.075V with Turbo LLC to try and minimize droop at lo

you lost me there, I have no idea what Turbo LLC is
I have noted down your post, I did already try undervolting once, but that was with the stock cooler still. Scores did get worse but temps stayed sucky.

I just want to chime in. If you have AIDA64 with a license (I do to, need renew it next month) you don’t need HWInfo as well

I do have AIDA with a licence, but it shows me much less in terms of temps than hwinfo does. Probably I just don't know where to look for it, but under "Sensor", it is only showing mobo,cpu,pch,gpu and ssd temps. It doesn't show me the case temps, nor the dim temps.
It has been discussed in this thread before that hwinfo polling rate can be set in the options if there's a need for it.

@ecopsorn
whats your min cpu perf level in the power profile? change it to 20 or 30%, if its higher.
yes, it is supposed to be 99%, but it only makes the cpu much more jumpy on boost clocks (and v/temps),
for virtually no increase in "snappiness" of win.
its not like a pro sprinter runs around the house at max speed to get things done quicker,
same way you dont need to run 8 cores at full boost for 5s just because you opened a browser.
even when gaming, i never seen the need for more than about 30% for min.

install aida (trial) and turn of any other monitoring sw, just to check if thats causing it.
to see if you might need a bit higher min load once rig is running, install afterburner (turn off any power related monitoring in settings) ,
play your fav games, and after closing game, check the cpu load (%) and look at the lowest (peak) on the graph.
adding 5-10% to that and set as min cpu per. this makes sure your not increasing v/t at low loads, while getting enough min output for gaming.

I will try to set it to 20% until the new mobo is installed. Once I got it all running with the new setup, I can check the cpu load% peak. thanks.

Yea your NVMe temps looked a bit high but that might be due to different factors including internal case temperature.

If I recall correctly typically the first NVMe slot under the CPU is the preferred one to use first as those get the PCIe lanes from the CPU instead of from the chipset. Check your motherboard manual to be sure. As you stated your SSD is under your GPU so I imagine it's not getting the coolest air that it can. I kinda miss having side intake fans.
From the list you indicated you're using Corsair QL120's. I wonder if those fans might not be good enough to pull very well through your case dust filter and limited glass clearance.
Try removing the glass and dust filter to see what different that makes in your airflow. You might consider getting fans that can deal with a higher static pressure for your intake on that case.

Yes the SSD is in the correct - CPU controlled slot, but with the godlike that I'm getting, the "best" slot to put it in is in the lowest available slot according to this thread.

@everyone
so I removed the glass in front of the case and have AIDA open (no other background apps and no other hw monitoring). The temps are pretty much identical after 20min. win running than with the glass in front.
at 20min. I noted down the temps and removed the dust filter. Without dust filter, everything gets about 2-3°C cooler (again after 20min.) but to be I'm never gonna run it without dust filter . I might just increase the front fan speed a bit.

 

[A Computer Guy]

...

View attachment 155701

....

Just a general question? Does Cinebench even care about RAM speed?

I can pull a 4940 in CRr20 even with these, perhaps not so good timings in the below screenshot.

https://imgur.com/a/NWzrtIi

 

[ecopsorn]

Just a general question? Does Cinebench even care about RAM speed?

at least what I learned through this all and can be confirmed by my ram timing testing, that it cares a little bit but not much at all - more important for me now are the read/copy/latency results in Aida64. I'm gonna focus on those when the mobo arrives.
And if I won't be happy with the results, I will still be able to return my expensive g.skill royals .

 

[INSTG8R]

do have AIDA with a licence, but it shows me much less in terms of temps than hwinfo does. Probably I just don't know where to look for it, but under "Sensor", it is only showing mobo,cpu,pch,gpu and ssd temps. It doesn't show me the case temps, nor the dim temps.

Seriously as a paying customer I would bring up the lack of sensor data here and I assure you they will try to fix it very quickly with a little help from you probably just requiring some data from your MoBo. I had my VRAM issue fixed in under a week by just sending a debug file they asked for AIDA created.

AIDA64 Discussion Forum

 

[ecopsorn]

and for the fun I disabled all case fans now for 15min .
CPU temp seems unaffected, mobo got 3°C hotter, chipset 10°C, GPU 6°C, Disk 6°C

 

[A Computer Guy]

...
so I removed the glass in front of the case and have AIDA open (no other background apps and no other hw monitoring). The temps are pretty much identical after 20min. win running than with the glass in front. at 20min. I noted down the temps and removed the dust filter. Without dust filter, everything gets about 2-3°C cooler (again after 20min.) but to be I'm never gonna run it without dust filter . I might just increase the front fan speed a bit.

In a prior case mod I used an automotive filter (didn't have dust filters on Antec 900) and that worked quite well and produced a nice diffuse pattern with the rgb light. That might have less air flow restriction but don't put that kind of filter it near something that can cause fire. You could wire frame something with balsa wood and hot-glue to make a better replacement dust filter but well that's a whole other exercise PC customization.

 

[ecopsorn]

Seriously as a paying customer I would bring up the lack of sensor data here and I assure you they will try to fix it very quickly with a little help from you probably just requiring some data from your MoBo. I had my VRAM issue fixed in under a week by just sending a debug file they asked for AIDA created.

AIDA64 Discussion Forum

will change the board anyway, I assume the new board is covered with the sensor data than