Transferred 5900X / RTX3060ti build to different case, runs warmer

[An0maly_76]

So, I started out with...

 

AMD R9 5900X

Scythe Mugen 5

Asus Tuf B550-PLUS board

Asus KO RTX3060ti-8GB-OC

Ballistix BL16G32C16U4B.16FE CL16 DDR-3200 16-18-18-36

WD Blue SN570 1TB NVME M.2

WD Black 6TB HDD

Corsair RM850x PSU

Corsair 4000X case (incl. SP120 fans x3)

Corsair iCUE Commander Core XT

Corsair LL120 fans x3)

 

So, try to keep a long story short, about every 8-10 cold boots, this machine would randomly show a blank screen, no confirmation of boot. Was partially disassembled several times (12) to check for the problem, eventually caught a DRAM light in spite of DIMMs testing okay. About this time debug lights started going nuts. Swapped RAM and GPU from another machine, no change. Board was RMA'd, but could not duplicate issue. RAM since has appeared to fail in another machine, GPU works fine with known good RAM and has been ruled out. My temps were 38C-43C idle, 57C-65C nominal load, 70-72C gaming, 76C-87C in Cinebench R23, PBO dependent.

 

I decided to make some changes, since the RAM is not on that board's QVL list and it appeared to be faulty anyway. Reassembled today with:

 

R9 5900X

Scythe Mugen 5

Asus Tuf B550-PLUS

Patriot Viper 4 Blackout PVB432G320C6K  - On TUF B550-PLUS QVL

WD Blue SN570 1TB NVME

Corsair RM850x

Fractal Design Pop Air XL

Corsair iCUE Commmander Core XT

Corsair LL120 x 6

Pioneer BDR-212DBK

 

Applied three dots of thermal paste in a triangle, then tightened cooler down. Very minute showing of paste squeezing out around edges, not enough to worry with wiping, didn't even get off of the IHS. Machine seems to run okay, albeit the temps are higher, about 50-55C idling, sometimes 60-61C. Hasn't overheated yet though, even with 30 cars chasing me in BeamNG.drive and American Truck Simulator on moderately high detail. I'm stumped as to why it's running 20C warmer at idle. Can a bigger case actually have a detrimental affect on cooling? Or does a smaller case generate higher positive pressure for airflow?

Only other thing I can think of is the that mix of three SP120s and three LL120s was somehow cooling better. And apparently, the LL120s (I had forgotten this from before) do not support PWM, depending on which version one is referring to. I've noticed slight differences in advertised RPM and CFM, which could explain it I guess, but I just don't think this would amount to a 20C difference at idle. Does aynone have any ideas?

 

Edited August 16, 2022 by An0maly_76
Revised, more info

[DoctorNick]

9 minutes ago, An0maly_76 said:

So, I started out with...

 

AMD R9 5900X

Scythe Mugen 5

Asus Tuf B550-PLUS board

Asus KO RTX3060ti-8GB-OC

Ballistix BL16G32C16U4B.16FE CL16 DDR-3200 16-18-18-36

WD Blue SN570 1TB NVME M.2

WD Black 6TB HDD

Corsair RM850x PSU

Corsair 4000X case (incl. SP120 fans x3)

Corsair iCUE Commander Core XT

Corsair LL120 fans x3)

 

So, try to keep a long story short, about every 8-10 cold boots, this machine would randomly show a blank screen, no confirmation of boot. Was partially disassembled several times (12) to check for the problem, eventually caught a DRAM light in spite of DIMMs testing okay. About this time debug lights started going nuts.

 

Edit:

Pretty sure you would have solved this issues, if you had set timings manually, but hard to say. Also what is your load temps? But if they tested bad, then...

Idle temps on ryzen is not trustworthy because the way ryzen runs. To be sure you must run a load. I suggest running R23 or the same program as you tested with last time. 

[freeagent]

2 hours ago, An0maly_76 said:

Can a bigger case actually have a detrimental affect on cooling?

Yes. You need more air to cycle the chamber.. and your fans are farther away from your components. 

2 hours ago, DoctorNick said:

Idle temps on ryzen is not trustworthy because the way ryzen runs

Sure they are. Most people run theirs under cooled, which is why the vast majority of people complain about how hot they run. The reality is they run just fine. Cool as a cucumber. 

[DoctorNick]

28 minutes ago, freeagent said:

Yes. You need more air to cycle the chamber.. and your fans are farther away from your components. 

Sure they are. Most people run theirs under cooled, which is why the vast majority of people complain about how hot they run. The reality is they run just fine. Cool as a cucumber. 

What i meant is that idle temps are not "static" as they usually are on intel. With PBO on that is

[freeagent]

27 minutes ago, DoctorNick said:

What i meant is that idle temps are not "static" as they usually are on intel. With PBO on that is

For me they are.. no spikes or any other complaints other people have..

[An0maly_76]

9 hours ago, DoctorNick said:

Edit:

Pretty sure you would have solved this issues, if you had set timings manually, but hard to say. Also what is your load temps? But if they tested bad, then...

Idle temps on ryzen is not trustworthy because the way ryzen runs. To be sure you must run a load. I suggest running R23 or the same program as you tested with last time. 

 

I'm not seeing where manual timings would have mattered if DOCP had the same values that were listed on the DIMMs. It's irrelevant, however, as these DIMMS will no longer even boot and they've been replaced with the RAM in the second spec list. So either they were defective to start with, or they just won't work well with any board it's not QVL'd for.

 

Remember, these DIMMs tested fine in the machine that apparently still had issues with it, and caused issues to the point of their own failure in another, and neither board listed them as QVL. The ASRock B450M-HDV / 3600X test bed is now happily humming along with QVL-certified RAM.

 

While the Asus TUF-B550-PLUS / 5900X machine's random POST failures seem to be gone, the debug lights are still acting like a traffic signal gone haywire (Asus said they couldn't duplicate the issue). I can say that if the board fails, I'll be getting a Gigabyte or something. Not impressed with Asus or their "customer service / repair process" at this point. The only logical explanation at this point would be that the board is failing and the Asus tech was too clueless to see it. The CPU seems to run fine.

I'm just wondering why with the same amount of fans, possibly larger case volume, and all fans running full-speed (either they don't support PWM or there is a problem with my Commander), I'm seeing 53-55C idle, where I used to see 38C-43C. Also seeing a constant hourglass cursor in Winblows, where I didn't see it before. Even more strange is that the RAM I bought specs out at 1.35V, 16-18-18-36, yet DOCP doesn't pick up on that profile and I can't find anything in the BIOS that will let me set those timings. I'm going to try a fresh Windows install next, but I don't see that fixing it. If it keeps aggravating me, it's going away from here.

 

 

6 hours ago, DoctorNick said:

What i meant is that idle temps are not "static" as they usually are on intel. With PBO on that is

 

5 hours ago, freeagent said:

For me they are.. no spikes or any other complaints other people have..

I've always noticed a slight fluctuation, PBO or not. Not much of a swing, but never consistent.

Edited August 16, 2022 by An0maly_76
Revised, more info

[DoctorNick]

3 hours ago, An0maly_76 said:

 

I'm not seeing where manual timings would have mattered if DOCP had the same values that were listed on the DIMMs. It's irrelevant, however, as these DIMMS will no longer even boot and they've been replaced with the RAM in the second spec list. So either they were defective to start with, or they just won't work well with any board it's not QVL'd for.

 

Remember, these DIMMs tested fine in the machine that apparently still had issues with it, and caused issues to the point of their own failure in another, and neither board listed them as QVL. The ASRock B450M-HDV / 3600X test bed is now happily humming along with QVL-certified RAM.

 

While the Asus TUF-B550-PLUS / 5900X machine's random POST failures seem to be gone, the debug lights are still acting like a traffic signal gone haywire (Asus said they couldn't duplicate the issue). I can say that if the board fails, I'll be getting a Gigabyte or something. Not impressed with Asus or their "customer service / repair process" at this point. The only logical explanation at this point would be that the board is failing and the Asus tech was too clueless to see it. The CPU seems to run fine.

I'm just wondering why with the same amount of fans, possibly larger case volume, and all fans running full-speed (either they don't support PWM or there is a problem with my Commander), I'm seeing 53-55C idle, where I used to see 38C-43C. Also seeing a constant hourglass cursor in Winblows, where I didn't see it before. Even more strange is that the RAM I bought specs out at 1.35V, 16-18-18-36, yet DOCP doesn't pick up on that profile and I can't find anything in the BIOS that will let me set those timings. I'm going to try a fresh Windows install next, but I don't see that fixing it. If it keeps aggravating me, it's going away from here.

 

 

 

I've always noticed a slight fluctuation, PBO or not. Not much of a swing, but never consistent.

I see, yeah didn't read this properly I'm sorry. Most of the time DOCP only sets primary timings, this why it would make a difference, but if ram cant even post then, never mind.

I agree with you, Asus' support is horrible. I actually had a case a fair while back with an artifacting GPU . They couldn't find anything, even though I could replicate the issue in multiple PC's and providing them with complete documentation. Ended up sending it for RMA 3 times. And the third time it came back completely broken. (Broken crossfire connector)

Hard to say anything about the temps. Just be sure you are testing it with the same settings, voltage etc and make sure cooler is mounted correctly