Discussion Ryzen 9 5900/5950XT owners come talk to me

[punkncat]

If you own one of these 16C/32T CPU please tell me what you are doing as a cooling solution, case/case fans.

What kind of idle temps are you getting? Load temps?

I don't really wish to muddy the water too much, but I picked one of these up to use for my personal gaming rig. I am having some pretty considerable temperature control issues in a Lian Li 011D Mini with a HUGE Dark Rock Pro 4 cooler (it has 2 fans) and (7) case fans. Taking covers off actually makes the problem worse, as unlikely as that sounds.
(2) 140 intake (front side)
(2) 120 intake (bottom)
(3) 120 exhaust (back and top)

Power supply is an SF750 Platinum that has been rock solid in previous hardware profile on an i9. new in '23
B550 Tomahawk Max WiFi - newest BIOS
SP DDR4 3200 X4 8GB

Adrenalin is installed and all chipset and graphics drivers are updated, including that latest release chipset. Windows 11, fully updated.

I have this paired with an RX7800XT. The system is shutting down at random under a gaming load and will either soft crash to desktop or black screen crash. HWI is reporting temp hot spot at ~81C right before the crash. I know these are supposed to throttle at ~90C. That same GPU has been rock solid in another system since I purchased it on release. I do not suspect an issue with it.

I am having an exceptionally hard time with reporting programs as they are almost all showing different temps, voltages, and utilization. I understand aspects of it being due to the type of reporting which is the default for the specific program but between Task Manager, RIVA, HWI, Core Temp, and Ryzen Master there is barely anything aside from voltage coming anywhere close to each other.

I have on order some Thermal Grizzly Kryonaut Extreme. I am not expecting miracles but is regarded as one of the best pastes on the market right now. I will also be taking a look at the previous application to be sure I have a good spread and the cooler tightened down evenly. This situation is being a major PITA and not really sure where to go aside from a possible RMA.

I have had very good system stability with a 5600 and a 6600XT, so don't suspect anything wrong with motherboard (related) and RAM.

 

[beyondlogic]

If you own one of these 16C/32T CPU please tell me what you are doing as a cooling solution, case/case fans.

What kind of idle temps are you getting? Load temps?

I don't really wish to muddy the water too much, but I picked one of these up to use for my personal gaming rig. I am having some pretty considerable temperature control issues in a Lian Li 011D Mini with a HUGE Dark Rock Pro 4 cooler (it has 2 fans) and (7) case fans. Taking covers off actually makes the problem worse, as unlikely as that sounds.
(2) 140 intake (front side)
(2) 120 intake (bottom)
(3) 120 exhaust (back and top)

Power supply is an SF750 Platinum that has been rock solid in previous hardware profile on an i9. new in '23
B550 Tomahawk Max WiFi - newest BIOS
SP DDR4 3200 X4 8GB

Adrenalin is installed and all chipset and graphics drivers are updated, including that latest release chipset. Windows 11, fully updated.

I have this paired with an RX7800XT. The system is shutting down at random under a gaming load and will either soft crash to desktop or black screen crash. HWI is reporting temp hot spot at ~81C right before the crash. I know these are supposed to throttle at ~90C. That same GPU has been rock solid in another system since I purchased it on release. I do not suspect an issue with it.

I am having an exceptionally hard time with reporting programs as they are almost all showing different temps, voltages, and utilization. I understand aspects of it being due to the type of reporting which is the default for the specific program but between Task Manager, RIVA, HWI, Core Temp, and Ryzen Master there is barely anything aside from voltage coming anywhere close to each other.

I have on order some Thermal Grizzly Kryonaut Extreme. I am not expecting miracles but is regarded as one of the best pastes on the market right now. I will also be taking a look at the previous application to be sure I have a good spread and the cooler tightened down evenly. This situation is being a major PITA and not really sure where to go aside from a possible RMA.

I have had very good system stability with a 5600 and a 6600XT, so don't suspect anything wrong with motherboard (related) and RAM.

hello i have a 9 5900x what i did was drop the voltage the voltage on these cpus is insane it keeps smacking the 1.5v ceiling i undervolted it till i got it down to 1.35v and left the clocks on auto. only ever gets up to 70c max that ive ever seen.

also to point out the ryzen 9 5900x is a 12 core chip. but i had issues with this chip also and it was the only work around i could get it without the pc turning off.

DeepCool LS520 is what im using to keep it cool and locked down. tried with 7 heatpipe cooler deepcool assassin iv and it just wasnt manly enough to keep it under control so switched to water.​

 

[beyondlogic]

If you own one of these 16C/32T CPU please tell me what you are doing as a cooling solution, case/case fans.

What kind of idle temps are you getting? Load temps?

I don't really wish to muddy the water too much, but I picked one of these up to use for my personal gaming rig. I am having some pretty considerable temperature control issues in a Lian Li 011D Mini with a HUGE Dark Rock Pro 4 cooler (it has 2 fans) and (7) case fans. Taking covers off actually makes the problem worse, as unlikely as that sounds.
(2) 140 intake (front side)
(2) 120 intake (bottom)
(3) 120 exhaust (back and top)

Power supply is an SF750 Platinum that has been rock solid in previous hardware profile on an i9. new in '23
B550 Tomahawk Max WiFi - newest BIOS
SP DDR4 3200 X4 8GB

Adrenalin is installed and all chipset and graphics drivers are updated, including that latest release chipset. Windows 11, fully updated.

I have this paired with an RX7800XT. The system is shutting down at random under a gaming load and will either soft crash to desktop or black screen crash. HWI is reporting temp hot spot at ~81C right before the crash. I know these are supposed to throttle at ~90C. That same GPU has been rock solid in another system since I purchased it on release. I do not suspect an issue with it.

I am having an exceptionally hard time with reporting programs as they are almost all showing different temps, voltages, and utilization. I understand aspects of it being due to the type of reporting which is the default for the specific program but between Task Manager, RIVA, HWI, Core Temp, and Ryzen Master there is barely anything aside from voltage coming anywhere close to each other.

I have on order some Thermal Grizzly Kryonaut Extreme. I am not expecting miracles but is regarded as one of the best pastes on the market right now. I will also be taking a look at the previous application to be sure I have a good spread and the cooler tightened down evenly. This situation is being a major PITA and not really sure where to go aside from a possible RMA.

I have had very good system stability with a 5600 and a 6600XT, so don't suspect anything wrong with motherboard (related) and RAM.

also couple of other things i did

i dropped my ram speed to 3000. as i was running 4 sticks at it wasnt playing nice.

i never used ryzen master this made the chip unstable and unreliable i undid ryzen master settings then removed it from the pc.

also make sure msi doesn't have some annoying push the cpu to hell and back override of amds original boost as ASUS did and it was this that was pushing voltage even further then was allowed.

also to point out that there are 2 ccds unlike the 5600 which is 1 ccd. so 1 will get way hotter then the other as they fight for cooling.

air coolers are fine for 1 cmd but when multiple are involved the air cooler is essentially trying to fight off 2 cpus. which it cant as it gets saturated with heat to quickly and the air heatsink cant get rid of it fast enough.

causing my gpu to get quite hot as it now had a massive radiator of heat next to it. this vanished when i went to water cooling dropped down by 10c. 2 reasons 1 the radiator kicking out all the heat was further away at the top of the case not only removing the heat source further from gpu but also allowed for better removal of heat. which is why i still favor water cooling for anything more then 1 ccd.

if your case can handle it go higher then the 240 version and go 360. even if your top can handle 3x 120mm fans make sure there's like 2 cm-3cm of space for the actual radiator ends as they are a bit bigger.

its important the paste covers the whole ihs on these duel chips

 

[punkncat]

I was not at all pleased with the reliability side of using the Curve Optimizer and that auto PBO thing. I got a bit of a clock speed increase from 4800 to 4950 but it just wanted to give issues in various ways. It was set to write to BIOS so I did have to reset CMOS and get fan curves back where I wanted them. In fairness, I have the fans all set to the "Smart Fan" option on MSI mobo and it is really nice on desktop. Setting the fans up to fun a higher performance curve doesn't help a lot with temps and is far too loud.
edit- of note here, the Dark Rock tower being used has a performance and a silent switch that will limit fan speed down. I have it set to the performance side.

There is a gaming boost tab in the BIOS, but it is off. Only have the "XMP" profile loaded.

I have a 240 AIO sitting here but it is also older than 5yo. I had no issues with it when it was taken from the last build about a month ago, but really don't want to count on it. I was of the hope that I could get this air cooling to work not only from a noise standpoint, but reliability and safety in relation to failure.

 

[beyondlogic]

I was not at all pleased with the reliability side of using the Curve Optimizer and that auto PBO thing. I got a bit of a clock speed increase from 4800 to 4950 but it just wanted to give issues in various ways. It was set to write to BIOS so I did have to reset CMOS and get fan curves back where I wanted them. In fairness, I have the fans all set to the "Smart Fan" option on MSI mobo and it is really nice on desktop. Setting the fans up to fun a higher performance curve doesn't help a lot with temps and is far too loud.
edit- of note here, the Dark Rock tower being used has a performance and a silent switch that will limit fan speed down. I have it set to the performance side.

There is a gaming boost tab in the BIOS, but it is off. Only have the "XMP" profile loaded.

I have a 240 AIO sitting here but it is also older than 5yo. I had no issues with it when it was taken from the last build about a month ago, but really don't want to count on it. I was of the hope that I could get this air cooling to work not only from a noise standpoint, but reliability and safety in relation to failure.

i dont blame you its just bad lol i spent about 3 weeks going over several youtube videos on how to get this work around to work.

cpu voltage> offset mode> cpu offset mode use minus - > offset voltage 0.17500 was what i set mine to your milage may vary you may only get as low as 0.15 because of the 16 cores vs my 12. so i would do 0.1 then try 0.15.

View: https://www.youtube.com/watch?v=X21gGLS4h9k


this will limit the voltage and drive the temps down.

other thing you can do if your desperate to do air cooling do the above.

and then set a max temp limit the cpu cant go beyond for whatever reason msi and asus boards seem hell bent on running it up to 90 degree shut off. so setting it to 82 or 85 would be safest.



To set a temperature limit in the MSI BIOS, navigate to the "OC" (Overclocking) section, then "Advanced CPU Configuration" and look for "Set Thermal Point" or similar options

 

[punkncat]

i dont blame you its just bad lol i spent about 3 weeks going over several youtube videos on how to get this work around to work.

cpu voltage> offset mode> cpu offset mode use minus - > offset voltage 0.17500 was what i set mine to your milage may vary you may only get as low as 0.15 because of the 16 cores vs my 12. so i would do 0.1 then try 0.15.

View: https://www.youtube.com/watch?v=X21gGLS4h9k


this will limit the voltage and drive the temps down.

other thing you can do if your desperate to do air cooling do the above.

and then set a max temp limit the cpu cant go beyond for whatever reason msi and asus boards seem hell bent on running it up to 90 degree shut off. so setting it to 82 or 85 would be safest.



To set a temperature limit in the MSI BIOS, navigate to the "OC" (Overclocking) section, then "Advanced CPU Configuration" and look for "Set Thermal Point" or similar options

Thank you.

About to jot some notes and check that video out. May as well try.

 

[beyondlogic]

Thank you.

About to jot some notes and check that video out. May as well try.

no problem let me know how it goes. also the fan in your top near the front next to side fans turn the other way round as intake. as the side fans air will just get exhausted out the case before it can cool anything. see if it makes a diffrence. as for cpu clocks just leave on auto the board and cpu will decide what clocks it can handle at what power less likely to crash. good luck hope it works

 

[punkncat]

Will report back in a moment.

 

[QyInvoLing]

I roughly searched "PSU" in this post and found nothing. Can you borrow a PSU from your friend? All your problems look like to be caused by the weak PSU, not overheating.
Remember to change the cable, don't simply connect the old PSU's cable to the new one.

 

[punkncat]

no problem let me know how it goes. also the fan in your top near the front next to side fans turn the other way round as intake. as the side fans air will just get exhausted out the case before it can cool anything. see if it makes a diffrence. as for cpu clocks just leave on auto the board and cpu will decide what clocks it can handle at what power less likely to crash. good luck hope it works

I did a few "quick and dirty" adjustments as was the video subject.

.050 turned in 1384 on Cinebench 24 at a max temp of 64.9

.075 turned in 1385 at 62.9 max temp.

Both of those allowed the CPU to hit 4950 and max voltage was 1.406 on the latter.

For sure going to mess with this some more. Going to load into some BG3 for a bit and see how that goes. Thank you very much!

edit- max temp on a 20 min or so BG3 run was 70C. For sure going to have to play with this in little increments in order to see where it is ideal.

 

[beyondlogic]

I did a few "quick and dirty" adjustments as was the video subject.

.050 turned in 1384 on Cinebench 24 at a max temp of 64.9

.075 turned in 1385 at 62.9 max temp.

Both of those allowed the CPU to hit 4950 and max voltage was 1.406 on the latter.

For sure going to mess with this some more. Going to load into some BG3 for a bit and see how that goes. Thank you very much!

edit- max temp on a 20 min or so BG3 run was 70C. For sure going to have to play with this in little increments in order to see where it is ideal.

that's not terrible for a air cooler. the issue is long term with air coolers i find can get overwhelmed quicker i wouldnt go below 1.3 cause i think youll see instability or drastic lower clocks.

i recomend hw monitor as it shows the 2 dies that make up the cpu. temps. ccd1 being the die that gets the hottest.

 

[CelicaGT]

Just a quick bit of advice, having done this process before a few times. Instability will crop up at idle, or low load, not under benchmark load. So if you start to see stability issues on the desktop, browsing, or watching video, for example, you might need to add a little voltage back.

 

[beyondlogic]

Just a quick bit of advice, having done this process before a few times. Instability will crop up at idle, or low load, not under benchmark load. So if you start to see stability issues on the desktop, browsing, or watching video, for example, you might need to add a little voltage back.

yeah i noticed that as well there's only so low you can go before something doesn't get enough power.

 

[punkncat]

Thank you for that @CelicaGT

 

[alceryes]

Another thing to note, 16 core CPUs can get a boost in performance and lowered heat generation by turning SMT off.
Unless you regularly do work that will load up the 16 real cores and 16 shadow (SMT) cores, it's better to keep SMT off. SMT/HT is basically a next job cache-fill smart technology. It definitely can help performance if you really need 32 cores, but can be a waste otherwise.
The benefit of having SMT/HT off is that your 16 real cores no longer share the power profile with shadow cores anymore and there's no extra heat generation from SMT/HT.

 

[punkncat]

I would have to double check again, but last night while I was fooling around, games, synthetic benchmarks and so on I noticed a spot in HWI where it shows active core stats. I didn't check again after several games but one of the times back on desktop I noted it saying that the maximum core use was 4.5. Lol, how does it do half a core?

I was also noticing in RIVA that the core voltage was only bumping around 100-109W whereas on stock it would occasionally go as high as 130W. I am going to have to play with the fan curves a little bit. They are very rangy now, like I can move the mouse and hear whirr-whirr as it ramps up and down.

 

[beyondlogic]

I would have to double check again, but last night while I was fooling around, games, synthetic benchmarks and so on I noticed a spot in HWI where it shows active core stats. I didn't check again after several games but one of the times back on desktop I noted it saying that the maximum core use was 4.5. Lol, how does it do half a core?

I was also noticing in RIVA that the core voltage was only bumping around 100-109W whereas on stock it would occasionally go as high as 130W. I am going to have to play with the fan curves a little bit. They are very rangy now, like I can move the mouse and hear whirr-whirr as it ramps up and down.

the wirling is quite normal on air cooling its usually when its one of ccds getting hotter even on water depending on what application it is good think about hw monitor it tells me when it does spike like that whats going on. it happens less on water it really is a juggling game. how low have you got the voltage down ?

 

[Eugene$$]

i am using thermalright si 100

 

[punkncat]

the wirling is quite normal on air cooling its usually when its one of ccds getting hotter even on water depending on what application it is good think about hw monitor it tells me when it does spike like that whats going on. it happens less on water it really is a juggling game. how low have you got the voltage down ?

ATM I haven't gone below a .075 negative offset. I have some Thermal Grizzly on the way so will check the current paste, put in some better paste, and play a bit more with that.

 

[punkncat]

i am using thermalright si 100

On a 5900XT?

 

[beyondlogic]

ATM I haven't gone below a .075 negative offset. I have some Thermal Grizzly on the way so will check the current paste, put in some better paste, and play a bit more with that.

you can go to 0.1 id keep going down till you hit about 1.33 or 1.35v

my min voltage is 950v

on water I found 1.35v is sweet spot

I have 0.175 dialed in. i recon you safe at 0.15 and could get down to mine you might shave off a bit of clock speed but it will be stable.

image below and ive ran 3d mark/ and cpu profile and kept 7 tabs open and several appilcations just to show how solid it is. my memory is set to 3000 as it wasnt stable at 3200 or 3600.

https://imgur.com/a/M9hxM3a

View: https://imgur.com/a/M9hxM3a


on air it would hit just about 80c

 

[punkncat]

It turns out that my motherboard selection B550 Tomahawk Max WiFi will only allow the offset up to 0.075V. Any setting higher than that just defaults back to that setting. Luckily, it seems to have given me enough headroom with the additional fans that I have not experienced a crash since setting it that way.

I got the good Thermal Grizzly paste in so will take that on this weekend and see if it changes much. The "chart" seems to show I could expect up to about a 3C difference lower. We shall see.

I appreciate the feedback given.

 

[beyondlogic]

It turns out that my motherboard selection B550 Tomahawk Max WiFi will only allow the offset up to 0.075V. Any setting higher than that just defaults back to that setting. Luckily, it seems to have given me enough headroom with the additional fans that I have not experienced a crash since setting it that way.

I got the good Thermal Grizzly paste in so will take that on this weekend and see if it changes much. The "chart" seems to show I could expect up to about a 3C difference lower. We shall see.

I appreciate the feedback given.

that is unfortunate. the bios up to date ?.

thermal grizzly paste it does knock down temps by 3-5 degrees long term compared to other pastes and has roughly a curing time or settle time of about a week. its strenght i find is it seems to keep the temps swings lower it is like concrete one its set so allow for the pc to heat up if you ever need to remove the cooler.

e.g on mx4 my temps would swing between 40-50 on thermal grizzly temps would be lower 37c swing 45c
when properly pushed at the max whack on a 5700x it topped out at 60c on thermal grizzly and 65c on mx4 ( last time i used thermal grizzly got sick of the cost and the unspreadability).

i find it difficult to spread however i use thermalrights tf8 as its a bit easier to spread.

( tf9 however is like clay zero spreading ability).

 

[punkncat]

Yeah, the tip of the paste has a spatula sort of appearance to it.

I don't really anticipate being back under the cooler for some time. This is within 3% or so of being the best CPU that could be sourced for this board. No real reason to go in outside of a problem with the motherboard or RAM. Hopefully it won't be an issue in the future as for sure these AM4 chips love to come out with the cooler.

edit- yes, BIOS updated to latest

 

[beyondlogic]

Yeah, the tip of the paste has a spatula sort of appearance to it.

I don't really anticipate being back under the cooler for some time. This is within 3% or so of being the best CPU that could be sourced for this board. No real reason to go in outside of a problem with the motherboard or RAM. Hopefully it won't be an issue in the future as for sure these AM4 chips love to come out with the cooler.

edit- yes, BIOS updated to latest

yeah hopefully all works out I still recommend water cooling these chips that would be the next step.

im happy with my 5900x i doubt ill be jumping to any new amd builds for at least another 4 years its serving me well. at 1440p.