Ryzen 5600G temperature too high?

[framebuffer]

Just put together my first little Ryzen system but I'm not sure about the temperatures.
Based on what I can see around in reviews it should stay in the 50s at max, but mine passes easily the 70s with CB23

I'm using a Noctua NH-U12S and have a well ventilated case.
If I touch the heatsink while CB is running it's pretty much cold (at ambient temperature I'd say).
I even dismounted it to make sure the thermal paste was doing contact and of course it was perfect



Not sure what else should I try, this thing is pulling only 70W, it should be cold, with the same cooler I used to keep a 10850K (with more than double of the power) under 80c
(The integrated Vega GPU is disabled)

Any idea?

 

[BlaezaLite]

I think you might be worrying about nothing. Cinebench is a stress test, it's going to heat up your CPU. Nice system by the way.!

 

[freeagent]

7nm can be a bear.. some of my biggest heat sinks have been brought to their knees by it.

 

[BlaezaLite]

7nm can be a bear.. some of my biggest heat sinks have been brought to their knees by it.

It's at 4.45ghz too. Noctua or not, there will be heat. Try Linpack extreme, ooh toasty!

 

[Bomby569]

you have this to compare, but it's a higher end noctua

AMD Ryzen 5 5600G Review - Affordable Zen 3 with Integrated Graphics

The Ryzen 5 5600G is AMD's most affordable entry to the Zen 3 architecture. Priced at $260, this six-core, twelve-thread processor achieves excellent performance that rivals older eight-core processors from both Intel and AMD. The integrated Vega graphics cores are much more powerful than even...

www.techpowerup.com

i would expect a bit less heat, but i do not own one. Anyway expecting 50's, no way in hell.

 

[tabascosauz]

Just put together my first little Ryzen system but I'm not sure about the temperatures.
Based on what I can see around in reviews it should stay in the 50s at max, but mine passes easily the 70s with CB23

View attachment 254218

I'm using a Noctua NH-U12S and have a well ventilated case.
If I touch the heatsink while CB is running it's pretty much cold (at ambient temperature I'd say).
I even dismounted it to make sure the thermal paste was doing contact and of course it was perfect

View attachment 254221

Not sure what else should I try, this thing is pulling only 70W, it should be cold, with the same cooler I used to keep a 10850K (with more than double of the power) under 80c
(The integrated Vega GPU is disabled)

Any idea?

70 is way too hot for a 5600G on a large cooler, especially when stock (4450MHz is the stock clock ceiling, PBO goes up to 4650MHz max).

On a dinky L9x65 I never exceeded 65C on my 4650G and 5600G. Only the 8-core APUs run closer to the temps of the other desktop Ryzens. The 6-cores are 10-15C cooler under load, even under max PBO.

Stop using HWmonitor and switch to HWInfo as you should be doing for any Ryzen. Already I can see that the software has no clue what it's doing, look at its numbering of your cores.

Actually, are you running a 4.45GHz all-core, or any sort of "performance enhancement" in BIOS? I've accidentally hit their Game Boost button before. HWmonitor thinks Vcore is at 1.3V during the test. Is oddly high for stock clocks on these APUs. Either that or it's time to get on a newer BIOS.

 

[eidairaman1]

Wait a sec pull the heatsink and make sure there is no plastic film on the base (i check every heatsink for that before installation)

I dont think hwmonitor is even being revised anymore, so yeah use hwinfo64.

 

[framebuffer]

7nm can be a bear.. some of my biggest heat sinks have been brought to their knees by it.

OK but why the heatsink is cold?
The other day I was playing with a Xeon 1650v3 with this same NH-U12S and when the CPU was pulling around 180+W the heatsink was quite warm at touch (and btw CPU temps were in the 70's)

I'm starting to think that my sample could have very bad die-to-IHS contact

70 is way too hot for a 5600G on a large cooler, especially when stock (4450MHz is the stock clock ceiling, PBO goes up to 4650MHz max).

On a dinky L9x65 I never exceeded 65C on my 4650G and 5600G. Only the 8-core APUs run closer to the temps of the other desktop Ryzens. The 6-cores are 10-15C cooler under load, even under max PBO.

Stop using HWmonitor and switch to HWInfo as you should be doing for any Ryzen. Already I can see that the software has no clue what it's doing, look at its numbering of your cores.

Actually, are you running a 4.45GHz all-core, or any sort of "performance enhancement" in BIOS? I've accidentally hit their Game Boost button before. HWmonitor thinks Vcore is at 1.3V during the test. Is oddly high for stock clocks on these APUs. Either that or it's time to get on a newer BIOS.

I was using HWinfo64, switched to hwmonitor to see if there was any difference and happen to take the screenshot with this one, but the temps are equal across all the sw
Actually I have a slight undervolt of the PBO Curve, which is giving a little boots, but the power draw (and temps) are the same as with everything at default

Wait a sec pull the heatsink and make sure there is no plastic film on the base (i check every heatsink for that before installation)

I dont think hwmonitor is even being revised anymore, so yeah use hwinfo64.

heheh the plastic film has been removed long time ago
hwmonitor has been updated on April 29th, 2022 but OK, i'll not use it anymore

 

[tabascosauz]

I'm starting to think that my sample could have very bad die-to-IHS contact

Well, that's a wrap it sips power, but the cooler still shouldn't be cold after a few runs of R23.

Even if it was the cooler/mounting, you barely need any cooler contact at all with Renoir/Cezanne. There's only 1 medium-ish sized die (~150mm^2(?)), dead center on the substrate. My old '21 5700G had a badly uneven IHS, so something like 1/4 of the heatspreader (on one corner) never made contact regardless of the cooler I used or technique I tried. Never actually impacted temps in any way, when compared to my 4650G, 5600G, or new 5700G.

IHS contact is rare, but not the first time. Generally the Cezanne APUs have excellent internal IHS contact (<5C across all cores on my replacement '22 5700G), but the chiplet CPUs can be all over the place (13-17C delta on my 5900X just between neighbouring cores), so AMD's manufacturing isn't infallible.

I hope you've only just put together the rig, and can still use the return policy, because trying to convince AMD to RMA in this condition might be a bit of a problem lol. The core deltas are excellent, and to the uninitiated 72C won't look like a problem.

BIOS you are using right now? You haven't shown any system specs anywhere, I'm guessing X570 Tomahawk? iirc Cezanne needs AGESA 1200 or later to function properly, which would be quite far down the pipeline for an earlier production X570 board.

Last but not least, it wouldn't hurt just to clear CMOS and double check that the board knows what it's doing. You mentioned using CO, so you're clearly not stock.

 

[framebuffer]

Well, that's a wrap it sips power, but the cooler still shouldn't be cold after a few runs of R23.

Even if it was the cooler/mounting, you barely need any cooler contact at all with Renoir/Cezanne. There's only 1 medium-ish sized die (~150mm^2(?)), dead center on the substrate. My old '21 5700G had a badly uneven IHS, so something like 1/4 of the heatspreader (on one corner) never made contact regardless of the cooler I used or technique I tried. Never actually impacted temps in any way, when compared to my 4650G, 5600G, or new 5700G.

IHS contact is rare, but not the first time. Generally the Cezanne APUs have excellent internal IHS contact (<5C across all cores on my replacement '22 5700G), but the chiplet CPUs can be all over the place (13-17C delta on my 5900X just between neighbouring cores), so AMD's manufacturing isn't infallible.

I hope you've only just put together the rig, and can still use the return policy, because trying to convince AMD to RMA in this condition might be a bit of a problem lol. The core deltas are excellent, and to the uninitiated 72C won't look like a problem.

BIOS you are using right now? You haven't shown any system specs anywhere, I'm guessing X570 Tomahawk? iirc Cezanne needs AGESA 1200 or later to function properly, which would be quite far down the pipeline for an earlier production X570 board.

Last but not least, it wouldn't hurt just to clear CMOS and double check that the board knows what it's doing. You mentioned using CO, so you're clearly not stock.

CPU/MB are new (bought 5 days ago) so I could return them but... yes, while this is certainly odd, I think it's not so bad to ask a return, is it?
The MB is a MSI MAG X570 TOMAHAWK WIFI, flashed with latest BIOS (AGESA 1.2.0.7)

I tested temperatures under CB23 as first thing after Windows updates and drivers install, so with pretty much everything in [Auto] except boot order and they were exactly the same as now
After that, the only thing I did is to tune the memory settings and apply a PBO undervolt, which as I said did not changed the temps because the amount of watts pulled is the same, although frequency raised about 200MHz thanks to the lower voltages

Btw temps while gaming are around 50c-54c
What should be the VCore of the 5600G, is 1.3V too high in your opinion?

 

[tabascosauz]

CPU/MB are new (bought 5 days ago) so I could return them but... yes, while this is certainly odd, I think it's not so bad to ask a return, is it?
The MB is a MSI MAG X570 TOMAHAWK WIFI, flashed with latest BIOS (AGESA 1.2.0.7)

I tested temperatures under CB23 as first thing after Windows updates and drivers install, so with pretty much everything in [Auto] except boot order and they were exactly the same as now
The only thing I did is to tune the memory settings and apply a PBO undervolt, which as I said did not changed the temps because the amount of watts pulled is the same, although frequency raised about 200MHz thanks to the lower voltages

Btw temps while gaming are around 50c-54c
What should be the VCore of the 5600G, is 1.3V too high in your opinion?

Store return would be a no-brainer if it's no-questions-asked. No reason to RMA if you can just return. Unless you get a really lax RMA agent, I doubt they'd agree to an RMA anyway based on these symptoms ("technically" nothing out of spec). They hardly know their own hardware.

Gaming temps about normal for the APUs.

1.3V Vcore for all-core is a bit high for any stock Ryzen 5000 SKU. That was common for Ryzen 3000 and was part of why they ran pretty hot under load. Ryzen 5000 generally comes in at 1.2-1.25V at stock power limits Cinebench, if not lower.

If you're up for it, I'd try a few recent BIOSes to see if anything changes. In the period of time between AGESA 1200 and 1207, I've found sometimes BIOS updates can oddly affect APUs' performance a bit more than other Ryzen CPUs.

 

[BlaezaLite]

A lot of people have said 1.2.0.3 was a good bios for performance.

 

[freeagent]

OK but why the heatsink is cold?

Because it’s still lower wattage compared to old cpus, but in a much smaller space. Heat pipes that aren’t covering anything are just wasted pipes..

The last CPU I had that actually got big heat sinks warm was my old x5690. Nice and chunky at 32nm. Even my 22nm 3770K at 4600 didn’t get my heat sinks warm. The only time it got my Le Grand Macho RT warm was when it was running Linpack Xtreme at 4400MHz completely passive on the motherboard box. Just the fans on the GPU.

 

[damric]

The heat flux is very dense, but very small, so yeah it might be running hot while not really heating up the heat sink. This becoming more and more of a problem for both AMD and Intel as the transistors get smaller.

As long as it's not throttling, then your cooling is adequate enough.

If you were wanting to push clocks higher then an AIO could be justified perhaps since the cold plate tends to stay cold until you heat the liquid up to steady state. I prefer a cheap asf ali express custom loop though. Cezanne can usually be pushed to 4650MHz very easily with just the +200MHz offset and/or 104 base clock.

 

[The red spirit]

Because it’s still lower wattage compared to old cpus, but in a much smaller space. Heat pipes that aren’t covering anything are just wasted pipes..

The last CPU I had that actually got big heat sinks warm was my old x5690. Nice and chunky at 32nm. Even my 22nm 3770K at 4600 didn’t get my heat sinks warm. The only time it got my Le Grand Macho RT warm was when it was running Linpack Xtreme at 4400MHz completely passive on the motherboard box. Just the fans on the GPU.

lol I ran FX 6300 stock passively cooled with Mugen 4 cooler in enclosed case and it was fine. Didn't reach 60s in prime95 small FFTs

 

[framebuffer]

Because it’s still lower wattage compared to old cpus, but in a much smaller space. Heat pipes that aren’t covering anything are just wasted pipes..

The last CPU I had that actually got big heat sinks warm was my old x5690. Nice and chunky at 32nm. Even my 22nm 3770K at 4600 didn’t get my heat sinks warm. The only time it got my Le Grand Macho RT warm was when it was running Linpack Xtreme at 4400MHz completely passive on the motherboard box. Just the fans on the GPU.

The heat flux is very dense, but very small, so yeah it might be running hot while not really heating up the heat sink. This becoming more and more of a problem for both AMD and Intel as the transistors get smaller.

As long as it's not throttling, then your cooling is adequate enough.

If you were wanting to push clocks higher then an AIO could be justified perhaps since the cold plate tends to stay cold until you heat the liquid up to steady state. I prefer a cheap asf ali express custom loop though. Cezanne can usually be pushed to 4650MHz very easily with just the +200MHz offset and/or 104 base clock.

I'm new to Ryzen 5000 (or any Ryzen) but I'm also starting to think that the heat is happening in an extremely small area that's hard to move very efficiently to the heatsink

I tested Blender, which is what W1zzard used in he's review for temperature reporting

Considering that I have 26c ambient temp (I guess he had 20-22), running a lower end cooler and have everything in a closed case (ventilated, but closed) maybe things are not so terribly off
In fact Core3 and 4 seems to be the ones that get a little too hot, the others are not bad

1.3V Vcore for all-core is a bit high for any stock Ryzen 5000 SKU. That was common for Ryzen 3000 and was part of why they ran pretty hot under load. Ryzen 5000 generally comes in at 1.2-1.25V at stock power limits Cinebench, if not lower.

You are right, uhm, I may try manual Vcore offset "undervolt" to land around 1.225 and see what it does, I guess PBO frequency will drop considerably

 

[silkstone]

Just put together my first little Ryzen system but I'm not sure about the temperatures.
Based on what I can see around in reviews it should stay in the 50s at max, but mine passes easily the 70s with CB23

View attachment 254218

I'm using a Noctua NH-U12S and have a well ventilated case.
If I touch the heatsink while CB is running it's pretty much cold (at ambient temperature I'd say).
I even dismounted it to make sure the thermal paste was doing contact and of course it was perfect

View attachment 254221

Not sure what else should I try, this thing is pulling only 70W, it should be cold, with the same cooler I used to keep a 10850K (with more than double of the power) under 80c
(The integrated Vega GPU is disabled)

Any idea?

I recently built a 5700G system. It was reaching over 90C with the stock cooler. I think these chips just run hot.

 

[tabascosauz]

The heat flux is very dense, but very small, so yeah it might be running hot while not really heating up the heat sink. This becoming more and more of a problem for both AMD and Intel as the transistors get smaller.

I'm new to Ryzen 5000 (or any Ryzen) but I'm also starting to think that the heat is happening in an extremely small area that's hard to move very efficiently to the heatsink

I tested Blender, which is what W1zzard used in he's review for temperature reporting

View attachment 254298

Considering that I have 26c ambient temp (I guess he had 20-22), running a lower end cooler and have everything in a closed case (ventilated, but closed) maybe things are not so terribly off
In fact Core3 and 4 seems to be the ones that get a little too hot, the others are not bad

You are right, uhm, I may try manual Vcore offset "undervolt" to land around 1.225 and see what it does, I guess PBO frequency will drop considerably

I recently built a 5700G system. It was reaching over 90C with the stock cooler. I think these chips just run hot.

Like I already said, there is no comparison between the APUs and the chiplet products. They simply run tens of degrees cooler. The 5600G is also not the 5700G. The 8-cores run considerably (10C+) warmer even stock.

I'll have to run some more R23 on my 5700G today to see where Vcore lands.

As for ambient, mine is also 25C because my 5700G is upstairs and my 5900X downstairs.

Managed to dig up a 5600G R23 result from last year. If you are stock like you say then it seems like your board is just pretty aggressive on the Vcore, considering mine was PBO -15 all cores.

 

[framebuffer]

Managed to dig up a 5600G R23 result from last year. If you are stock like you say then it seems like your board is just pretty aggressive on the Vcore, considering mine was PBO -15 all cores.

I'm also at PBO -15
can you post your all core boost frequency/ Vcore during CB/Blender?
based on the score I see in the screenshot you should be at around 4.40 and if you do that with just 1.22V or so, well I probably have the worst 5600G in the world, since it needs 1.3 for the same freq.

 

[tabascosauz]

I'm also at PBO -15
can you post your all core boost frequency/ Vcore during CB/Blender?
based on the score I see in the screenshot you should be at around 4.40 and if you do that with just 1.22V or so, well I probably have the worst 5600G in the world, since it needs 1.3 for the same freq.

Sorry mate, the 5600G is the one CPU that's no longer with me. The only result that I could still find. But it was certainly not over 4.4 all core, and not over 1.25V.

I don't think it's the fault of your 5600G. When stock they're all the same. The SP quality differences between chips only show when you manual OC or PBO (+200, +power limits).

It's the boards that can have a big impact. Gigabyte usually forces higher Vcore for higher scores, across both Intel and AMD, even at stock. I'm surprised with your Tomahawk; my experience with MSI is they're generally average to conservative at stock. I got rid of a B550I Gaming Edge Wifi because clocks and scores were abnormally weak. And even on a single board, different BIOS can affect the APUs CPU and iGPU perf a lot.

Which is why I suggested you could shop around with BIOSes a little, you never know if it's just one BIOS that's subpar. Generally I haven't seen this kind of variation from the regular CPUs, 3700X and 5900X.

edit: just ran some R23 on my 5700G - 4.35-4.4GHz @ 1.21-1.25V, reached 81C max @ 90W. PBO disabled, but I think I left it on -12 or something. So 1.3V definitely seems a bit high.

 

[eidairaman1]

Because it’s still lower wattage compared to old cpus, but in a much smaller space. Heat pipes that aren’t covering anything are just wasted pipes..

The last CPU I had that actually got big heat sinks warm was my old x5690. Nice and chunky at 32nm. Even my 22nm 3770K at 4600 didn’t get my heat sinks warm. The only time it got my Le Grand Macho RT warm was when it was running Linpack Xtreme at 4400MHz completely passive on the motherboard box. Just the fans on the GPU.

This is why I prefer baseplate heatsinks and not direct pipe