5800x3D Higher Temps with Cinebench R23

[doryuu13]

Good afternoon all,

I recently purchased a 5800x3D to replace my old 2700x as Asrock released BIOS compatibility. I installed the 5800x3D with an EK basic 360 AIO and got everything up and running fine. During gaming or CPUZ stress test temps are hovering between 65 and 75c however, when running all core Cinebench R23 for 10 minute loop, my temps will hit almost 90c on the package and spike above 90 on CCD 0 (highest I saw was 91c). Frequency remains fairly high even at high temps around 4.2 gigahertz but the temps are causing concern. I have seen conflicting information about the subject where some people say 90c is higher than usual and some are saying even with 420mm AIO they will hit 90c on synthetic benchmarks. I have already reseated cpu, repasted the block and made sure contact from the block to the IHS was even but temps stay fairly consistent. I was worried I would need to RMA the chip, does anyone have experience with this chip that can weigh in?

Build:
Case: Lian Li PC 0-11 Dynamic --> 3 bottom intake 120mm fans, 3 side intake 120mm fans, radiator top mounted with 3 Vardar s fans acting as exhaust.
Mobo: Asrock x370 Killer Sli/Ac BIOS 7.40
Cooler: EKWB basic 360 AIO (full pump speed and fans at 100% at 80c)
GPU: MSI RTX 2080 Gaming X Trio

 

[v12dock]

You are throwing an unrealistic synthetic load at it, and it’s still at or under the Tj-Max of 90c. There is no cause for concern, you don’t need to RMA because you didn’t get a “golden sample”.

 

[doryuu13]

You are throwing an unrealistic synthetic load at it, and it’s still at or under the Tj-Max of 90c. There is no cause for concern, you don’t need to RMA because you didn’t get a “golden sample”.

Thank you very much for the reply. I was thinking that as well just had no experience with a chip this hot so I wasn't sure. Would there be any benefit to changing coolers? It seems most coolers perform roughly the same as far as I can tell.

 

[DemonicRyzen666]

cinebenche23 isn't as unrealistic as Y-cruncher.
Run Y-cruncher & see if it's higher.
Most likly has poor contact with it's I.H.S it's that bad.
My 5600x has poor contact under the I.H.S

 

[lightning70]

5800x3d is usually a hot cpu Have you tried lowering the voltage?

 

[doryuu13]

5800x3d is usually a hot cpu Have you tried lowering the voltage?

I haven't tried doing anything other than keeping stock settings as of yet. I played some games last night to try to simulate real world use and the processor sat around 65-68c with a few momentary spikes to 72c or so. I have read that undervolting does wonders for the 5800x3D though I'm just not sure how to do it.

cinebenche23 isn't as unrealistic as Y-cruncher.
Run Y-cruncher & see if it's higher.
Most likly has poor contact with it's I.H.S it's that bad.
My 5600x has poor contact under the I.H.S

I will run Y-Cruncher after work tonight and see what kind of numbers I get.

 

[freeagent]

The loads you guys call unrealistic are actual loads Just because your game or video conversion doesn’t reach such temps doesn’t mean it’s unrealistic. The biggest problem with older coolers is that they are not tuned for todays CPUs. Which is why you hear things like

My 5600x has poor contact under the I.H.S

And

5800x3d is usually a hot cpu

 

[doryuu13]

The loads you guys call unrealistic are actual loads Just because your game or video conversion doesn’t reach such temps doesn’t mean it’s unrealistic. The biggest problem with older coolers is that they are not tuned for todays CPUs. Which is why you hear things like




And

You stated older cooler in your post. Would you consider the EK basic 360 an older cooler? I know the hotspot is in a slightly different area than previously on the 5800x3D and that some coolers have started offering an offset bracket to help with this, that being said, would you suggest a different cooler for my use case? Air/water?

 

[GerKNG]

why do people think that even larger radiators cool a hot chip better when the water in the AIO is literally cold?
it's a high density chip with another high density cache die on top of it.
the heat is barely through the silicon at the solder and the chip temperature is already in the 70s.
my 5800X3D runs at 80°C in R23 with a dirt cheap 25€ cooler.
my Kraken X73 at 78°C.

 

[freeagent]

You stated older cooler in your post. Would you consider the EK basic 360 an older cooler? I know the hotspot is in a slightly different area than previously on the 5800x3D and that some coolers have started offering an offset bracket to help with this, that being said, would you suggest a different cooler for my use case? Air/water?

Nope I wouldn’t call it old at all. I strongly dislike AIOs though. What would I recommend? I have had great luck with Thermalright coolers lately. FC140 is their best cooler for modern CPUs right now. PA120 is ok too but I prefer FC140. I have a True Black on my 5600X, good cooler for that cpu. They also have a PS120 coming out, similar to PA120 but with 7x 6mm pipes instead of 6.

 

[Deleted member 185158]

lol.

CBR23 is weak man.

The go to bullshit heat maker.

 

[freeagent]

lol.

CBR23 is weak man.

The go to bullshit heat maker.

Linpack Xtreme should get an extra 20-25c

 

[Deleted member 185158]

Linpack Xtreme should get an extra 20-25c

Yes, with at least 80% memory utilization. Pound that cooler.

And then it's going to be, "under-volt" instead of address the cooling issue.

 

[freeagent]

Yes, with at least 80% memory utilization. Pound that cooler.

And then it's going to be, "under-volt" instead of address the cooling issue.

Undervolting just cuts GFlop output, you don’t want that

 

[doryuu13]

Nope I wouldn’t call it old at all. I strongly dislike AIOs though. What would I recommend? I have had great luck with Thermalright coolers lately. FC140 is their best cooler for modern CPUs right now. PA120 is ok too but I prefer FC140. I have a True Black on my 5600X, good cooler for that cpu. They also have a PS120 coming out, similar to PA120 but with 7x 6mm pipes instead of 6.

Great I'll look into that, I really appreciate your suggestions

lol.

CBR23 is weak man.

The go to bullshit heat maker.

forgive my ignorance but I don't have a ton of experience with computer building so far so still a novice. Bullshit heat maker as in not even that stressful or produces a bunch of heat for no real reason?

 

[Deleted member 185158]

Undervolting just cuts GFlop output, you don’t want that

The claims are = A game doesn't utilize that many threads, it's synthetic.

Rather see core reduction to reduce temps and raise frequency for more performance.

Are the SMT threads synthetic? No? Don't need 16+ threads? Hate the heat? Run it as a quad with SMT 5ghz and probably be under 80c with Linpack.

This kind of testing seems super rare to the point it would be a random adventure.

forgive my ignorance but I don't have a ton of experience with computer building so far so still a novice. Bullshit heat maker as in not even that stressful or produces a bunch of heat for no real reason?

All good, we all started somewhere.

CB R23 is mainly the go to show off benchmark. Many claims a cpu runs hot with this benchmark, realistically it's not going to fill CPU Cache like a burn in test such as OCCT Linpack as an example. Or if you wanted a hot benchmark, try Y-Cruncher 2.5b. It'll give a number when finished just like cinebench, the nice thing is you don't have to stare at some lame picture. Pure calculation. But it's pretty short benchmark period, so you don't have to wait to get the heat. At 72% through, your chip should throttle nicely.

 

[doryuu13]

The claims are = A game doesn't utilize that many threads, it's synthetic.

Rather see core reduction to reduce temps and raise frequency for more performance.

Are the SMT threads synthetic? No? Don't need 16+ threads? Hate the heat? Run it as a quad with SMT 5ghz and probably be under 80c with Linpack.

This kind of testing seems super rare to the point it would be a random adventure.


All good, we all started somewhere.

CB R23 is mainly the go to show off benchmark. Many claims a cpu runs hot with this benchmark, realistically it's not going to fill CPU Cache like a burn in test such as OCCT Linpack as an example. Or if you wanted a hot benchmark, try Y-Cruncher 2.5b. It'll give a number when finished just like cinebench, the nice thing is you don't have to stare at some lame picture. Pure calculation. But it's pretty short benchmark period, so you don't have to wait to get the heat. At 72% through, your chip should throttle nicely.

yeah someone had mentioned y-cruncher earlier so I was going to explore that tonight. I guess what I'm getting from all this is 5800x3D is really hard to cool if used outside its intended use case

 

[freeagent]

I guess what I'm getting from all this is 5800x3D is really hard to cool if used outside its intended use case

No not at all. It’s a very easy cpu to cool.

 

[Deleted member 185158]

yeah someone had mentioned y-cruncher earlier so I was going to explore that tonight. I guess what I'm getting from all this is 5800x3D is really hard to cool if used outside its intended use case

It's the cache. It uses energy when utilized. So it's hotter than the 2D counter parts for that reason. But even a 5800X (no 3D cache) is a pretty hot chip, the same complaints. Another reason is the cpu is a single core complex unlike a 5900X which spreads the heat over 2 dies instead of one. More cores, but less thermal output only because of surface area.

These top chips should be cooled with a custom loop. Nice heavy all copper waterblock and D5 pump with a minimum 360 radiator. Some blocks are really decent such as the all copper HeatKiller IV. And there are some blocks that are in the 200$+ range.

Most AIO's area a simple cold plate. You'd buy the top AIO and it may be comparable to a custom loop, even expandable.

Generally, people throw a lot of money at a board and CPU and then lack the funds to properly cool the darn thing.

No not at all. It’s a very easy cpu to cool.

For you.... but I also know your ambient temps are probably 20c lower than most people HAH!!! XD

 

[doryuu13]

It's the cache. It uses energy when utilized. So it's hotter than the 2D counter parts for that reason. But even a 5800X (no 3D cache) is a pretty hot chip, the same complaints. Another reason is the cpu is a single core complex unlike a 5900X which spreads the heat over 2 dies instead of one. More cores, but less thermal output only because of surface area.

These top chips should be cooled with a custom loop. Nice heavy all copper waterblock and D5 pump with a minimum 360 radiator. Some blocks are really decent such as the all copper HeatKiller IV. And there are some blocks that are in the 200$+ range.

Most AIO's area a simple cold plate. You'd buy the top AIO and it may be comparable to a custom loop, even expandable.

Generally, people throw a lot of money at a board and CPU and then lack the funds to properly cool the darn thing.


For you.... but I also know your ambient temps are probably 20c lower than most people HAH!!! XD

It seems I have a lot to learn about cooling hahaha I'll look in to higher end AIOs then and maybe start some research into custom water cooling in the future. I thought EK would have been a top brand, not really sure what other models to look at.

 

[Deleted member 185158]

It seems I have a lot to learn about cooling hahaha I'll look in to higher end AIOs then and maybe start some research into custom water cooling in the future. I thought EK would have been a top brand, not really sure what other models to look at.

Only if you're concerned about those full load testing temps really. If it works as it is, stick with what you got. No reason to throw money at it, if gaming temps are decent and that's the primary concern.

 

[doryuu13]

Only if you're concerned about those full load testing temps really. If it works as it is, stick with what you got. No reason to throw money at it, if gaming temps are decent and that's the primary concern.

Okay great thank you so much for your help, the thermals during gaming are completely fine and as such I'll probably just run with it for now

 

[Deleted member 185158]

Okay great thank you so much for your help, the thermals during gaming are completely fine and as such I'll probably just run with it for now

Right on. Have a good one!

 

[doryuu13]

Right on. Have a good one!

you too

 

[DemonicRyzen666]

The loads you guys call unrealistic are actual loads Just because your game or video conversion doesn’t reach such temps doesn’t mean it’s unrealistic. The biggest problem with older coolers is that they are not tuned for todays CPUs. Which is why you hear things like




And

It is not hotter.
that is 100% false. The review on here on techpower shows 2-4C difference, from the original 5800x. Many chips just have terrible contact.
My 5600x has 12C difference on core pairs, it's not a hot chip. It' just has terrible I.H.S contact, on part of it.