3700x temps

Dervious · 18 Aug 2019 at 17:32

My R5 3600 idles at around 29C. Running Prime95 for a few minutes and peak temperate is 55C (just under 50C form most of the tests). OK, it's a couple of cores down, but I wouldn't have thought a couple of extra cores at idle would increase idle temp by much. I can't see why a 3700X would idle at much above 30C with a decent cooler.

Tell a lie. I've just run Prime95 for half an hour. Peak temp is 76C with core voltage set at 1.35v. Stopped the test and a minute later temp is 30C.

theone8181 · 18 Aug 2019 at 20:04

Which part of the UK are you in to get ambient of 35 in a room? Oh scratch that you meant CPU temps lol. I'm idling low 30s on stock cooler, with ambient around 23ish. There's a lot of stuff running that causes it to though.

Overclocker321 · 29 Aug 2019 at 10:17

Please touch the end of the heatpipes on your coolers (if it's stock or air) during CPU full load and tell us if yours are hot ?

There are reports that some batches of Ryzen have poor heatspreader implementation, too deep machining or contact issues for correct heat transfer from the chiplet.

theone8181 · 29 Aug 2019 at 12:55

Overclocker321 said:
Please touch the end of the heatpipes on your coolers (if it's stock or air) during CPU full load and tell us if yours are hot ?

There are reports that some batches of Ryzen have poor heatspreader implementation, too deep machining or contact issues for correct heat transfer from the chiplet.

How hots hot, (i've no idea what temp they should be). I'd have thought idle temp and load temp would be enough to tell you if alls ok.

Overclocker321 · 29 Aug 2019 at 13:05

I mean, not warm

, but rather uncomfortably hot to hold the finger on them.

If under full load of the CPU like a stress test with Prime 95 or CPU Z the heatpipes are just warm, might indicate heat transfer issues inside the CPU assembly if the cooler was properly installed.

TrixX · 29 Aug 2019 at 18:51

I got two 3700X together. One idled at 29C, the other at 39C. Under any load the 2nd one would just hit 89-92C before hard throttling and performing like nutsack. The first one would hit and hold high clocks at ~72C and slowly start heat saturating the stock cooler over time. It took 2hrs to hit 89C with Aida64 Stress test going.

The 2nd one was basically DOA for any viable use case and has been RMA'd. The first clocked to 4.5GHz with manual overclocking of the CCX's. If I had only got one, diagnosing the crap one would have been a much more laborious process.

theone8181 · 29 Aug 2019 at 20:31

TrixX said:
I got two 3700X together. One idled at 29C, the other at 39C. Under any load the 2nd one would just hit 89-92C before hard throttling and performing like nutsack. The first one would hit and hold high clocks at ~72C and slowly start heat saturating the stock cooler over time. It took 2hrs to hit 89C with Aida64 Stress test going.

The 2nd one was basically DOA for any viable use case and has been RMA'd. The first clocked to 4.5GHz with manual overclocking of the CCX's. If I had only got one, diagnosing the crap one would have been a much more laborious process.

I'm assuming though, that the it should be night and day like that. One think I was surprised at was that the cooler isn't a very flat surface, I'd have thought a better cooler would be a flat copper block with the pipes pressed into the back.

TrixX · 30 Aug 2019 at 05:54

theone8181 said:
I'm assuming though, that the it should be night and day like that. One think I was surprised at was that the cooler isn't a very flat surface, I'd have thought a better cooler would be a flat copper block with the pipes pressed into the back.

Cheaper to make the current method. You are correct a flat copper block/shim would be better. Crappy mount system aside it still cools very well.

fizzy · 1 Sep 2019 at 18:33

Overclocker321 said:
Please touch the end of the heatpipes on your coolers (if it's stock or air) during CPU full load and tell us if yours are hot ?

There are reports that some batches of Ryzen have poor heatspreader implementation, too deep machining or contact issues for correct heat transfer from the chiplet.

Curious about this myself. Been using stock cooling and idle has been in 30's to 40's most of the time. Fan wavers all over the place. Finally managed to get hold of an AMD mounting kit for my TRUE 120 black - a bit old school but should be better than the stock one, if only a bit quieter. Cpu still idleing in 40's, and temps skyrocket if I try using prime95 to push cpu. Reportedly hitting high 80's. Double checked everything and seems to have good contact with cpu heatspreader - yet when at the high temps the heatpipes are very cool. Even tried lapping the cooler to make sure it was flat, but no difference.

Running CPU stress test in CPUZ, and a litttle better than with stock cooler (4.1 all core at 75c vs 4.15 all core at 71.5c) and a fair bit quieter, but rather puzzled with heat pipe temps.

This range of CPU's has some puzzling aspects - and the auto overclocking just confuses things. Or I could just be getting old.

LtMatt · 12 Sep 2019 at 12:12

Speaking of 3700X cooling, i switched out the stock thermal paste that comes with the Wraith Prism for Thermal Grizzly Conductanaut and my temps have dropped 3-5 c which is nice, as every little helps on the stock cooler.

R3X · 12 Sep 2019 at 15:06

Overclocker321 said:
Please touch the end of the heatpipes on your coolers (if it's stock or air) during CPU full load and tell us if yours are hot ?

There are reports that some batches of Ryzen have poor heatspreader implementation, too deep machining or contact issues for correct heat transfer from the chiplet.

TrixX said:
I got two 3700X together. One idled at 29C, the other at 39C. Under any load the 2nd one would just hit 89-92C before hard throttling and performing like nutsack. The first one would hit and hold high clocks at ~72C and slowly start heat saturating the stock cooler over time. It took 2hrs to hit 89C with Aida64 Stress test going.

The 2nd one was basically DOA for any viable use case and has been RMA'd. The first clocked to 4.5GHz with manual overclocking of the CCX's. If I had only got one, diagnosing the crap one would have been a much more laborious process.

Sounds like General AMD teething early on issues, the 3700x is a packed chip for sure but does appear its a hot one !

TrixX · 12 Sep 2019 at 19:01

LtMatt said:
Speaking of 3700X cooling, i switched out the stock thermal paste that comes with the Wraith Prism for Thermal Grizzly Conductanaut and my temps have dropped 3-5 c which is nice, as every little helps on the stock cooler.

Why? Conductonaut on Copper isn't good. If it was nickel plated not a problem, but copper reacts to Gallium...

However they do like water cooling from what I'm seeing, so nickel plated cooling on the way

LtMatt · 12 Sep 2019 at 19:06

TrixX said:
Why? Conductonaut on Copper isn't good. If it was nickel plated not a problem, but copper reacts to Gallium...

However they do like water cooling from what I'm seeing, so nickel plated cooling on the way

The instructions say its only aluminium that it cannot be used on.

This isn’t news, but don’t use liquid metal with aluminum – it will embrittle the aluminum and form an alloy with the coldplate, resulting in chalky disintegration of the two connecting metals. For a nickel-plated IHS, liquid metal is fine. The heavier gallium composition makes combination with nickel tenable, and is a non-issue for corrosion and performance. Copper is also mostly OK. It will stain with Conductonaut (heavily), but performance remains the same. We have not tested other liquid metals with copper yet.
https://www.gamersnexus.net/guides/...cts-copper-nickel-and-aluminum-corrosion-test

TrixX · 12 Sep 2019 at 19:29

LtMatt said:
The instructions say its only aluminium that it cannot be used on.

This isn’t news, but don’t use liquid metal with aluminum – it will embrittle the aluminum and form an alloy with the coldplate, resulting in chalky disintegration of the two connecting metals. For a nickel-plated IHS, liquid metal is fine. The heavier gallium composition makes combination with nickel tenable, and is a non-issue for corrosion and performance. Copper is also mostly OK. It will stain with Conductonaut (heavily), but performance remains the same. We have not tested other liquid metals with copper yet.
https://www.gamersnexus.net/guides/...cts-copper-nickel-and-aluminum-corrosion-test

My Threadripper water block would like to have a bar fight with whoever wrote that! It doesn't just stain, it absorbs it over time and if you break the seal you'll find there's no 'liquid' aspect left.

I need to get either a new copper base for that block or sand it down. Personally I think it's time to replace with a Heatkiller block, the Swiftech has served me well though.

LtMatt · 12 Sep 2019 at 19:53

TrixX said:
My Threadripper water block would like to have a bar fight with whoever wrote that! It doesn't just stain, it absorbs it over time and if you break the seal you'll find there's no 'liquid' aspect left.

I need to get either a new copper base for that block or sand it down. Personally I think it's time to replace with a Heatkiller block, the Swiftech has served me well though.

Well, worst comes to worst its only the stock cooler. I am not worried about staining, as long as performance is not affected as i don't spend time looking at the bottom of a cooler heatsink.

In your scenario having paid for a block though, i can understand your concern.

TrixX · 13 Sep 2019 at 08:06

LtMatt said:
Well, worst comes to worst its only the stock cooler. I am not worried about staining, as long as performance is not affected as i don't spend time looking at the bottom of a cooler heatsink.

In your scenario having paid for a block though, i can understand your concern.

Eventually it'll absorb into the block, so depends on the length of time between cooler removal and cleaning. Mine was ~1yr and after removal I couldn't re-use it without machining or getting a replacement copper base. The other issue I ran into was the nickel plating of my GPU block actually failing and separating from the copper into the water and clogging my filter

Probs just a defective bit of plating but still not something one expects from such an expensive part.

EDIT: BTW sent a PM in regards to iRacing and AMD support, just wondering if there's something that can be done in that regard.

Jimmy Weirdarms · 24 Sep 2019 at 19:17

fizzy said:
This range of CPU's has some puzzling aspects"

You can say that again! I have a 3800X on a 240mm AIO radiator/pump cooler. It idles between 35 and 50C, randomly dancing about. Running prime95 it started in the mid 60s and heads up to 95C... which is a bit worrying!

Woodsta888 · 24 Sep 2019 at 19:20

Jimmy Weirdarms said:
You can say that again! I have a 3800X on a 240mm AIO radiator/pump cooler. It idles between 35 and 50C, randomly dancing about. Running prime95 it started in the mid 60s and heads up to 95C... which is a bit worrying!

What are you monitoring the temps with?

Jimmy Weirdarms · 24 Sep 2019 at 19:55

Woodsta888 said:
What are you monitoring the temps with?

HWMonitor, but Ryzen Master agrees with it.

Looking at it again, I think I may have stuffed up and put my worst (but prettiest!) case fans on push duty for the radiator. Probably time for a rethink there.

Woodsta888 · 24 Sep 2019 at 21:07

Just looked at my 3800X and it's 42- 60 in afterburner, Ryzen master with histogram on shows 35 temps at idle. Prime 95 will make it go really high and will get 100% load, but gaming you'd never get anywhere near that. Mine gaming is 65 degrees.

Also as the heat under 7nm lithography is concentrated in smaller areas and heats up quicker any cooler air or water has a harder job to draw out the heat from that area, higher flow rates of water may help but there will definitely be some lag in removing such intense heat so there will be some high maximums. My concern is that if it hits 95 degrees or whatever the throttle temp is then how long does it stay throttled or is it second to second or fraction of?