Ryzen cooling (Surprised this isn't addressed)

[TemjinGold]

So with XFR only being 100 mhz and there being so little headroom on the cpus, is it actually worth it to even go for good air/water cooling? Couldn't basically crappy or midrange cooling theoretically get you the same thing? Do the motherboard power phases even matter?

 

[TemjinGold]

Wow... no one has an opinion on this?

 

[Malogeek]

Probably just too early for a decent amount of folks to have their systems and tested it. Overclocking itself isn't exactly reliable at the moment for CPU and memory so once motherboard BIOS are more stable, there will likely be a lot more opinions out there?

 

[WhoBeDaPlaya]

That's what I suspect. Have two "cheap" ($279) R7-1700s coming in from eBay, and have yet to decide on mobos (probably B350).

TBH, I haven't found the need for exotic / crazy cooling in awhile (the fanciest thing I have is an Antec Kuhler 920).
It's not like we're back in the Nehalem / Westmere days where you really needed monsters like the Megahalem.
Probably a cheap CM Hyper212+ (or EVO) would be adequate to get you up to the Ryzen ~4GHz ceiling.

 

[TemjinGold]

Figured. I returned the Corsair H115i (Corsair told me their AM4 brackets were delayed a few weeks) and ordered a Noctua U12S AM4 edition instead. Thought about the 15 but didn't like the idea of 3 lbs hanging vertically when I probably don't need that much cooling.

 

[Crono]

Temperature reporting is a little suspect, as well.

 

[guskline]

My Corsair H110i GT is doing a great job of cooling my Ryzen 7 1800x on an Asus Prime B350 plus mb inside of a Fractal Define S case.

 

[james1701]

Depends, would you like a setup you can run the fans at only 800RPM for an almost silent system?

 

[IEC]

Temperature reporting is a bit suspect. I will see if I can attain the same max OC 4.1GHz+ with the Noctua D15 AMD special edition HSF I have on the way. I had an overkill custom water cooling loop to test OCs and with the immature state of the BIOS still managed my 4.1GHz. 4.0GHz is readily attainable with very safe voltages, at least on my sample. I suspect thermals will be more than fine overclocked with the Noctua D15.

Edit:
Some users reporting it's tough to get 4.1GHz+ with even a Noctua D15 and you need AIO or water to pipe away the heat quickly enough.

Also, my order for Noctua D15 still hadn't shipped so I canceled it. In light of other user reports I'll probably reconfigure my custom loop for 24/7 operation and reinstall my water block instead of going air.

FWIW I was able to hit 4.0GHz all-core on my Hyper 212+ without issue (well, except mobo reporting temps in 50s which doesn't seem possible with such a weak cooler...)

 

[IEC]

The primary temperature reporting sensor of the AMD Ryzen™ processor is a sensor called “T Control,” or tCTL for short. The tCTL sensor is derived from the junction (Tj) temperature—the interface point between the die and heatspreader—but it may be offset on certain CPU models so that all models on the AM4 Platform have the same maximum tCTL value. This approach ensures that all AMD Ryzen™ processors have a consistent fan policy.

Specifically, the AMD Ryzen™ 7 1700X and 1800X carry a +20°C offset between the tCTL° (reported) temperature and the actual Tj° temperature. In the short term, users of the AMD Ryzen™ 1700X and 1800X can simply subtract 20°C to determine the true junction temperature of their processor. No arithmetic is required for the Ryzen 7 1700. Long term, we expect temperature monitoring software to better understand our tCTL offsets to report the junction temperature automatically.

Source: https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update?sf62107357=1

Weird temperature arithmetic, 1700X and 1800X don't actually run hotter than 1700 at same clocks, they just report hotter. Also explains why under certain BIOS settings my reported temps dropped exactly 20C...

 

[TemjinGold]

Temperature reporting is a bit suspect. I will see if I can attain the same max OC 4.1GHz+ with the Noctua D15 AMD special edition HSF I have on the way. I had an overkill custom water cooling loop to test OCs and with the immature state of the BIOS still managed my 4.1GHz. 4.0GHz is readily attainable with very safe voltages, at least on my sample. I suspect thermals will be more than fine overclocked with the Noctua D15.

Edit:
Some users reporting it's tough to get 4.1GHz+ with even a Noctua D15 and you need AIO or water to pipe away the heat quickly enough.

Also, my order for Noctua D15 still hadn't shipped so I canceled it. In light of other user reports I'll probably reconfigure my custom loop for 24/7 operation and reinstall my water block instead of going air.

FWIW I was able to hit 4.0GHz all-core on my Hyper 212+ without issue (well, except mobo reporting temps in 50s which doesn't seem possible with such a weak cooler...)

Thanks, so basically, go extreme or don't bother. There's no sense in going with something in the middle.

 

[Valantar]

Thanks, so basically, go extreme or don't bother. There's no sense in going with something in the middle.

I think there is. Ryzen is very efficient (as long as you don't push it past ~3.9 on all cores. So a mid-range air cooler could do a great job keeping temps and noise low without crowding out other components like the NH-D15 and the like.

 

[TemjinGold]

Sounds like my NH-U12S was a good choice then. Okay, I better power down and actually build the sucker now haha...

 

[Valantar]

Sounds like my NH-U12S was a good choice then. Okay, I better power down and actually build the sucker now haha...

Just remember that if your temps look ~20C too high (and you have a 1700X or 1800X), you're absolutely right, and that they're actually 20C lower.

 

[DrMrLordX]

I will tell you this much: At voltages below 1.35v, Ryzen produces very little heat. If you can be happy at clockspeeds of 3.8 GHz and lower, you probably do not need that much cooling.

Above 1.35v things get intense.

Also, I have noticed that there is a very big difference between keeping Tctl down and keeping CPU/socket temp low. It's easier to cool the IHS than it is the die. Check out this screenshot I took of data gathered during a y-cruncher run:

https://www.dropbox.com/s/pt9id67e6e7fpa2/ycrunch.png?dl=0

As you can see, the max CPU/socket temp was 45C. Even with -20C to account for the offset, Tctl still hit 68.1C. That's a ~23C difference in temp between the die and the IHS/socket.

Does anyone have a ballpark figure for the average bondline thickness of indium solder used between integrated heatspreaders and CPU dice?

 

[Valantar]

*snip*

Your picture embed is broken, and right-clicking and opening it in a new tab leads to a page telling me that

Sorry, you don't have the required permissions to download attachments. More information about this is available in the Attachment Permissions FAQ.

 

[DrMrLordX]

Your picture embed is broken, and right-clicking and opening it in a new tab leads to a page telling me that

Woops, try this link:

https://www.dropbox.com/s/pt9id67e6e7fpa2/ycrunch.png?dl=0

 

[Raduque]

Depends, would you like a setup you can run the fans at only 800RPM for an almost silent system?

I would. I am looking at parting up a new gaming rig as my laptop is getting a bit underpowered. I am looking at the 1600x and was wondering if the Cryorig H7 was enough to run it at stock.

 

[Valantar]

I would. I am looking at parting up a new gaming rig as my laptop is getting a bit underpowered. I am looking at the 1600x and was wondering if the Cryorig H7 was enough to run it at stock.

Yes. Yes it is. The H7 is a 140W TDP cooler, and even the Ryzen 7 1800X doesn't exceed 125W peak (and averages out at ~90W unless overclocked). It's more than good enough.