[mongoled]

snipp......In reference to the EK understanding of the Zen2 chip...Im not convinced that AMD support understands it. Too many variables to manage and not a lot of clarity on why bios voltages are so high when they work with such a low voltage.

Although im partially in agreement here,

this is the main reason (below response from them) I said that EK support staff are not well informed with regards to Ryzen 3000 series

According to what you have sent us (screenshots) we can tell, that your Maximum voltage is 1.5V (too much) which then the cause of this can be the rise of the temperature like in your case 85.5°C.

​

As you mentioned already "Peak temperature of 85C @ 1.3 volts", this is not true, but you have this temperature with 1.5V.

Yup, the above is what they told me, I answered back to them to first educate themselves on how Ryzen 3000 series CPUs work with regards to PBO and PB2 and specifically with regards to voltages/temperature/frequency and how many cores were active

 

[AlphaC]

You should be looking at CPU package power.


69% of 128W is 88W

 

[mongoled]

You should be looking at CPU package power.

69% of 128W is 88W

Hi!

Sorry but im not entirely sure what you are eluding to.

Are you telling me that I should have told EK to look at the CPU power package ??

 

[AlphaC]

Yes the voltage is irrelevant. Nobody rates coolers based on voltage. It just happens that CPU's power normally scaled with the square of voltage with fixed voltage overclocks.


See article: https://www.computerbase.de/2018-08/cpu-tdp-verbrauch-amd-intel/


65W TDP 3rd gen Ryzen = 88W power limit , so you're basically similar to stock clocks.


For 2nd gen the TDP was obtained by 0,189 °C/W with a tCASE target around 60°C.

 

[Kutalion]

Temperature also depends on power dispersion. If most of the TDP is centered around 2-3 cores, the thermal density will be very high, which is naturally hard to cool as it's 7nm process so every core is very small. And Ryzen only shows the hottest core temp.

If you have any question, EK has a forum rep with a good understanding on Ryzen 3000 if the support is not yet familiar with the complex mechanism it works on. Just poke @EKJake

 

[AlphaC]

If you mean power density, as long as the microfins have coverage I don't see why there would be a difference. All of the microfins have contact with the working fluid/coolant and the fluid is in motion due to the pump.


The temperature may be higher because it's more concentrated, but I mean block to block the main thing is the microfin coverage.

 

[mongoled]

Yes the voltage is irrelevant. Nobody rates coolers based on voltage. It just happens that CPU's power normally scaled with the square of voltage with fixed voltage overclocks.

See article: https://www.computerbase.de/2018-08/cpu-tdp-verbrauch-amd-intel/

65W TDP 3rd gen Ryzen = 88W power limit , so you're basically similar to stock clocks.

For 2nd gen the TDP was obtained by 0,189 °C/W with a tCASE target around 60°C.

Thanks for clarifying.

I really hope that these BIOS are very immature and that we will see improvements in the way PBO and PB2 work.

As it seems (apart from the few erroneous results floating around with regards to temps) either the Ryzen 3000 series chips with current BIOS implementation are running 'hot' or all our water blocks are doing a poor job in cooling the CPU's.


** EDIT **
Forgot to say, my CPU is a 3600x, which is 95W TDP.

So from what you have said above, my CPU is not even using its power limits 'correctly' .....

 

[Kutalion]

If you mean power density, as long as the microfins have coverage I don't see why there would be a difference. All of the microfins have contact with the working fluid/coolant and the fluid is in motion due to the pump.


The temperature may be higher because it's more concentrated, but I mean block to block the main thing is the microfin coverage.

Because it's harder to pull heat away from small concentrated spots than when it's spread around across bigger surface. It's not the same if you have 40W of power on 2 cores, and 40W spread across 6 cores. I'd argue that you could probably have double the power consumption if it's spread across both CCDs and all cores and achieve almost the same temps as having 40W power consumption on 2 cores on let's say single CCD.

 

[AlphaC]

Yet people are cooling on cheaper mainstream blocks such as Raystorm Neo and Eisblock XPX just fine with R9 3900X or R7 3800X (which has more cores per CCD). Also that is stock power limit (actually lower for than stock for R5 3600X). There goes that theory.


That's nearly stock cooler performance, yet you're defending it?

https://www.techspot.com/review/1871-amd-ryzen-3600/


The EK response reminds me of clueless tech support people that have never taken thermo or heat transfer. Might as well ask people if they turned on their pump too while you're at it. Did you go defend EK when they messed up their Threadripper block too?

 

[Bart]

EK sucks, this is the conclusion I have come to after years of buying their crap. They are decent for radiators (not the best), decent for pump tops and reservoirs, but I will NEVER buy another EK CPU or GPU block again. Their quality control is an absolute JOKE, and has been for years. They're kinda like Corsair, making cheap products with high price tags, and they succeed because they have been VERY smart with product distribution. You can get their crap EVERYWHERE, while the good stuff like Watercool is harder to come by, depending on what country you live it. Plus, they are very good at being the FIRST to market with wide availability. Look at the Radeon 5700XT, EK had a water block out the door for that thing before AIB partners came out with their own air cooling solutions. It probably came with bad thermal pads, loose screws (or STRIPPED ones, that's an EK specialty), an awful suffocating back plate that looks pretty, costs tons, and makes things worse, etc. You just cannot trust their QC / QA.

Back on topic, my HeatKiller IV CPU block seems to be doing a *fantastic* job of cooling my 3900X, die design be damned. Watercool is quality kit, always has been.

 

[mongoled]

I just had this confirmed to me by EK support.

The R&D guys checked in details the things again and unfortunately, the temps that were displayed in the graphs are correct and this is sort of expected for the monoblock that has the older Supremacy core design (fins), so the best possible option would be to use the Velocity block on the CPU as it has the new core design and that could have made the temps bit lower than with the monoblock.

Now I am off to research the differences between the supremacy and velocity, to see if I can modify my monoblock to improve performance.

 

[zGunBLADEz]

This... TDP is what it needs to take care off

Btw with a monoblock you always gonna have higher cpu temps than a cpu block thats just how it works.


Cpu cooling is the most tedious thing to cool on any loop. You can have the best loop ever and the cpu is going to drag.

 

[mongoled]

This... TDP is what it needs to take care off

Btw with a monoblock you always gonna have higher cpu temps than a cpu block thats just how it works.

Cpu cooling is the most tedious thing to cool on any loop. You can have the best loop ever and the cpu is going to drag.

Yeah, of course temps are going to be higher, though looking at other results from custom loops and CPU used etc, im not doing too bad.

Regards the advice from EK, looking at comparative reviews the velocity does not show any significant gains compared to its predecessor.

Probably why the advice from EK was carefully worded to say the following

snip......as the older Supremacy core design (fins), so the best possible option would be to use the Velocity block on the CPU as it has the new core design and that could have made the temps bit lower than with the monoblock.

 

[AlphaC]

Saw this today:
https://hothardware.com/reviews/all-amd-ryzen-9-3900x-liquid-cooled-gaming-rig-build-with-maingear
3900X PBO 3 4.13GHz on liquid (probably Bitspower Touchaqua kit) = 69.57°C in CB R20
Power draw per Ryzen master = 12% of 1200W so around 144W
EDC current draw per Ryzen master = 26% of 600A which means ~156A

TDC Current draw per Ryzen master = 18% of 540A which means ~97A

 

[BeeDeeEff]

What websites/blogs/forums other than this one should I follow if I'm still interested in comparative waterblock performance for zen 2? Haven't heard much discussion on this topic in a few weeks but I'm still waiting for the 16-core chip to come out to finalize things.

 

[Nighthog]

Just for some useful information I checked what kind of temperature my setup would get when running fully topped out 3800X with a Alphacool Eisblock XPX.
Mounted in the regular orientation with intake to the left and outflow on the right. That makes the channels run up and down vertically on the block crossing the dies on their short sides up and down. Water coming in just between the IO-die & core-chiplet.

3800X 1.525V @ 4450Mhz CPU with 4x8GB 4266Mhz MEM & LLC ULTRA EXTREME (no vdroop)

CB R20 ~160W, 86A = 90-91C
IBT Very High 10x ~180W, 100A = 95-100C

I pushed full RPM on all fans to reach 28C water temperature with 24-25C Ambient.
Though I have to note my GPU says 26C so maybe water temp is even lower.

1x360 & 2X120mm 45mm thick radiators with a DDC-1T Plus.

Wattage = PPT
Amperage = TDC

 

[jfrob75]

Looking for some feedback

moved to separate thread

 

[jakemfbacon]

So I am in need of a new cpu waterblockfor my 3900x. Which should I go with and what is the best site to buy from these days? I'd like to get on amazon if possible.

 

[Haldi]

What websites/blogs/forums other than this one should I follow if I'm still interested in comparative waterblock performance for zen 2? Haven't heard much discussion on this topic in a few weeks but I'm still waiting for the 16-core chip to come out to finalize things.

Yeah... i'm stuck on the same one here.
Using Google you find almost nothing. Seems like all those cheap AiO's totally killed the last bit of surviving Waterblocks reviews.

Seems like the "Watercool HEATKILLER® IV PRO" is the thing to go. (Difference Pro vs Basic: https://watercool.de/wbb/index.php/Thread/4564-IV-basic-vs-Pro/)
Price difference is like 10.-

Other things that fit AM4 would be:
Bykski AMD AM4 CPU-RYZEN-X-MK-S
XSPC RayStorm Neo WaterBlock
Aquacomputer cuplex kryos NEXT AM4
Alphacool Eisblock XPX

Though i'm not sure others than the Bykski and XSPC RayStorm have a big enough groundplate to fill the whole CPU....


Anything i missed?

 

[Bart]

Seems like the "Watercool HEATKILLER® IV PRO" is the thing to go. (Difference Pro vs Basic: https://watercool.de/wbb/index.php/Thread/4564-IV-basic-vs-Pro/)

Anything i missed?

Nope, you listed the only block that really matters. But seriously, Watercool makes amazing gear. EK's cheap crap cannot compare to anything in the HeatKiller line. Once you hold one in your hands, you'll understand. They might not be as widely available than other supposedly 'top' brands (lulz), their quality is LEAGUES above stuff like EK and Corsair. I'd order directly from them if you can't find them anywhere else. If they can ship DHL to Canada in a decent time frame, and not let Canada Customs ruin it, they can ship anywhere.