[SOLVED] AIO Pump speed too high ?

tek3195

Prominent
Feb 7, 2021
67
4
545
I have an Alphacool Eisbaer 240 Black AIO that both operating systems show pump speed over 3100 RPM. Everything I have found lists it as a 2600 RPM pump. I am wondering if I need to cut some power to slow it down to increase life expectancy, let it fly and hope for the best or if both systems are reading it wrong. I can run it 80% and get idle temps 30 C and high 63 C in GTK-StressTest, which should be sufficient I think. Should I slow it down to avoid abuse or will it hurt it to run faster ?

Asrock B365M Phantom Gaming 4
Core i7-9700
TeamGroup T-Force Vulkan 4x8GB @2666MHz
MSI GTX 970 Gaming X
Samsung 970 EVO m.2 500GB
SK Hynix Gold S31 1TB SSD
WD Blue 2.5" 2TB HDD
Intake: Alphacool Eisbaer 240 black push/pull w/ Arctic Bionix P120
Exhaust: NexXxoS ST30 120mm pull w/Arctic Bionix P120, 2x120mm Be Quiet Pure wings 2
Rosewill Photon 550W
CoolerMaster Q300L
 
Solution
Sometimes, using more than 1 reporting software simultaneously can give false results, so that's always a consideration when asking 'how' it's read, like having control by motherboard software setup to read a fan on CPU1 and pump on CPU2, but open up Cam to check and it's assuming the pump on CPU1 and a fan on CPU2 and not knowing which software is pulling priority. Simple oversight and not knowing exactly how much you have actually invested into setup, ppl have to assume the worst case scenario until details are clarified, like you did in subsequent posts.

As to pump speeds, the original Eisbaer was released/reviewed back in 2016, and thats when all the documentation on the websites was released and historically web ppl are very slow...

Paperdoc

Polypheme
Ambassador
My first thought is: how do you know the pump speed? The system design, if I understand its manual properly, is that the PUMP should plug into the CPU_FAN header of some type using its 3-hole fan connector. Using the supplied PWM fan Splitter, the RAD FANS plug into a different mobo fan header that NEEDS to be guided by the CPU internal temperature sensor if at all possible. Exactly which mobo fan headers you use depends on what your mobo has.

On your mobo, the ideal place to connect the PUMP would be the header labelled CPU_Fan2 / W Pump header. See your mobo manual p. 7, item 5. Then in BIOS Setup (see p. 80-81) configure that this way: Switch to Pump; Control Mode to PWM (even though this is a 3-pin connection); Fan Setting to Standard or to Full Speed or Turbo; Temp Source to CPU. This will run that pump at full speed all the time.

On your mobo, the ideal place to connect the RAD FANS would be the header labelled CPU_Fan1 header. See your mobo manual p. 7, item 2. Then in BIOS Setup (see p. 80) configure that this way: Control Mode to PWM (IF you even have a choice for this item!); Fan Setting to Standard. This will have the rad fans' speeds controlled automatically according to the CPU's internal temperature.

With those connections, you can observe the speed of the PUMP in BIOS Setup on the CPU_FAN2 / W Pump header. the speed of one of the RAD FANS will be shown on the CPU_FAN1 header (the second fan's speed will not be shown).

If you have been using some third-party utility to "see" fan and pump speeds, you need to realize two things. First, some of those require calibration to display speeds properly. Secondly, the labels they apply may NOT be correct - with your own knowledge of WHERE you plugged in which, you have to decipher which fan label in the utility is which device. The utility has NO way of knowing those details. On the other hand, what you see in BIOS Setup is reliable - it is done by the mobo maker for this mobo. You MAY have a utility supplied by the mobo maker - see your manual p. 38 regarding Phantom Gaming - that runs under Windows so you can use it during normal operations, and that can show you fan and pump details. But again, remember to look for the speeds of your items at the header labels you have them plugged into.
 

tek3195

Prominent
Feb 7, 2021
67
4
545
I am seeing pump speed via Asrock tuning utility and in bios. Can also see it in fancontrol in Linux which gives me the same reading(+/- 50RPM) as BIOS and Asrock Tuning utility in windows. I have pump plugged to CHA_FAN1 for accessibility and 5 aio fans plugged to CPU_Fan2. Phantom Gaming 4 allows changing cpu_fan1,cpu_fan2 and cha_fan1 between fan or water pump in the bios. Also allows switching between cpu or mobo monitoring for each connection. I have them configured correctly for the way I have them plugged in.

The odd connection choices were needed with total of 7x 120mm fans, 2 rads and vertical gtx 970 in a matx mini tower case. No way of disconnecting pump lead to run pump only for filling loop with everything installed, CPU_Fan1 can't be seen or touched from any angle.

My question was more towards wanting to know whether running the pump that much faster than the normal 2600rpm was bad for it. Not whether there was skepticism of my ability to "see" the pump speed.
 

Karadjgne

Titan
Ambassador
Sometimes, using more than 1 reporting software simultaneously can give false results, so that's always a consideration when asking 'how' it's read, like having control by motherboard software setup to read a fan on CPU1 and pump on CPU2, but open up Cam to check and it's assuming the pump on CPU1 and a fan on CPU2 and not knowing which software is pulling priority. Simple oversight and not knowing exactly how much you have actually invested into setup, ppl have to assume the worst case scenario until details are clarified, like you did in subsequent posts.

As to pump speeds, the original Eisbaer was released/reviewed back in 2016, and thats when all the documentation on the websites was released and historically web ppl are very slow (to never) about going back to adjust info, unless there's a different sku, model etc. In the last 5 years it's entirely possible that the motor was changed/upgraded and now runs at 3100rpm ± instead of the documented (and possibly out of date) 2600rpm.

Or it could be a built in surplus, the 2600rpm as a stock default value that changes to a higher rpm when a higher load (added components) is applied.

I know my EK 3.2 DDC pwm is rated for a minimum of 1725rpm, yet according to any reported reading, it has no issues going as low as 1400rpm.
The built-in pump is similar in design to Alphacool’s DC-LT Ceramic Ultra Low Noise pump. PWM control allows for pump power as low as 7V, for quiet operation

The pump here hit 2592 RPM at 12 V which is within error margins of the rated 2600 RPM and nice to see, and went all the way down to 524 RPM at 2.75 V before completely stopping (it hit 829 RPM at 3 V) and then restarted again at 3.25 V.

Rated 7v-12v, yet would spin all the way down to 3v. 2 different website reviews.

So is the 3100rpm accurate? Probably. Could you lower that to a constant 2600rpm? Definitely, or set it on a fan curve between 1000-2600rpm pwm variable? Definitely.

I can't see it as an issue, but it is a pump and vital hardware, so limiting its ability or setting its ability might be in your best interests if it's a concern.
 
  • Like
Reactions: Paperdoc
Solution

tek3195

Prominent
Feb 7, 2021
67
4
545
Sometimes, using more than 1 reporting software simultaneously can give false results, so that's always a consideration when asking 'how' it's read, like having control by motherboard software setup to read a fan on CPU1 and pump on CPU2, but open up Cam to check and it's assuming the pump on CPU1 and a fan on CPU2 and not knowing which software is pulling priority. Simple oversight and not knowing exactly how much you have actually invested into setup, ppl have to assume the worst case scenario until details are clarified, like you did in subsequent posts.

As to pump speeds, the original Eisbaer was released/reviewed back in 2016, and thats when all the documentation on the websites was released and historically web ppl are very slow (to never) about going back to adjust info, unless there's a different sku, model etc. In the last 5 years it's entirely possible that the motor was changed/upgraded and now runs at 3100rpm ± instead of the documented (and possibly out of date) 2600rpm.

Or it could be a built in surplus, the 2600rpm as a stock default value that changes to a higher rpm when a higher load (added components) is applied.

I know my EK 3.2 DDC pwm is rated for a minimum of 1725rpm, yet according to any reported reading, it has no issues going as low as 1400rpm.




Rated 7v-12v, yet would spin all the way down to 3v. 2 different website reviews.

So is the 3100rpm accurate? Probably. Could you lower that to a constant 2600rpm? Definitely, or set it on a fan curve between 1000-2600rpm pwm variable? Definitely.

I can't see it as an issue, but it is a pump and vital hardware, so limiting its ability or setting its ability might be in your best interests if it's a concern.

Thanks for the clarification on assuming worst case scenario. That eases the sour taste quite a bit. I see where it would have helped to have included that info to start with, I'll get it eventually.
Slowing it down was what I decided to do. Alphacool finally replied that it is 2600rpm(+ -10%) and that 3180 would just be extra performance not to be worried about. My background is mechanical and 50 years have taught me that all things mechanical when pushed beyond their limits will break. 10% margin is understandable but at 22% I was a bit concerned. Lowering speed to 2500 increased temps by 1 degree for 1-hour stress test, average temperature 63 C sounds good enough. Especially when running WFO only drops it by 1 degree.
I want to replace the whole loop with bigger better and stronger for absolutely no good reason other than that's just me. But, I want to replace it on my terms, when I want to, not due to failure at the most inopportune moment possible. Thanks for the insight, I appreciate it.
 
  • Like
Reactions: Paperdoc

Karadjgne

Titan
Ambassador
Optimus Foundation has about the best cpu blocks there is, solid construction and top thermal performance. Xspc raystorm pro is very decent too.

For gpu blocks, I'm a fan of the Watercool HeatkillerIV, just seems to do a better job than most.

Tubing is upto you, but I prefer the thicker 10/16 to the more standard 10/13, it runs less risk of kinking or sagging and there's more meat for the fittings to grab.

Do your homework on rads. They aren't all the same, especially when it comes to fan needs. Some do better with high rpm, some do better at lower rpm, all depending on where the peak (if there is one) falls and resultant diminishing returns.

The Eisbaer is a decent idea, but the relatively weak pump (compared to a D5) and smaller tubing doesn't lend well to decent flow or headpressure
 

tek3195

Prominent
Feb 7, 2021
67
4
545
Optimus Foundation has about the best cpu blocks there is, solid construction and top thermal performance. Xspc raystorm pro is very decent too.

For gpu blocks, I'm a fan of the Watercool HeatkillerIV, just seems to do a better job than most.

Tubing is upto you, but I prefer the thicker 10/16 to the more standard 10/13, it runs less risk of kinking or sagging and there's more meat for the fittings to grab.

Do your homework on rads. They aren't all the same, especially when it comes to fan needs. Some do better with high rpm, some do better at lower rpm, all depending on where the peak (if there is one) falls and resultant diminishing returns.

The Eisbaer is a decent idea, but the relatively weak pump (compared to a D5) and smaller tubing doesn't lend well to decent flow or headpressure

Thanks for recommendations and info.

The Eisbaer was a rookie move. This is first PC build. Not a bad move though, decent quality and with additions I have great performance. It allows me to run i7-9700 @4.5GHz all cores for hours on end with power limit set at 300W to stop all throttling and high of 63 C. Chasing any cooler temps wouldn't be anything but a waste of money for this setup. Or at least that's what I'm thinking, performance can't get any higher that I'm aware of.
 
Last edited:

Karadjgne

Titan
Ambassador
Most AIO 240mm rads are rated @ 250w, 280mm @ 300w and 360mm @ 350w.

The 9700k can boost to 4.6GHz all core without any OC, only hitting @ 170w, so your 4.5GHz locked is not only well below your 300w power limit, but also below default turbo core speeds for all cores.

You've shortchanged the cpu, which is why you get such low temps, if you were to bump that to a 5.0GHz all core, you'd still only be @ 200w-220w, and within the Eisbaer specs, but would be seeing temps at least in the higher 70's, possibly low-mid 80's.

Temp curves are not linear, they are logarithmic, the closer you get in wattage output to maximum, the more temperature climbs per watt. So at 170w you are still pretty low on the curve, 1 watt added might = 1°C, but at 220w 1 watt added might = 3°C instead, that curve going almost straight up.

Using a larger radiator, that has higher capacity, like a 280mm, 170w is only halfway there, you are still on the flat part of the curve. Even 220w is barely equitable to 170w on a 240mm. Better efficiency, lower temps.

Custom loop rads aren't aio rads. Some 280mm can reach 500w+ range, depending on the fans and thickness of the rad used, figure a 240mm can reach 400w+ if specd right.

So yes, you can lower temps significantly and/or get some performance out of that i7 without going much (if any) higher than you currently are at.

Loops aren't stuck, except for gpu blocks, even the cpu block is movable if you stick with Intel, as a 9th gen is the same as 11th gen mounting. Even if 12th gen changes its mount (doubtful) most Intel blocks will have adaptable or changeable brackets, so your loop can pretty much last as long, if not longer than air or aio cooling.

Options 💪
 
Last edited:

tek3195

Prominent
Feb 7, 2021
67
4
545
Most AIO 240mm rads are rated @ 250w, 280mm @ 300w and 360mm @ 350w.

The 9700k can boost to 4.6GHz all core without any OC, only hitting @ 170w, so your 4.5GHz locked is not only well below your 300w power limit, but also below default turbo core speeds for all cores.

You've shortchanged the cpu, which is why you get such low temps, if you were to bump that to a 5.0GHz all core, you'd still only be @ 200w-220w, and within the Eisbaer specs, but would be seeing temps at least in the higher 70's, possibly low-mid 80's.

Temp curves are not linear, they are logarithmic, the closer you get in wattage output to maximum, the more temperature climbs per watt. So at 170w you are still pretty low on the curve, 1 watt added might = 1°C, but at 220w 1 watt added might = 3°C instead, that curve going almost straight up.

Using a larger radiator, that has higher capacity, like a 280mm, 170w is only halfway there, you are still on the flat part of the curve. Even 220w is barely equitable to 170w on a 240mm. Better efficiency, lower temps.

Custom loop rads aren't aio rads. Some 280mm can reach 500w+ range, depending on the fans and thickness of the rad used, figure a 240mm can reach 400w+ if specd right.

So yes, you can lower temps significantly and/or get some performance out of that i7 without going much (if any) higher than you currently are at.

Loops aren't stuck, except for gpu blocks, even the cpu block is movable if you stick with Intel, as a 9th gen is the same as 11th gen mounting. Even if 12th gen changes its mount (doubtful) most Intel blocks will have adaptable or changeable brackets, so your loop can pretty much last as long, if not longer than air or aio cooling.

Options 💪

I have i7-9700 locked non-K processor. It's a 65W processor with 3.0GHz base speed and boost speed of 4.7GHz single core and 4.5GHz all core. The 300W long and short duration power limit is where I had to set it to stop all throttling with my mobo. Stepped it up 10W at a time till I got throttling free run. With it set at 300W I have seen high draw of 201W but normally high is closer to 175W with avg 150W.

I don't want to shortchange anything or anybody, myself in particular. How can I bump it past 4.5GHz when it's locked ? I can take the heat if I can get to it.

Edit: My 240 is more like 360. There is an additional 120mm NexXxoS 30 rad in loop at top of case. Tube length for short tube is about 3.75" with around 16" total for all tubing.
 
Last edited:
  • Like
Reactions: thequn

thequn

Distinguished
Nov 4, 2012
393
0
18,810
hacked bios but that's a whole other can of worms to open.

or check your bios updates a few had them from that gen but was patched out. long time ago for remember all the details.
 

tek3195

Prominent
Feb 7, 2021
67
4
545
hacked bios but that's a whole other can of worms to open.

or check your bios updates a few had them from that gen but was patched out. long time ago for remember all the details.

Yeah, I don't think I want to open that can. Not wanting to fish quite that deep yet. Hell, a couple months ago when I started this venture all I knew was it had twice the cores as my i5-3740. Knew nothing of base or boost clocks or anything else. It's all new to me so I'm thinking it's not too bad for first build. Beats the re-furbed Dell Optiplex 7010SFF by a long shot.
 

Karadjgne

Titan
Ambassador
Ahh, ok. Sorry lol, so used to seeing 9700k that I skipped right over that part with a 9700. Also not used to seeing a locked cpu under liquid cooling ☕

Ok, so you are about as good as you'll get, especially with that extra 120mm. A full loop would be better, but at this point you are right, it's not worth the cost, especially not on a gtx970, that's expense that's not required.

Maybe when you replace the gpu etc.
 

tek3195

Prominent
Feb 7, 2021
67
4
545
Ahh, ok. Sorry lol, so used to seeing 9700k that I skipped right over that part with a 9700. Also not used to seeing a locked cpu under liquid cooling ☕

Ok, so you are about as good as you'll get, especially with that extra 120mm. A full loop would be better, but at this point you are right, it's not worth the cost, especially not on a gtx970, that's expense that's not required.

Maybe when you replace the gpu etc.

No worries, kinda figured you missed that part. This started out as a way to build Android ROMs and Recoveries that didn't take 12 hours like my 7010 Optiplex with i5-3470 and 16G DDR3. I started with air cooling and it was in the 80's and throttling across the board in XTU. There was no turbo boost to be had and running 3.0GHz was completely unacceptable. So I started with thermal throttling and went to water cooled, then on to dealing with power limit throttling and edp limit throttling. I honestly don't know if I would be able to run max boost speed all core for unlimited time on air or not, I doubt it, but don't know.

I can not think of any reason to replace the gpu though. I don't play any games and the GTX 970 is overkill when internal graphics would work fine. The biggest reason the gpu is there is looks. Red theme and MSI red and black gpu work well together. The vertical mount falls in line with the added 120mm rad, which was just to see if I could. Neither are supported by the case and required a bit of modification to make them happen. Felt sort of like breaking the rules when doing it and that's a good feeling for me. Like we say at the race track, there's cheaters and losers, and I hate rules and being told I can't do something. Could have gotten the same results for half the cost if it was just a straight forward build. Now that I know what I would have to do to stop throttling I could do it cheaper but I already have this one, and would never buy a locked processor again. It's been one hell of a learning experience.

On second thought, I may replace gpu. I have an r9 270 that is way smaller that would allow easier access for maintenance. It would also let me see some of the money I spent as the 970 hogs the scene and hides a lot of stuff. It's no sff build but still packed fairly tight in matx mini tower. The r9 would still be overkill for checking email and watching a few how-to vids.
 

Karadjgne

Titan
Ambassador
You can run that cpu flat out on air, just requires the biggest aircooling, like Darkrock 4 Pro or NH-D15, because they have the same 250w capacity as the Eisbaer, but you'd not get the benefits of the extra 120mm rad.

But a lot depends on the software. 100% load like Prime95 small fft (no AVX) is one thing, running that with AVX or AVX2 (or equivalents) is just downright brutal on a cpu, seeing loads equitable to upto 130%, buh bye wattage.

So with your use case, you might think of it as a rookie error, but realistically it's a very viable solution and basically follows the intended purposes of why the Eisbaer was designed the way it was in the first place.
 

tek3195

Prominent
Feb 7, 2021
67
4
545
You can run that cpu flat out on air, just requires the biggest aircooling, like Darkrock 4 Pro or NH-D15, because they have the same 250w capacity as the Eisbaer, but you'd not get the benefits of the extra 120mm rad.

But a lot depends on the software. 100% load like Prime95 small fft (no AVX) is one thing, running that with AVX or AVX2 (or equivalents) is just downright brutal on a cpu, seeing loads equitable to upto 130%, buh bye wattage.

So with your use case, you might think of it as a rookie error, but realistically it's a very viable solution and basically follows the intended purposes of why the Eisbaer was designed the way it was in the first place.
I don't really use windows very often. Mine is portable on a usb drive and in fact, in a drawer right now since I'm not using it. I don't know what kind of load make puts on it when building. It hits it with 100% cpu usage but will have to look and see what kind of power it draws. I believe I can use powertop to read wattage real time.

The reason I said rookie move was that there is higher quality better performing equipment that I was unaware of at the time. I don't think of it as an error and agree with you, actually, I think it's a kickass solution for first build with no experience.

The Eisbaer being designed to be expandable was great. Once I replaced fittings and tubes it lost its identity somewhat with no quick disconnects. Just a loop with a pump/res combo now. They say it can be taken apart and customized any way you want to make it your own. Guess I've made it my own now, no turning back.