Jump to content
Corsair Community

CoPro & Hydro X Performance Settings


Alex B

Recommended Posts

Hi All – I skimmed the other threads looking for any that pertained to these issues but didn’t see anything… please correct me if I’m wrong.

 

I have a few questions after just finishing my first custom loop: XD5 pump/res -> corsair 2080 ti block -> corsair intel CPU block -> 360 mm rad -> 360 mm rad -> pump with 14 mm acrylic hard line. See attached pic.

 

1.) I’m getting the message “you need to plug temp sensor into #1 slot on CoPro” in the performance tab when clicking on “fan #6” (pump/res). I currently have the temp sensor in the back port of the XD5 and plugged into the #1 T slot on the CoPro. Is the presence of this message signifying that the software thinks I don’t have it plugged in? Shouldn’t this message NOT be there and instead a pump/speed curve that I can adjust? I’ve assumed that the message is there for everyone and that in order to adjust the pump speed with respect to the temperature I just need to assign a custom curve to the pump. That brings me to my second issue.

 

2.) the pump speed seems slow if set to the “hydro x series pump” setting. If I set my rad fans to a fixed 1200 RPM (just for example) and let the hydro x series profile control the pump speed, my GPU gets up to the 70s C during heaven benchmarking. If I assign a custom profile to the pump that gets up to 3000-4000 rpm or higher, the temp quickly falls to mid 40s – low 50s. I plan on overclocking the GPU and CPU here soon, and I’m just not putting much confidence in the default hydro x series pump speed profile. What setting/profile are you guys using?

 

3.) Is the aforementioned default hydro x pump speed curve setting basing its curve off the coolant temp? In my mind this makes sense to do, but I’ve found that the delta T is pretty narrow… ranging 24C – 27C. In theory I could make a custom curve along the lines of 24C = 1000 RPM, 25 = 2000 RPM, 26 = 3000 RPM, 27 = 4000 RPM… BUT the problem I foresee in this is that the coolant temperature climbs MUCH slower than actual component (GPU and CPU) temps because of the coolant/water’s high specific heat. Because of this, wont the pump speed significantly lag behind the actual GPU/CPU temps? I’d really prefer not to have the GPU and CPU running at 90 degrees for 15 minutes before that’s enough time for the component to heat the coolant enough for the pump/fans to change RPM. Isn’t this a problem for everyone?

 

4.) In my testing, the most important thing for cooling in my loop is pump speed. I understand/agree that its fan speed and radiators that should what’s important over some minimum pump speed (0.5 – 1.0 gpm)… but it just doesn’t seem to be the case in reality for me. If I max out my 6 radiator fans (pushing out) and my 3 intake fans (not paired to rads), this minimally affects my temps… varying the pump speed drastically affects my temps… even at higher RPMs above some minimum threshold. What gives? Based off what I’m seeing I should do one of two things: #1, just set the pump to a fixed 4000 rpm and set the fans to fixed 1000 rpm (where I can barely hear them) since this keeps the components in the 40-50C range, or #2 create a custom curve for the pump speed based off coolant temp or GPU/CPU temp and leave the fans to a low fixed speed (since I’m seeing diminishing returns with higher fan RPMs). Option #3 could just be to set the pump to fixed maximum 100% of the time… since the pump makes hardly any noise, but I’d be afraid of burning out the pump.

 

5.) Should I put the fans paired with rads on the same CoPro as the Hydro X series stuff? I think the answer is probably yes. I have two separate CoPros daisy chained together with the 6 radiator fans on the 1st CoPro and the hydro X series stuff + 3 intake fans on the 2nd CoPro. I did this because the XD5 pump takes up a fan slot on the CoPro and there wouldn’t be enough room on one CoPro for the hydro X series stuff plus all six fans (I suppose I could get a PWM fan splitter for two of the fans then I’d have enough room). BUT, I can control the rad fans from the other CoPro based on the T inputs from the coolant, GPU, or CPU. So if I don’t care to have fan speeds based on the default “hydro x series fan” curve and instead a custom curve of my own choosing, having it all on the same CoPro doesn’t matter—and, I’m inclined to do it this way because I’m seeing pretty crap performance if I leave the “hydro X series pump/fan” curves to do their thing instead of my custom curves.

 

This is absolutely a ton of information/questions, and I really appreciate whoever is reading through all of this to provide some feedback. Thanks in advance.

 

AB

setup.jpg.e51972c211a60c11175b8df7851cd514.jpg

Link to comment
Share on other sites

Check out the default curves here: http://forum.corsair.com/v3/showpost.php?p=1011691&postcount=14.

 

1) Yeah, I get this message as well. It's confusing (IMHO) and it doesn't indicate that it's not configured that way - just reminding you to configure it that way. However, the default curves are NOT adjustable. You can, however, make your own.

 

2) I find the pump speed a little slow for my liking as well so I created a custom curve for the pump with a min at 3000 RPM.

 

3) It is based on the coolant temp. And you can create your own.

 

4) Keep in mind that fans that are exhausting are impacted by internal temperatures as well. Even though you don't have the GPU dumping heat into the case (good), you still have the VRMs ... and they can get hot at load. Intake does cool better.

 

5) Yes. The CoPro can run fan curves based on the onboard sensors without the software, so you get your fan curves running at all times (even at boot). This is definitely recommended.

Link to comment
Share on other sites

Thanks for your reply!

 

1) glad to know nothing is wrong with my temp sensor.

 

2) gonna end up creating a custom curve as well then.

 

3) Do you base your custom pump curve on coolant temp then? How do you get around the problem of the components heating up faster than the coolant itself? Or does this not happen to you? Seems like it takes forever for my coolant to change temps and my components run hot for a long time before it kicks in.

 

4) I'll probably use the temp sensors that came with the CoPro and put them in various places inside the case and vary the intake fans RPM based on their temps. Or just set it to also vary with the coolant temp vs. set it to a fixed value.

 

5) Shouldn't whatever "profile" saved to the hardware run at boot without the software loaded up yet, irrespective of which CoPro the fans are connected to vs. the hydro X series?

 

Thanks again!

Link to comment
Share on other sites

3) Yes, the pump curve is based on the coolant temp using the sensor in the pump. Component temps will increase more than the coolant temperature - water as a much higher specific heat than a processor or GPU. This is normal. Heat is generated at the pins when voltage is applied. At load, there's just no way that this can be transferred to the cold plate fast enough to not heat up the components. That's normal. However, as coolant has a higher specific heat, it heats up more slowly and serves as a heat trap. It also cools down more slowly. The more liquid in the loop, the longer it takes - and that's one of the advantages of a custom loop. An AIO will heat up quicker because there's just a lot less coolant in the system (oh, and the radiator is usually aluminum which isn't as good a conductor of heat as the copper radiators in the HydroX series). With all of this said, the coolant temp is the baseline temp for your components - they won't get cooler than the coolant - and your heat should be viewed as a delta from the coolant temp. A 70C proc temp with a coolant temp of 25C is vastly different from a 70C proc temp with a coolant temp of 50C. So - base your speeds on what it is that is being cooled. With the loop (as with an AIO), the thing you need to cool is the coolant - so that should be your control. Also, take note of the intake temps for the radiator. That'll impact your overall coolant temperatures quite a bit.

And define what you mean by "run hot".

 

4) See above - a bit. But yes, this is a good plan. The case fans need to control interior temps, especially if your radiator flow is exhaust. If its intake, it's less important if both your GPU and CPU are cooled by your HydroX loop. And if your radiator is configured as exhaust, be especially cognizant of the intake temperature for the radiator's fans - that will have a direct impact on the coolant temperature, which will have an impact on your component temperatures. It's all interrelated.

 

5) The CoPro can only run the hardware profile based on the temperature sensors on that very same CoPro. If the temp sensors are on a different CoPro, that will require the software to run. CoPros do not "talk" to each other nor do they directly read the values of the various system sensors that iCUE can. If the fan/pump curve is based on ANY temperature not on the CoPro itself, it'll require the software to apply it.

 

Another tip: Set the fan header for the pump to 4-pin in iCUE, not Auto.

Link to comment
Share on other sites

I ran a few tests. See the attached screenshot. For all but one of the scenarios I just ran the 3 intake and 6 exhaust/radiator fans at 1200 rpm then jacked them up a little to 1700 for the last test. Otherwise, the only thing I changed was pump speed to demonstrate that the lower pump speeds really affected the difference between my coolant temp and GPU temp. I assume the idling coolant temp increased throughout my tests because I wasn't giving the system a ton of time for the coolant to cool significantly between test runs.

 

3.) By run hot I Just mean 50-60 C or so difference between the coolant temp and the GPU temp with the pump at lower flows (using default Hydro X settings). Agreed though, I'll base the pump and fan speeds off of the coolant temperature. Because my coolant never changes more than 3 degrees between idling and load (does this also seem lower than expected to you?), I'll have to set a custom curve with baseline at 2500 rpm and going up 500 rpm every half degree of the coolant temp. Based on your link outlining the default settings for the pump, it looks like my coolant never gets hot enough for the pump to realize it needs to run faster... and that's the problem I was having. Still though, there's about a 20 degree difference between coolant temp and GPU temp at load... and from what I've read that is pretty crappy. Agree?

 

4) Yes I agree. Running hot air over a radiator doesn't make much sense.

 

5) I'll change the layout between the commander pros. Just ordered a H75 H105 Y-header cable off eBay for a few bucks and will Y two of the fans together to create space on the same CoPro. And, I'll change the pin to 4 instead of auto, thanks.

249328199_ScreenShot2019-07-18at3_06_58PM.thumb.png.c8fdb82f051d36e1dd6d5ab39a44fc5d.png

Edited by Alex B
Link to comment
Share on other sites

3) Depends on the GPU and how you are creating the load. If it's a stress tool, that's not bad - they exist to push things to the limit and you are still well under thermal limit. If it's a game - then I'd expect less. From what I hear, the 2080 Ti puts out monster heat.

 

Good to see that you are basing results on data and tests! I've seen the same behavior, which is why my minimum pump speed is at 3100 RPM. It may also depend on how much you have in the loop - I have 2 GPUs and 2 radiators plus the CPU. That does create additional restriction - that's my hypothesis, at least.

Link to comment
Share on other sites

Hi! I pretty much have the same setup as you, what is the vertical GPU bracket that you used that allowed the installation of fans on the bottom? I got the Lian Li one but that blocks the fans on the bottom for me to use as intakes.
Link to comment
Share on other sites

temp and the GPU temp with the pump at lower flows (using default Hydro X settings). Agreed though, I'll base the pump and fan speeds off of the coolant temperature. Because my coolant never changes more than 3 degrees between idling and load (does this also seem lower than expected to you?), I'll have to set a custom curve with baseline at 2500 rpm and going up 500 rpm every half degree of the coolant temp. Based on your link outlining the default settings for the pump, it looks like my coolant never gets hot enough for the pump to realize it needs to run faster... and that's the problem I was having. Still though, there's about a 20 degree difference between coolant temp and GPU temp at load... and from what I've read that is pretty crappy. Agree?

 

That is too hot for a GPU to coolant differential. These damn 2080 Ti are the worst I have ever used in this regard, but you still should looking at a GPU temp +12-20 (at most) over the coolant. +50C means you are in the 70-80s and that should never happen on water. It seems like you might have a contact issue on the GPU.

Link to comment
Share on other sites

That is too hot for a GPU to coolant differential. These damn 2080 Ti are the worst I have ever used in this regard, but you still should looking at a GPU temp +12-20 (at most) over the coolant. +50C means you are in the 70-80s and that should never happen on water. It seems like you might have a contact issue on the GPU.

 

The +50C was at low flow rates using the default pump curve in iCUE. It seems that these curves are about as useful as the default fan curves (Quiet/Balanced/Performance). He gets about 20C with 3000 RPM on the pump.

Link to comment
Share on other sites

Ok, then don’t go ripping things apart yet. Sounds like things were in a no flow state. Funny enough, I just finished a 011 build with Hydro X stuff. I’ll have to try and find my minimum possible flow speed, although I have a lot of unorthodox twists and turns. I like how you handled the pump re-entry. I fiddled with for a while and may do so again. I was one 90 short and had to drop it straight into the top.
Link to comment
Share on other sites

Hi! I pretty much have the same setup as you, what is the vertical GPU bracket that you used that allowed the installation of fans on the bottom? I got the Lian Li one but that blocks the fans on the bottom for me to use as intakes.

 

Hey -- thought I had an alert in place for replies not this thread... I guess not.

 

I used the cooler master option. You will DEFINITELY not have enough space for fans PLUS radiator, FYI. To minimize/prevent sagging, I put the rubber block that came with the case (designed to support a 2nd PSU) and put it below the cooler master bracket near the middle of the case. It does fine on its own, but just mentioning it in case that interests you.

Link to comment
Share on other sites

Ok, then don’t go ripping things apart yet. Sounds like things were in a no flow state. Funny enough, I just finished a 011 build with Hydro X stuff. I’ll have to try and find my minimum possible flow speed, although I have a lot of unorthodox twists and turns. I like how you handled the pump re-entry. I fiddled with for a while and may do so again. I was one 90 short and had to drop it straight into the top.

 

Yeah I’ve figured it all out now. Problem all along was not realizing pump flow was so low on the hydro x setting, which was also why it was taking so long for my coolant temp to change.

 

Here’s a better pic of how I handled two rads in the case and the re-entry. Higher quality pic of the front too. It would be nice if you could buy single packs of fittings/adapters. Would also be nice if Corsair made some of the more exotic pieces that bitspower offered (T-piece, white ball valve, white stop plugs, etc), but I get why they don’t (yet?!).

1601EE56-0947-480A-862D-10E30322372F.thumb.jpeg.f4eceb52344c20a322979c2cf31a4715.jpeg

AA6CFC9C-BA8E-4FD3-B462-930C8390EC52.thumb.jpeg.062c177fdb02c09cd5665f7889daa500.jpeg

Link to comment
Share on other sites

  • Corsair Employee

Hi Alex,

I'm providing some input from our side as well.

Your description with high GPU temperatures did raise some concern and discrepancy with our findings so we did a few verification runs on our end.

 

Background:

Fin stack used on GPU and CPU water block works well even with very low flow rates. It's also one of the reasons we've configured Hydro X pump profiles in a way they are implemented in iCUE.

We've ruled out bad contact between GPU water block and GPU. It would be reflected in overall high GPU temperatures and not react so well with higher coolant flow rates.

 

System setup:

  1. 2080 Ti (XG7 water block)
  2. ASUS B450 + 2700X (XC7 water block)
  3. XD5 pump unit
  4. XR5 360 Radiator + ML120 fans

 

Verification with varying flow rate:

OW5rXQz.png

Delta between coolant and GPU in range of 10 – 14°C which is also confirmed with our CFD analysis.

Above 1LPM (0.25 GPM) the water block shows almost no gains on coolant/GPU delta temperatures.

 

While performing the coolant/GPU temp testing by varying flow rate we did had one single run where the GPU temperature readings reached similar values as with your original findings but we were not able to replicate it.

High GPU temperature readings went away after disassembly of cooling system and making a new one to perform validation with Hydro X pump profile (below).

 

Verification with pump controlled by Hydro X pump profile:

dA0Fw8l.png

Delta between coolant and GPU stays within 11°C with pump speed below 2k RPM.

 

Superposition benchmark at 'low pump speed' and full speed. Difference within expectations.

wZZytmA.png

 

FLIR images.

Pump at full speed:

uHV0RlH.png

 

In case of GPU overheating, here we should see a massive difference compared to the full pump speed flir image.

Pump at 1900RPM:

H10P3yT.png

 

Currently, we suspect that there comes to some kind of irregularities with our software and GPU temperature readings as they are accessing the same SDK. Could be related to 2080 Ti as it's the common point with your particular system.

At this point, I would suggest to replug and reinstall your iCUE and try if you'll be able to replicate high GPU temp readings.

We will also continue with our efforts to replicate and resolve the issue.

 

It's safe to use Hydro X pump profiles with a water cooled 2080 Ti. There is no need to manually adjust pump settings unless you would like to run a different performance/noise curve.

Edited by Corsair GregX
Added FLIR images and
Link to comment
Share on other sites

Thanks for looking into it so seriously.

 

I’ve tried reinstalling icue and forced updates on all drivers. My commander pros are daisy chained and I recently switched the hydro x stuff (temp control, pumps, rad fans) to the opposite CoPro to see if that did anything and nothing changed. In doing that I basically unplugged all the the connections involving the hydro x loop. I didn’t unplug any of the fans, any of the power supply, or any of the USB connections. Do you think I should? I also did not unplug any of the daisy chained CPU -> GPU -> pump LEDs... but I suppose I could do that and/or change their order if you recommend it.

 

Approximately which RPMs correlate to liters or gallons per minute? It would be nice to know so I can compare apples to apples. Thanks again.

Link to comment
Share on other sites

  • Corsair Employee

We definitely want to assure you that our hardware and software provides flawless out of the box operation. No tweaking should be necessary unles the user wants to run it differently.

 

Could you please tell me which version of iCUE are you using?

If possible, I would recommend to try it out with the most basic setup.

1 CoPro with daisy chained pump -> GPU -> CPU LEDs. Temp probe to temp#1, pump 4-pin to fan#6.

Other five fan ports on CoPro used for fans (5 + 1 Y splitter).

 

I'll update with approximate RPMs to achieve certain flow rates for your setup. Will run the calculations in the office tomorrow.

Edited by Corsair GregX
Flow rates update.
Link to comment
Share on other sites

Hi — sorry for the late reply. Was out of town. I’m running the latest version of icue, whatever that may be... I can check it when I get back home from work. Doing things w the most basic setup doesn’t seem to make a difference. Maybe it’s a wonky temp sensor? Sort of running out of other ideas. Maybe the GPU contact to the block is sufficient but “not great”. The GPU never gets above 55c at least...

 

Thanks again.

Link to comment
Share on other sites

×
×
  • Create New...