EVGA brand 3080 GPU fan percentage incorrect

Hi, I have an EVGA XC3 3080 (NVidia) GPU and hwinfo correctly reports the fan RPM, but incorrectly (as compared to EVGA's own Precision X1 utility) reports the percentage. My fans show a max RPM of over 3,000 and from a post I made on EVGA's forums, it would appear that most of their fans are 3,000 RPM maximum on the new 3080 line.

I have attached two screenshots, taken at the same time and then cropped and highlighted to show what each program is reporting.

Here's a link to a reply from an EVGA employee confirming the fans running at or around 3,000 RPM: https://forums.evga.com/EVGA-Precision-X1-11011-for-3020-Series-ONLY-m3120459.aspx#3120629

Please let me know if there's anything else I can provide. I generated the report file, but only want to upload if necessary, as I will have to go through it and scrub out things like serial numbers.

Thanks for your time.
 

Attachments

  • hwinfo.png
    hwinfo.png
    32.2 KB · Views: 17
  • px1.png
    px1.png
    56.8 KB · Views: 16
This is quite strange as HWiNFO retrieves the GPU fan speed % via standard NVIDIA interface. So perhaps EVGA is using some additional customization.
Please try the latest HWiNFO Beta build v6.33-4280. That has additional support of iCX on RTX 3xxx series, so you might see an additional sensor if present.
 
I'm sorry, I meant to mention that in my post. I'm using the latest beta build of hwinfo (4280). I did find some other posts on guru3d talking about what I think is the same issue? Maybe the post will make more sense to you: https://forums.guru3d.com/threads/rtss-6-7-0-beta-1.412822/page-129#post-5833980

Is it possible the percentage hwinfo is reporting is a "fake" percentage, and the EVGA software knows this and is silently correcting it from "fake" to "real"? That is assuming that person posting knows what they're talking about, I don't think they work for any GPU vendor.

Thanks for taking the time to read my post and reply, I appreciate it.
 
I wouldn't say that the percentage is fake, it's a real value. But as the Alex's post above says - NVIDIA might not know about ODM-specific fan properties, so the fan might be running at a different (min/max) scale than NVIDIA thinks it is.
Let me check this with EVGA and come back to you once I have a clarification.
 
The third fan also doesn't show up in hwinfo, but it doesn't show up in any other programs either, including MSI Afterburner. EVGA might be doing something unusual there, too.
 
I'm not so sure. I don't have my own tachometer, but I set the fans like this (see screenshot of EVGA's Precision X1 software). I popped the panel off my case and from what I can tell they're each definitely running at a different speed, going from slowest to fastest. I used the highly scientific method of "gently touch the spinning fan with my finger and listen to what pitch sound it made." I think that means each fan is controllable separately, at least by EVGA's own software.

Edit: I can also set each fan to 0% and it will stop independently of the other 2.
 

Attachments

  • onlyfans.png
    onlyfans.png
    95.1 KB · Views: 7
Could you please attach a Debug File when the 3rd fan is being reported at a different speed than the others? In the previous file it was most likely same.
Also, have you checked the iCX sensor values in HWiNFO?
 
Last edited:
My card isn't an EVGA FTW model, so I only have a single temp sensor, the one built into the GPU. So I can't help with those additional temp sensor points. I would like an FTW card, but alas. I'm lucky to have one at all.

Here are two debug files. The first I set the fans to 0/0/50% and the second I set them to 0/0/100%, which resulted in speeds of approx 1600 and 3300 respectively. I'm sure many other things changed automatically between runs, but that was the only setting I manually changed. Hopefully that can help isolate the 3rd fan.
 

Attachments

Thanks. I don't see an issue here, you should be able to see the 3rd fan under the "GPU [#0]: NVIDIA GeForce RTX 3080: EVGA iCX" sensor.
 
Ah, I wasn't looking at that other iCX section. Completely missed it. I apologize.

I did notice that the iCX section fan does appear to get the percentage right, though it's presented as a raw number. It shows 23.000, which matches the 23% from Precision X1, while the 2 fans in the other section still read 46.0% at the same speed.
 

Attachments

  • icx.png
    icx.png
    26.8 KB · Views: 7
I will adjust this to not show the decimal digits for percentage, but regarding the first two fans still waiting for response from EVGA.
 
This different fan % is apparently a problem in NVIDIA drivers and will be fixed in the next release.
 
Ah, so not a bug with hwinfo at all. I appreciate you looking into it, and apologize for taking up your time with something that was out of your control.
 
Sorry to keep bumping this thread, but NVidia updated the post saying that the new driver is how things are going to be done, and it's the apps (Precision X1 and Afterburner) that need updating. So it's probably safe to say that however the readings are being presented in driver version 457.09 is the "right" way going forward.

Hopefully it's easy for you to implement. If you need any further debug data or information from me, please just ask!
 
Thanks for the update. Other tools included patches to fix reporting of fan speed based on the range, so I guess this is what's causing troubles now and these patches should be removed.
Since HWiNFO didn't include such patch (as I was aware that this is going to be fixed in drivers), and since HWiNFO doesn't perform GPU fan control, I assume it should be OK now. Can you confirm this or if not, what's the actual situation/problem here?
 
With HWiNFO v6.34-4300 (and the new version of PX1) here's what I'm seeing:

RPMs are correct for the two main fans and the iCX fan.
Percentage value is incorrect for the main two fans.
Percentage value is correct for the iCX fan, but the number is formatted as a raw integer without the percent sign.

Let me know if you need more info, and thank you again for the time you've spent looking into this.
 

Attachments

  • fans.png
    fans.png
    738.4 KB · Views: 6
Back
Top