PDA

View Full Version : G73SW GPU fan/HWiNFO64 issues



TheeSheriff
12-01-2012, 04:18 AM
Since opening my G73 for a DIY repair on a bad DC jack pin, my GPU-side fan now ramps up to a very loud level, even when sitting idle. It has been barely audible for over a year, unless gaming, and even then, it's only what I would call "loud" for several minutes at a time. For the last two days, it randomly go's from loud to quiet, with no apparent regard to how hard the GPU is actually working.

I have been sifting through these forums and find my issue is not unique. From this site, I learned to install HWiNFO64 as part of the diagnostic process. That in itself proved a challenge and the issues I've had with that program may be a symptom to my fan issues.

Upon install of HWiNFO64, only the CPU fan was shown under the "ASUS G73 EC" heading. Try as I may, I could not get the GPU fan to show up.

By chance, I started the game, Eve online, and within several minutes, the GPU fan speed (5000+) was suddenly shown in HWiNFO64. Once the fan speed was recognized in HWiNFO64, the GPU fan began to operate "Appropriately" and throttle up and down as it had before. As soon as I left the game, the "Current" fan speed would drop to "0", and after several minutes, the fan would again get loud. I repeated these results several times, re-starting between.

Loading WOT produced the same results. I could surf or do anything else and the fan would go from loud (5000+ RPM) to quiet on random intervals. When I loaded WOT, HWiNFO64 again suddenly recognized the GPU fan and it immediately began throttling the fan temp appropriate. Only difference is while typing this and doing other things, I have been out of WOT for over 20 minutes and the fan is still quite at HWiNFO64"s reported 2200 RPM.

So why does HWiNFO64 only see my GPU fan when I start a game? What might be causing the GPU fan issue. Thanks.

dstrakele
12-01-2012, 01:17 PM
How does the temperature of the GPU compare to the various fan speeds? In other words, do you see high GPU temperatures when the fan is running at 5000+ RPM? What's happening with the CPU Fan and its temperatures?

I suspect your DIY repair of the DC jack may have damaged the GPU Sensor.

TheeSheriff
12-01-2012, 08:21 PM
No, as I said, there seems to be no connection to fan speed and GPU temp. Here's this mornings example....

Start computer. After 3-4 idle minutes, the GPU fan starts to ramp up and hits 5000 + RPM (based on sound). I opened HWiNFO64 and notice, like before, the GPU fan is not listed. The GPU fan stayed constant for the next 20 minutes while I surf the net (It did not do it this morning, but this is about where the fan "might" slow to barely audible for 5-10 minutes, before ramping back up again on unpredictable cycles) WIth the fan still at 5000+, the air blowing from the GPU side is COLD. Air movement on the CPU side is barely detectable (I have to wet my finger).

CPU cores at 36-40 and GPU at 34.

I start Eve online while monitoring HWiNFO, and after several minutes, the GPU fan icon appears. GPU fan at 5128 RPM, but it immediately starts to drop. After 5 minutes, CPU is 59 and GPU is 50. GPU fan has dropped below 3000 RPM and continues to drop (Once at this point, my experience has been the fan will continue to throttle normally).

I exit the game, and within several minutes HWiNFO reports the "Current" GPU fan speed as 0. CPU/GPU are now cooler than they were, yet the fan is throttling back up to 5000+, were it will usually stay until I reboot. I exit and re-start HWiNFO, and the GPU fan icon is gone again. I have repeated this a dozen times with the same results, with the exception of last night when the fan continued to work correctly for more than 20 minutes after exiting WOT , until I shut down.

I agree with you and the odds it was something I did are to great to ignore... the sensor making the most sense. What I can't explain is why my GPU fan does not show up in HWiNFO until I start a game, and why the fan seems to work perfectly as soon as it is detected by HWiNFO. That almost seems more like a software/driver issue?

dstrakele
12-01-2012, 10:29 PM
The fan speeds are controlled by the thermal profile in the BIOS, so it's unlikely the behavior you report is driver-related. You could confirm this by entering the BIOS (so no OS is loaded) and observing GPU fan speed behavior - just by listening and feeling air flow and temperature, since you won't have any HWINFO64 data.

I'm guessing the GPU fan shows up in HWINFO64 and runs normally once a game is launched may be due to a higher voltage overcoming resistance or thermal expansion completing a contact to the Sensor. As the GPU cools, the contact is lost or becomes erratic, causing abnormal fan speed behavior. I'm also guessing that in the absence of sensor data, the fan gets ramped up to full speed.

TheeSheriff
12-02-2012, 05:11 AM
dstakele,

What you said makes perfect sense to me, but this afternoon I was prepared to disagree because I could not duplicate my results unless HWINFO was running, and believed there to be some odd link. Since then, I have managed to achieve the same results with the game alone, so your theory looked promising. After several tests, the sensor seems to get "lost" when the GPU drops to 39.

What still confused me about your theory, why would the fan run slow (Normal) for the first 3-5 minutes after start up when everything is cold?

Now for a new turd in the punchbowl. With my fan running at 100%, I did a re-start to BIOS. The fan went silent within a minute. I went back to windows, expecting the fan to ramp up at any second, but it did not. I went into HWINFO and saw the GPU fan listed, without a game running, for the first time. RPM at 2200 RPM. GPU temp 40. It has been working correctly for the last 25 minutes. I have only done the BIOS check once, so I can't yet say it's repeatable. I can't help but think if the temp drops another degree or two, the fan will max again. The house is a bit warmer now, perhaps just enough?

More testing :(

Edit: Oh, and thanks for your help,
Brad

dstrakele
12-02-2012, 06:18 AM
Well, that sounds like promising news! You might try "Load Optimized Defaults" in the BIOS if the problem should reappear.

TheeSheriff
12-02-2012, 07:15 AM
Thanks, I saw that advice in other threads dealing with this issue and have done so several times without success. It's been several hours now and the fan is still quiet at 2750 RPM. GPU has not dropped below 43 during that time, so that might be enough heat to keep the sensor working, based on your theory.

The fan speed reduction when I entered BIOS is still confusing and I will try to duplicate that in the morning.

Do you have any info on the location of the GPU sensor? I'm curious since my first thought is to question I caused any damage, even though the odds would seem otherwise (I agree it's too coincidental the problem started after my repair). Not to sound arrogant.... I know #((%* happens, but I have built my own comp's for over a decade, including a rebuild of an old Dell lappy, so I'm not new to this, but I would like to see this sensor for myself to possibly gauge the likely hood I actually damaged it.

TheeSheriff
12-02-2012, 07:49 PM
UPDATE

I successfully repeated the BIOS test 3 times in a row, but if my GPU temps are to be believed, it still appears to be a heat issue.

All three tests had nearly identical results. Several minutes after entering the OS, the fan gos to max. HWINFO reports the GPU at 35. Re-start and enter BIOS. Fan slowly throttles down over several minutes to nearly inaudible. Enter the OS and open HWINFO. GPU reports 39 and falling. As the temp heads back down to 35, the fan is ramping back up. If HWINFO displays the fan RPM before I conclude the test, it will usually be around 4400 at this point. If you wait, the RPM will eventually read "0" (Loss of sensor?) and the fan will get just a bit louder.

So it would appear the BIOS, or the restarting process, or both is heating the GPU above that magic 38-39 threshold. Or, at least HWINFO thinks it is.

Any other thoughts?

dstrakele
12-02-2012, 11:45 PM
It I were in your situation, and looking at a possible motherboard replacement, I would probably attempt to update the BIOS (or backrev it if I was already running the latest version) as a last resort. The idea being a power surge or static electric shock has somehow corrupted the BIOS, affecting the thermal profile. By flashing the BIOS, this would hopefully be restored.

G73JH laptops had an issue where a BIOS update corrupted the portion of the BIOS that controls the keyboard backlights (see http://forum.notebookreview.com/asus-gaming-notebook-forum/513264-keyboard-light-fix-gary-key.html), so it is possible that BIOS corruption can affect a single feature of the laptop.

If you are currently running the latest BIOS version, see http://rog.asus.com/forum/showthread.php?24602-G75VW-AHCI-stopped-working&p=176342&viewfull=1#post176342 for information on how to backrev the BIOS. In the worst case scenario, the BIOS flash will brick your laptop and you'll have to send it to ASUS anyway.

This is another Wild Ass Guess on my part. You may wish to wait until some more knowledgeable folks weigh in on your issue.

TheeSheriff
12-05-2012, 01:00 AM
dstrakele,

Thank you for your help. At this point I'm not inclined to do anything that may make things worse. The only time the fan noise is an issue is when I turn on the comp and go strait to surfing, or some other low GPU use task. For anything else, it doesn't take long to warm the GPU above 37-38; the point at which things work normally. 85% of the time comp is on, if I'm not actively gaming, I'm gaming in the background.

I would still be curious to know the location of the GPU sensor(s) on this MB, if that info becomes available. I was surprised I could find nothing on this subject online, even though others have asked the exact same question.