speedfan aversion?

[ermi]

HW Monitor can actually control fans? May you explain how?

Only HW Monitor Pro, which costs money.

Speedfan is not easy/intuitive to setup, so maybe that's why some people give up.

[Modo]

I concur on the "not intuitive" part. But the biggest problem is that it won't read the temperatures of many GPUs. This is an issue if you're trying to have an inaudible rig that only ramps the speeds up for games (i.e. based on the GPU temperature).

[quest_for_silence]

If you click on manual > 60% the fan will run at 60%..simple. It only work for 4-pins (pwm) fans indeed

Well, I think I understand why you use it rarely (and why you prefer a rheobus).

If the only feature needed is to give a fixed speed to a (PWM) CPU fan, the smaller (and cheaper I mean) Argus Monitor might work even better (and probably cheaper).

Personally I wouldn't ever talk about actual "fan control": speaking in metaphors, it would be like saying that DOS and Linux do about the same job just because they both use a CLI (it's just a light joke, I'm not mocking you).

[quest_for_silence]

This is an issue if you're trying to have an inaudible rig that only ramps the speeds up for games (i.e. based on the GPU temperature).

How may you address this issue? With a rheobus? With another software? And is this issue strictly Radeon-related? Can't help you the CPU temp?
However, SpeedFan may control also a T-Balancer.

[Modo]

I haven't seen other software that performs the speed control tricks SpeedFan can, so I'm still using it. I already have set it up to use the CPU and hard disk temperatures for fan speed control, but that misses an important variable. The GPU tends to get the most load in games, while the CPU can remain close to the idle level. I have had situations where the fans didn't ramp up at all, because a game would not stress the CPU past what a Flash movie would, while using the GPU a lot. And since the CPU cooling on my rig has lots of headroom, I really don't need the fans to go faster when I'm watching YouTube or ripping CDs.

Yes, as far as I know, SpeedFan doesn't see the Radeon temperature sensors. Support for NVidia GPUs was added last year, I think.

[tim851]

This is an issue if you're trying to have an inaudible rig that only ramps the speeds up for games (i.e. based on the GPU temperature).

How may you address this issue? With a rheobus? With another software? And is this issue strictly Radeon-related? Can't help you the CPU temp?

SpeedFan can read the temp sensors of nVidia cards, but not AMD cards. Apparently the developer has not received the appropriate documentation from AMD.

I contemplated getting a Radeon 58xx before settling on a GeForce. I had already ordered an adapter cable that let's me connect a regular case fan to the (smaller) graphics card fan header. I would have then used the software MSI Afterburner to implement a smart speed control for that fan.

[ermi]

SpeedFan can read the temp sensors of nVidia cards, but not AMD cards.

Wow, this is disappointing. Is it true for all AMD cards?

Will this change anytime soon? (edit: apparently he's working on it)


If not, is there an alternative to Speedfan for automating fans based on AMD cards' temps?

[dhanson865]

I just set my speed fan based on CPU/chipset/HD/case temps and consider the video card separately but I'm currently using a Geforce 6600.

When I install a AMD HD 5000 series card this week or next I'll probably play around with AMD OverDrive which has fan controls in it. I also use RivaTuner on my 6600, I'll play with the latest version of it to see how it works with the HD 5000 card.

BTW my HD 5000 series card is fanless so I'm not going to be controlling a fan on the card I'll just be curious to see how case air flow in general affects my fanless GPU card.

Assuming nothing works better take note of your Idle temps and your load temps in the program that you use most that gets the video card hot at a fixed fan rate (disable automatic fan rates while testing).

Right now as I'm posting this (which isn't true idle as I have at least 10 instances of firefox running plus a HTML editor and an email client open) my temps are

CPU ~38 (its bouncing around 38 on average)
HD 29
Chipset 29
Case 25
GPU 0 (not reading it apparently or only reads it when it's higher)

Now lets test with say rthdribl.exe running in a window

CPU ~45
HD 28
Chipset 29
Case 25
GPU 0 (definitely not reading my card)

So since nothing else changed I'd just set the fans that I think affect the CPU and Video card to monitor the CPU temp. Then I'd run some tests with RivaTuner graphing the temp of the video card while I played a real game for 15 minutes and alt-tab out to check temps.

Change the fan controls in speedfan and game another 15 mins, check temps again

repeat as much as needed.

Oh btw RivaTuner says my GPU is at 58c now. And if I open rthdribl.exe it slowly rose to 72c before I got bored and closed them both.

I'd especially want to test again this spring or summer as right now my computer is in a room that is several degrees cooler than its worst case scenario. It's about 20c / 67f in here now, worst case is probably if it got to about 100f/38c outside I might let it get up to 78f/25c in the house. An extra 5c ambient can be more than 5c inside the PC case (increased temps lead to increased power consumption which also increases temps it's a feedback loop that can cause a runaway if it gets out of hand so do pay attention to temps if you aren't sure).

I'd want to test again on a warm day to be sure all is good but I sure wouldn't let speedfan not knowing my GPU temp slow me down on making my system work the way I wanted.

I'll use more than one tool to check temps any time I change a major component. But once I'm comfortable with the setup I just leave speedfan running unmonitored for months at a time.

[dhanson865]

heh, I just installed the HD 5570 today and had problems getting Catalyst Control Center to work. I uninstalled RivaTuner (two versions), cleaned up some temp files, uninstalled some old ATI drivers, uninstalled old NVidia drivers, uninstalled and reinstalled the 11.1 driver a few times. Installed some more versions of .NET. Eventually figured out what it took to make CCC happy.

Why do I mention this here? Well before today I couldn't get speedfan 4.4 or above to install. I was stuck on 4.39. I just installed 4.42 flawlessly. Somewhere in my cleaning up old support files and uninstalling unneeded apps I fixed a conflict with the 4.4x installer of speedfan.

[dhanson865]

viewtopic.php?f=8&t=57528&p=532051#p532051 has some other details but I'm on a different HS and fan now so my speedfan settings have changed.

The stock AMD heasink had a 70mm fan that 29% in speedfan gave me 690RPM and 85% gave me 2800RPM, I'm not sure what 100% was but it was over 3000 I guess.

The Tranquillo stock fan was PWM and whether it was the motherboard fan control circuit or the fan I couldn't get anything less than 670RPM out of the Tranquillo PWM fan. At 100% speed it was over 1400 RPM. After playing with it I decided the PWM range that made sense was 35% min and 73% max. At 73% the fan noise was noticeable but not as obnoxious as the 1400RPM level. I didn't write down the RPM at 73% but 35% got me that 670 RPM I mentioned.

I took that fan out and put a Yate Loon D12SL-12 3 pin fan in and was able to get speed fan to control the Yate Loon to 0%, and from 20 to 100%. 20% on this fan was around 340 RPM. The fan won't start at 20% so if I allow it to stop it takes a much higher percentage to start. To avoid that lag I decided to run with speedfan set to 20% minimum and 100% maximum. fwiw the Yate Loon at 100% is over 1400 RPM as well but is much quieter than the Tranquillo. In fact I'd rather have the Yate Loon at 1400 RPM than the Tranquillo at 73%.

Also because the Yate Loon is so smooth sounding I've changed my mind about the Delta value % in the speedfan options. I used to think (from lack of experience) that a 120mm fan would need a lower percentage than the 70mm fan. I was running on 3% with the AMD stock HS/Fan, I've moved it up to 5% on the 120mm yate loon and I might be willing to run it at a higher percentage.

I ran 3D mark06 last night and all though the speedfan graph shows me the fan varied from 300 to 700 RPM during the benchmark I never noticed the noise change at 4%. I'm thinking with this HS/Fan combo that even 10% might be reasonable but I'm too impatient to bother running the bechmarks enough times to find out.

The big advantage to a higher percentage Delta is that during boot my system ramps the CPU fan to 100% and when speedfan loads it starts to slow the fan down. A higher Delta % means less time until the system gets down to the under 400RPM range where it'll sit most of the time.

I could use a lower maximum like say 75% to shorten the trip on the way down but it's so smooth sounding I figure I'm better off just leaving the full range there for speed fan to use if I ever play a game that truly heats things up.

Maybe if I ever get the last hard drive out of the case when SSDs are cheaper I'll change my mind and throttle that fan down some more but for now I'm happy with how it works.