View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001712 | SpeedFan | Hardware support | public | 2010-12-20 15:54 | 2011-08-28 11:07 |
| Reporter | norman02us | Assigned To | alfredo | ||
| Priority | normal | Severity | feature | Reproducibility | have not tried |
| Status | resolved | Resolution | fixed | ||
| Platform | PC | OS | Win7 64bit | OS Version | 7600 |
| Fixed in Version | 4.45 beta | ||||
| Summary | 0001712: support for AMD/ATI GPUs | ||||
| Description | Hi Alfredo, I would like to know if support for AMD graphic cards is any closer. Do you have the necessary documentation by now? | ||||
| Tags | No tags attached. | ||||
| Motherboard Model | EP45-UD3P | ||||
| Video Card Model | |||||
|
|
I purchased a video card, but AMD didn't give that much. I still need to find the right person. |
|
|
AMDs first (lame) response was: "maybe you can find the necessary information here: http://developer.amd.com/pages/default.aspx" Since I assume that won't help you, can you please write me an email with the information what exactly you're looking for? |
|
|
I also have problems to get the temperature of my ATI HD 5870 into Speedfan. Im wondering cause a lot of other tools like Everest does show it. I just want to adjust my case fans, if the gpu gets hot. Maybe a workarround would be nice, somthing like to import the temp values from a virtual sensor driver. Plz let me know how i could help you with your great work, thanks a lot for speedfan! I did run your ACPI Explorer, maybe that report will help, should i send it in ? cheers nimbo123 |
|
|
Heck, even OCCT can read GPU temp on my HD4850 with it's internal source, so it cannot be that hard. Please alfedo, ask for any help you need, let's make this work :) I will try that with ADL SDK and post results here. ED: Hey, I managed to do that! It's not that hard at all, alfredo contact me if you'd like to include that in SpeedFan! I'd like that very much! wlw_wl - at - tlen.pl |
|
|
hey can u publish here your results ? i still have problems. Thanks |
|
|
I did it in my own application, using ATI/AMD ADL ASK, which is an API for ATI/AMD video card driver. If alfredo will be kind enough to implement it in SpeedFan, then we all can profit from that :) Until then, no Radeon temps in SpeedFan. Without alfredo's commitment, I would have to simulate a device which reads temps from GPU, that device then can be discovered by SpeedFan and provide it with temperature. If possible at all, it's a wrong way to go. |
|
|
Gentle bump. Upgraded to 6850, no reading of course. |
|
|
Hello all :-) I purchased the hardware a few months ago. I was looking for the best method to read temperatures and I am still trying. I will implement the SDK method, hopefully, in 4.44. If everything goes fine, 4.43 Final should be almost ready. |
|
|
thanks a lot |
|
|
Do you know of any other method? The only one I found was through the driver (ADL SDK), so naturally there is at least one more method - the way the driver gets the information from the video card. I'm not sure if it's worth going that way. Thanks a lot! |
|
|
I purchased some video cards to add support for ATI, but it looks like I purchased some that do not have all the monitoring features I expected. Can you suggest a model (not too expensive, please) that exposes a hardware monitor chip on the I2C BUS? Thank you. |
|
|
In the meantime I can tell you that SpeedFan 4.44 will show ATI video card temperatures :-) |
|
|
Do you want to use I2C on cards that can't be read with ADL or to avoid using ADL? ADL_Overdrive5_Temperature_Get() works from R600 up (HD2000), and since R500 and below are cards from the X1000 series and below, they aren't used much anymore. I tried reading the temp from I2C some time ago on some older card (I think it was my Radeon 9550 Pro), but couldn't get it to work. Guess Ring0-direct-PCI-access-stuff was a bit too much for me back then. I've read a lot about the topic back then, many people were trying to implement it on driver level in Linux, but I haven't encountered any successful attempts. I looked at Steam Hardware Survey (which I think is a comprehensive source) and video cards unsupported by ADL SDK are way way less than 1% of all cards. So maybe leaving them unsupported wouldn't be a great sin ;) Also, ADL lets you set GPU's fan speed easily and safely. As for the cheap card, what models have you bought? I'd say try old 9000 series, you'll find one for almost nothing (couple of Euros here). Of course it's hard to say which will let you talk with the temp sensor on I2C directly, if any at all. The fastest way to check is to scan the bus from within AtiTools, I guess. I'd love to see that in next beta and test, test, test, repeat (if there will be any beta before 4.44). Not knowing your schedule, so I'm just saying... :) All in all, this is probably the other method I was talking about - the way the driver gets the info in the first place. I seriously think the amount of work required to support such a small fraction of graphics cards used nowadays, could be diverted to generally more useful areas, like enhanced fan speed control discussed in another ticket (1746) :) |
|
|
One of my video cards is an Asus EAH5770 CuCore. On that video card the fan is using a 3-wire connector, but I can't read its speed. |
|
|
If the fan speed shows up in Catalyst Control Center on Overdrive tab, then you can read it with ADL SDK easily (ADL_Overdrive5_FanSpeed_Get();). If you mean I2C, well, don't :) |
|
|
As far as I know, none of the ATI 9xxx series had temperature sensors (unless some custom models had it). I believe ATI x800 models started to get temp sensors. jonib |
|
|
Bugger, must have been a different GPU then, sorry for the confusion. |
|
|
*bump* :> 4.44 is here :> |
|
|
I am working on this. 4.44 has already more new features than I planned to add. It is not easy to beta test them all at once :-) |
|
|
Ok, ok, no pressure... BUT JUST DO IT! kidding ofc., you're doing very good job, I'm waiting for GPU temp eagerly, but take your time to do your stuff :) |
|
|
@alfredo: If you're not sure if your AMD graphic cards have monitoring sensors the easiest method to check is the 'sensors' tab of the GPU-Z program from techpowerup: http://www.techpowerup.com/gpuz/ |
|
|
Hello Thomas, thank you for the link. What I actually would like to do is to buy a video card with an "external" hardware monitor chip that can be accessed using the I2C bus. I bought tow with no luck :-( |
|
|
I doubt that such a card exists, since AMD HD series GPUs, even the HD-2xxx seem to have on-chip monitoring of temperature, core voltage, and fan speed PWM control logic (although some cards don’t use the PWM logic). I doubt any manufacturer will add an external I2C chip when it's already in the GPU core. And the AMD ADL SDK allows reading the temperature and maybe other sensors. |
|
|
why not contact msi afterburner creator maybe they can help you much than amd. Maybe they can explain you how they have made the thermal table mode (it's a really good feature in riva tuner the first graphic card overclocking application they made)... maybe ray adams can help you (Ati tray tool maker)... look in guru3d forum there's a section for all that application... I think it can be a good start... |
|
|
Reading the GPU temp and fan RPM is fairly easy and it has been said many times, Alfredo wants to read it thorugh I2C, I guess for backward compatibility reasons. I personally doubt that many people still use ATI cards that don't support those features on driver level. Unwinder in RivaTuner (and consequently in MSI AB I guess) used I2C on nVidia cards, I wrote two or three plugins for RT of which one was for my then owned GPU. The built-in plugin for LM63 used I2C to read temp, fan speed and fan duty cycle. It didn't work even though it had LM63, hence my own plugin. Example board that allowed interfacing through I2C was reference Radeon X800. There was also NVThermalDiode for nVidia cards, but I'm not sure if it used I2C. Using I2C nowadays for AMD cards is pointless IMO, as all the cards provide required information throught driver, giving uniform interface regardless of the actual IC used on the card. Everything is available in RivaTuner SDK which you get along with the RivaTuner itself. It's old stuff however, that's why I so strongly vote for AMD ADL SDK which is a walk in the park, 10 minutes and done. |
|
|
Besides, RivaTuner was open hobby project, MSI AB is commercial and closed and so Unwinder will not provide any info on it's source, I doubt he is even legally allowed to do so. |
|
|
I will add ATI support in SpeedFan 4.45, after 4.44 Final will have been released. |
|
|
Wasn't that supposed to be in 4.44...? Que sera, sera; fingers crossed ;) |
|
|
I have a beta that reads temperature and fan speed from ATI RADEONS. If you want to give it a try, contact me at alfredo [at] almico.com. |
|
|
I just added fan control capabilities on Radeons. It works great! |
|
|
It works! Finally :-) |
|
|
Awesome. Works great on my Asus HD6870. If I set it to 100%, it runs with almost 4000 rpm :) It's almost as loud as a vacuum cleaner. Alfredo, before I take control of the fan speed, Speedfan tells me it's at 15%. If I change it manually, there is no difference in rpm below 30%. The maximum I can bear to listen to is 40%, so that's for gaming. I never understood why GPU manufacturers use fans with such high rpms. @edit Anywhere between 0 and 30% the speed stays at 1375 rpm. |
|
|
Because it must have certain CFM but the diameter of the fan is restricted, so in order for the GPU not to fry itself before the warranty expires, they have to make a headroom (taking radiator fouling into account) in terms of the cooler performance, and the only option left is more RPM... ...or these are the only BLDC motors they can get in China for cheap ;) BTW the fan speed is set through driver API, if the driver won't allow certain settings, there's nothing to be done about it in SpeedFan. It is very likely that the driver doesn't allow slowing the fan too much. On my card "0%" actually means 1060RPM. The VRM is also cooled by the air stream and I bet this is the main reason why they won't let you go passive. The core and memory would last, the VRM would not. |
|
|
I agree that some head room is required, but 4000 rpm seems a little overkill. It starts getting noisy at 40% already and anything above 50% would make me return the card ;-) |
|
|
You are right: the driver has a minimum percentage. I thought it would have returned it when a lower value was entered, but that's not true. Anyway: I think that it makes little difference, because whenever you configure SpeedFan for automatic fan control you need to experiment a little to find the minimum and the maximum values to use. Thank you for testing. I am glad to read that everything is working fine. I will release the beta to the public in a few hours. |
|
|
Mine works in the same way, below 40% it's inaudible, 60% is loud and 75% is a helicopter. It's not an overkill when you OC, the difference between 80% RPM and 90% RPM is exactly 1MHz more on the core 8-) |
|
|
I just added voltage reporting. |
|
|
You could add the GPU load as well, since there is CPU load in SpeedFan and there is such a feature in ADL SDK... :) BTW does this work well with multi-GPU configurations (CrossFire and/or multicore - 5970, 6990)? |
|
|
I will, in the future, but at the moment the CPU LOAD is not integrated with other sensors :-( |
|
|
Why would it be? |
|
|
I mean that the CPU Load is not taken from the TCpuSensor class. I am working to add support for new sensor types to the base TSensor class. Obviously the only thing that slows down this change is the GUI :-) |
|
|
Yeah, there isn't much space to work with. But if you ask me, the init/status window (TMemo) at the top along with the "Minimize" and "Configure" buttons are the first things to say goodbye to... |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2010-12-20 15:54 | norman02us | New Issue | |
| 2010-12-20 15:54 | norman02us | Status | new => assigned |
| 2010-12-20 15:54 | norman02us | Assigned To | => alfredo |
| 2010-12-20 16:14 | alfredo | Note Added: 0005552 | |
| 2010-12-20 16:14 | alfredo | Status | assigned => acknowledged |
| 2010-12-22 00:59 | norman02us | Note Added: 0005606 | |
| 2010-12-22 01:12 | norman02us | Note Edited: 0005606 | |
| 2011-01-21 03:13 | nimbo123 | Note Added: 0005708 | |
| 2011-01-21 03:15 | nimbo123 | Note Edited: 0005708 | |
| 2011-02-04 02:11 | wlw_wl | Note Added: 0005800 | |
| 2011-02-04 02:13 | wlw_wl | Note Edited: 0005800 | |
| 2011-02-04 03:13 | wlw_wl | Note Edited: 0005800 | |
| 2011-02-04 03:38 | nimbo123 | Note Added: 0005802 | |
| 2011-02-04 13:31 | wlw_wl | Note Added: 0005804 | |
| 2011-02-04 13:57 | wlw_wl | Note Edited: 0005804 | |
| 2011-02-16 00:09 | wlw_wl | Note Added: 0005852 | |
| 2011-02-16 00:37 | alfredo | Note Added: 0005853 | |
| 2011-02-16 01:16 | nimbo123 | Note Added: 0005854 | |
| 2011-02-16 02:07 | wlw_wl | Note Added: 0005855 | |
| 2011-02-16 14:39 | alfredo | Note Added: 0005863 | |
| 2011-02-16 15:56 | alfredo | Note Added: 0005864 | |
| 2011-02-17 00:11 | wlw_wl | Note Added: 0005871 | |
| 2011-02-17 00:21 | wlw_wl | Note Edited: 0005871 | |
| 2011-02-17 00:35 | alfredo | Note Added: 0005873 | |
| 2011-02-17 01:11 | wlw_wl | Note Added: 0005875 | |
| 2011-02-17 14:05 | jonib | Note Added: 0005878 | |
| 2011-02-17 19:37 | wlw_wl | Note Added: 0005879 | |
| 2011-04-03 04:07 | wlw_wl | Note Added: 0006012 | |
| 2011-04-03 09:32 | alfredo | Note Added: 0006015 | |
| 2011-04-03 12:33 | wlw_wl | Note Added: 0006019 | |
| 2011-06-20 10:35 | Thomas | Note Added: 0006251 | |
| 2011-06-22 11:17 | alfredo | Note Added: 0006253 | |
| 2011-06-22 11:48 | Thomas | Note Added: 0006254 | |
| 2011-06-22 12:50 | el_sauvageon | Note Added: 0006255 | |
| 2011-06-22 13:08 | wlw_wl | Note Added: 0006257 | |
| 2011-06-22 13:11 | wlw_wl | Note Added: 0006258 | |
| 2011-06-22 14:14 | alfredo | Note Added: 0006259 | |
| 2011-06-22 14:17 | wlw_wl | Note Added: 0006260 | |
| 2011-08-26 14:44 | alfredo | Note Added: 0006415 | |
| 2011-08-26 17:34 | alfredo | Note Added: 0006416 | |
| 2011-08-26 18:05 | wlw_wl | Note Added: 0006417 | |
| 2011-08-26 18:36 | norman02us | Note Added: 0006418 | |
| 2011-08-26 18:39 | norman02us | Note Edited: 0006418 | |
| 2011-08-26 18:41 | wlw_wl | Note Added: 0006419 | |
| 2011-08-26 18:45 | wlw_wl | Note Edited: 0006419 | |
| 2011-08-26 18:51 | norman02us | Note Added: 0006420 | |
| 2011-08-26 19:23 | alfredo | Note Added: 0006421 | |
| 2011-08-26 19:54 | wlw_wl | Note Added: 0006422 | |
| 2011-08-27 12:03 | alfredo | Relationship added | has duplicate 0001845 |
| 2011-08-27 12:04 | alfredo | Note Added: 0006424 | |
| 2011-08-27 12:21 | wlw_wl | Note Added: 0006426 | |
| 2011-08-27 12:25 | wlw_wl | Note Edited: 0006426 | |
| 2011-08-27 12:35 | alfredo | Note Added: 0006427 | |
| 2011-08-27 12:37 | wlw_wl | Note Added: 0006428 | |
| 2011-08-27 12:46 | alfredo | Note Added: 0006429 | |
| 2011-08-27 12:51 | wlw_wl | Note Added: 0006430 | |
| 2011-08-28 11:07 | alfredo | Status | acknowledged => resolved |
| 2011-08-28 11:07 | alfredo | Resolution | open => fixed |
| 2011-08-28 11:07 | alfredo | Fixed in Version | => 4.45 beta |
