«

[Japo]

The topic is not about which manufacturer or model gives better performance, but any other aspect, specially it's stability (driver bugs) that I'm concerned with. But feel free to talk about anything you know or have experienced.

My computer came with an ATI Radeon HD 4670 512 MB GDDR3. I wasn't particularly worried about graphic performance, I don't play the newest games. The computer was a somewhat but not too high-end package two years ago, aimed at the professional (as opposed to gaming or home) market, and the graphic card was one of the things that I didn't opt for improving above the default package (for a supplement they offered a nVidia with 1 GB I think).

The specs are still today more than enough for me. The problem is that the retards at AMD are unable to publish drivers that are acceptably bug-free. I have experienced the following problems myself (I have Windows x64):

- Some very particular Adobe Flash content freezes the system completely and permanently, forcing a hard reset.

- One game might cause the mouse pointer graphic to get corrupted. It might return to normal without rebooting, but it's not clear what might make it do so--some posts I googled advise to move the mouse very fast, heh.

- Enabling and disabling extended desktop screws Windows Aero or something, causing the Windows Explorer to stop responding for a minute, and making the Vista Windows theme unavailable.

How can AMD provide this product with a straight face? Of course when I went to their website to download the latest drivers, which they update every few months, the changelogs talked about adding support for this and that game, and when I try and install the updates, it the basic bugs that prevent Windows from working aren't fixed! It's unfortunate for me that game performance are their priority, but inevitable. The real problem is the quality of their work, that software should never have been released without further bugfixing work.

Now I have un-installed all ATI software, reverted to the latest drivers approved by Windows Update--which of course have bugs too, since AMD hasn't made anything that works completely well--and I don't plan to update these drivers outside this channel.

I have vowed not to buy an ATI card ever again, but I regret very much having an ATI card in my main computer, even if I don't know the situation with nVidia or others; I'm even glad that I don't have AMD processors here or in the netbook. Dunno to me AMD has given the impression of a company overwhelmed by its competition. Currently Intel is very ahead of them in PC processors; they don't have a lot in mobile devices as far as I know; and on the graphic card market they have stiff competition too.

[Eagle of Fire]

I suggest that you try looking for the Omega Drivers, which are created and maintained by a dedicated man who knows his stuff about graphic card drivers.

Most of his work goes around using only drivers which solve problem and overlooking drivers which cause problems. This mean that sometimes he adds the latest drivers while other specific drivers are left behind. This give great stability and better performance for computers which don't aim to be a bomb of the moment, like your computer (and mine).

This might be just what you were looking for.

[KrazeeXXL]

ah yea, the crappy new AMD drivers. :/

A friend and myself (we both use a 3850 btw) also experienced these bluescreens since the 10-11 drivers while using flash and ff. They occured 1-3 times a day.

So I contacted AMD support to tell them...

Got a message some days later and the service guy told me to go to the Sapphire Site (as I'm using a Sapphire card) and download the drivers there...

Ok I did so and downloaded the drivers there. Got some archive (rar, zip whatever) and extracted it. The file somehow looked farmiliar so I did an md5 check of both files. The one of the AMD site and the one from the Sapphire Site. And both of them were identical.

So I contacted the AMD support once more and asked the guy how the same driver, yea even the exact same file should solve my problem... (I also sent the md5 check )

As i didn't want to bother the guy more I also wrote that my only intention from the beginning was that this error could somehow be fixed...

He answered that he'll refer it to the responsible section.

So I checked 10-12 and then the new 11-01. error still there. So I stick with the 10-10 which works flawless. As I also don't play the latest games (would be impossible with my old machine here, too ) - I don't care anymore tbh.

My friend with the same card + problem still tries the latest drivers for some reason. But both of us think that this problem won't be solved and AGP support will be cancelled soon anyway (but I think it already was at least since AMD took over ATI...)

I didn't try for a long time but next to Omega Drivers I recommend the DNA drivers. They've helped me in the past.

btw I'll wait for the AMD Bulldozer. I hope it'll live up to it's name though

[Japo]

I now think it's overheating, that's why I haven't been able to reproduce any of this. This card has always been hot when idle, but oddly enough it's not playing games what caused most problems (but Adobe Flash and extended desktop), so I didn't think about it. The card may be defective, but it's out of warranty now, and besides they wouldn't admit it on the basis of temperature, I already talked to Dell tech support once.

If this card is doing NOTHING it has always been at almost 70º C, but this didn't usually cause too many problems. Enabling extended desktop alone raises it 10º C (WTF?!), and that's what has been causing more consistent problems lately (with Windows Explorer). The problem is that I almost never used extended desktop before I got a big TV recently. Now I have the case open on one side, which makes it around 10º C cooler.

This is a desktop with a huge and very well ventilated case, the CPU is at 35º C. And I don't have any other PCI cards so it's not crowded and the air flow around the card must be as good as it can be. I think I'm going to have to get a new card.

[Eagle of Fire]

Did you try the alternative drivers we mentioned?

Stupidly enough, the fact that drivers are so badly made nowaday might just be why you are experiencing those problems. The fact that everything else in your rig works 110% make me believe there is a good chance it might be the case.

[Japo]

I didn't know about those, but there aren't any ATI drivers for Vista or any x64 OS in the Omega website. Anyway as I understand those alternate packages use a version of the original drivers and only add registry fixes and custom installers and utilities.

After all this time, the fact that the same problem happens with any driver version, that hardware diagnostics are passed with flying colors, and there's no reproducible cause except temperature (which I have to confirm further) makes me think it's a hardware fault.

[Maxor127]

Did the problem start happening after a graphics card update? If so, I'd say it's a driver problem. Otherwise, I agree it sounds like a hardware problem. My Nvidia 8800 was having graphic corruption a few months ago. I put in an old ATI card, and graphics were fine so I knew it was a problem with the card.

[Japo]

When the problem started I tried system restore, and un-installing the drivers and installing old versions, to on avail. I think the problem with the card existed before, but I didn't really detect it until I started using extended desktop because of the new TV. Before I had seen occasional system hangs when playing some flash video, but they may also be because of overheating.

I re-installed the ATI Catalyst Control Center, because without it Windows alone doesn't let me do basic things like choosing the aspect ratio, or cloning the desktop as well as extending it. (Although during the installation I opted out of installing the actual driver, and left the older version taken from Windows update.) The good thing is that the Catalyst shows the fan speed as well as the temperature. Come to think of it, I don't think I've ever seen it above the minimum speed ("0%"). Maybe I could sort the problem out, whatever the cause, if I re-calibrated the fan or set it manually.

I'll monitor the temperature, and when the problem appears again, if the fan speed is at the minimum, I'll re-calibrate it. Though I have to disable the factory preset and accept a disclaimer (heh). I may end up buying a new card all the same, depending on what happens; this one has been nothing but trouble.

[KrazeeXXL]

whoa, 70°C is way too much. Even now with the warm weather my 3850 is at "only" 50 °C.

So this is definitely not a driver problem imo.

At first you should check the fan and the heat sink and clear it off from all the dust. I recommend to do this every quarter of a year. Dust is known to destroy fans. So take the test. Remove the card and spin the fan a bit. It should spin w/o problems for some seconds. Ah well b4 I forget it.

Very important: If you use a vacuum cleaner to remove the dust from fans be sure you jammed them because of the dynamo effect.

Another thing about monitoring the temps and stuff.
I suggest ati tray tools ( I never install ccc it's crap imho ). there you can f.e. setup a systray icon where you can monitor the temperature easily. Within this great tool there's also a monitoring display with graphs in which you can setup all kinds of sources from Clockrate (in Mhz) to GPU-Activity (in %).

if nothing had changed after removing the dust... and the fan spins fine for some seconds...
you should check the heat-transfer paste on the GPU. My guess is that it melted away. If so remove any leavings of the paste and put on some new one. You can buy it in every good computershop for under 5€.

Ofc, you have to remove the fan-cooling-system first which can be tricky if you do it the first time. Be careful when doing so and don't simply rip it off of by force. My old 3650 had lots of screws and @ the beginning it was a bit of pain in the ass to remove this thing.

If you have some bad luck and the fan is broken (also happend with my old 3650)... there are cooling kits for graphic cards out there for something around 20 €. But that should be last solution if nothing else works.

As I said b4, i don't think it's a driver problem. This doesn't make any sense. You should get over 70°C with a full load of 100% and not with 1 to 10%... so check the cooling system of your card b4 you waste your time with other things that don't make any sense regarding this problem

gl and if you have some problems doing the things I mentioned feel free to write me a pm as I don't check in here every day.

Greetz Krazee

ah well I re-read your post and seen the thing with recalibrating the fan speed. Now I remembered that there were cards with this specific problem (yea always happens after I wrote all this stuff above ) - besides all that removing dust and checking the fan, etc... still is important...

In att (ati tray tools) you can setup fan speeds for every temperature you like. Also you can setup to run the fan always at a specific percentage.

Either way b4 I would waste my money on a new card I'll check out this "little" tool if I were you. It has so much to offer. I could write pages about it's "pros"

edit: besides the fan speeds you can also change the PCI-Latency in att. Try a value of 64

[Japo]

Thanks. I had already cleaned the dust. I even unplugged the card completely so I could look at it carefully and remove every bit or dust from the inside. It wasn't really dirty, but now the heatsink is as clean as a mirror. Before with the case open I could see the fan spinning, no noise or vibration, and when I took the card out it span OK by hand.

Right now I have closed the case again, with no activity whatsoever I have around 50º C (ambient 20º). Enabling double desktop and reproducing a video raises it to around 76º and the fan spins at around 15% (over the minimum), but I haven't had the problem again. These 76º C are below the middle of the temperature gauge shown in Catalyst, not in the red or orange, as if it's a safe value.

Anyway I think the problem happened in the past after a long time, for example when I was watching many episodes in a row. But I think the temperature was more or less similar (now the ambient may be slightly cooler than a couple of days ago), although it's now that I'm starting to monitor it more closely.

Tinkering with the card itself would be the last resort for me, besides I loathe spending more time with this. Since the card's working (most of the time) I can't expose myself to being left with no card at all. I would try tweaking the fan before.

At any rate when I had the case open I touched the copper heat sink and it felt really hot, which is good. It would be bad if the circuitry were hot but the heatsink weren't.