I have recently assembled a new desktop, and after a while, I started to experience VPU recover issues in a certain game. Let me start with my computer specs first: Asus P5Q Deluxe motherboard, Kingston Hyperx 2x 1 gb RAM 1066 MHz (KHX8500D2/1G), Intel Core2 Duo E8500 CPU, and a HIS HD4850 IceQ4, and all of these backed by a OCZ 500W Modxstream PSU (doesn’t beat my previous Enermax Liberty, but I can’t be bothered with removing it from my previous desktop now) in a Enermax Chakra case.
Before I start, I must say I am not completely an expert on the subject though I’m quite experienced with such issues. So, I might neglect some aspects when writing this. While most of these information here is already available on the web, it is mostly scattered among pages and they might be hard to reach with improper keywords. But still, let’s get started:
So, what is VPU recover? It is a feature in ATI software that let’s the GPU (graphics processing unit) to restart itself when it faces a problem and stops responding that would otherwise freeze the computer if it didn’t restart. So, if you don’t have this selected in your Catalyst Control Center settings, then your PC is likely to freeze with a blank screen. First of all, this isn’t an error itself; rather the outcome of an error that has occured. VPU Recover let’s you continue your work without causing you lose the work you are working on, or whatever.
Then, what causes the GPU to stop responding? There might be a couple of reasons. The generic reply you will get from ATI/AMD is the following:
Based on our experience, lockups and hangs issues are usually due to improper
system configuration, driver issues, faulty hardware, or a known issue between
a given application, driver and hardware in use.
Please click the link below for more information:
Before going to other possible reasons, the following is a good checklist:
- Update your drivers. To update, you should uninstall them first. After uninstalling, make sure you clean everything else with Driver Sweeper. If you are unable to remove the remaining files from your Windows, then start Windows at Safe Mode and run Driver Sweeper. Once you are sure its all clean, you can reinstall your new drivers.
- Overheating: Overheating of the GPU itself and the RAMs on the GPU can cause the GPU to stop responding. To see if this is the problem, monitor your GPU and RAM temperature with GPU-Z. This little software will help you monitor the maximum temperature your graphic card has reached and also can log that. In my experience, graphic cards will be ok to run at 90 celcius degrees. With GPU-Z running, you can start playing your game and go back to GPU-Z if and when GPU recover occurs. Alternatively, you can also put your GPU under stress using software made for this. While running 3DMark can help, I suggest you try Furmark and Video Card Stability Test. If it turns out that overheating is the problem, I suggest you don’t do any overclocking, if you are already doing, and arrange your case in a way to maximise air flow with proper exhaust and intake coolers. If your graphics card cooler isn’t that good, you can give better ones a try. I suggest you have a look at Arctic Cooling GPU coolers, which provide good performance at a relatively cheap price. Do not forget your RAM modules on your GPU.
- PSU (power supply) failing to provide enough power for GPU: Old computer setups didn’t really need a lot of power. However, now a modern computer setup will require at least 400 W. And that is also not enough. There are a lot of cheap PSUs out there that will boast hundreds of watts of power but will just not provide that. I suggest you look for known brands like Enermax, OCZ, Zalman, Thermaltake, Tagan, and so on while choosing your PSU. Additionally, its not only the power they provide, but also the amount of current on each rail. In other words, a GPU might require a certain amount of amperes but the PSU might be failing to provide that. Under stress, GPUs will start using more power. So, if you are suspicious of your PSU, then borrow a good PSU if you can and try with that. If you can’t (and if you have already done what else I say here), consider getting a good PSU. Yes, a good PSU is expensive, but in most cases, its a one time investment and its required to keep your computer safe.
These are the two well-known issues that might cause you to get VPU recover. The others that helped me solve my issue might not be the case with you, but it won’t hurt if you try.
- RAMs (not your GPU RAMs this time): I was also surprised to see RAMs as components of the computer that can cause this issue. But what I read made me convinced. Basically, RAMs (sort of random access memory) are where your computer writes and reads data on the fly, because they are real fast, they are used instead of harddrives. How does this relate to your GPU? Well, your GPU drivers and games can be in RAM when an error occurs, resulting in a failure to read. To make sure your RAMs are in good condition, you can try Memtest86. Microsoft also has a memory test software called Memory Diagnostic. You will need to burn these ISO images to a blank cd and boot your computer with these to start testing. Let the test run over and over. If it is your RAMs, then you can remove one of the sticks and test them one by one to find the faulty one. Remember that RAMs these days find with lifetime warranty.
- RAM compatibility: Does your motherboard like your RAMs? In my case, it was a no, because there were a couple of BIOS updates that were increasing compatibility with a lot of RAMs. Fortunately, my RAMs were in the qualified vendor list – RAMs that are known to run with no problems with this motherboard. This doesn’t mean others are not supported. It is just a list of RAMs that will work for sure. So, check BIOS version and look if there are any new updates relevant to RAM compatibility or anything else that is similar. Please note that, updating BIOS can be risky. Though it is a very easy thing to do, you might want to consult online forums or your motherboard manufacturer if you aren’t comfortable about it.
- RAM timings: Some people report that running RAMs at wrong/unintended timings was the problem for them. So it was in my case: My RAMs were running at 800 MHz though they should be running at 1066 MHz. DDR2 has a limit of 800 MHz and over that speed is only acquired through overclocking. Some RAMs are produced to be able to run at that speeds and timings safely. So, check your BIOS if your RAMs run at desired speeds. Some motherboards will set the speeds and voltages automatically. Mine didn’t, so I had to set my timings to 5-5-5-15 at 2.2 Volts. These will change from RAM to RAM, so make sure you check your RAM manufacturer site for correct timings and voltages.
- USB Legacy Support: You will find this setting in your BIOS. Disable it and see if it works for you.
- Turn off Windows Page file, which is the virtual memory stored on your harddrive.
If all of these don’t help you, you should test your graphics card in another system. If not, contact your GPU manufacturer and let them know about this, and maybe RMA it.
I hope these were helpful to you.