Intel HD4600, Is this a driver or hardware issue?

D

Dk975

Gawd
2FA
Joined
Sep 24, 2005
Messages
872
I put together a i3-4330, 8GB Crucial Ballistix 1600, Asus H87M-E, Samsung 120GB SSD and using any of the Intel HD4600 video drivers for Win7 x64, I am getting "Display driver igfx stopped working, and has successfully recovered". I can get this watching a local video, online video or even just browsing or sitting on the desktop. It can happen multiple times within a couple of minutes or 15-30 minutes apart. I ran Furmark, and the driver never crashed.

Also, when this crash occurs, I have GPUZ open, the graphics freq shoots to 1150 (max), gpu power and load spike up, and there is a 5 C increase in temps, whereas graphics memory plunges, and this lasts for 3-5 seconds.

I have run Memtest86+ 6 passes, no errors. System is only using 26-65W, and I have a 300W PSU that is 3 years old. No OC, and everything in BIOS is Auto and BIOS updated.

So, I installed Ubuntu on my data drive (not the SSD), everything seemed okay, so I figured I would install the Intel drivers that they brought out with the updater. Played some local videos using VLC, and had Firefox open, and the system froze up,but apparently I was still able to open things because when the system came back, things I had clicked on when the screen froze were open.

I bought a AMD HD5450 card on the forums and will test that out in Windows. But seriously, is this a driver/software problem or a hardware (PSU, Mobo, bad video on the cpu) problem?
 
If it's a driver problem, you seem to be the only one reporting it. Considering Intel's gfx market share, if it were a wide problem it would turn up more often.

The freezing under Linux and driver instability under Windows makes me want to suspect memory problems, or other hardware instability, are behind it.
 
What brand PSU? I don't believe any of the upper tier make a 300W. Have you checked your rails? maybe the 12v is choking.
I have 7-64 using HD4600 for now with no issues but I also have a 1 year old Seasonic 660w platinum with dead solid rails which is Haswell compatible.
 
I have the original 300W Powerman Psu that came with the case, but it is a SFX power supply, so I don't have many options, and this system draws so little power I don't think I need the 450W Silverstone SFX PSU. Here is what I have tried over the past few hours:

Was using Ubuntu 13.04, watching local videos, online videos, playing Flash and not one crash of the drivers in 2 hours. Uninstalled drivers on Win7, used standard VGA and no crash. Installed the graphics drivers from the Mobo CD, and just got the same message as before in Win7 about graphic driver stopping and recovering. It looks like there are no good drivers for me under Windows. I am going to look into pushing the TDR delay up a lot, maybe that can help.

I found this Microsoft hotfix on the Asus Mobo CD, KB979903, if you look it up it addresses an error when large amount of contiguous memory is addressed. Might this work?

Update: I double clicked on the install file for the hotfix, and it said the update was not for my system. I find it strange that it is buried in the VGA folder under x64 driver on the Mobo CD. Still getting crashes.
 
My rails look good, +3.344, +5.05, and +12 ranges from 11.90 to 12.00. I had a crash occur browsing on Firefox and had HWInfo64 open with graphs on. I saw a 4C rise in temp, and a sharp decrease in memory usage by gpu. The VCCIn (1.744 )and +12 were steady. Again, just like in GPUZ, gpu usage and power spike up either during or slightly before the crash occurs.
 
Just buy a cheap HD6450 or something to test. And get a real PSU.
 
If you are having problems in windows and Ubuntu I would assume that your problem is software and not hardware. Take your board out of the case and try another PSU to see if it fixes your problem.
 
Memtest is horrible at finding memory errors. It's worse than using nothing at all since it doesn't find errors most of the time and you just end up wasting your time.
 
RamonGTP said:
Memtest is horrible at finding memory errors. It's worse than using nothing at all since it doesn't find errors most of the time and you just end up wasting your time.
Click to expand...


Which program do you recommend to test RAM with? I have run both Memtest86+ 5RC and the 4.3.3 Passmark version and nothing came up. I bought a cheap HD5450 from someone on [H] the other night and it should be here this week. Maybe I could swing a new PSU. It's either go for a Seasonic SFX or maybe get an ATX and have it sit just outside of the case?
 
I would use Prime 95 blend test. Not saying your memory is necessarily the issue, but if I'm going to test it, I would start with that. You ultimately need to use multiple programs because they'll all stress the system differently and some may generate errors while others do not.
 
Considering I was getting temps from 90-95C, I only let Prime95 blend on all four threads run for 20 minutes and no errors, and I didn't get a display driver stopped working message. After finishing Prime95, 5 minutes later, I had Firefox open just browsing, and went to click the X to close it, and the display driver crashed.
 
Last edited:
Run OCCT GPU test. Its entirely possible you have a bad processor as well.
 
If I am getting these errors, it almost feels like a bad cpu, and not the RAM, but now that Intel puts IMC, Graphics, IVR on cpu it is harder to tell what is wrong. I will check out OCCT GPU. Crazy, I went to download OCCT and got the save file dialog box in Firefox, went to click on Save, and the graphics driver stopped and recovered.
 
Ran OCCT GPU for 20 minutes at 1680x1050 (my monitor res), DX11 and it finished without any errors showing up in the event log. Also ran OCCT CPU but it stopped after 4 minutes due to CPU temp getting to 86C, and it said the maximum was 85. I monitored VCCin during this test, and it dropped 1.744 to 1.728V, if that is significant. The +12 rail, dropped to 11.808V during the 4 minutes, which is lower than I have seen.
 
The voltage rail readings in software are completely inaccurate. If you want to know voltage you need to measure it using a multimeter. I'm at a loss as to the issue. I'd think if the IGP in the processor was faulty, it would have surly manifested itself during the OCCT GPU test.
 
What Intel video driver(s) have you tried?

I checked the ASUS website and, of course, theirs is horribly outdated. Try this one.
 
I second the use of Prime95 for testing memory. I had faulty RAM that passed Memtest just fine, but failed very quickly with Prime95.
 
is it a clean install? i'm having similar issues using an external card too, but upgraded from SB to Haswell on the same install, i used a different drive with a fresh install and magically all the problems are gone.
 
I installed on an SSD, but all I did was destroy the partition from the old setup through windows Setup. Is there a way or need to do a Secure Erase on the SSD?
 
hey, I was glancing through the thread here.
According to USPS your card is scheduled for delivery tomorrow.
But did you say your temp was 90~95c?
Are you really overclocked?
That's too hot. What are you idling at?
Whats your cooling setup?
Maybe need to reseat your cooler or reapply the thermal paste.
I put an h60 on and found I was not making full contact due to some other components. rotated 90 degrees and moved it as far to the back as it would go and it dropped 30 degrees at idle and 15 under load.
 
apopleptic said:
hey, I was glancing through the thread here.
According to USPS your card is scheduled for delivery tomorrow.
But did you say your temp was 90~95c?
Are you really overclocked?
That's too hot. What are you idling at?
Whats your cooling setup?
Maybe need to reseat your cooler or reapply the thermal paste.
I put an h60 on and found I was not making full contact due to some other components. rotated 90 degrees and moved it as far to the back as it would go and it dropped 30 degrees at idle and 15 under load.
Click to expand...

I am not overclocked, running everything Auto in bios using an Intel stock cooler in a SFF case. I hit 90-95c while running Prime95. My idling temps are around 35-40c, and idling power draw is only 26W according to Kill-A-Watt.
 
SolidBladez said:
What Intel video driver(s) have you tried?

I checked the ASUS website and, of course, theirs is horribly outdated. Try this one.
Click to expand...

Tried these drivers, and just experienced first driver crash just playing a 1080p local video and browsing on Firefox.
 
Something I find interesting, using HWInfo64 latest, monitoring the iGPU VID, playing a 1080p video, this value bounces around from 0.000V!!?? to 0.735V, yet on the min/max readings, it always displays 0.735V. Just had another crash, and the iGPU went to 0.835V along with a spike in cpu temp, gpu temp and gpu power, but gpu memory usage plummets to near 0. Didn't I read that due to adaptive voltage that Haswell uses, it adds an extra 0.1V? Let me put a cpu core voltage graph on and monitor. The VCCIn is always about the same 1.744, but sometime goes down to 1.728.
 
Last edited:
Nothing strange happens to the cpu core voltages when the crash occurs, only the gpu voltage has that extra 0.1V spike.
 
I couldn't stand the crashing in Win7, so I booted into Ubuntu and everything was going fine, except while I was playing a video at full screen, it froze, there was like smearing of the picture, then the desktop came back and the video was still playing. The screen didn't go black like on windows, and this happened only once, as opposed to many times in Win7.

I will try the HD5450 that I have coming today, hopefully no driver crashes. Then I still need to decide whether to RMA CPU/Mobo if something really is wrong.
 
nevermind, new install crashes too, have you tried bios update manually? i just updated mine to the latest h87m-e bios 0604 it wouldn't do it automatically had to manually select the file, we'll see if that helps at all.
 
That is how I updated the bios, I put it on a flash drive, used the ez bios 2 utility in the UEFI interface and flashed it like that. I have put a Radeon HD5450 into my machine, installed the latest drivers in W7 x64 and so far no crashes. I will have to see if this lasts. So if this video card gives no crashes, then I could only suspect that it is either the Intel drivers, the igpu in the cpu is faulty, or some conbination of drivers and the bios conflicting. Don't know.
 
After reading some more about this "crashed and recovered" problem with all types of video cards, like here about Nvidia, http://answers.microsoft.com/en-us/...751-a1c8-12067c4125c5?page=1&tm=1379465622078, I think it is safe to assume it is some kind of driver issue from any manufacturer and Windows 7 &8. There is no way all these people could have bad iGPU or discrete cards. Here is a search about the problem on Microsoft's website, http://answers.microsoft.com/en-us/...rdware&askingquestion=False&ContentTypeScope=, it just is too suspicious to be bad hardware across so many different configurations. So I guess I won't RMA mobo+cpu, and keep using the radeon HD5450 until it gets solved for Intel HD4600.
 
Dk975 said:
After reading some more about this "crashed and recovered" problem with all types of video cards, like here about Nvidia, http://answers.microsoft.com/en-us/...751-a1c8-12067c4125c5?page=1&tm=1379465622078, I think it is safe to assume it is some kind of driver issue from any manufacturer and Windows 7 &8. There is no way all these people could have bad iGPU or discrete cards. Here is a search about the problem on Microsoft's website, http://answers.microsoft.com/en-us/...rdware&askingquestion=False&ContentTypeScope=, it just is too suspicious to be bad hardware across so many different configurations. So I guess I won't RMA mobo+cpu, and keep using the radeon HD5450 until it gets solved for Intel HD4600.
Click to expand...

I am sorry to hear your troubles, but this post is spot on..There are literally millions of people that have experienced this issue while using nearly every different GPU/MB/CPU combo out there..I think that there is some underlying issue in Windows code that causes this to crop up from time to time..I have experienced it myself, with an older Nvidia and AMD card, go figure:rolleyes:..
 
I have re-enabled the Intel GPU along side the HD5450, but Intel is set to primary. At first I lowered my RAM to 1333 and saw no crashes, then I raised it to 1600, but without XMP enabled. Finally I have enabled XMP and been using windows for about an hour, playing local and streaming video with web browsing (stuff that usually causes the crash) but nothing yet. I am using the 3257 drivers from Intel. If this holds up, I will uninstall the Radeon HD5450 because the Intel actually looks better in videos and text, plus it runs OCCT GPU at 17fps vs only 6fps for AMD.
 
Physically uninstalled HD5450 and drivers, had 1080p video playing, Firefox open and HWInfo. I got the driver crash again, plus Firefox crashed because I have it set to use the memory as the cache. Next thing I will try is taking off XMP Profile from Ram.
 
I don't think it is the RAM's inability to run at 1600 because I have put it to 1333 and I am getting the driver crashes again using Intel 3257 drivers. Just with Firefox open, playing a 1080p video and HWInfo. Strange thing is that when I had Intel graphics alongside HD5450 card, I couldn't generate a driver crash. Looks like I will put the card back in and use that.
 
only a few people mentioned the 90 degrees C, think you need to do something with your temps, that is dangerous to run at that temp
 
I have given up on trying to use iGPU because there are no drivers that don't cause the issue. I have tried altering RAM timings, registry tweaks, running torture tests. I have reinstalled the HD5450 with Catalyst 13.4 and not a single hiccup yet. Therefore I can only assume my RAM and PSU are good. The question remains, is it the Intel drivers, the Asus mobo/bios, or is the iGPU bad somehow? Should I RMA the board and get another brand, and get a CPU replacement just to make sure?
 
Dk975 said:
I have given up on trying to use iGPU because there are no drivers that don't cause the issue. I have tried altering RAM timings, registry tweaks, running torture tests. I have reinstalled the HD5450 with Catalyst 13.4 and not a single hiccup yet. Therefore I can only assume my RAM and PSU are good. The question remains, is it the Intel drivers, the Asus mobo/bios, or is the iGPU bad somehow? Should I RMA the board and get another brand, and get a CPU replacement just to make sure?
Click to expand...

are you ignoring the multiple posts about your temp being too high?
 
Digital Viper-X- said:
are you ignoring the multiple posts about your temp being too high?
Click to expand...

In fairness, he did say the temps were only during a Prime95 run, although, it does make sense that if such a low power chip hits 95C at anytime, it's probably a bad HSF application.
 
You must log in or register to reply here.
Back
Top