P9X79-E WS - Intermittent Hardware detection

[Rassilon]

I just built a brand new computer about a week ago:

Motherboard model: P9X79-E WS
BIOS Version: 1501
CPU: Intel Core i7-4820K Ivy Bridge-E
Memory: G.SKILL Ripjaws Z Series 32GB (4 x 8GB) Model F3-1866C9Q-32GZH

GPU (in PCIE Slot 1): Sapphire Radeon HD7850 2G OC ver. (SKU 11200-01-20G)
GPU (in PCIE Slot 5): Asus Radeon HD7850-DC2-2GD5-V2
(crossFireX mode)

SSD/HDDs/Optical drives:
OS Drive: SanDisk Extreme II 240GB SSD
Drive 2: Crucial C300 64 GB SSD
Drive 3: WD Black 640 GB HDD
Drive 4: WD Black 640 GB HDD
Drive 5: WD Black 1 TB HDD
Drive 6: Lite-On ihas424-08 DVD Burner
(No Raids)

PSU: Seasonic SS-760XP2

USB Devices:
Keyboard: Logitech G110
Mouse: Mionix Naos 5000
Printer: Epson RX580
Printer: Brother HL-5240

Monitor: BenQ G2400WD
CPU Cooler: Noctua NH-U12S
PC CASE: SilverStone Fortress Series FT02B-USB3.0

Operating system: Microsoft Windows 7 Pro x64 (non-uefi), Activated

Drivers Installed:
Intel(R) Management Engine Interface V9.0.0.1287
Intel Chipset Software Installation Utility V9.3.2.1015
Realtek Audio Driver V6.0.1.7023 (which apparently causes a security audit failure in sxs.dll)
Intel LAN Driver V18.1.59.0
Asmedia USB 3.0 Driver V1.16.10.0
Asmedia SATA Controller Driver V1.3.4.1
Marvell SATA 6G Driver V1.2.0.1027

Any third Party temp/voltage software installed: No
System Overclocked (provide details): Only XMP profile #1 loaded


Issue:
Hauppauge HVR2250 (x1 PCIE) TV Tuner card (which is on page 11 of the QVL list) not reliably detected in pcie slot 4.

Out of about 20 boots the card has been detected in Windows only 3 times if it is in pcie slot 4.
It is always detected when in slots 3 & 7 (tried about 10 boots with it in these slots).

When it IS detected in slot 4, I noted that in Windows Device Manager (view by connection) it appears directly under:
Intel(R) Xeon(R) E7 v2/Xeon(R) E5 v2/Core i7 PCI Express Root Port 1a - 0E02

When it is NOT detected in slot 4, the root port still shows as working properly, but if I right click on the root port and select "scan for hardware changes" it still does not detect the card.

I was hoping it is just a bios setting. I have tried:
disabling Fast Boot
disabling PCIE Spread Spectrum
PCIEx16_4 Link Speed - tried all settings
enabling CSM (Compatability Support Module)

None of these settings have helped

I have run CPU-Z reports during the following scenarios:
1.) PCIE slot 4: empty
2.) PCIE slot 4: card inserted and functional
3.) PCIE slot 4: card inserted but not detected in windows

Pertinent results from CPU-Z, Slot 4 section:
Scenario 1, PCIE slot 4 (empty):
PCI capability
Presence detect: no
Link width: 0x (max 8x)

Scenario 2, PCIE slot 4 (card inserted and functional):
PCI capability
Presence detect: yes
Link width : 1x (max 8x)

Scenario 3, PCIE slot 4 (card inserted but NOT detected in windows):
PCI capability
Presence detect: yes
Link width : 1x (max 8x)

The PCI registers are also different for each scenario and I can post them if needed.

When the card is functional, CPU-Z also has a section detailing the card itself under:
Location: bus 5 (0x05), device 0 (0x00), function 0 (0x00)
But this section is missing when the card is not detected.

Googling this issue, some sites suggest it is a bios issue, perhaps the board booting too fast.

Other, perhaps related, anomalies:
1 week after completing this build, windows suddenly installed a firewire driver for no apparent reason.
Today I had Device Manager scan for hardware changes, and it suddenly installed a Thermal Management Controller.
Every time the computer is booted, the Event Viewer lists errors with both LANs being disconnected, then 1 second later in the log they are shown to connect.

Also I should add that when the card is detected in slot 4 it works without flaw. I have watched TV on it for many hours and did not notice any problems at all.

And just in case someone suggests reinstalling drivers: The drivers for the TV card cannot be installed if the hardware is not detected. And yes, I tried re-seating card, and physically inspecting the slot. I did not, however inspect the rear of the board for cold solder joints, and shudder at the thought of having to disassemble this system.

[Raja]

Let me report this to HQ and see if they have anything to suggest.


Thanks for taking the time to fill out the bug report form - what a breath of fresh air that is! 🙂

-Raja

[Rassilon]

Awesome, Thank you very much!

Sadly I don't have any other pcie cards to test the slot with, other than the graphics cards. And I didn't think using graphics cards in suboptimal configurations would yield very good data.

I did find a thread regarding a specific bios incompatability with this card on an intel mb:
https://communities.intel.com/thread/29254?start=45&tstart=0

[Necrosan]

I wonder if your card has the same issue as mine.

Try forcing lane speed the Hauppauge card is in to PCIE 1.0.

Is it reliably detected then?

[Rassilon]

When I first installed the Hauppauge card all the slots were set to auto, which didn't work well. Then I tried the other settings (Gen 1, 2, & 3). None of those settings helped. So I have left it on "Gen 1" since I gave up troubleshooting and it was only detected once since then (out of 5 boots?).

The board seems very stable otherwise. I have played long hours of an mmo in windowed mode with all graphics settings on max, and had the TV on in another window (when it works) with no crashes or stuttering.

[Necrosan]

Rassilon wrote:
When I first installed the Hauppauge card all the slots were set to auto, which didn't work well. Then I tried the other settings (Gen 1, 2, & 3). None of those settings helped. So I have left it on "Gen 1" since I gave up troubleshooting and it was only detected once since then (out of 5 boots?).

The board seems very stable otherwise. I have played long hours of an mmo in windowed mode with all graphics settings on max, and had the TV on in another window (when it works) with no crashes or stuttering.

Could you possibly send me a PM with your email address?
There are a couple of other things that could be causing your issue & I'd like to forward you some email correspondence I had with their lead driver developer.
Thanks!

[Rassilon]

Thanks for the info Necrosan. But the motherboard that I have doesn't share pcie lanes like that, it has some PLX chips which enable 4 slots to run at x16 at the same time, and even all 7 slots can run at 8x simultaneously. The slot 4 that I am concerned with has a direct connection to the CPU, it is not shared and does not run through a PLX chip. It would also be the best Slot to put a 3rd card in, ventilation-wise.

I found some very interesting info here:
http://en.community.dell.com/support-forums/desktop/f/3514/t/19535771.aspx

Some relevant snippets from there:
The original HVR-2250 card was marked "REV C3F2" while the newest one from Dell is labeled "REV C4F2". I contacted Hauppauge to ask them about the problem. Their reply was: "The older revision will not be recognized on newer motherboards which use Gen2 or Gen3 PCIe slots. . . There was only a small run of HVR-2250 with this particular issue. When it started to occur we simply swapped out the affected cards. . . The problem occurs when the PCIe root complex (North bridge of the chipset on the motherboard) and the PCIe end point (in this particular case the NXP chip on our cards) do not finish the training sequence (TS1,TS2) and the bus traffic between the root complex and the end point are keeping busy in a retrain loop. This is caused by the North Bridge register "Target Link Speed" (TLS ) setting not auto changing from 5Gb/s to 2.5Gb/s, which is required for 1.x backward compatibility. When the problem cropped up a few years back some motherboard manufactures released new BIOS which corrected the problems but many didn’t bother since the older Gen1 PCIe chipsets were being phased out by manufacturers such as ourselves."

After reading this I promptly checked the pcb revision on my Hauppauge card and, yes, I have the older Rev C3F2.

So it would seem that setting the slot to "Gen 1" in the bios should fix this, but it doesn't.