Member Avatar for cogribin

Hey guys, i'm at my wit's end with this problem.

Getting freezing in D3D applications, and then a report from windows that the gpu recovered from a serious error (amdkmdap stopped working) event viewer shows the same thing. This will occur multiple times until it will fail to recover, showing a 0x000000EA (Thread_stuck_in_device_driver) BSOD.

I have done the following:

-uninstalled drivers completely, reinstalled various versions, including stable and beta.
- replaced the card. (I now have two, both are behaving the same , together in crossfire and seperately.)
- replaced my RAM, I have two different sets, but the issue persists when only one pair is in the board.
- memtest shows now errors after 11 hours of testing.
- originally watching video would cause the problem to manifest, but now its only in D3D/gaming and also hardware accelerated google chrome windows and such.
-replcaced the PSU with a beefier one (up from 580W to 750W).

My Setup:

core i5 2500k, currently at stock.
Samsung 840 Pro 128GB
4x other HDD (assorted)
12GB Kingston ram (DDR1600 XMP)
ASUS P8P 67 PRO rev1.1 mobo with the cougar point design flaw. (note: I am running a PCI-E SATA board in the PCI-E x1 slot as a result).

Asus 3GD5 R9 280x
Sapphire R9 280x.

Ive downclocked the asus to match the Sapphire stock clocks to rule things like clock mismatches out of the picture, even though I know this isn't an issue anyway.

OS: Win 7 x64 ultimate, patched up to day zero.

Currently disabled Aero as I was told this might be problematic.
Also tried the timeout registry fix for 0x000000EA BSODs.

My latest thought is that either there is some sort of PCI-E conflict from the SATA expansion that's causing these interruptions (shouldn't this be in a logfile somewhere if so?)

OR

the motherboard has had it and I should just replace.

Kind of loath to do this after forking out for a new card and ram and psu D:.

Any other thoughts would be welcome.

  • 2 Contributors
  • 2 Replies
  • 2K Views
  • 21 Hours Discussion Span
  • Latest Post Latest Post by cogribin

Dani AI

Generated

Great catch by . What you were seeing (amdkmdap resets, then an occasional 0x000000EA) is classic Windows GPU Timeout Detection and Recovery (TDR): the OS gives the GPU about 2 seconds to respond, tries to reset the driver, and if recovery fails or a driver thread is spinning, you get a bugcheck. That is why Windows first showed “Display driver stopped responding and has recovered,” then progressed to a BSOD. (learn.microsoft.com)

On the ASUS P8P67 PRO specifically, the black PCIe x16_3 slot and the two PCIe x1 slots share bandwidth with other onboard controllers (USB3_34 and eSATA). Populate those x1 slots or hammer a PCIe x1 SATA card and you can starve the GPU just enough to trigger TDRs. If you need the expansion card, move it to the black x16_3 slot and set “PCIe X16_3 slot (black) bandwidth” to x4 in BIOS; that setting disables the x1 slots (and some ports) but gives the add‑in more headroom and typically restores GPU stability. (pcper.com)

A few quick checks that help others with similar rigs:

  • Update the board BIOS (e.g., 2303 lists “Improve system stability”). After any CMOS reset like suggested, re‑check the X16_3 bandwidth setting because it may revert to Auto. (asus.com)
  • Prefer the Intel 6 Gb/s (gray) ports and the board’s Marvell 912x SATA ports if you are on an early Cougar Point (B2) revision; avoid the affected 3 Gb/s ports that were part of the 2011 SATA degradation issue. (anandtech.com)

Bottom line: your root cause (PCIe bus contention from the SATA card) fits both the platform’s lane‑sharing design and how TDR behaves under I/O‑induced GPU stalls. (learn.microsoft.com)

Recommended Answers

All 2 Replies

Member Avatar for rijve

try to reset ur bios if any option or unplug mobo battery for 1hur and try again i am using dp55wb try to use gigabyte r9 270x boot gone so i reset bios by mobo jumper and unplug battery for long time

Member Avatar for cogribin

got it sorted. For those that are curious

It was indeed a PCI-E bus conflict of some sort. As soon as I removed my SATA expansion card from my pci-e x1 slot , the problem disappeared.

Seems like some of the traffic from the PCI-E x1 slot powering my SATA expansion card was causing latency spikes from my graphics card, even though the mobo manual specifically mentions that running an x1 card alongside to x8 or two x16 cards is fine.

[x] lame.

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.