7970 does not fold when headless

[rhavern]

Myself and another user on the #fah IRC channel have noted that folding does not start if the system is video headless. I am using version 13.4 drivers with 7.3.6 on an ASUS M3N-HT Deluxe AMD64x2 4000+ 780a SLI chipset 2G RAM 2x XFX 7970 W7-32 SP1 CM M1000 PSU. I don't get any errors, just no folding. Additionally, not all the sensors appear in MSI Afterburner when headless. I use logmein to access the box when headless. No CPU folding running.

If I start the box with monitor attached, all the sensors appear and folding starts normally. As a workaround, I've installed a dummy plug I had from years ago when this previously was an issue, and everything works as expected on both GPUs, even though I only have one dummy plug installed. Everything in AMD control center left at default, power control is +20, core at 1100, memory at 1500, custom fan profile for 100% @ 70C, current temps ~65C, case is open, no VPU recovers, no bridge installed.

Any ideas?

[bruce]

Post the hardware detection page at the top of the log. Which video device(s) are present in your system?

[rhavern]

As requested:

Code: Select all

*********************** Log Started 2013-08-08T02:13:52Z ***********************
02:13:52:************************* Folding@home Client *************************
02:13:52:      Website: http://folding.stanford.edu/
02:13:52:    Copyright: (c) 2009-2013 Stanford University
02:13:52:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
02:13:52:         Args: 
02:13:52:       Config: C:/Users/Rick/AppData/Roaming/FAHClient/config.xml
02:13:52:******************************** Build ********************************
02:13:52:      Version: 7.3.6
02:13:52:         Date: Feb 18 2013
02:13:52:         Time: 15:25:17
02:13:52:      SVN Rev: 3923
02:13:52:       Branch: fah/trunk/client
02:13:52:     Compiler: Intel(R) C++ MSVC 1500 mode 1200
02:13:52:      Options: /TP /nologo /EHa /Qdiag-disable:4297,4103,1786,279 /Ox -arch:SSE
02:13:52:               /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qopenmp /Qrestrict /MT /Qmkl
02:13:52:     Platform: win32 XP
02:13:52:         Bits: 32
02:13:52:         Mode: Release
02:13:52:******************************* System ********************************
02:13:52:          CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 4000+
02:13:52:       CPU ID: AuthenticAMD Family 15 Model 107 Stepping 1
02:13:52:         CPUs: 2
02:13:52:       Memory: 2.00GiB
02:13:52:  Free Memory: 1.18GiB
02:13:52:      Threads: WINDOWS_THREADS
02:13:52:  Has Battery: false
02:13:52:   On Battery: false
02:13:52:   UTC offset: 1
02:13:52:          PID: 3672
02:13:52:          CWD: C:/Users/Rick/AppData/Roaming/FAHClient
02:13:52:           OS: Windows 7 Professional Service Pack 1
02:13:52:      OS Arch: X86
02:13:52:         GPUs: 2
02:13:52:        GPU 0: ATI:5 Tahiti XT [Radeon HD 7970]
02:13:52:        GPU 1: ATI:5 Tahiti XT [Radeon HD 7970]
02:13:52:         CUDA: Not detected
02:13:52:Win32 Service: false
02:13:52:***********************************************************************

[7im]

Dummy plugs were needed for a long time, and then at a certain driver level, that problem went away because the OEMs fixed it in their drivers. It's not inconceivable the problem could return, or only happens with certain driver versions. And time to lobby the AMD forum again to put the fix back in to the next driver.