1. Computers
  2. Display Drivers
  3. Graphics Cards
  4. Memory
  5. Motherboards
  6. Processors
  7. Software
  8. Storage
  9. Operating Systems


Facebook RSS Twitter Twitter Google Plus


Phoronix Test Suite

OpenBenchmarking.org

Intel's New Clarkdale IGP On Linux Not So Pleasant

Intel

Published on 22 January 2010 05:54 AM EST
Written by Michael Larabel in Intel
33 Comments

Last week we shared that Intel Core i3 Linux benchmarks were being worked on at Phoronix with an Intel Core i3 530 "Clarkdale" processor that we had purchased. This recently released Intel Core i3 processor features an integrated graphics processor on the CPU itself. This next-generation Intel graphics processor is supported by the Linux 2.6.33 DRM and was actually being worked on publicly within the Intel Linux driver stack for months under the name of IGDNG. While Intel has been working on the Linux support for more than six months, the experience is still not ideal.

A daily development build of Ubuntu 10.04 LTS was loaded up on the Intel H55 motherboard we were using for testing in conjunction with the Intel Core i3 530 processor. Ubuntu 10.04 uses the Linux 2.6.32 kernel, Mesa 7.7, and right now is using the older xf86-video-intel 2.9 X.Org driver. Even with the older DDX, the Ubuntu 10.04 live environment started up fine from a USB flash drive. Kernel mode-setting was there and even Compiz was working "out of the box" from the live mode and then once installed to the hard drive.

After installing this Ubuntu 10.04 snapshot everything continued working fine on the desktop with KMS and Compiz. The Phoronix Test Suite was then installed and a few OpenGL benchmarks were carried out. They ran fine. However, with time problems have started to arise. While running some OpenGL programs, the display would simply shut off and no signal was being sent to the monitor. In most cases the system though was still online and could be remotely connected in via SSH. When doing so, many Intel DRM errors were exposed through dmesg. The Intel Clarkdale GPU was getting hung under Linux and seems to originate with its kernel memory management, the Graphics Execution Manager (i915_do_wait_request comes from i915_gem.c).

[ 545.947427] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 545.947437] render error detected, EIR: 0x00000000
[ 545.947454] [drm:i915_do_wait_request] *ERROR* i915_do_wait_request returns -5 (awaiting 657028 at 657024)
[ 545.947464] [drm:drm_crtc_helper_set_config] *ERROR* failed to set mode on crtc ffff88006e982800
[ 548.197399] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 548.197407] render error detected, EIR: 0x00000000
[ 548.197420] [drm:i915_do_wait_request] *ERROR* i915_do_wait_request returns -5 (awaiting 657033 at 657024)
[ 548.947404] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 548.947412] render error detected, EIR: 0x00000000
[ 548.947430] [drm:i915_do_wait_request] *ERROR* i915_do_wait_request returns -5 (awaiting 657036 at 657024)


The Linux kernel on Ubuntu 10.04 was then updated to the latest Linux 2.6.33 code, libdrm was also updated, and then the xf86-video-intel 2.10 DDX (the latest release with official Clarkdale / Arrandale support) was installed. These GPU hangs continue to occur and in some cases even more frequently than using the slightly older driver stack. So far these GPU hangs is the most common problem we have encountered with this new Intel on-CPU graphics processor, which results in the need for a reboot and then hoping that when it reboots the OpenGL benchmark will run without crashing. Within the next week or two an article will be on Phoronix with our formal findings as well as the first Linux benchmarks from the graphics processor found on the Intel Clarkdale / Core i3 processor.

About The Author
Michael Larabel is the principal author of Phoronix.com and founded the web-site in 2004 with a focus on enriching the Linux hardware experience and being the largest web-site devoted to Linux hardware reviews, particularly for products relevant to Linux gamers and enthusiasts but also commonly reviewing servers/workstations and embedded Linux devices. Michael has written more than 10,000 articles covering the state of Linux hardware support, Linux performance, graphics hardware drivers, and other topics. Michael is also the lead developer of the Phoronix Test Suite, Phoromatic, and OpenBenchmarking.org automated testing software. He can be followed via and or contacted via .
Latest Linux Hardware Reviews
  1. Rosewill RS-MI-01: An Ultra Low-Cost Mini-ITX Chassis
  2. D-Link DCS-2330L HD Wireless Network Camera
  3. Gigabyte AM1M-S2H
  4. AMD's New Athlon/Semprons Give Old Phenom CPUs A Big Run For The Money
Latest Linux Articles
  1. AMD Catalyst 14.4 Brings Few Linux Performance Improvements
  2. The Performance Of Fedora 20 Updated
  3. Clang Fights GCC On AMD's Athlon AM1 APU With Jaguar Cores
  4. Ubuntu 14.04 LTS vs. Oracle Linux vs. CentOS vs. openSUSE
Latest Linux News
  1. PC-BSD Is Developing Its Own Desktop Environment
  2. Valve Is Bringing VOGL To Windows & Working On Regression Tests
  3. Canonical Is Taking Over Linux 3.13 Kernel Maintenance
  4. Google Web Designer Is Now Natively Available On Linux
  5. Ubuntu 14.10 Is Codenamed The Utopic Unicorn
  6. Audacious 3.5 Lightweight Audio Player Released
  7. Steam Updated For Ubuntu 14.04 LTS, SteamOS
  8. DNF 0.5 Yum Replacement Now Supports Groups
  9. Red Hat Enterprise Linux 7.0 Is Looking Fantastic
  10. Intel Is Launching An Interesting Bay Trail NUC Next Week
  11. Another X.Org EVoC Proposed For OpenGL 4+ Tests
  12. The Best Features Coming With Qt 5.3
Latest Forum Discussions
  1. Linux Kernel Developers Fed Up With Ridiculous Bugs In Systemd
  2. The Most Amazing OpenGL Tech Demo In 64kb
  3. Announcing radeontop, a tool for viewing the GPU usage
  4. HTPC-upgrade advice: AMD Richland A8-7600 or Kaveri A10-6700T ???
  5. New card. Open source drivers only.
  6. The GNOME Foundation Is Running Short On Money
  7. Script for Fan Speed Control
  8. Torvalds Is Unconvinced By LTO'ing A Linux Kernel