From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <4EDE5AFA.2050503@domain.hid> Date: Tue, 06 Dec 2011 19:12:10 +0100 From: Gilles Chanteperdrix MIME-Version: 1.0 References: <4ED8EEB1.1040904@domain.hid> In-Reply-To: <4ED8EEB1.1040904@domain.hid> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Xenomai-help] Xenomai installation: high latencies List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Kurt Geebelen Cc: xenomai@xenomai.org On 12/02/2011 04:28 PM, Kurt Geebelen wrote: > Dear all, >=20 > I've tried installing Xenomai, following these instructions: >=20 > http://www.xenomai.org/index.php/Building_Debian_packages >=20 > but then for the latest version (xenomai 2.6.0, and kernel 2.6.38.8 and= =20 > for 64 bit) . >=20 > For configuring the kernel, I followed these instructions: >=20 > http://www.xenomai.org/index.php/Configuring_x86_kernels >=20 > All that is stated on these websites works without problems, and I am=20 > able to boot in the installed xenomai kernel. However, when I perform=20 > the latency-test (latency in /usr/lib/xenomai), I get latencies of more= =20 > than 1400 microseconds. >=20 > The system I have is the following: >=20 > Dell=E2=84=A2 Precision=E2=84=A2 T1500 > One=C2=AE Intel=E2=84=A2 Core i7-860 (2.8GHz, 8MB L3, 95W, Quad Core) > 6GB DDR3 non-ECC Memory (1333MHz, 3x2GB) > 512MB PCIe x16 NVIDIA Quadro FX 580 Graphics Card > Intel Solid-State Drive 320 Series 120GB >=20 > I have tried reinstalling the kernel several times, but I always end up= =20 > with high latencies. > I also have consulted with others, who followed the same guidelines as = I=20 > did, and their latencies are about 100 times lower. >=20 > Is there somebody who knows what the problem might be? Might it be a=20 > hardware-related problem? It is already written in the URL you mention, but I will repeat it nevertheless, if enabling SMI workaround, have you checked the kernel logs to see whether the SMI workaround actually works? --=20 Gilles.