Networking :: Programming Using DMA And RDMA On RedHat With Intel Nehalem?
Mar 24, 2010
I've read there is a way to communicate without tradtional TCP/IP programming on Redhat Linux and Nehalem CPU with ethernet. I've tried to collect some related information with sample program codes, but failed to collect useful information.
I've got a dual Xeon 5570 (Nehalem) server. It clocks down properly, but never, under any circumstances that I've found, clocks up past its TSC. Even when running basically nothing but cat /dev/zero > /dev/null, nothing gets clocked higher than the TSC. Below is the turbostat output, which clearly knows what the hardware is actually capable of, but it never actually achieves it. I'm running Lenny, with backported kernel 2.6.30-bpo.2-amd64. I tried the 2.6.32 backport and there wasn't any difference. know where I should look next? BIOS settings? Do I need to change some kernel option and recompile?
# ./turbostat Nehalem multiplier 22, TSC frequency 2933 MHz Nehalem 4 cores active: 24 mult, max turbo frequency = 3200 MHz Nehalem 3 cores active: 24 mult, max turbo frequency = 3200 MHz Nehalem 2 cores active: 25 mult, max turbo frequency = 3333 MHz Nehalem 1 core active: 25 mult, max turbo frequency = 3333 MHz
I just installed a new HP DL360 with 32 GB of RAM and a Intel Xeon X5550 which supports speed stepping and turbo mode. I've enabled turbo in the BIOS, but cannot seem to get CentOS 5.4 to make use of either of these features. I've been googling for a solution for a while now, and tried various things, such as installing cpufreq. However, when I type cpufreq-info, I get this:
# cpufreq-info cpufrequtils 005: cpufreq-info (C) Dominik Brodowski 2004-2006 Report errors and bugs to cpufreq@vger.kernel.org, please. analyzing CPU 0:
Are Intel wireless cards compatible with non-intel-based laptops? In my case I wish to upgrade the current Atheros-based mini pci express wifi card with the Intel 4965agn. It is an Asus 4520 with an AMD Athlon X2 processor and nvidia nForce chipset.
I'm running Ubuntu Server 9.10 on a dual socket X5550 platform with hyperthreading disabled. There should be 8 cores available for running processes: dual-socket, quad-core. When I start 8 single-threaded, cpu-intensive processes running, top shows six processes running at 100% and two at 50% (presumably sharing the seventh core). No other processes are reported as having >1% cpu, and yet the 1-minute load average is listed as over 11.0?!
1) Why is it not running each process on its own core? 2) Why is the load average so much greater than 8? 3) Is there anything I can change in the configuration to fix this?
I am customizing the CentOS and I want to change the complete look and feel of the OS. I know all the installation wallpapers and images comes from the rpm mentioned above. Even I extracted the said rpm using rpm2cpip and have replaced the my customized images but not sure how can I prepare my new RPM to include that in a my build.
Does any have any clue or I should go for src.rpm?
As a part of a school project I am working on monitoring context switches. I'm working on redhat 8.0, kernel version 2.4.18-14 and for some reason I keep getting unclassified reasons which causing context switches from a process to itself. I am trying to determine each context switch, what was the reason that causes it, whether time slice ended, or yielded, or went for waiting or returned from waiting or parameters changed or task ended to task forked or any possible reason.
But all the time I find that the only switches that I cannot determine the reason happens from a process to itself, all this unrecognized switches from process to itself happens only on interactive processes, and sometimes the need_resched flag is on and sometimes is off. What could possibly cause a context switch from a process to itself? Can a returning of a process with a lower (higher by number) priority from waiting cause a context switch?
On Debian, if I want to compile some package X myself, but I don't want to spend time hunting down and installing its dependencies, I can issue the following command:
Code: apt-get build-dep X
And apt (so long as it knows about package X) will install all the dependencies for me.Does the yum package manager on RedHat-based systems have a similar feature?
I had Windows XP working. I then put in a new hard disk and installed Redhat 5 on it. rom then i could only boot into Redhat. Further, everyting is very slow. Anothe rproblem is that Redhat does not see my netwrok so I cannot get onto the Internet. I need to have the choice of booting into XP or Redhat.
Question: how to use user space i2c/SMBus when I cannot open + utilize devices on the smbus?
For example: smbus_fd = oepn("/dev/i2c-o", O_RDWR) cannot be done since '/dev/i2c-o' does not exists nor is it attached to the i2c-bus (see tree below).
I cant mount a directory on red hat 5 from red hat 4 using fstab.I keep getting permission denied.I mount directories on the same red hat 5 sys from other red hat 5 and 2 old HPUX systems using fstab.When I try to telnet to it i get "no route to host."
Does anyone know what the Inet/AMD instruction 'nopl' is or does anyone know where I can find a list of hex values for the Intel/AMD instructions. nothing important is waiting on this reply.
Kernel 2.6.21.5, slackware 12.0 as 2.17 [I am speaking here about 80x86 processors, with their segment registers SS, DS, CS, ES.] In a source file for as, I wrote
Code:
.section _STACK .rept 4096 .word 0 .endr
Now, in the .text section, I want to initialize SS, the stack segment, to point to the _STACK section. In other assemblers, there was a special section for the stack, and the linker cared about initializing SS. Here I use a common section for the stack. For a segment such as DS, I could write
mov ax,_DATA mov ds,ax
But I don't think mov ax,_STACK would do with as/ls.how do I initialize SS?
I am running some muti-threaded code that uses a lot of memory (~6GB). I have it compiled on a cluster that uses Xeon 5160 quad core CPUs with 4MB cache/core, 8GB of RAM and kernel 2.6.18 (x86_64) and I have also the same code compiled on another cluster that uses Opteron 2214 quad core CPUs with 1MB cache/core, 8GB of RAM and kernel 2.6.28 (x86_64). On the Intel cluster the code generates barely any system load when I look on top but on the AMD cluster it can generate over 50% of system load. On the intel machines my process can generate a constant load of 3, while the load generated by the process on the AMD machines is about 1.6.
Does anybody know what could cause this? Could it be the version of the kernel or the difference between the CPUs (for example, the much smaller cache of the AMD CPUS)? My software and all its dependencies are the same on both clusters. I am also using the exact same version of GCC to compile everything.
We have Redhat 9 proxy server with 4 lan cards using 1 static and 3 internal private ips,as there is no pci slot in the pc for fifth lan card so i have to purchase usb lan card.
My question is that can the redhat9 will detect the usb lan card if it will detect the usb lan card then I will purchase the lan card
I changed the complier PGI to Intel. But i had some error message p4_error: interrupt SIGSEGV: 11 This error message was occured using intel compiler I don't know what is the ploblem.
i m kapil sharma, i want to know about linux server configration in redhat linux 5.0? how i configure ICS service in rdhat linux 5.0? i have two lan cards in the system and one ip address assigned by ISP.I want to configure linux sever to share this ip address in network approx 100 pc's of widow's xp.
I'm an it student and a newbie in linux OS. We have this final project that we have to set-up a firewall in a linux redhat OS which i totally have no idea .. i did some research but haven't found the right one.
I have installed redhat linux5.I have wireless internet connection at home. wireless internet connection working fine in window 7 but not in redhat linux . Also wireless LED is not lighting in redhat . I have inbuilt intel wifi 5100 AGN card which is not detecting in redhat.rehat comman output which will help you
How to do PTCL broadband settings in redhat.in windows i just do followingRight Click My Network Places on desktop and select Properties Right Click Local Area Connections andselect Properties.In the General Tab; double click on Internet Protocol (TCP/IP) and made the following settings:
had a box lose it's virtual IP address on the eth0 card. I rebooted the box but it just would not give the card the second IP address. We rebooted several times in fact.Today admin ran 'ifup ifcfg-eth0:1' and the virtual IP was there. Nothing was changed, just that command was ran.I thought there might be a problem with the automagical stuff so I rebooted the box again, virtual IP is there and happy.Anyone know why it would not have come back on reboot and why it now does?
1. im working on several distros, sometimes i dont remember the syntax of the ethernet settings (i.e those directives in /etc/sysconfig/network-scripts/ifcfg-eth0,or /etc/network/interfaces ) in redhat its IPADDR,NETMASK etc..
so,where exactly is the man page for this information, i tried (in centos) to type man -k "IPADDR" to find some reference but i come up with nothing.
2. sometimes im logged as user "user1" and then i do "su user2" and then again "su user3", i would like to know how far the nesting is and who was first.. i.e is there any command to give me this output: "user1->user2->user3"?
How To Enable networking in redhat i have tried all the possible way to enable it but it doesnt work..it shows my localhost name and details but it cant connect to it.i have tried system-config-network command but blue screen splashes.
I'm using a program on a group of servers which contact each other using SSH. I've got the authentication keys setup with no password but when one box tries to connect to another it wants the password for the user it's trying to communicate with. So the main problem is there is a lot of communicating within this program so it's important to allow each box communicate with each other without the use of a password. I'm using OpenSSH 4.3 with OpenSSL 0.9.8 on Red Hat Enterprise Linux Server 5.3 (Tikanga). I'm also using SSH on Ubuntu(Intrepid) to connect to these servers at the mo as i don't have direct access to the boxes.