topic Even with the best cards in Software Archive

MPSS Installation micflash

Aurelio_Rodriguez — Fri, 09 Nov 2012 09:58:20 GMT

Hi,

I am trying to install the last version software stack on a rhel 6.3 but when I run the flash update tool I am getting the following message:

[root@localhost /home/software]# /opt/intel/mic/bin/micflash -Update

VERSION: Copyright 2011-2012 Intel Corporation All Rights Reserved.
VERSION: 4346-16

Intel(R) Xeon Phi(TM) Coprocessor stack initialization failed

The DeviceList option seems ok:

[root@localhost /home/software]# /opt/intel/mic/bin/micflash -DeviceList

VERSION: Copyright 2011-2012 Intel Corporation All Rights Reserved.
VERSION: 4346-16

deviceId | bus# | pciDev# | hardwareId
---------|------|---------|-----------
       0 |    0 |       0 |   00000000
       1 |    0 |       0 |   00000000
       2 |    0 |       0 |   00000000
       3 |    0 |       0 |   00000000
       4 |    0 |       0 |   00000000
...

any clue about what is happening?

Regards

What does it say when you run

Frances_R_Intel — Fri, 09 Nov 2012 21:13:07 GMT

What does it say when you run: micctrl –w Does it say the card is in the ready state? If not, try micctrl -r and then again micctrl -w

I thought I left a second

Frances_R_Intel — Mon, 12 Nov 2012 03:38:41 GMT

I thought I left a second comment here but I don't see it - so anyway - also remember you must reboot the host processor whenever you update the flash memory on the coprocessor. So if you haven't already, try rebooting the host.

In my experience, the most

TimP — Mon, 12 Nov 2012 04:30:42 GMT

In my experience, the most recent mpss upgrade required a full power down both before and after deleting old and installing new rpms, even without a flash update.

hi,

Aurelio_Rodriguez — Mon, 12 Nov 2012 09:07:39 GMT

hi, thanks for your suggestions, about micctrl –w I have done a full power down and now I getting: [root@localhost /home/software]# micctrl -w No MIC cards found in the system Please verify using the 'lspci' utility [root@localhost /home/software]# lspci | grep 2250 02:00.0 Co-processor: Intel Corporation Device 2250 (rev 10) [root@localhost /home/software]# /opt/intel/mic/bin/micinfo MicInfo Utility Log Created Mon Nov 12 09:58:49 2012 This is a new installation, not an update. Before de powerdown the micinfo utility gave the following: [root@localhost /home/software]# /opt/intel/mic/bin/micinfo MicInfo Utility Log Created Fri Nov 9 12:07:33 2012 System Info Host OS : Linux OS Version : 2.6.32-279.el6.x86_64 MPSS Version : 2.1.4346-16 Host Physical Memory : 12320 MB CPU Family : GenuineIntel Family 6 Model 26 Stepping 5 CPU Speed : 1600.000 Threads per Core : 1 *************************** MPSS Driver is not loaded ************************** **************** MicInfo will display all available information **************** Device No: 0, Device Name: Intel(R) Xeon Phi(TM) coprocessor Version Flash Version : 2.1.0.ffff Board Vendor ID : 8086 Device ID : 2250 MIC Processor Stepping ID : f MIC Processor Model : 0x0f MIC Processor Model Ext : 0x0f MIC Processor Type : 0x03 MIC Processor Family : 0x0f MIC Processor Family Ext : 0xff MIC Silicon Stepping : Undefined Core Total No of Active Cores: 63 Voltage : 4294967295 uV Thermal Fan Speed Control : NotAvailable SMC Firmware Version : NotAvailable FSC Strap : NotAvailable Fan RPM : NotAvailable Fan PWM : NotAvailable GDDR GDDR Vendor : ��' GDDR Version : 0xf GDDR Density : 0 Mb GDDR Size : 524287 MB GDDR Technology : GDDR5 GDDR Speed : 4294.867188 GT/s *************************** The information above may not be complete ************************** ********* Please run the /opt/intel/mic/bin/miccheck utility to begin troubleshooting ********** Checking again the readme, I think the problems cames from the hardware requirements, I guess at least a Motherboard S5520SC (5600 family) is needed. am I right? Thanks again for your help Regards

Hi,

TaylorIoTKidd — Mon, 12 Nov 2012 14:16:04 GMT

Hi, I enountered this issue once and found it was because the card was not seated correctly in the slot. I suggest that you open up your machine, and carefully remove and then reinstall the card. Also make sure to use the support bracket at the end of the card. Regards -- Taylor

Even with the best cards

TimP — Mon, 12 Nov 2012 14:46:00 GMT

When this happens, it is possible to recover by a full power shutdown and restart, possibly repeated. The latest Crown Point Sandy Bridge with the current Intel Xeon Phi Coprocesor card works most reliably We have not seen support brackets for the KNC cards. Consequently, it may be advisable to remove them when shipping. As you indicated, only the specific Westmere (5600) and Sandy Bridge host platforms are adequately tested with MIC cards. There are specific BIOS requirements for the host platform. The Westmere required a user BIOS upgrade, followed by a required change of memory map setting, to support the KNC card. Such BIOS support may not be available on earlier Xeon platforms.

Ok, lets try to help you.

Stephan_W_1 — Mon, 17 Dec 2012 20:07:44 GMT

Ok, lets try to help you. First some questions Is your card a KNF or a KNC? What Version do you have? Whats the output of uname -a Which linux distribution do you use? RedHeat/CentOs? Have you checke the PCI-E power cables? You need!!! a 8 and a 6 pin Power-Connector. When you use just 2 6 Pin, the can card show undefined behaviour. The pins REALLY have to fit exactly. So check it twice.. Have you used the bracket screws? If not, use them, the card is heavy. What´s the output of: lspci -vv for the coprocessor Have you enabled 64bit PCI-mode in the BIOS settings? I think this is enough for the beginning.

Just as a note, we no longer

TaylorIoTKidd — Mon, 17 Dec 2012 20:14:03 GMT

Just as a note, we no longer support KNF cards and any recent stack and SW releases, say for the last several months, will not work with them.

Aurelio -

Frances_R_Intel — Thu, 03 Jan 2013 02:21:21 GMT

Aurelio - where does this problem stand? Did you get the flash to update? Are you able to use your coprocessor?

Thanks all of you,

Aurelio_Rodriguez — Mon, 07 Jan 2013 08:48:26 GMT

Thanks all of you, everithing ok using a Sandy Bridge host platform. Regards Aurelio