azertyqwerty
Technical User
In this post I’m going to describe the memory upgrade process on the switch-fabric 8690 DS1411016.
I was interested in the idea of the memory upgrading with existent DIMM modules in the computer. And that is why I decided to try. I didn’t do it alone and further I will use “we”.
So, at first we decided to do everything in haste. First of all we took some DIMM modules of 256 Mb and put them into the switch-fabric.
The result was very bad. The boot loader was defining correctly the memory volume, but after module initialization in the slots the basic code began to reboot. We began to investigate the problem. On account of some reasons we had two types of modules that worked correctly in the switch-fabric. One of them was 64 Mb and the other – 128 Mb. We noticed that they both were of the PC100 standard while those non-working modules were of the PC 133 standard.
So we decided to look for some modules with the lesser speed. But everything was in vain. The unique thing that we found, were some DIMMs of 128 Mb. At first we though to order them, but our hardware specialist said that it was possible to re-burn the work speed information in the SPD chip adjusted on the DIMM module. Besides he mentioned the instrument, which we had to use.
After the fast access retrieval in Runet we obtained a reference on the Guidance about the information re-burning in the SPD chip.
Then we were thinking about information to be burned in the chip. We found the following decision: after downloading some module documentation from the site of one working DIMM module I discovered well documented description of differences between the module burnings in different speed.
The bytes responsible for the speed are the following: 9, 10, 23, 24, 27, 28, 29, 30, 32, 33, 34, 35, 41
and the checksum is 63.
After that we took Kingston bilateral modules with eight chips from every side and downloaded the existent burning from them. Then we modified and put it back. Then we checked up work stability and burning accuracy with CTSPD program on the computer. And after that we began to check up in the switch-fabric. Everything was successful and at present the modules work well.
I would like to warn those people, who has two switch-fabrics. NORTEL recommends using the same switch-fabrics in chassis with the same memory volume. I want to add that it is not in vain. In the case of different volumes the switching between switch-fabrics is not correct.
Should any questions arise, please do not hesitate to contact me.
P.S. As the article about SPD information burning is written in Russian, so you have to know the used instrument.
I was interested in the idea of the memory upgrading with existent DIMM modules in the computer. And that is why I decided to try. I didn’t do it alone and further I will use “we”.
So, at first we decided to do everything in haste. First of all we took some DIMM modules of 256 Mb and put them into the switch-fabric.
The result was very bad. The boot loader was defining correctly the memory volume, but after module initialization in the slots the basic code began to reboot. We began to investigate the problem. On account of some reasons we had two types of modules that worked correctly in the switch-fabric. One of them was 64 Mb and the other – 128 Mb. We noticed that they both were of the PC100 standard while those non-working modules were of the PC 133 standard.
So we decided to look for some modules with the lesser speed. But everything was in vain. The unique thing that we found, were some DIMMs of 128 Mb. At first we though to order them, but our hardware specialist said that it was possible to re-burn the work speed information in the SPD chip adjusted on the DIMM module. Besides he mentioned the instrument, which we had to use.
After the fast access retrieval in Runet we obtained a reference on the Guidance about the information re-burning in the SPD chip.
Then we were thinking about information to be burned in the chip. We found the following decision: after downloading some module documentation from the site of one working DIMM module I discovered well documented description of differences between the module burnings in different speed.
The bytes responsible for the speed are the following: 9, 10, 23, 24, 27, 28, 29, 30, 32, 33, 34, 35, 41
and the checksum is 63.
After that we took Kingston bilateral modules with eight chips from every side and downloaded the existent burning from them. Then we modified and put it back. Then we checked up work stability and burning accuracy with CTSPD program on the computer. And after that we began to check up in the switch-fabric. Everything was successful and at present the modules work well.
I would like to warn those people, who has two switch-fabrics. NORTEL recommends using the same switch-fabrics in chassis with the same memory volume. I want to add that it is not in vain. In the case of different volumes the switching between switch-fabrics is not correct.
Should any questions arise, please do not hesitate to contact me.
P.S. As the article about SPD information burning is written in Russian, so you have to know the used instrument.