Dell PowerEdge 2650
Memory Technology
The PowerEdge 2650 supports up to six PC200 DDR SDRAM DIMMs in 128MB, 256MB, 512MB, and 1GB capacities for a total system memory capacity of 6GB. The memory interface runs a 200MHz double data rate, meaning that two signals are transmitted on each clock cycle to provide the equivalent of a 400MHz interface. |
|
 |
DDR Memory
DDR (Double Data Rate) is the next generation SDRAM standard, using a double data rate clocking technique to push its peak burst bandwidth to 1.6 GB/s (as compared to 1 GB/s for PC133 SDRAM). This performance advantage will immediately be appreciated in high-end graphics applications and other memory intensive deployments.
The manufacturing process for DDR SDRAM is very similar to that of SDRAM, except for higher-speed testers, and therefore does not require new equipment, training, or validation. This means that the price for DDR-SDRAM will generally be very close to its SDRAM predecessor.
Interleaving
The memory system supports 2-way interleaved memory configurations to reduce latency compared to non-interleaved implementations. The 6 DIMM sockets are divided into three memory banks. To support memory interleaving, the memory capacity must be identical within a given bank. However, the capacities for two separate banks can be different. The following chart relates the DIMM Socket ID to the memory banks.
 |
|
NOTE: Non-interleaved mode is supported when a single DIMM is populated in Socket DDR0. This means the system will complete POST and boot to an OS if only one memory module is installed in the first slot in the first bank. |
 |
DIMM Position |
Socket |
Logical Row |
Silkscreen Label |
 |
|
|
Right DIMM |
DDR0 |
Row 0 |
Bank 1 DIMM A |
|
|
|
|
DDR1 |
Row 0 |
Bank 1 DIMM B |
|
|
|
|
DDR2 |
Row 1 |
Bank 2 DIMM A |
|
|
|
|
DDR3 |
Row 1 |
Bank 2 DIMM B |
|
|
|
|
DDR4 |
Row 2 |
Bank 3 DIMM A |
|
|
|
Left DIMM |
DDR5 |
Row 2 |
Bank 3 DIMM B |
|
|
Memory Module Installation Guidelines
Starting with the connector nearest the power supplies, the memory module connectors are labeled "DIMM A1" through "DIMM C2". When installing memory modules, follow these guidelines:
- You must install memory modules in matched pairs if you want to use memory interleaving.
- The system will work with only one memory module installed in Slot DDR0, but it will run in non-interleaved mode.
- Install a pair of memory modules in connector DIMM A1 and A2 before installing a second pair in connectors B1 and B2, and so on.
 |
Total Memory |
DIMM A |
DIMM B |
DIMM C |
 |
|
|
256 MB |
128 MB |
128 MB |
None |
None |
None |
None |
|
|
|
512 MB |
256 MB |
256 MB |
None |
None |
None |
None |
|
|
|
1 GB |
512 MB |
512 MB |
None |
None |
None |
None |
|
|
|
2 GB |
512 MB |
512 MB |
512 MB |
512 MB |
None |
None |
|
|
|
2 GB |
1 GB |
1 GB |
None |
None |
None |
None |
|
|
|
3 GB |
1 GB |
1 GB |
512 MB |
512 MB |
None |
None |
|
|
|
6 GB |
1 GB |
1 GB |
1 GB |
1 GB |
1 GB |
1 GB |
|
|
Redundant Memory Features
The PowerEdge 2650 supports the following Redundant Memory Features.
ChipKill - The ServerWorks GC-LE chipset supports ChipKill on x4 (4-bit) DRAM memory modules, so for the PE 2650 only the 512MB and 1GB DIMMs provide ChipKill support. ChipKill support enables the system to avoid uncorrectable data errors if a DRAM device fails. This is accomplished by spreading 2 144-bit ECC works across 4 DIMMs (32 bytes). Each ECC work has the capability of correcting any single bit error on any four adjacent bits.
Spare Bank Memory - This feature is supported if the system is populated with either 2 or 3 banks of identical DIMMs and the feature has been enabled through BIOS setup. When enabled, one of the memory banks is designated as the 'Spare Bank', meaning the two DIMMs are not actively used by the system. If several (correctable) single-bit errors are generated from a memory bank in a short period of time, then the BIOS will initiate a copy of all data from the faulty DIMM bank to the spare bank. Spare Bank support prevents a DIMM that is creating multiple single bit errors from continuing to degrade into a fatal multi-bit error that could bring the system down. By using this feature, you can continue to use the server until a scheduled downtime can be instituted to replace the failed DIMMs instead of experiencing an unplanned server outage.
|