Intel® X38 Express Chipset
Memory Technology and
Configuration Guide
White Paper
January 2008
Document Number: 318469-002
INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL
IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT
AS PROVIDED IN INTEL’S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY
WHATSOEVER, AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF INTEL
PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY,
OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT. Intel products are not intended
for use in medical, life saving , or life sustaining applications.
Intel may make changes to specifications and product descriptions at any time, without notice.
Designers must not rely on the absence or characteristics of any features or instructions marked "reserved" or "undefined." Intel
reserves these for future definition and shall have no responsibility whatsoever for conflicts or incompatibilities arising from
future changes to them.
The Intel
product to deviate from published specifications. Current characterized errata are available on request.
Contact your local Intel sales office or your distributor to obtain the latest specifications and before placing your product o rder.
Initial release
Corrected DDR3 DIMM Module Support
Added Configuration details
October 2007
January 2008
§
4 White Paper
Introduction
1 Introduction
This document details the Intel® X38 Express Chipset system memory technology
enhancements, supported memory configurations, and memory organizations. It is
intended for a technical audience interested in learning about the performance
enhancements and simplified population rules offered by Intel
and Intel
Chipset.
®
Flex Memory Technology in the platforms based on the Intel X38 Express
®
Fast Memory Access
White Paper5
Technology Enhancements of Intel® Fast Memory Access (Intel® FMA)
2 Technology Enhancements of
®
Intel
(Intel
Fast Memory Access
®
FMA)
This chapter details Intel® X38 Express Chipset technology enhancements of its
memory controller known as Intel
controller is located on the chipset’s 82X38 Memory Controller Hub (MCH) component.
With the growing reliance on faster and less latent memory technologies for today’s
high performance platforms, it has become necessary to not only increase system
memory transfer rate speeds, but to also streamline usage of the memory controller
protocol in novel and intelligent ways to decrease latency and optimize memory
bandwidth. To do this, several Intel technologies, known collectively as Intel
have been included in this generation of Intel’s chipsets.
The following sections outline and explain the technology enhancements: Just In Time
Scheduling, Command Overlap, Out of Order Scheduling, and Opportunistic Writes.
®
Fast Memory Access (Intel® FMA). The memory
2.1 Just in Time Command Scheduling
The Intel X38 Express Chipset has an advanced command scheduler where all pending
requests are examined simultaneously to determine the most efficient request to be
issued next. The most efficient request is picked from all pending requests and issued
to system memory Just In Time to make optimal use of Command Overlapping. Thus,
instead of having all memory access requests go individually through an arbitration
mechanism forcing requests to be executed one at a time, they can be started without
interfering with the current request allowing for concurrent issuing of requests. This
allows for the optimization of bandwidth and reducing of latency while retaining
system memory protocol.
®
FMA,
2.2 Command Overlap
Command Overlap allows for the insertion of the DRAM commands between the
Activate, Precharge, and Read/Write commands normally used, as long as the inserted
commands do not affect the currently executing command. This allows for situations
where multiple commands can be issued in an overlapping manner, increasing the
efficiency of system memory protocol.
6 White Paper
Technology Enhancements of Intel® Fast Memory Access (Intel® FMA)
2.3 Out of Order Scheduling
Leveraging Just In Time Scheduling and Command Overlap, the Intel X38 Express
Chipset continuously monitors pending requests to system memory for the best use of
bandwidth and reduction of latency. If there are multiple requests to the same open
page, these requests would be launched in a back to back manner to make optimum
use of the open memory page. This ability to reorder requests on the fly allows the
Intel X38 Express Chipset to further reduce latency and increase bandwidth efficiency.
This is especially important for helping overcome the in-order manner of the Front
Side Bus between the Intel X38 Express Chipset and the processor to minimize
processor starvation.
2.4 Opportunistic Writes
Processor requests for memory reads usually are weighted more heavily than writes to
memory to avoid cases of starving the processor of data to process while the writes
are issued to system memory. Instead of having writes issued to a pending queue to
be flushed to memory when certain watermarks are reached, which could starve the
processor of data while it waits for the write flush to finish, the Intel X38 Express
Chipset monitors system memory requests and issues pending write requests to
memory at times when they will not impact memory read requests. This allows for an
almost continuous flow of data to the processor for processing.
White Paper7
Supported Memory Technologies and Configurations
3 Supported Memory
Technologies and
Configurations
3.1 Memory Technology Supported
The X38 Express Chipset supports the following DDR2 and DDR3 Data Transfer Rates,
DIMM Modules, and DRAM Device Technologies:
• DDR2 Data Transfer Rates:
⎯ 667 (PC2-5300) and 800 (PC2-6400)
• DDR3 Data Transfer Rates:
⎯ 800 (PC3-6400), 1067 (PC3-8500), and 1333 (PC3-10600)
• DDR2 DIMM Modules:
⎯ Raw Card C - Single Sided x16 un-buffered non-ECC
⎯ Raw Card D - Single Sided x8 un-buffered non-ECC
⎯ Raw Card E - Double Sided x8 un-buffered non-ECC
⎯ Raw Card F - Single Sided x8 un-buffered ECC
⎯ Raw Card G - Double Sided x8 un-buffered ECC
• DDR3 DIMM Modules:
⎯ Raw Card A - Single Sided x8 un-buffered non-ECC
⎯ Raw Card B - Double Sided x8 un-buffered non-ECC
⎯ Raw Card C - Single Sided x16 un-buffered non-ECC
• DDR2 and DDR3 DRAM Device Technology:
⎯ 512 Mb and 1 Gb
For DDR3 the X38 Express Chipset does NOT support ECC, does not support ECC unbuffered DIMMs, and it does not support any memory configuration that mixes nonECC with ECC un-buffered DIMMs.
For DDR2 the X38 Express Chipset does support ECC and ECC un-buffered DIMMs but
it does NOT support any memory configuration that mixes non-ECC with ECC unbuffered DIMMs.
Rate
tCL t
t
RCD
WL Units
RP
See Section
10 White Paper
3.1 for un-buffered DIMM support details.
Supported Memory Technologies and Configurations
3.4 Valid Front Side Bus and Memory Speeds
The X38 Express Chipset supports the following Front Side Bus (FSB) and system
memory speed configurations.
Note: The X38 Express Chipset does not support system memory frequencies that exceed
the frequency of the Front Side Bus. If memory with higher frequency capabilities than
that of the FSB is populated, the memory will be under-clocked to align with the FSB.
3.5 System Memory DIMM Configuration Support
The X38 Express Chipset directly supports one or two channels of DDR2 or DDR3
memory with the following DIMM configurations:
⎯ Supports one or two DDR2 or DDR3 DIMM modules per channel.
White Paper11
§
Memory Organization and Operating Modes
4 Memory Organization and
Operating Modes
The Intel X38 Express Chipset memory interface is designed with Intel® Flex Memory
Technology so that it can be can be configured to support single-channel or dualchannel DDR2 or DDR3 memory configurations. Depending upon how the DIMMs are
populated in each memory channel, a number of different configurations can exist for
DDR2 or DDR3.
The following sections explain and show the different memory configurations that are
supported by the X38 Express Chipset.
4.1 Single-Channel Mode
In this mode, all memory cycles are directed to a single channel.
Single channel mode is used when either Channel-0 or Channel-1 DIMMs are
populated in any order, but not both.
4.2 Dual Channel Modes
4.2.1 Dual Channel Symmetric Mode
This mode provides maximum performance on real applications. Addresses are pingponged between the channels after each cache line (64 byte boundary). If there are
two requests, and the second request is to an address on the opposite channel from
the first, that request can be sent before data from the first request has returned. If
two consecutive cache lines are requested, both may be retrieved simultaneously,
since they are ensured to be on opposite channels.
Dual channel symmetric mode is used when both Channel-0 and Channel-1 DIMMs are
populated in any order with the total amount of memory in each channel being the
same, but the DRAM device technology and width may vary from one channel to the
other.
Table 4-1 is a sample dual channel symmetric memory configuration showing the rank
organization.
In this addressing mode addresses start in channel-0 and stay there until the end of
the highest rank in channel-0, and then addresses continue from the bottom of
channel-1 to the top.
This mode is used when both Channel-0 and Channel-1 DIMMs are populated in any
order with the total amount of memory in each channel being different.
Table 4-2 is a sample dual channel stacked asymmetric memory configuration
showing the rank organization.
In this addressing mode the lowest DRAM memory is mapped to dual channel
operation and the top most DRAM memory is mapped to single channel operation. In
this mode the system can run at one zone of dual channel mode and one zone of
single channel mode simultaneously across the whole memory array.
This mode is used when both Channel-0 and Channel-1 DIMMs are populated in any
order with the total amount of memory in each channel being different.
Table 4-3 is a sample dual channel L-shaped asymmetric memory configuration
showing the rank organization.