iXBT Labs - Computer Hardware in Detail






DDR Memory Module Analysis. Part 5: Corsair TWINX2048-3500LLPRO Modules

December 6, 2005

We proceed with the low-level analysis of the most important characteristics of DDR memory modules (which haven't got due attention for a long time already) using our RightMark Memory Analyzer. ASUS and Corsair have recently launched together their products — A8N32-SLI DELUXE/WIFI and A8N32-SLI DELUXE motherboards and Corsair TWINX2048-3500LLPRO memory modules — the first DDR-400 modules of large capacity (1 GB each), offering very low latencies (2-3-2-6, 1T command rate). These modules will become an object of our today's review.

Manufacturer Information

Module manufacturer: Corsair Memory
Manufacturer of module chips: unknown
Web site of the module manufacturer: http://www.corsairmemory.com/corsair/xms.html

Module Exterior

Photo of the memory module

The exterior of TWINX2048-3500LLPRO modules is quite usual for XMS PRO series, which was reviewed previously by the example of DDR2 modules of the XMS2 PRO series — CM2X512-4300C3PRO. Distinctive features of this series — massive aluminum heatsink (our modules have a black heatsink) and LEDs — activity indicators (these modules use 18 LEDs, while XMS2 PRO series modules have only 24), you can make them out on the photo below (end view). I remind you that these LEDs do not display activity of individual logical banks, but the overall load of the memory bus in a module — from minimal, indicated by the first three couples of green LEDs, to medium (three couples of yellow LEDs in the middle) and up to maximum (the last three couples of red LEDs).

Module Part Number

The manufacturer's web site does not publish the DDR Part Number expansion of XMS PRO series memory modules. Datasheet on these modules states that this product is a complex of two DDR-400 modules, adjusted to each other, with the total capacity of 2GB. These modules can operate at up to 218 MHz (DDR2-437 mode) and extremely low (for modules of such a large capacity) timings 2-3-2-6-1T. According to the manufacturer, these very timings are written in SPD. There is nothing said about the voltage, so let's assume that these modules can work in these conditions even with the standard voltage (2.5 V).

SPD chip data

Description of the general SPD standard:
JEDEC Standard No. 21-C, 4.1.2 - SERIAL PRESENCE DETECT STANDARD, General Standard

Description of the specific SPD standard for DDR:
JEDEC Standard No. 21-C, — Appendix D, Rev. 1.0: SPD’s for DDR SDRAM

Parameter Bytes Value Expansion
Fundamental Memory Type 2 07h DDR SDRAM
Number of Row Addresses on this assembly 3 0Dh 13 (RA0-RA12)
Number of Column Addresses on this assembly 4 0Bh 11 (CA0-CA10)
Number of DIMM Banks 5 02h 2 physical banks
Data Width of this assembly 6, 7 40h, 00h 64 bit
Voltage Interface Level of this assembly 8 04h SSTL 2.5V
SDRAM Cycle time (tCK) at maximum supported CAS# latency (CL X) 9 50h 5.00 ns (200.0 MHz)
DIMM configuration type 11 00h Non-ECC
Refresh Rate/Type 12 82h 7.8125 ms — 0.5x reduced self-refresh
Primary SDRAM Width (organization type) of the memory module chips 13 08h x8
Error Checking SDRAM Width (organization type) of the memory chips in the ECC module 14 00h Not defined
Burst Lengths Supported (BL) 16 0Eh BL = 2, 4, 8
Number of Banks on SDRAM Device 17 04h 4
CAS Latency (CL) 18 04h CL = 2.0
Minimum clock cycle (tCK) at reduced CAS# latency (CL X-0.5) 23 00h Not defined
Minimum clock cycle (tCK) at reduced CAS# latency (CL X-1.0) 25 00h Not defined
Minimum Row Precharge Time (tRP) 27 28h 10.0 ns
2, CL = 2.0
Minimum Row Active to Row Active delay (tRRD) 28 28h 10.0 ns
2, CL = 2.0
Minimum RAS to CAS delay (tRCD) 29 3Ch 15.0 ns
3, CL = 2.0
Minimum Active to Precharge Time (tRAS) 30 1Eh 30.0 ns
6, CL = 2.0
Module Bank Density 31 80h 512 MB
SDRAM Device Minimum Active to Active/Auto Refresh Time (tRC) 41 37h 55.0 ns
11, CL = 2.0
SDRAM Device Minimum Auto-Refresh to Active/Auto-Refresh Command Period (tRFC) 42 46h 65.0 ns
13, CL = 2.0
Maximum device cycle time (tCKmax) 43 30h 12.0 ns
SPD Revision 62 00h Revision 0.0
Checksum for Bytes 0-62 63 ABh 171 (true)
Manufacturer’s JEDEC ID Code (only the first significant bytes are shown) 64-71 7Fh, 7Fh,
Module Part Number 73-90 CMX1024-3500LLPRO
Module Manufacturing Date 93-94 00h, 00h Not defined
Module Serial Number 95-98 00h, 00h,
00h, 00h
Not defined

SPD contents look acceptable, if not standard. Only one CAS# latency is supported - 2, which is not surprising for top memory modules — such modules will hardly need compatibility with "older" modes like DDR-333. So it's only natural that the main and only tCL = 2 corresponds to the cycle time of 5 ns, that is DDR-400 mode. The timings scheme strictly matches the specification value — 2-3-2-6, which is a nice surprise, as it theoretically should guarantee that BIOS versions in the majority of motherboards would set these very timings by default. Peculiarities of the SPD contents in the modules under review include the not-defined SPD revision "0.0" as well as the lack of information on the manufacturing date and the serial number. Considering our previous reviews of Corsair memory modules, the last two peculiarities seem to pertain to all modules from this manufacturer — both DDR as well as DDR2.

Testbed Configurations and Software

Testbed 1

  • CPU: AMD Athlon 64 4000+, 2.4 GHz (ClawHammer rev. SH-CG, 1 MB L2)
  • Chipset: NVIDIA nForce4 SLI X16
  • Motherboard: ASUS A8N32SLI Deluxe, BIOS 0502 dated 10/06/2005
  • Memory: 2x1024 MB Corsair XMS PRO PC3500, "DDR-437" (in DDR-400 mode)
  • Video: Leadtek PX350 TDH, NVIDIA PCX5900
  • HDD: WD Raptor WD360, SATA, 10000 rpm, 36Gb
  • Drivers: NVIDIA Forceware 77.72, NVIDIA nForce4 X16 6.82, DirectX 9.0c

Testbed 2

  • CPU: AMD Athlon 64 4000+, 2.4 GHz (ClawHammer rev. SH-CG, 1 MB L2)
  • Chipset: NVIDIA nForce4 SLI X16
  • Motherboard: ASUS A8N32SLI Deluxe, BIOS 0502 dated 10/06/2005
  • Memory: 2x512 MB Corsair XMS PC3200, DDR-400
  • Video: Leadtek PX350 TDH, NVIDIA PCX5900
  • HDD: WD Raptor WD360, SATA, 10000 rpm, 36Gb
  • Drivers: NVIDIA Forceware 77.72, NVIDIA nForce4 X16 6.82, DirectX 9.0c

Test Results

Performance tests

As these Corsair memory modules were manufactured specially for the new ASUS A8N32SLI Deluxe motherboard (labels on the boxes with the memory modules as well as with the motherboard), we logically decided to test these modules on this very motherboard (Testbed 1).

The modules were tested in the standard mode — DDR-400. For comparison purposes, we performed the same tests with a couple of 512 MB Corsair DDR-400 modules (used in our test lab for a long time already) with extremely low timings 2-2-2-5 (Testbed 2).

Parameter Testbed 1 Testbed 2
Average memory read bandwidth, MB/sec
Average memory write bandwidth, MB/sec
Max. memory read bandwidth, MB/sec
Max. memory write bandwidth, MB/sec
Minimum Pseudo-Random Access Latency, ns
Maximum Pseudo-Random Access Latency, ns
Minimum Random Access Latency*, ns
Maximum Random Access Latency*, ns

*16 MB block size

As usual, memory timings in the first series of tests were set by default (in BIOS Setup — Memory Timings: "by SPD"). In both cases ASUS A8N32SLI Deluxe motherboard set these values correctly, that is strictly according to SPD. What's important, in both cases BIOS set Command Rate to 1T (a faster mode), which can be supported by these modules, according to the manufacturer.

The new 1 GB Corsair modules are expectedly almost no worse in memory bandwidth than much older 512 MB modules with a tad lower timings — in all the four cases (average/maximum memory read/write bandwidth) the spread is so small that it can be written off to the measurement error. Strange as it may seem, 1 GB modules do not differ in pseudo-random latencies either. Real differences, against the slightly slower (to be more exact, offering a tad higher latencies) 1 GB modules, can be seen only in random access latency. The difference is about 8%.

Stability tests

Timing values were adjusted "on the fly" due to the built-in RMMA feature that allows to change dynamically memory settings supported by the chipset (in this case - by the memory controller integrated into Athlon 64). Memory operating stability was evaluated with an auxiliary utility RightMark Memory Stability Test, included into RMMA.

Parameter Testbed 1 Testbed 2
Average memory read bandwidth, MB/sec
Average memory write bandwidth, MB/sec
Max. memory read bandwidth, MB/sec
Max. memory write bandwidth, MB/sec
Minimum Pseudo-Random Access Latency, ns
Maximum Pseudo-Random Access Latency, ns
Minimum Random Access Latency*, ns
Maximum Random Access Latency*, ns

*16 MB block size

The minimal timings scheme, allowed by the new 1 GB Corsair modules, differs by one in tRAS value (note — it's quite a disputable difference, because the majority of modules are indifferent to its changes, as we know from our multiple previous tests). Yep, unfortunately, we didn't manage to squeeze the 2-2-2 scheme, usual for 512 MB modules. To be more exact, it immediately resulted in errors (quite crude, on the level of losing OS stability), even if we tried to raise memory voltage to 2.8 V inclusive. Not to make the timing changes look so insignificant in this series of tests (in the second case — no changes at all), we decided to expand the scheme by adding two extra parameters — SDRAM Device Minimum Active to Active/Auto Refresh Time (tRC) and Minimum Refresh to Active/Refresh Command Period (tRFC). Moreover, we did it taking into account that changing these parameters would result in changing memory system operation (for example, reducing tRFC to minimum = 9 in 1 GB modules resulted in immediate freeze of the system; in case of 512 MB modules, it resulted in hard-to-detect errors :)).

The results of this low timing overclocking are expectedly low as well — we can skeptically speak of a slightly increased memory bandwidth, but the most reliable change reveals itself as a very insignificant reduction of random access latencies in both cases, by about 0.3-0.4 ns, that is less than 1%.

Bottom line

Corsair TWINX2048-3500LLPRO memory modules can be considered the first high-performance DDR-400 modules of a large capacity, which can operate with low latencies — the fast 2-3-2-6 timings scheme. Timing overclocking potential of these modules is practically out of the question — that's seems to be the "minimal minimum", currently attainable for memory module chips of such high density. So these modules are definitely a success.

Corsair TWINX2048-3500LLPRO modules are kindly provided
by Corsair Memory representative office in Russia

Dmitri Besedin (dmitri_b@ixbt.com)
November 26, 2005.

Write a comment below. No registration needed!

Article navigation:

blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook

Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.