Logo   Banner   TopRight
TopUnder
Transparent
Corsair TWINX1024-3200XLPRO - Low Latency DDR400 RAM
Transparent
Date: Jul 22, 2004
Section:Misc
Author: Marco Chiappetta
Transparent
Introduction, Specifications & Gallery

Corsair's XMS (Xtreme Memory Speed) memory modules have long been a favorite amongst the enthusiast community. Corsair has memory kits available in many different sizes and speeds, and all of them are regarded as some of the best modules available in their respective categories.  Unfortunately, with the demise of Winbond's BH5/BH6 chips, low-latency memory kits from Corsair (and other manufactures) virtually disappeared.  Low-latency memory offers better overall performance than similarly clocked memory with higher latencies.  Chips capable of much higher clock speeds quickly surfaced, but true low-latency memory was very tough to come by - until now.  Today on HotHardware, we'll be shining the spotlight on Corsair's new TWINX1024-3200XLPRO kit (1GB, 2x512MB CMX512-3200XLPRO DIMMs).  These double-sided PC3200 DIMMs sport new Samsung chips that are capable of running with a CAS Latency of 2T, a RAS to CAS Delay (tRCD) of 2T, a Row Precharge (tRP) of 2T and an Active to Precharge Delay (tRAS) of 5T (2-2-2-5) at 200MHz (DDR400).  And with less aggressive timings, Corsair claims they'll even run at DDR500 speeds.

    

Specifications of the Corsair TWINX1024-3200XLPRO - Low Latency DDR400 RAM
Tight Timings are Back!
Features:
1024 Megabytes of memory
- Two CMX512-3200XLPRO memory DIMMs

High-efficiency aluminum XMS heatsink
- 95% greater surface area than standard memory heatspreaders

- Thermally bonded to the module

18 activity LED's show level of memory activity
- Benchmarked over multiple PC3200 based chipsets, processors, and motherboards

- Approximately 1.75" tall x 5.41" long x 0.30" wide; 2.9 oz

- Lifetime warranty
Corsair QA Test Specifications:
- Tested as a matched pair in multiple Intel and AMD PC3200 based test setups

- Packaged together immediately following system test

- Tested at DDR400

- Tested at X-treme Low latency settings (2-2-2-5)

- Test voltage: 2.75V

- Plug-n-Frag Auto-Configuration

- Boots automatically using the X-treme Low latency values stored in the SPD

    

  

The Corsair CMX512-3200XLPRO DIMMs are fairly unique, from a physical standpoint.  Aside from Corsair's other "PRO" branded products, no other memory modules feature the signature activity LEDs.  The 18 LEDs blink rapidly to show the level of memory activity, much like the sound meters found on some audio equipment.  Unless you have a windowed case the activity LEDs will go unseen, so make sure to feature them prominently if a set of these DIMMs is in your future.  The activity LEDs are a definite eye-catcher.

Aside from the activity LEDs, there isn't much to talk about besides the physical size of these modules.  As you can see, when stacked next to a traditional DIMM the CMX512-3200XLPROs are quite a bit taller.  This won't be an issue in most circumstances, but these sticks may not fit in some small form factor systems that have limited clearance.  Make sure you've got the room before you make a purchase.  The Samsung TCC4 chips that populate the PCBs are covered with large, finned heatsinks clearly labeled "XMS".  The small fins add a significant amount of surface area, when compared to the more commonly used flat heat spreaders, and should be more than capable of keeping the RAM cool provided your case is well ventilated.

Transparent
Compatibility & Our Test System

 

Compatibility Testing: Before we sat down in front of our test machines to benchmark Corsair's TWINX 1024-3200XLPRO memory modules, we installed them in some of the systems we had available at the time to see if there were any compatibility problems to speak of.  We tried the sticks in a total of 7 different systems, powered by various popular chipsets, and didn't encounter a single problem.  All of the systems booted into Windows and worked without a hitch.  We tested these modules on the following platforms...

With the exception of the Abit IC7-MAX3 and the Asus A8V Deluxe, which we used for the benchmarks in this article, to quickly test these modules all we did was install them, power up the systems, made sure Windows booted and we ran SiSoft SANDRA's burn-in wizard for a few minutes.  This limited compatibility testing should obviously not be considered gospel, but it does bode well for the TWINX 1024-3200XLPRO memory modules.  The fact that they worked properly on the popular platforms that we tested here, means most consumers should have good luck as well.

HotHardware's Test Systems
Not all are created equal...
SYSTEM 1:
S939 - AMD Athlon 64 3800+ (2.4GHz)


Asus A8V Deluxe Motherboard
VIA K8T800 Pro Chipset

Corsair TWINX 1024-3200XLPRO
Kingson HyperX PC3500
Generic PC3200

Radeon 9800 Pro
On-Board 10/100 Ethernet
On-Board Audio


WD "Raptor" 36GB Hard Drive
10,000 RPM SATA

Windows XP Pro SP1
ATi Catalyst 4.6
DirectX 9.0b
SYSTEM 2:
Intel P4 3.2GHz "Northwood"


Abit IC7 Max3
Intel 875P Chipset

Corsair TWINX 1024-3200XLPRO
Kingson HyperX PC3500
Generic PC3200

Radeon 9800 Pro
On-Board 10/100 Ethernet
On-Board Audio


WD "Raptor" 36GB Hard Drive
10,000 RPM SATA

Windows XP Pro SP1
ATi Catalyst 4.3
DirectX 9.0b

How we configured our test systems:  When configuring our test systems for this review, the first thing we did was enter the system BIOS and set each board to their "Optimized" or "High-Performance Defaults".  The hard drives were then formatted, and Windows XP Professional (SP1) was installed. When the installation was complete, we hit the Windows Update site and downloaded all of the available updates, with the exception of the ones related to Windows Messenger.  Then we installed all of the necessary drivers, and removed Windows Messenger from the system altogether.  Auto-Updating and System Restore were also disabled, and we setup a 768MB permanent page file on the same partition as the Windows installation.  Lastly, we set Windows XP's Visual Effects to "best performance", installed all of our benchmarking software, defragged the hard drives and ran all of the tests.  Throughout all our benchmarking, we had the memory voltage set to 2.8v.

Transparent
Performance @ 2-2-2-5

 

We began our testing with SiSoftware's SANDRA, the System ANalyzer, Diagnostic and Reporting Assistant. SANDRA consists of a set of information and diagnostic utilities that can provide a host of useful information about your hardware and operating system.  We ran SANDRA's Memory Bandwidth test on our Intel and AMD powered test beds with three different types of memory installed...

Performance Comparison with SiSoft SANDRA 2004
Raw Bandwidth

 

Just in case it's not clear by the description in our graphs, these tests were conducted with the each set of DIMMs (2x512MB) installed into our test beds, with their timings set by SPD.  The SPD (Serial Presence Detect) basically gives the motherboard's BIOS simple information about speeds, timings, manufacturer, etc.  In this configuration, the Corsair modules were running with 2-2-2-5 timings, the Kingston modules at 2.5-3-3-8, and the generic PC3200 modules were running at 3-3-3-8.  As you can see, even though the timings were quite different from module to module, performance in this test was quite similar.  Less than 100MB/s separate the first from the last place finishers.  The Corsair sticks were the fastest on both the Intel and AMD test systems, followed by the Kingston sticks.  The generic sticks that we picked up at a local computer show have no identifiable markings, but they put up some decent numbers, even though they finished last in this test.

Transparent
Performance @ 2-2-2-5 Continued

For our next round of benchmarks, we ran the Memory performance module built-into Futuremark's PCMark04. For those interested in more than just the graphs, we've got a quote from Futuremark that explains exactly what this test does and how it works...

Performance Comparison with PCMark04
Overall Memory Score

"The Memory test suite is a collection of tests that isolate the performance of the memory subsystem. The memory subsystem consists of various devices on the PC. This includes the main memory, the CPU internal cache (known as the L1 cache) and the external cache (known as the L2 cache). As it is difficult to find applications that only stress the memory, we explicitly developed a set of tests geared for this purpose. The tests are written in C++ and assembly. They include: Reading data blocks from memory, Writing data blocks to memory performing copy operations on data blocks, random access to data items and latency testing."

Like the SANDRA tests, all three sets of memory modules performed at similar levels in the PCMark04 memory performance module as well.  The performance delta was a bit more pronounced on the Intel test system (51 vs. 44 point spread), but for all intents and purposes these scores here are identical, and fall well within the margin of error in this test.

In-Game Performance Comparisons
System Memory Affects Framerates?  You Bet!

We continued our testing with Novalogic's combat helicopter simulation, Comanche 4.  Despite the fact that this is a game benchmark that can be used to test the relative performance of video cards, frame rates are strongly influenced by processor speed and available memory bandwidth, especially at low resolutions with sound disabled, which is how we ran the tests to get the frame rates listed below.

We didn't see any dramatic performance differences with Comanche 4 either.  The Corsair modules were once again the fastest of the bunch, but not by much.  On the Intel i875 test bed, the Corsair sticks were about 1.2% faster than the Kingston modules and on the AMD system they were about .4% faster.

Transparent
Performance @ Highest Stable Clock Speed

 

For our next set of numbers, we focused on the overclockability of Corsair's TWINX1024-3200XLPRO modules on our Intel test bed.  Using an unlocked Pentium 4 3.2GHz CPU, we raised the Front Side Bus speed while concurrently lowering our processor's multiplier.  We tried to keep the CPU as close to it's stock 3.2GHz clock speed as possible.  For these tests, we locked the AGP/PCI speeds at 66MHz/33MHz, raised the CPU voltage to 1.65v, and raised the memory voltage to 2.9v.

Overclocked Performance with SiSoft SANDRA 2004
Raw Bandwidth

These scores illustrate the highest stable clock speeds we were able to hit with the Corsair modules configured with 2-2-2-5 timings, and then again with 2.5-3-3-8 timings.  The "stock" scores are the DDR400 numbers with timings set to 2-2-2-5.  With the fastest timings, we were able to take these Corsair sticks all the way up to 220MHz (DDR440).  At DDR440, bandwidth scores in SANDRA jumped by about 9% in both the ALU and FPU tests.  With less aggressive timings, the Corsair sticks actually hit 247MHz (DDR494), which resulted in roughly a 19% increase in bandwidth.  With even less aggressive timings, DDR500+ speeds are a definite possibility with these modules.

Transparent
Performance @ Highest Stable Clock Speed Continued

We also re-ran the PCMark04 Memory performance and low-res Comanche 4 benchmarks with the Corsair TWINX1024-3200XLPRO memory overclocked to DDR440 (2-2-2-5 timings) and DDR494 (2.5-3-3-8)...

Overclocked Performance with PCMark04
Overall Memory Score

We saw performance increases similar to SANDRA's scores with PCMark04's Memory Performance module.  At DDR440, the 3200XL Pro's PCMark04 memory score jumped by about 7.3%.  The DDR494 scores were significantly higher, with an increase of over 650 points, or 13.4%.

In-Game Performance while Overclocked
System Memory Affects Frame rates?  You Bet!

Performance in Comanche 4 is affected more by CPU speed than memory bandwidth, but we did see some increased frame rates in this test, nonetheless.  We should note, however, that because we did not have the exact same CPU speed in these overclocked tests (3.2GHz vs. 3.3GHz. vs. 3.23GHz), the scores will be somewhat skewed.  In the end though, overclocking the RAM increased the frame rate slightly in both configurations.

Transparent
Conclusion

 

Benchmark Summary: While it's clear that having aggressive memory timings will increase system performance, the differences were not dramatic.  The biggest speed gains come with increased clock speeds.  Fortunately, Corsair's TWINX1024-3200XLPRO modules will also run at relatively high clock speeds when less aggressive memory timings are used.  With that said, the Corsair TWINX1024-3200XLPRO modules did outpace both of the other brands we tested in every benchmark we ran.

If you're in the market for some high-end RAM, Corsair's TWINX1024-3200XLPRO memory modules should definitely be on your "short list".  Corsair guarantees this memory will run with aggressive 2-2-2-5 timings at DDR400 speeds at 2.75v.  With a little bump in voltage, however, we found them to be perfectly stable with these timings at speeds as high as DDR440.  Raise the latency a bit, and DDR500+ speeds should be no problem as well.  Performance was good, and we experienced no compatibility or stability issues until we overclocked these modules way out of spec.  The activity LEDs are a bit of a novelty, but they'd be a cool addition to a windowed case and will surely draw some attention at your next LAN party.  About the only negative aspect to this RAM is the cost.  The TWINX1024-3200XLPRO kit is currently selling for $330+ dollars.  You can save about 10% and go with the non-Pro version of this RAM without activity LEDs, but even then they're a bit more expensive than most other high-end PC3200 RAM.  Then again, the versatility of these modules definitely makes them worth more than most other memory kits.  We're giving Corsair's TWINX1024-3200XLPRO 1GB memory kit a solid 9 on Heat Meter...

Get into HotHardware's PC Hardware Forum Right Now!



Content Property of HotHardware.com