Crucial Ballistix DDR3 PC3-12800 Review

I’ve always been a bit of a fan of Crucial memory, so when the opportunity came to test out the latest addition to their Ballistix line, I jumped at the chance.
Crucial Technologies is a division of Micron and use Micron chips exclusively on their Ballistix, high performance line of memory. As some of you may or may not know, Micron memory IC’s are highly regarded by many people as the best overclocking DDR2 chips available. They can be found on most, if not all, of the higher end (higher binned) 2GB memory kits from all the top manufacturers such as Corsair, OCZ, G.Skill, Kingston, Cellshock and TeamGroup. With such a reputation for high performance, I was certainly looking forward to seeing if Micron and Crucial could continue this trend with DDR3. Let’s begin with a closer look at the specifications and the modules themselves…

Product: Crucial Ballistix DDR3 PC3-12800
Author: Paul Mercer
Category: Enthusiast/Overclocker
Date: November 2007
Product cost: MSRP $569.99 US
Manufacturer: Crucial
Spelling and Grammatical Editor: PlanetX64 Staff

{mospagebreak title=Features and Packaging}

FEATURES

• 240-pin, unbuffered dual in-line memory module (UDIMM)
• Physical dimensions are compliant to JEDEC MO-269B
• Fast data transfer rates: PC3-12800
• 1GB (128 Meg x 64)
• Exclusively designed for high-performance systems
• Critical performance parameters tested for functionality
• Speed verification performed in-system
• VDD = VDDQ = +1.8V ±0.075V
• VDDSPD = +3.0V to +3.6V
• Reset pin for improved system stability
• Nominal and dynamic on-die termination (ODT) for data, strobe, and mask signals
• 8 internal device banks for concurrent operation
• Fixed burst length of 8 (BL8) and burst chop of 4 (BC4) via the mode register
• Adjustable data-output drive strength
• Serial Presence-Detect (SPD) EEPROM
• Gold edge contacts
• RoHS compliant
• Fly-by topology
• Terminated command, address, and control bus
• Aluminium heat spreader
• 1.18” (30.00mm) PCB Height

TIMING PARAMETERS

* Performance is verified during testing 

PART NUMBERS

Crucial® Ballistix™ products are warranted to meet the datasheet specifications as found on its web page www.crucial.com/ballistix.
Some system configurations may not be designed to operate or may not operate at the published Crucial Ballistix memory speed and timing settings. Overclocking, running your system faster than the speed for which it was designed or the published speed, or otherwise modifying your system timing may result in damage to computer components and Crucial disclaims any and all liability for such damage.

©2007 Micron Technology, Inc. All rights reserved. Information is subject to change without notice. Micron, Crucial, the Crucial logo, Ballistix, and the Ballistix logo are trademarks of Micron Technology, Inc. All other trademarks are the property of their respective owners.

As you can see, the Crucial Ballistix DDR3 arrived in the usual fancy memory packaging  from Crucial. This consists of a cardboard box, a simple piece of cardboard used as a separator and antistatic bags (one per module). This method of shipping may look a little bland, but it is quite eco-friendly and incredibly secure as I’ve never received any damaged modules from Crucial and I’m glad to see its continued use.

The included leaflet is rather simplistic but it details instructions on how to install your memory modules correctly, some minor troubleshooting tips, contact details for technical support & customer service and a copy of the “Terms and Conditions of Sale” including warranty details.

Opening up the anti-static bags released that familiar smell of new memory again (mmmm…) and also reveals the new snazzy heatspreader

Ain’t they purdy?

The PCBs used on the DDR3 ballistix are still made by Levin.

Density: 1Gb
Part Status: Sampling
RohS: Yes
Depth: 128Mb
Width: x8
Voltage: 1.5V
Package: FBGA
Pin Count: 86-ball
Clock Rate: 533 MHz
Cycle Time: 1.875ns
Op. Temp.: 0C to +85C
CL: CL = 7
Data Rate: DDR3-1066

These chips, despite having a manufacturer rating of DDR3-1066MHz, CAS7, can be found on higher specced memory such as the CAS 7, DDR3-1800 kits from Corsair. Because of this, even though at DDR3-1600, they’re already overclocked way over the stock speed, I’m still expecting a little extra room to play with.

More details on XMP can be found here:
http://www.intel.com/personal/gaming/extremememory.htm 

{mospagebreak title=Performance}

As per usual, a clean installation of Windows was performed before benchmarking began, the latest drivers installed and no tweaks applied. As well as the benchmarking software, a number of applications were also installed to simulate a new system owned by an average user and then drives were defragmented.
Benchmark tests were run multiple times and the average or middle score recorded.

Software installed included the following:
NOD32 Antivirus
Microsoft Office 2007
Diskeeper 2008 (Trial)
Adobe Acrobat Reader
Cyberlink PowerDVD
Firefox

I dislike seeing memory reviews, claiming that because the memory doesn’t crash the PC when running  benchmarks or a couple of games, that the kit is stable.
I came across a review recently in which, despite the memory’s inability to complete a simple benchmark such as Everest or Sandra, it was still called stable because it passed an undisclosed number of hours of Orthos and Super PI. The reviewer also tried telling the readers that it was not the memory and somehow the fault of the benchmarking software used not being able to cope with the extended speeds of DDR3 above DDR3-1900.

I could do the same thing and throw up all kinds of screenshots and benchmarks. But for the purposes of a product review, I think that it is misleading to you the readers and gives unfair expectations to potential customers who will be disappointed when they do not reach similar levels of performance. I would rather show slightly slower, more stable and untweaked results to give a more realistic picture that can very likely be improved upon.

To show at least a small modicum of stability, final speeds were checked for 5hrs with HCI Memtest as well as having to successfully complete the ALL of the review benchmarks. If any tests fail, the setting is considered unstable and different settings are found/tested. I still would not call my results fully stable though as I personally run HCI Memtest for 24-48hrs on my own systems before use as well as running various other stress tests. Complete stability testing would take much longer, but due to time constraints, this obviously had to be cut short.

In another bid to reduce the time spent testing, the secondary latency timings of the Crucial Ballistix DDR3 were set to “Auto” for the duration of the review, only the primary timings and performance levels (Transaction Booster) were manually configured (Performance Level was configured to PL7 for all tests). Write to Precharge Delay and Write to Read Delay were set to disabled, and the BIOS option for Command Rate was set to 1T.

Because of the way current DDR3 motherboards work, the highest memory ratio currently available is 1:2. This means that in order to run the Crucial Ballistix at their rated speed of DDR3-1600, the Front Side Bus (FSB) will have to be overclocked to at least 400MHz.

I began testing by checking the memory for the lowest Voltage useable while maintaining stability at stock speeds and timings. Despite the specifications of DDR3-1600 with 1.8V, this kit was perfectly happy to run at stock speeds and timings with just 1.65V set in the BIOS.

You may notice in a couple of the memtest screenshots, that the performance level is set to level 6. This was a small mistake and in all recorded benchmarks this was set to level 7.

Stock Speed Tests – DDR3-1600

Everest:
 

Sandra: 

Memory Bandwidth:

Everest: 

Sandra: 

Memory Bandwidth:

 
Dual:

DivX Encode:

{mospagebreak title=Overclock}

Keeping the CPU speed constant at the various memory speeds proved impossible with the E6750 processor used in the review, as the x8 multiplier was the limiting factor. With a x9 multiplier, it would have been possible to test at 9×400 and 9×450, keeping the CPU at 3600MHz. Even this would not strictly be comparable though, because as the FSB is increased so is the available bandwidth… Which, although unavoidable with the standard memory dividers, also affects the test results.

Unfortunately, there has to be some kind of limit to the amount of voltage you can safely pump into memory (or any chip) before it starts to error or degrade quickly. The absolute maximum voltage allowable for Micron DDR3 is a relative unknown at this stage, as people have not been running these chips for any significant length of time to gather conclusive data yet. Although I have seen people using more than 2.0V for benchmarks and claiming stability with a single or dual Super PI 32M, I decided from the outset that I would not use more than 2.0V at this time. Mainly because I didn’t want to run the risk of killing the kit before the review was finished, but also because I would not feel comfortable running any more than that 24/7.

Raising the FSB to 450MHz while still keeping the 1:2 Memory ratio pushes the kit to 1800MHz, far beyond the capabilities of DDR2. With 1.80V set in the BIOS for memory and 8-7-6-18 timings, the kit happily passed an HCI Memtest session and all review benchmarks.

450×7(3150MHz), DDR3-1800, 8-7-6-18, 1.80Vmem, 1.55VNB, 266 Strap, PL7

Everest: 

Memory Bandwidth:

 
Dual:

DivX Encode: