- Back to homepage
- Related articles
  Related articles

Lights out Performance! Crucial Tracer Memory

Memory by JNav89GT @ 2005-03-17

Do you want your cake and to eat it too? Do you have a need for speed and bling? Well step inside as we?ll skewer a few sticks of Crucial Ballistix Tracer DDR and DDR2 over our crustacean Barbie and see if these memory sticks are true speed demons, or posers with Type R badging.

Test Setups & Methodology

Test Setups

For DDR2:

JNav89GT's Intel Test Setup
CPU	Intel Pentium4 550 3.4Ghz LGA 775 CPU-Retail
Mainboard	Asus P5AD2-E Premium
Video	ATI Radeon X700Pro
Cooling	AlphaCool NeXXos Bold
Power Supply	Zippy HP2-6460P
Operating System	Windows XP SP2

For DDR1:

JNav89GT's AMD Test Setup
CPU	AMD Athlon FX55 S939
Mainboard	MSI K8N Neo4 Platinum
Video	ATI Radeon X700Pro
Cooling	Modified Prometeia Mach1 R402a by Chilly1
Power Supply	Zippy PSL-6701P
Operating System	Windows XP SP2

Testing/Methodology

Stability testing:

Test setups were established and tested with various memory modules to insure stability at speed levels tested. To verify memory stability at stated speeds, I first ran Memtest86+ through the full series of tests, for several loops.

Timings were altered or frequencies adjusted until a maximum passing frequency was achieved. After Memtest86+ produced error free passes, the system was then put through a series of tests/benchmarks (outlined below) that I have found to be particularly effective for assuring reasonable stability. After all this is still a Bill Gates OS ;)

If a system can pass all of these tests without errors, I find very little reason to believe the system is not stable at that speed. I have used this method of testing for well over a year, with excellent results, and have yet to experience an instance where an application of game would experience a memory related error after passing the testing outlined above.

Testing parameters DDR1 (AMD):

After maximum frequencies were determined from each kit, I then decided to attempt to stratify scores in an effort to determine how each kit would perform at both higher frequency and slightly more lax timings versus lower frequency at tighter timings, while at the same time, attempting to keep the CPU speeds relatively constant.

For the AMD/Ballistix DDR testing, the sweet spots for each kit happened to correspond with the maximum stable frequency at both CL2.5 2:2:8 and CL3 3:3:8 timing areas. I did attempt to test at CL3 4:4:8 timings, but the headroom over the CL3 3:3:8 setting was only 7Mhz, and the performance actually decreased.

DDR1 was tested at 240Mhz at CL2.5 2:2:8, and at 260MHz at CL3 3:3:8. Using memory dividers I was able to keep the CPU speed at 3120Mhz for both settings.

Testing parameters DDR2 (Intel):

For Intel testing I found my max stable CPU speed which happened to be about 4.1GHz, prior to memory testing. I then began testing the Tracer DDR2 at the 3:4 ram divider up until my max CPU speed was reached.

Using the default CPU multiplier of 17 for the Pentium4 3.4GHz CPU, this corresponded to a 240MHz front side bus overclock. Timings used at this level were CL4 4:4:8 as set in bios and ram voltage was never pushed over 1.9v, as there was no need. SPD timings were disabled due to problems I have encountered with ALL DDR2 I have tested in this particular motherboard. While running SPD, this Asus motherboard seems to vary timings outside of actual SPD tables as programmed by the manufacturer. When testing for high bus speed at 1:1 memory on the Intel setup and Asus motherboard, our testing was somewhat complicated by motherboard peculiarities and instabilities past 285MHz bus. I have seen several people get well over 285MHz on this motherboard, but I was limited to the 285MHz bus speed in testing. I do not believe the system is CPU or Memory bottlenecked at this point as I could run the CPU higher in other testing, and the memory while running 1:1 at 285MHz is far below its rated specifications. Running at 285MHz the system was stable in benchmarking and in Memtest86+, yet would periodically display erratic behavior in shutdown and posting, which again I attribute to the motherboard. At 285Mhz front side bus and the memory running at CL3 3:3:8 timings, the final CPU speed produced was 3990MHz which was 90MHz below the speed produced during testing at 240MHz front side bus using the 17x Multiplier.

While this is not an ideal situation, it did provide me the opportunity to see if the added front side bus speed could offset the CPU MHz deficit of 90MHz. I also tested the CPU at the 14x Multiplier and the 3:4 divider to find the maximum overclock on the memory in total memory frequency. While benchmarking results are not performed at this level due to how the scores would be skewed lower from the marked drop in CPU frequency, I did include the maximum frequency in the overclocking charts seen on the following page.

Benchmarks Overview

Super Pi 1mb

SiSoft Sandra Memory Bandwidth

Everest Home Memory Bandwidth

PCMark 2004

Futuremark 3DMark2001SE

Futuremark 3DMark2003 Build 360

Doom3-Low detail default settings/benchmark

Far Cry 640x480, 1280x1024

Quake3 640x480 Low details

All information and graphics contained in Madshrimps are sole property of the Madshrimps crew and may not be reproduced or copied in any manner without written permission from us.