I came up with some answers to the question of what memory should be paired with an AMD Thoroughbred processor. Here's the system I used in my analysis.
CPU: AMD 2400+
MB: Asus A7N8X-D
Memory: 512MB Corsair XMS 2700C2PT
Video: ATI Radeon 9800 Pro w/ Omega 3.6 set to Optimal Performance
HDD: Western Digital 80GB, 8MB cache SE
OS: Windows XP Pro
The procedure I used to get these results was to vary the FSB and multiplier on my CPU so the clock speed would be at or near 2.0GHz, and then I would run the benchmarks. I used multipliers from 17 to 10, but I couldn't go lower since my memory isn't stable above 200MHz and I didn't want to compromise the timings. I used the UT2003 Demo, 3DMark2001 SE, and Splinter Cell to get my numbers that I have to share here. Graphs of my results and the raw numbers are in the links below.
3DMark2001 SE Graph
Splinter Cell Graph
UT2003 Flyby Graph
UT2003 Botmatch Graph
MS Excel Table
My opinion from these results is that performance scales steadily for the most part, though slight tapering is evident at higher FSB speeds, with added bandwidth up to 200MHz. At 200MHz FSB, the performance increase over 133MHz was between 10% and 12%.
Here are a few things to note before I wrap this up. Results of this analysis cannot be directly applied to Bartons since an equally clocked Barton will require less access to the FSB since it has a larger L2 cache and therefore less cache misses. For CPU's clocked higher than 2.0GHz the performance gains you will see when running at high FSB speeds will likely be greater than my results, while systems clocked lower than 2.0GHz will see less benefit. This being because the amount of bandwidth a CPU can use depends of the clock speed and architecture of that processor.
If you want to maximize the performance of your Thoroughbred system, get the best memory that you can afford and run your memory and FSB at the highest speed you can reach with stability. On the other hand, the performance increase may not be enough to justify the price you will pay. I hope this was helpful to you. Thanks for reading. Peace
CPU: AMD 2400+
MB: Asus A7N8X-D
Memory: 512MB Corsair XMS 2700C2PT
Video: ATI Radeon 9800 Pro w/ Omega 3.6 set to Optimal Performance
HDD: Western Digital 80GB, 8MB cache SE
OS: Windows XP Pro
The procedure I used to get these results was to vary the FSB and multiplier on my CPU so the clock speed would be at or near 2.0GHz, and then I would run the benchmarks. I used multipliers from 17 to 10, but I couldn't go lower since my memory isn't stable above 200MHz and I didn't want to compromise the timings. I used the UT2003 Demo, 3DMark2001 SE, and Splinter Cell to get my numbers that I have to share here. Graphs of my results and the raw numbers are in the links below.
3DMark2001 SE Graph
Splinter Cell Graph
UT2003 Flyby Graph
UT2003 Botmatch Graph
MS Excel Table
My opinion from these results is that performance scales steadily for the most part, though slight tapering is evident at higher FSB speeds, with added bandwidth up to 200MHz. At 200MHz FSB, the performance increase over 133MHz was between 10% and 12%.
Here are a few things to note before I wrap this up. Results of this analysis cannot be directly applied to Bartons since an equally clocked Barton will require less access to the FSB since it has a larger L2 cache and therefore less cache misses. For CPU's clocked higher than 2.0GHz the performance gains you will see when running at high FSB speeds will likely be greater than my results, while systems clocked lower than 2.0GHz will see less benefit. This being because the amount of bandwidth a CPU can use depends of the clock speed and architecture of that processor.
If you want to maximize the performance of your Thoroughbred system, get the best memory that you can afford and run your memory and FSB at the highest speed you can reach with stability. On the other hand, the performance increase may not be enough to justify the price you will pay. I hope this was helpful to you. Thanks for reading. Peace
Facebook
Twitter