Test Setup
Test System |
---|
CPU: | Intel Core Duo L2400 - 1.66 GHz |
---|
Motherboard: | Lenovo Thinkpad X60s, i945 Centrino |
---|
Memory: | 2x 512 MB F2-4200PHU1-512SA |
---|
Video Card: | Intel GMA950 12.1" XGA |
---|
Harddisk: | IBM SATA 60GB |
---|
Software: | Windows XP SP2 |
---|
The first the shows how the stock Lenovo memory performs in the notebook. Please note this is single channel, one 512 MB module.
As second test we have replaced the stock memory with one stick of G.SKILL 512SA which shows how a single module would perform in single channel mode. Here we see that the memory Read Bandwidth is down a little bit, which is because the G.SKILL memory runs at a lower clock, yet all the other benchmarks seem to be pretty much the same except for 3DMark which is down 5%.
As third test we have installed two 512 MB modules to get additional performance from running dual-channel DDR2. Now the Read bandwidth is increased compared to the stock configuration and 3DMark gains over 15% compared to single-channel. However, the overall differences are very slim and nowhere near what's gained on regular PC platforms.
This really surprised me, I specifically got two modules to benefit from dual-channel's extra performance at the cost of a limited upgrade path. I don't think this is an issue with the G.SKILL memory but more of how the notebook's memory controller works.
G.SKILL F2-4200PHU1-512SA |
---|
CPU Clock & Memory Ratio | Memory Speed | Memory Timings | Everest Read | Everest Write | Everest Latency | Quake 3 Timedemo | 3DMark 2001SE | SuperPi Mod 1M |
---|
10 x 166 1:2 | 333 MHz | 5-5-5-13 | 3873 MB/s | 3037 MB/s | 96.2 ns | 50.0 fps | 5141 | 36.36 s |
---|
10 x 166 5:8 | 266 MHz | 4-4-4-12 | 3665 MB/s | 3034 MB/s | 97.0 ns | 50.0 fps | 4838 | 36.38 s |
---|
10 x 166 5:8 | 266 MHz | 4-4-4-12 | 3996 MB/s | 3037 MB/s | 95.7 ns | 50.0 fps | 5647 | 35.73 s |
---|