Testing:
All testing was done on the machine I use daily. Each step was checked for stability by running a combination of Prime95, 3dMark05, and Super Pi. The included BIOS version of Memtest86+ was also used to check for errors.
System Specifications:
• DFI NF4 Ultra-D motherboard (BIOS 310)
• AMD A64 3000+ (Winchester core)
• Thermaltake PIPE101 cooler
• Sapphire Radeon X850XT graphics
• Enermax EG701AX-VE (W) SFMA V2.0 PSU
• 36-gig Raptor hard drive
• 2 x Samsung 120 gig SATA in Raid 1
• NEC 3510A DVD burner
• LiteOn DVD player
The start of testing involved baseline bandwidth measurements using just the optimum defaults loaded into bios. This resulted in the Corsair SPD timings of 2.5,4,4,8, but at a CPU default of 200MHz x 9 for 1800MHz processor speed. PC3200 speeds do not even come close to the PC4400 speeds this ram is specified at, but for baseline information, I did test at these speeds.
Next, I took the ram up to its intended default speed with the same SPD timings. Knowing the limitations of my questionable Winchester cored CPU; I also dropped the multiplier to 7. This particular Winchester has never seen the higher side of 2300Mhz, let alone 2475MHz, which would have happened given the rated ram speed of 275MHz x 9. All testing from this point on was done with a multiplier of 7.
After running benchmarks at the rated speed of the ram, I decided to see just how high the ram would go with all DIMM settings set to auto. Without even blinking an eye, the Corsair TWINX1024-4400C25PT reached 290Mhz or DDR580 speeds without changing a single timing or even raising the voltage. This means with no labor involved, no high voltage worries, this ram could give even the most novice overclocker a huge jump in performance.
With all this testing at SPD timings over, it was time to go after the ceiling. Not having any luck with timings, I decided to look to the Internet and come up with a set of timings that might get this Corsair TCCD even higher. Below is an A64Tweaker screenshot showing those timings.
Now I realize that these timings are not very tight, as a matter of fact the base 2.5,4,4,8 hasn’t changed any. However, changing some of the other associated DIMM timings I was able to bring this memory up to 300Mhz or 600DDR. With some extensive tweaking of timings, this memory should be able to go even higher. As a side note, I booted into Windows at 308MHz with 2.8 volts, windows seemed to run fine, but none of the stability checkers were happy.
Results:
At each step of the testing detailed above I took Everest Home edition, and Sisoft’s Sandra memory bandwidth measurements. In the charts below you can see how the bandwidth increases nicely as the FSB is raised. Again, with lots of time and patience, the timings could be tightened up to increase bandwidth even more.
Please read on to the next page for more... Next
Page 1 | Page 2 | Page 3 | Page 4 | Home | Forum | Review Index
|
|
