| Kingston HyperX PC 4300 1GB Kit | |||||
|
Product: Kingston HyperX PC 4300 1GB kit When I was offered a chance to review a high-dollar item like Kingston HyperX PC4300 RAM, I’ll admit that I was scared. This product retails for 448 American dollars. That is more then almost all of my other system components and I have the average, perhaps slightly below average, overclocker’s system. This test rig has nothing over the top, and it also has nothing that people won’t use. Will having this high-quality RAM improve the average computer user’s experience? Let’s get this basic information out of the way before the rest of the review. The test rig includes the following:
The Kingston HyperX PC4300 I received is a dual channel kit consisting of two 512MB DIMMS (dual-inline memory modules). Being PC4300, they are tested to reach up to 533 MHz. At 533 MHz, these memory modules are rated to run at 3-4-4-8 timings. These are pretty generic numbers, and I’m hopeful the RAM can outperform its ratings. We’ll soon find out too. I first popped the RAM sticks in, as per the instructions that came with them. It’s a nice addition for Kingston to include directions for people who are not used to installing computer upgrades. While most people buying this high quality RAM will most likely know how to install a DIMM, the directions were well put for those who are less informed. Likewise, following the motherboard instructions, I placed the two modules in the correct sockets to enable dual channel memory. We will also be testing how much of an increase dual channel memory has to the memory bus, if any at all, today in this review. In addition, the effects of the memory timings and latencies will be shown through benchmarks. Right out of the box, I chose some low settings to make sure it worked. Running in a 1:1 ratio with my CPU’s front side bus, the RAM was running at 166MHz (333 DDR). I let it use the SPD settings for the timings, which was 3-4-4-8. SiSoft’s SANDRA was run right away. It showed a memory bandwidth of 2323 MB/s INT Buffered and 2180 Float Buffered. Dropping the timings to 2-3-3-5 yielded an increase to 2388 megabytes a second INT Buffered, and 2240 Float Buffered. That’s well over 50 megabytes each, and a great increase for not changing the speed of the memory.
The next step was to overclock the front side bus of my CPU to maintain the 1:1 ratio with the RAM. I put the FSB setting at 200MHz, and lowered the multiplier of my chip to maintain a similar CPU speed as with before. The multiplier used was 9.0, a 70 MHz overclock compared to the stock 11 multiplier and 166 front side bus of the former test. SiSoft’s SANDRA utility was run again to see what kind of results we would get. The bandwidth was 2792 megabytes a second INT Buffered. The Float Buffered topped out at 2579 megabytes a second. These numbers were obtained using the SPD timings of 3-4-4-8. Regardless of the timings, the bandwidth went up over half a gigabyte from the tight timings at 166 MHz. This shows that timings can’t beat raw speed in RAM.
Click image to enlarge Slight pushing the RAM this time, I kept the speed at 200 MHz, but made the timings better, to the same 2-3-3-5 that I used for the 166 MHz tests. The results this time were 2850 megabytes a second INT Buffered, and 2636 megabytes a second Float Buffered. This was a greater increase then the same timing changes at 166 MHz. Naturally, this proves tight timings and higher bandwidth combine for much greater throughput from computer memory.
Click image to enlarge Now it’s time to use another part of the SANDRA utility. This time, we’ll be looking at the memory and cache subsystem. There are slight changes in the way the benchmarks will be run however. Each benchmark will use the same multiplier with either a 166 or 200 front side bus. The multiplier will be 10x for these two tests. The timings of the RAM will remain at the SPD settings. This test is more CPU speed dependent, and this benchmark will hopefully show the significance of overclocking CPU speeds to help RAM speeds as well.After the test was run, it showed a combined index of 4050 megabytes per second. Compared to the bandwidth of only the RAM, this is a huge increase and is basically double. Raising the front side bus setting to 200 MHz, the CPU speed increased as well as the memory’s MHz speed. The timings were constant at 3-4-4-8. The benchmark yielded 4430 megabytes a second for the memory subsystem. That’s a tremendous increase over the 166 FSB setting. This benchmark clearly shows that increasing your front side bus, as a method of overclocking, truly helps system performance. |
|||||
