For the past few years, I've pretty much stuck with . Part of the reason is they have a good reputation of making some reliable memory, and considering I didn't change PCs all that often, I wanted ram that would last. Another reason is their ValueRam series was dirt cheap (at least here in Canada), and it did the job.

Thanks to VL, I have been bitten by the upgrade bug, and slowly, but surely I upgraded my box to a point where it isn't that embarassing to show off to the chicks. The question that ran through my mind a few months ago was what ram to buy? Why should I put cheap ram in a fast box? Why should I consider one brand over another? Why am I talking in questions?

Having never let me down, unlike one brand I tried out last year, I went back with Kingston. Luckily for me, Kingston released their , which believe it or not, is ram targeted at the enthusiast. Enthusiast ram, typically defined as faster than "normal" clock speeds, and fancy heatspreaders, used to be reserved for a few companies. You can now add Kingston to that list.

Specifications

- Designed and built by the world's largest independent memory manufacturer
- High performance DDR memory tested at speeds up to 434MHz
- CAS Latency 2 support for leading-edge performance
- HyperX memory modules based upon latest specifications and designs
- Kingston's memory chips were custom-packaged for Kingston and screened to support the rated speeds on multi-million dollar testers
- Backed by Kingston's reputation for quality and support
- Available in kits of 2 to meet Dual Channel architecture requirements

If you noticed in the specifications, Kingston makes the ram in kits of two for Dual Channel support. I want to let you know now that the two sticks I have were purchased separately. You can read Vipe's reviews to see the merits of Dual Channel Kits, but I've been using 1GB spread across two sticks on my nForce 2 for months, and I've never had any issues. Am I saying Dual Channel Kits are useless? Not at all, but I do think quality ram will work in Dual Channel mode just fine... at least mine does.

The Kingston HyperX PC3500 ship with some nice blue, aluminum heatspreaders. I pretty much share the opinion that heatspreaders don't really do enough to dramatically make a difference while overclocking, but they do get warm to the touch, so I suppose having some heatspreaders are better than having none at all.

On the heatspreaders, is a sticker indicating the ram type. We can see that the ram is rated for PC3500 operation, which is 434MHz. The timings are rated at 2-4-4-8-1 (CAS Latency 2), which is pretty good, though not great. Not to worry though, as the ram can do better, but those are the SPD settings detected in the BIOS, so we'll be fixing that when we get into testing.

Overclocking

The first test was to see if we could run the HyperX at its rated 434MHz, 2-4-4-8-T1 settings. This is always a good first step just to make sure your ram is working as it should be.

I decided to tweak the timings a bit since better timings result in better performance. Lowering the speed to 200MHz (400MHz DDR), I then moved the DDR speeds upwards, 1MHz at a time until we reached our maximum overclock. As we approached the higher speeds, stability was a concern, and we've had to increase the memory voltage up 2.7v to maintain stability. At 2-2-2-6, we managed a maximum overclock of 210MHz (420MHz DDR). You'll notice this is now 7MHz slower than the rated speed, but given the more agressive memory timings, this was the best I could manage.

The next overclocking test was to determine the best timings possible at stock speeds. At 217MHz, the best timings I managed was 2-3-3-7.

A 225MHz (450MHz DDR) OC was the highest clock speed we were able to manage while still maintaining stability at stock timings. I did try the same overclock again, and lowered the timings to 2-3-3-7, but the end result was a nice, steady beep tone from my PC speaker.

Therefore, in summary, here's our results of our best, and most stable, overclocks:

	2-2-2-6	2-3-3-7	2-4-4-8
Kingston Hyper X	210MHz (420DDR)	217MHz (434DDR)	225MHz (450DDR)

Test Setup

Abit NF7-S nForce2: Athlon XP 2500+ (10x200: 2000MHz), 2 x 512MB Kingston HyperX 512MB DDR PC3500, ATI Radeon 9500 Pro, 80GB Maxtor 7200RPM, Windows XP SP1, ATI Catalyst 3.4.

Test software will be:

SiSoft Sandra 2003
PC Mark 2002 Memory
PiFast
Quake 3
UT2003

We'll be presenting benchmarks at the ram's stock speed (434MHz), in both Single Channel and Dual Channel mode. We will also be presenting some benchmarks of the HyperX overclocked @ 225MHz in Dual Channel mode. The HyperX will be run asyncronous.

SiSoftware Sandra 2003

Although a synthetic benchmark, it's a popular one, freely available if you wish to make comparison benchmarks. We will be testing the memory speeds.

Some good scores across the board. What should be obvious here is if you have a Dual Channel motherboard, make sure you run your ram in that mode.

PC Mark 2002 - Memory

This is one synthetic benchmark that we here at VL don't exactly put a lot of emphasis on, but we're aware that many of our readers do use it.

As we should expect, memory performance increases in Dual Channel mode, and while overclocked.

PiFast

A good indicator of CPU/Motherboard performance is version 4.2, by Xavier Gourdon. We used a computation of 10000000 digits of Pi, Chudnovsky method, 1024 K FFT, and no disk memory. Lower numbers are better.

Dual Channel mode shaves about 3 seconds off the calculation time, and overclocking an additional 2 seconds.