Kingston HyperX 2GB PC3200 Dual Channel DDR Kit

**Doctor Feelgood**

The Kingston HyperX 2GB PC3200 Dual Channel DDR Kit offered surprisingly good performance in the face of another 2GB kit of performance DDR that seemed to have the advantage (at least on paper). For gamers and PC enthusiasts looking for an edge without the risks and time investment of overclocking, this kit is definitely worth considering. - The Review

**Doctor Feelgood** · Posted: Wed, 01 Feb 2006 10:08:49 Post Subject:

I am a little disappointed and confused with this review... But grateful to a new forum member, dracos, for pointing out a major issue!

I had requested the KHX3200K2/2G 2GB PC3200 Kit from Kingston and proceeded with the review thinking I had received it. There is nothing on the Kingston site describing a "newer" version of this kit, but apparently one exists (and I have it). Even though I asked a few questions of Kingston and showed them a preview of the review before releasing it, they didn't notice that I had conducted the review as if it was of the "KHX3200K2/2G", even though they sent me the "KHX3200AK2/2G".

I never noticed the A until dracos pointed it out this morning, and he had some information on the features that I hope to get Kingston to confirm today.

For one... Check the difference in timings:

KHX3200K2/2G Kit: 2.5-3-3-7-1
KHX3200AK2/2G Kit: 3-3-3-8 at +2.6V / 2-3-2-6 at +2.65V

Lower latency, if you throw just a little voltage at it! I asked them directly about my set being 3-3-3-8 when I thought it was supposed to be 2.5-3-3-7, and they said that was normal and didn't notice the bigger issue of me not knowing what I had.

That said... My testing results are still valid, whether I knew what I had in my possession or not. The default SPD values were 3-3-3-8. Thinking I had stuff spec'ed at 2.5-3-3-7 I was a bit disappointed, but was then pleased to see I could do 2-3-3-6 at elevated voltages. 2-3-3-6 seemed good when I thought the specs were worse, but now knowing I should have been capable of 2-3-2-6 reduces my excitement.

All in all, the kit did very well in testing at lower speeds, and the final marks stand. Now that I know 2.5-3-3-7 are not the true timings of this kit, I will run some new benchmarks to show what 2-3-3-6 can do for the scores.

When he is ready to publish, look for perhaps a bit more accurate review of the "A" version by dracos over at www.thinkcomputers.org

Thanks... and sorry,

Jason

**Doctor Feelgood** · Posted: Wed, 01 Feb 2006 10:35:49 Post Subject:

Attached is the PDF dracos got from Kingston. Too bad I just saw it now! Hopefully Kingston will list these on their site soon! Low latency 2GB kits are getting to be hot items, and they should be promoting these!!!

swank1 · Rated PG-13 Joined: 25 Oct 2005 Posts: 40

not your fault man, all memory companies do this. Memory is never clearly labeled and manufacturer web sites are rarely updated- they just make the memory depending on chip supply and forecasted consumer need...

Mushkin is really bad with this- try reading a review of theirs and then actually finding the product for sale or on their site Censored

nil123 · Rated PG Joined: 22 Apr 2006 Posts: 2

I just purcased a set of Kingston HyperX KHX3200AK2/2G for use in my Asus A8N-E. However, the DDR400 ram is detected as DDR266!, and the timings it automatically selects are 2-2-2-6 (at 266MHz)...

Strange. I've tried to force it to DDR400/2-3-2-6 at 2.65V and DDR400/3-3-3-8 at 2.6V (and some other combinations), but in all cases, the system just gets unstable.

Asus Award-Phoenix bios does have a lot of other timings to set than just these. Like - Row cycle time/Trc, Row refresh cycle time/Trfc, Read-to-write time/Trwt and Write recovery time/Twr - but I've found no indication anywhere of what they should be set to when manual.

It should be noted that I also have a set of KHX3200AK2/1G (2x512MB, PC3200/2-3-2-6/2.6V), these are detected fine. But the 2x1GB pair forces the system to 266 whether they are alone or together with the 2x512MB.

If anyone have suggestions to what I can do - tools for testing the bandwith/effects of latency etc. it might be helpful. Right now I've send a report to Kingson and awaits their response - if I can't get them to go 400MHz, I'll decide whether to return them or what...

/Nil

**Doctor Feelgood** · Posted: Sun, 23 Apr 2006 14:11:35 Post Subject:

Hi nil123... Hope you don't have bad modules there, but it doesn't sound good.

What happens if you up the voltage on the memory? Can you force it to DDR400 woth say, 2.8V?

Since I have the same stuff on the same motherboard, I know it should work. If after checking all the BIOS settings, and trying more voltage, and you get no where, I am guessing you might need to exchange it.

nil123 · Rated PG Joined: 22 Apr 2006 Posts: 2

Thanks for the reply Smile

but I managed to track down the problems, turned out to be more than one.

First the 400->266 MHz issue: The Asus A8N-E (and I assume all other mobos using nForce4) has a "4 rank limitation" (according to Kingstons site) - this mean that it only supports 400MHz up to 2GB! Use more memory and it can't go above 333MHz. (I can't seem to find this in the mobo manual or specs, but I grew tired of looking...)

That it keept the 266MHz setting when I removed the 2x512MB blocks was maybe because I didn't enter the bios. Later (after a lot of reboots) I noticed the 2x1GB blocks defaulted to 400MHz when alone.

Second problem; during all this I decided to update the bios, which apparently added some driver incompatility because I (being in a hurry - lot of reboots) forgot to update my nForce4 SATA drivers - this incompatability induced some random hanging during booting (and increased my confusion).

Finally, I had bad memory, but it wasn't the new blocks - the 2x512MB blocks (also Kingston HyperX) I have (had) was faulty! - nothing I really noticed when running - maybe a random crash now or then - but when I increased the voltage to 2.65V for the 2x1GB blocks, I must have destroyed the 2x512MB completly, because they're fried now (but still within warrenty! Smile

) - this took me some reboots to figure out, because they have been running just fine for over a year, so I (wrongly) assumed the fault was elsewhere...

In the end, the 2x1GB blocks works fine (alone!). When the bios is autodetecting they go 3-3-3-8 as in the review, but with 2.65V I could set them to 2-3-2-6 (T1). Haven't tried to overclock. But I did run Memtest86 3.2 overnight - 14 passes without errors - so I trust them now Smile

- So to sum it up for google (because it took me a while find the info, and someone else might find it useful): Running 2GB (2x1GB dual channel) Kingston HyperX KHX3200AK2/2G on an Asus A8N-E nForce4 motherboard, I'm using the following (non-overclocked, but stable) timings:
DDR voltage set to 2.65V. MHz: 400, Tcl:2, Tras:6, Trcd:3, Trp:2, Trc:11, Trfc:14, Trwt:5, Twr:3, 1T/2T Memory Timing:1T

**Doctor Feelgood** · Posted: Fri, 28 Apr 2006 13:15:50 Post Subject:

Wow! Quite alot to diagnose there! Glad you got it sorted out.