DDR3 desktop memory has been around for nearly two years, but it wasn’t until a few months ago that we finally got to see the first platform specifically designed to take advantage of its increased data rate. With an on-die memory controller that supports triple-channel mode, Core i7 processors have produced the biggest memory bandwidth improvement we’ve seen since RDRAM met Netburst. But once again, a transition that should have been smooth has been met by memory manufacturers who had other ideas.
The current problem is one of voltage. While DDR3 is specified to use 1.50 V, manufacturers found that the memory controllers of Core 2-generation chipsets could easily handle far more. Rather than try to produce the fastest possible memory within a relatively small voltage range, most vendors instead chose to use slower parts with extra voltage tolerance to produce highly-overclocked products for the enthusiast market. When Intel announced shortly before its Core i7 launch that the memory controller should not encounter more than 1.65 V, a quick look at the market revealed that only a single manufacturer was producing DDR3-1600 modules for standard-voltage configurations at that time.
What followed was a mad rush by memory brands to re-label "fast" memory at whatever slower speed was required to get it stable at the new voltage limit. In the process, this "lower-voltage" memory was packed in triple-channel kits to differentiate it from the heavily-overclocked dual-channel kits sold for previous-generation systems. Super-fast DDR3-2000 disappeared for a time, and DDR3-1866 took nearly a month to emerge in 6 GB triple-channel kits, as nearly every existing product was reduced by one or two speed grades to stabilize it under a lower-voltage ceiling.
December finally brought us high-capacity modules at DDR3-1600 and higher speeds in triple-channel kits, and we quickly rounded up as many of those high-end kits as we could for today’s mega-comparison.
What was always bothering me about ram reviews is how much memory speed/timing will really impact on system performance. I it is a lot of work I know, but it was never covered.
I could be wrong, but shouldn't it be ddr3 ? I'm not aware of i7 supporting ddr2?
Heheh, looks like it's been edited.
Kingston HyperX T1 Series Kits:
KHX16000D3T1K3/6GX
KHX16000D3ULT1K3/6GX
What does the UL signify?... would it benchmark the same as the former (which was tested in this article)?
I loled at that.. maybe it's too early in the morning for me
I guess many are a: Not wowed by i7 like they were core2 and b: Laugh and the prices of DDR3 at a time when DDR2 is not only still very fast relative to the software/games on the market but is dirt cheap.
They may have changed the name.
=======================================================================
Kingston part number KHX16000D3ULT1K3/6GX is a DDR3 PC3-16000 memory module running at 2000MHz.
Kingston part number KHX16000D3T1K3/6GX is not a valid Kingston part number. Did you mean to ask about Kingston part number KHX16000D3T1K3/3GX?
=======================================================================
Either Tom's had a typo, and this is really only a 3GB (1GBx3) kit, or it is not released for resale yet (e.g. the sales reps don't have that part number available to them). The *only* 6GB kit I saw had the UL designation.
I guess now you want a copy of the invoice? Tom's didn't make a mistake, but it's possible Kingston might have changed the part number after sending the kit.
Reason I'm asking: Building a new system this week and looking to purchase THIS ram (the Benchmark champion, Kingston HyperX T1 series 6GB kit per this article).
I know the article said you can't find this anywhere, I was merely asking Kingston for when this ram might become available.
But what about performance in areas that count? I want to know if DDR3 2000 gives me more FPS than DDR3 1066 in games (Crysis, COD: WaW, FC2, etc). Because what I've seen so far, the answer is no or very very little.
It does apply to memory modules, but to a lesser extent. If you overclock six sets of modules with identical specs, you usually get a varience of around 20MHz.
Because both are rated at 2000 MHz CAS 8-8-8 1.65V, the most likely scenario is that Kingston sent its test module set to THG first, then altered the model number BEFORE making them available. That would mean the original model number never even made it into the stocking system. Since that's the most likely scenario, I wouldn't have any problem buying the parts with the UL designation.
I are preparing a XY chart with Price-Performance based on that data, to post here soon.
Now, This article need a second part, Real world benchmarck, and bus overclocking limits on the wonderful P6T Deluxe.