Kingston vs OCZ DDR2-800MHz Memory

http://www.bit-tech.net/hardware/2007/02/23/kingston_ocz_ddr2_800_memory_review/1.html

 

funny that I just just picked a KHX6400D2LLK2/2G kit yesterday. I was all set to put together my core 2 duo system last night but I wasnt able to locate a 775 mounting bracket for my thermalright XP120 at any of the local computer shops. I was so disappointed that I was going to have to wait until tonight to put it together

 

So I guess the question is, then, what is your recommended 2GB kit for a C2D system in this sort of price bracket (£150-180), assuming I intend to OC to a moderate level (i.e. push the E6300 for whatever I can get out of it, but without having to add exotic cooling solutions or any other crazy OC stuff).

 

Any kit should do you well; as to hit a "moderate level" you need 500FSB, and at that, to get yourself DDR2-1000; provided you're locking and linking that is.

 

Ahh. Must be entertaining to tweak the settings over and over again and run the tests that are most likely not to give any breaking news. I didn't notice if you mentioned it, but do the water cooled thingys fit to the slots next to each other? I ran into problems with some aftermarked coolers... (On my mobo the sticks must be in slots next to each other for dual-channel)

Also great images! As some sort of photographer I really appreciate when people take their time taking the piccys and everything is accurate and nice. Good job!

 

Random question (which may well have been explained in the article) why are the tests carried out at different timings?
Surely comparing 3,3,3,x -> 4,3,3,x is going to lead to the tighter timings performing better? Likewise, TRAS values do have an impact on performance and didn't seem consistent between tests.
Fair enough, when testing for a max OC or tightest timings then run as tight (or loose) as you need to/can do for a set FSB otherwise run the timings that all modules can make at a certain speed. Least that way "like for like" comparisons can be made.

Maybe I've missed the point however (which is probably likely) and this was explained somewhere in the report!

 

We went for the fastest timings we could get that were stable, then stock timings to see if there was a difference between the modules themselves and the modules compared to others. Then we went for the highest overclock we could get with an upper limit of 5-5-5-15, as well as leaving the mobo to decide what was "optimum" and giving it a free rein do see the difference. A single timing makes SFA difference on performance, even less so in the real world. Unless you're a sandra obsessive I can't tell the difference between a P965 at 5500 and DFI at 5000.

Some of the modules were rated C4, some C5, so we have to measure the rated speed of the modules you buy. Why would we test all the modules at the same speed when some are rated faster than others? It's a "this is the maximum we could get from them and this is how they will perform in our system, and this is how much they cost: is it worth it?"

I see your point on like for likes in that it makes it easier, but the 680is have a limited number of memory bus adjustments: it goes from 1066 to 1142MHz as it clocks in the next ratio. I know of no one who has bought the DFI RD600 despite it having 3MHz memory adjustments and the P965 boards are even more tied in with FSB and memory bus making comparisons even harder.

Jipa: Took me 3 days of running again and again and again The OCZ's do fit next to each other fine, like on AMD boards in dual channel.