Interesting memory findings

11427

Senior member
May 9, 2003
412
0
71
These all came from the same rig (DFI SLI-DR and 3200+ Winchester OC'd to 2.550) and all settings remained the same except for some memory timings.

Mushkin Black @ 200 2.5-4-8-4-T1 2.7v
Everest: read=7156, write=2514, latency=41.1 Sandra: int=7095 / float=7026

Mushkin Green @ 166 3-4-8-4-T1 2.9v
Everest: read=7096, write=2436, latency=45.1 Sandra: int=7062 / float=6993

Kingston HyperX @ 180 2.5-3-7-3-T1 2.9v
Everest: read=6956, write=2902, latency=38.8 Sandra: int=6936 / float=6850

OCZ Platinum @ 180 2.5-3-8-3-T1 3.1v
Everest: read=6225, write=2158, latency=45.5 Sandra: int=6445 / float=6374

Now I don't really put a whole lot of stock in benchmarks, but I found this very interesting,.. the cheapest was just a bit off of the most expensive, and I really found the OCZ a bit disappointing (even after some tips from OCZAndy and 7 days of tweaking, it just was not in the same league). I did no tweaking with the Mushkin green, all timings were in auto, the black also had a week of tweaking and the HyperX about 12 hrs.
I have been considering some OCZ gold, but after the performance of the Platinum I'm not convinced.
Before anyone asks, this is OC'd so it is really not a fair comparison since not all the ram could perform at the same timings but this was the max I could get out of them, and it is for "MY Set-up" 3200+ Winchester @ 2.550 and DFI LP UT NF4 SLI-DR.
So my question is has anyone tried some great performing ram, and what was it?
Any other thoughts or opinions?
 

Shimmishim

Elite Member
Feb 19, 2001
7,504
0
76
is the second number listed the read or write speed? or is it the cpu speed?

or did you use 2.55 ghz 10 x 255?

cpu speed effects mem bandwidth...
 

11427

Senior member
May 9, 2003
412
0
71
Edited original post to showm more settings and make things a little clearer.
CPU speed was the same for all,... 10 x 255
ONLY changes were to some of the memory timings, for example the Mushkin black would run 1:1 at 255 ldt, the OCZ would not the best it could do at 1:1 was 231 so to get it up to speed (255) I had to use the 9:10 divider.
There were also some minor changes to trc, trfc, trrd, tref, and DRAM drive strength to get the best scores from each set of sticks. The Mushkin green also ran at cas3 while the others were all able to do cas2.5