2GB OCZ PC3200 Performance + DFI NF4 SLI-D
This memory is pretty old on P4 platform did 230-238mhz 3-3-3-8 at 2.65-2.8v vdimm. Now on AMD64 let's see :)
Product link:
http://www.ocztechnology.com/product...b_dual_channel
System:
3200+ CBBHD 0445 VPMW water cooled
DFI NF4 SLI-D 414-3 official bios
2x eVGA 6800U PCI-E SLI mode with 71.89WHL drivers
120GB WD 1200JB
Pioneer A05 Burner
650W Silverston Zeus ST65ZF PCI-E SLI psu
WinXP PRO SP2
- note SLI mode will probably reduce cpu related performance slightly so take that into a/c with the benchmark results.
- Tref of 0388 was the key to stability with 414-3 bios while 0648 was key for stability on 310 official bios both bios resulted in same max 225-230mhz 3-3-3-7/8 at 2.8v. With wrong Tref set test #3 in memtest would error once or twice every few passes.
2x 1GB OCZ PC3200 Performances series dual channel kit
@10x 230mhz 3-3-3-8 13-19-2-2-1-2 at 2.8v
chipset = 1.6v
LDT = 1.5v
LDT 4x
CPC = Enabled
CAS Latency(CL) = 3
RAS to CAS(Trcd) = 3
Min RAS# Active time(Tras) = 8
Row Precharge Time(Trp) = 3
Row Cycle Time(Trc) = 13
Row Refresh Cycle Time(Trfc) = 19
Row to Row Delay(Trrd) = 2
Write Recovery Time(Twr) = 2
Write to Read Delay(Twtr) = 1
Read to Write Delay(Trwt) = 2
Refresh Period (Tref) = 0388
Write CAS Latency(Twcl) = Auto
DRAM Bank Interleave = Enabled
Skew = AUTO
Skew Value = 0
DRAM Drive Strength = Level 8
DRAM Data Drive Strength = Level 1
Max Async Latency = 7ns
Read Preamble Time = 5ns
Idle Cycle Limit = 16
Dynamic Counter = Enabled
R/W Queue Bypass = 2x
Bypass Max = 2x
32 byte Granularity = Disable(8burst)
http://www.fileshosts.com/DFI/NF4_SL...LK/cpuz_tn.jpg
CPUZ validated: http://valid.x86-secret.com/show_oc?id=16254
Sandra 2005 Lite
Buffered = 6,384 / 6,306
Unbuffered = 2925 / 2,989
Everest Home
Read = 6,504
Write = 2,585
Latency = 44.5 ns
Pcmark04 = 4,384
Pcmark04 Memory = 5,657
SuperPi
1M = 37.312s
32M = 33min 03.875s
3dmark2003
@425/1100 stock speeds
http://www.fileshosts.com/DFI/NF4_SL...0_21625_tn.jpg
3dmark05
@425/1100 stock speeds
http://www.fileshosts.com/DFI/NF4_SL...00_9783_tn.jpg
Problems with the dividers
This weekend I tried out my Geil 2x1Gig PC3200 set and the initial results are not good. At default speeds they go to the advertised 3-4-4-8 timings. I clocked them up to 220 1:1 for a few runs through memtest and then into windows for a quick bandwidth check. Nothing earth shaking there of course.
The big problem seems to be the dividers. I can't get lower than the 166 divider. Anything lower results in no post into memtest. Using the 5:6 166 divider I can't get the HTT higher than 239HTT. As soon as I go to 240 (200 effective) it has problems. I can get into memtest but not into windows. All settings are on Auto except for the 3-4-4-8 and Interleave Enabled. It looks as though dividers = less than stock memory speed. Dang.
The only odd thing that I noted was that while using A64 Tweaker was everthing on the left column was showing the asterisks indicating a reserved value. Memory timings were showing up way higher than the 3-4-4-8 reported by CPUZ.
At stock 200HTT speed I got the timings down to 2.5-3-3-7 with 2.7v.
I finally settled on 3-4-4-8 224HTT 1:1 2.8v to try it out on EQ2. The extra memory seems to have eliminated the disk caching but I still have a little stutter on the part of the Port Qeynos map. Maybe because I'm running the CPU 900mhz slower than I do with my BH-5 or TCCD which have no problems with the 5:6 and 9:10 dividers. :( Although considering that, I'm surprised that the net effect in overall game play performance is small :confused: At least at 1280x1024 "Very High Performance" setting the difference is small. The performance heirarchy on this game looks like video card>memory>cpu.
I was thinking about returning these to NewEgg and getting another pair of the OCZ Plat. rev.2 to mate up with the pair I already have and see if I have better luck with 4x512 and 2T. But since these sticks do their advertised speed and timings (albeit only at 1:1) I'm reluctant to anger the RMA gods for fear they will send me more unspectacular memory.
EDIT: I'll try the latest bios. Haven't updated since the 3/10 released.