DDR2 memory CL4 @ 800 vs CL5 @ 1066

[Stewartm]

I have an E7200 on a P5Q pro running at 3.8G (400x9.5) if I go higher temp climbs rapidly
I have 2x corsair dominator cl5 8500 ram (4G)

what I whant to know and cannot seem to find is

am I better of

1/ running ram @ 400 (ddr800) and oc timing to cl4 or even cl3 if possible at 1:1 ratio

or

2/ running ram @533 (ddr1066) at cl5 with an uneven ratio

thanks

ps i am aware the difference may be insignificant.

 

[Deleted member 67555]

Run at 533 for 1066mhz really it makes a difference even with uneven settings,but not for all applications

 

[sneekypeet]

run 500 FSB with a 7.5 multi and just run the ram 1:1.

 

[gaximodo]

IMO, 400mhz CL4 will give you better experiences CL3 is impossible.
i was running my ram @ 510, 5-5-5-15-2T(uneven ratio)can't tell a difference compared to 800 5-5-5-18-2T even in benchmarks(3dmarks etc. not the Everest one)
but with 800 5-4-4-13-1T, i can even tell the difference even during daily computing.
I might be wrong tho..its just my feelings.

 

[Stewartm]

fsb 500

cannot seem to get above 430 stable

 

[sneekypeet]

IMO, 400mhz CL4 will give you better experiences CL3 is impossible.
i was running my ram @ 510, 5-5-5-15-2T(uneven ratio)can't tell a difference compared to 800 5-5-5-18-2T even in benchmarks(3dmarks etc. not the Everest one)
but with 800 5-4-4-13-1T, i can even tell the difference even during daily computing.
I might be wrong tho..its just my feelings.

I ran 1000MHz at 3-3-3-9 on an old 975X Abit board, I wouldnt say its impossible

cannot seem to get above 430 stable

IMHO either way is good, you will get very similar test results with 1066 at CAS 5 or 800 at CAS 5. I always go with the higher speed route if its easily achieved.

 

[Stewartm]

ok

I will stick with the 1066/cl5

I am trying 450x8.5 vcore 1.225 nb 1.2 ram 1081 (5:6)



ps 500 could not get stable nb 1.25

 

[Loboking]

You need to raise your NBv. 1.25v isn't enough for 2x2gb ram @ 500FSB. My P5Q pro+e8400+4gb ram @ 500FSB is ~ 1 hour p95 stable with 1.36v on the NB. It requires 1.4vNB for 8+hours p95 stability.

edit: Don't forget when you drop your multi it also raises the effective speed of your NB's FSB.The equation is stock Multi/new multi*FSB=NB effective FSB. i.e your e7200 has a stock multi of 9.5(i think correct me if I'm wrong) so 9.5/8.5*500=558.8 effective FSB. If you were to try 8x500 your NB FSB would be 593. You need alot more NBv and active cooling on the NB at those speeds.

 

[LagunaX]

Run:
1) Everest Cache and Memory benchmark
2) SuperPi 1 million

See which will give you getter numbers.

1066, huh?

 

[Wile E]

Which one is better depends on the mobo. Technically speaking, the 1066Mhz CAS 5 ram would outperform 800 CAS4 in both latency and bandwidth, but, since you would have to use a non 1:1 divider to achieve 1066Mhz, it may take a performance hit.

With 450fsb, see if you can get 900Mhz CAS4 or 1125 CAS5.

 

[LagunaX]

Here is a great article for you here:
Low Latency DDR2 800MHz versus 1,066MHz
http://www.bit-tech.net/hardware/memory/2008/10/27/low-latency-ddr2-800mhz-versus-1066mhz/1

 

[OnBoard]

Here's a good read also to help choosing. basicly what is says that board has a bandwith limit and unless you raise FSB you won't get performance gain on faster memory.
http://www.xbitlabs.com/articles/memory/display/core2duo-memory-guide_4.html#sect0

With this page OP question is answered:
http://www.xbitlabs.com/articles/memory/display/core2duo-memory-guide_8.html#sect0

800MHz 4-4-4-12 ~ 1000MHz 5-5-5-15, so CL5 @ 1066 wins.

Here is a great article for you here:
Low Latency DDR2 800MHz versus 1,066MHz
http://www.bit-tech.net/hardware/memory/2008/10/27/low-latency-ddr2-800mhz-versus-1066mhz/1

"We found that using the biggest performance difference for the popular P45 chipset was actually tRD - a chipset limitation, not memory. On the Gigabyte board this is the "Performance Enhance" option that can be changed between Standard, Turbo or Extreme - either this or adjusting precisely in the Advanced Timing Control section will greatly affect performance. Dropping down the performance level from Turbo to Standard with the 1,066MHz memory saw the performance drop significantly to sub-800MHz CAS-4 levels, effectively nullifying any point in buying and using faster memory settings."

I have that setting on Turbo, but had no idea what it actually does. Now I know I won't be standard (there is no Extreme setting on mine). Didn't have any stability issues with Turbo on my previous cheap memory either.

I should do something similar as OP is wondering, as it finally got a bit cooler. Can get back to overclocking after the week heat outside. Currently running 400FSB with 3:4 5-5-5-15 1066MHz 1.9V. Think I'll aim for something closer to 450FSB or see how much this memory can take with CAS 4. Still don't know can this board do 500FSB.

edit: hehe, just noticed you have E7200 too Post here how it goes for you, my CPU-Z validator link is pretty much max that I got with 1.3625v (not stable). Too much voltage needed to go any higher as higher multi does 4GHz with less voltage.

 

[APOLLOHERO]

I·m sorry i can`t help you
because my english is fool……