Strange issues with new OCZ PC5400 Memory

The BIOS settings for the memory dont reflect the same information in CPU-Z or SiSoft. Has anyone else run into these issues? I'm not sure if I can trust any of the 3.

Intel 640 HT (default 16x 200 = 3200MHz)
OCZ PC5400 4-4-4-8 (2 x 512MHz) Dual Channel
Abit AW8Max Mobo

Here's an example:

I set the following parameters in the BIOS

Multi - 16x
External Clock - 235MHz
Estimated New CPU Clock - 3760MHz
N/B Strap CPU - PSB800
Dram Spec - 3:5
Estimated New Dram Clock - 392MHz (DDR2-784)

CPU-Z shows the following: PC5300 (should be 5400)
Divider - 3:4
Memory Speed - 312MHz (2 x 312MHz) DDR2 624MHz

Sisoft '05 shows the following
4 x 156MHz (624MHz data rate)
Memory efficiency is only 54%

Any ideas? Should I suspect bad memory?

Just curious, why dont you run them memory 1:1? I mean 235MHz FSB would be only 470MHz effective RAM speed, piece of cake for that OCZ memory, right?

Anyway, usually I trust the BIOS readings more than any software, even though CPU-Z is quite accurate. Have you installed the latest CPU-Z version available?

Im running version 1.30 of CPU-Z. I went back and changed the dividers to 1:1, 2:5, etc.... and both Sandra and CPU-Z reported the same settings and speed as the BIOS.

It definitly has something to do with 3:5 divider.

As far as the 1:1 settings, that will cause me to under utilize my memory (paid extra money to have good fast ram). Maybe I'm missing something. Does a 1:1 setting with lower speeds on the RAM do better than higher dividers with higher memory speeds?

Thanks

Edit:

Mods, this looks like a Motherboard issue rather than a memory issue. Can you kindly move this to the ABIT Intel Motherboard section.

Thanks

Huh? A 1:1 divider will make the memory run at higher speeds than using something like 5:4 or 3:2 (333 and 266MHz), because it increases the memory speed "1:1" with the FSB. This will result in higher effective RAM speed (FSBx2) and therefore in higher Bandwidth, which is always better, especially for Intel rigs. With DDR2 memory timings arent that important, I doubt the difference between 3-3-3-8 and 3-4-4-8 is noticable for example.

Its a different story with high end DDR memory though. My BallistiX at 200MHz and 2-2-2-5 were just as fast as being overclocked to 240MHz and 2.5-3-3-8 timings, when using the same FSB. So good timings can make up for some MHz and increased bandwidth.

But in your case, as long as you can run the memory 1:1, I would do that. Its the fastest option and its always better to run everything "in sync" anyway, lol.

Edit: Just thought of something: Were you talking about running your memory at higher speeds than your FSB? If yes, just forget what I said before, lol. But I wouldnt do that, it causes instability in some occasions. Always run the memory "in sync" with the FSB or use a lower divider than 1:1. It wouldnt give you any performance increase anyway. Just overclock the CPU to a stable 24/7 setting (235MHz sounds good to me, try 3.8GHz though) and let the memory run at the same speeds. Should be the perfect mix of speed and stability!

So did i waste my money by getting the 667Mhz memory if the 1:1 divider is only going to run it at slower speeds?

Thanks