With the precipitous drop in RAM prices I purchased 2 Gigs of ADATA DDR-2 800mhz (4/4/4/12) RAM from Newegg, as it was $54 with a $15 mail in rebate.
The current system is an e6400 OCed to 3200mhz, with 2 gigs of Corsair XMS2 ram, running at 800mhz 4/4/4/12 stable. Asus P5B motherboard, OCZ GameXtreme power supply, ect..
I read on other overclocking forums that fully populating all 4 slots can be bad. Also, mixing the RAM can be bad as well.
Anyways, I dropped in the 2 sticks of new ADATA RAM and went straight to Memtest86. Immediately, there was a bazillion errors, and my heart sank. I figured I needed to RMA this dud RAM right back to newegg.
But...I tried removing the Corsair RAM, and the ADATA tested fine.
Anyways, after much tweaking and swapping sticks around to different slots I found out that I fail memtest86 at a CPU FSB frequency of 400mhz...BUT 399mhz works PERFECTLY! Why? Literally, it fails the very first test in Memtest86 with all four slots full, but at 399mhz it has run for an hour now and passed every time. I actually ended up crossing the pairs as well, bank 1 has one stick of Corsair and one of ADATA and vice versa. Seems to work fine.
Further data : voltage changes do nothing, it works fine at 2.0 and 2.1 volts at 399, but fails at 400. Nor do timing changes do anything : whether I time it at 5/5/5/18 or at 4/4/4/12 it fails at 400, works at 399.
I'm not too upset over losing 10mhz on my overclock...I just want to know why this is happening.
Theory : My North Bridge has some hard internal limit, either caused by software or literally a limit in hardware, that prevents it from addressing all four slots at 400mhz when the CPU is overclocked.
UPDATE : I just tested it again, and it still does the same thing. I changed nothing but adjusting the FSB frequency from 399->400, and ran memtest86. A screenfull of errors. HMM....Let me try clocking ABOVE it.
Nope, anything above 399 fails.
The current system is an e6400 OCed to 3200mhz, with 2 gigs of Corsair XMS2 ram, running at 800mhz 4/4/4/12 stable. Asus P5B motherboard, OCZ GameXtreme power supply, ect..
I read on other overclocking forums that fully populating all 4 slots can be bad. Also, mixing the RAM can be bad as well.
Anyways, I dropped in the 2 sticks of new ADATA RAM and went straight to Memtest86. Immediately, there was a bazillion errors, and my heart sank. I figured I needed to RMA this dud RAM right back to newegg.
But...I tried removing the Corsair RAM, and the ADATA tested fine.
Anyways, after much tweaking and swapping sticks around to different slots I found out that I fail memtest86 at a CPU FSB frequency of 400mhz...BUT 399mhz works PERFECTLY! Why? Literally, it fails the very first test in Memtest86 with all four slots full, but at 399mhz it has run for an hour now and passed every time. I actually ended up crossing the pairs as well, bank 1 has one stick of Corsair and one of ADATA and vice versa. Seems to work fine.
Further data : voltage changes do nothing, it works fine at 2.0 and 2.1 volts at 399, but fails at 400. Nor do timing changes do anything : whether I time it at 5/5/5/18 or at 4/4/4/12 it fails at 400, works at 399.
I'm not too upset over losing 10mhz on my overclock...I just want to know why this is happening.
Theory : My North Bridge has some hard internal limit, either caused by software or literally a limit in hardware, that prevents it from addressing all four slots at 400mhz when the CPU is overclocked.
UPDATE : I just tested it again, and it still does the same thing. I changed nothing but adjusting the FSB frequency from 399->400, and ran memtest86. A screenfull of errors. HMM....Let me try clocking ABOVE it.
Nope, anything above 399 fails.
Facebook
Twitter
Instagram