max RAM OC dropped with new CPU

[Aaron95]

Hello!
So I has a PC that ran with a C2D e4500 overclocked by 900Mhz with the memory (Kingston HyperX 1066) at 1110mhz. Since replacing the CPU, the memory will not clock to, or over 950Mhz without failing POST. The CPU was greatly down clocked (and tested stable) for testing the ram. I have tried changing MCH voltage with no luck..

So whats happeneing? D:

The motherboard is an ASUS P5B Deluxe.

Thanks.

 

[Mussels]

pretty sure the memory controller is inside the CPU, so its simply a different chip clocking different.


try different timings and voltages - as far as OCing goes, you've got a new system.

 

[BarbaricSoul]

what Mussels said.

but then again, I remember a CPU to memory multi from when I had socket 775 that controlled your memory speed in relation to the FSB your CPU is running at. did you check that?

what CPU did you upgrade to?

 

[Aaron95]

Its a 775 system, I'm pretty sure the memory controller is in the NorthBridge?

Anyway, neither CPU voltage or Northbridge voltage seem to effect it. The new CPU is an e8600.

The multiplier remains at a limit or 9 between the CPUs which means I'm running the chipset at 430Mhz at the moment. Can the higher base clock effect the RAMs limit?

The memory is supposed to run at 1066 @ 2.2v, but it won't do 950.. it did 1110 with the previous CPU though.

BarbaricSoul, that is correct, i am running the memory at a 1-1 ratio at the moment giving the effective frequency at 860Mhz

Thanks

 

[newtekie1]

The memory controller is on the Northbridge on the 775 platform.

However, the extra cache and higher FSB will put more strain on the Northbridge, which can lead to lower memory clocks.

 

[vega22]

did you knock the ram when you was swapping the cpu?

have you tried refitting the dimms since you noticed this issue?

other than something like this i can only think of 1 thing which would really cause the ram to stop running as fast today as it did yesterday, degradation

 

[Aaron95]

Thanks for the replies, so it's the Northbridge limiting me here? that does seem to make more sense.
Marsey99, I have reseated both DIMMs in different slots, no change. As for degredation.. I doubt it, sure its possible but its been running great for years, swap out CPU, and then nope.

 

[vega22]

it could be the higher fsb on the new cpu but i would bet it is a borked ram ratio in the bios.


have you tried it on any other ratios?

 

[Aaron95]

it could be the higher fsb on the new cpu but i would bet it is a borked ram ratio in the bios.


have you tried it on any other ratios?

Not.. really, except the next one up, which would run the RAM at 1080. With the same settings as I had it running at before, and it failed to POST. Both those failed when the RAM was clocked over about 950. I cant go too low on the Chipset clock because the CPU only has a multiplier up to 9 so maybe ill just have to live with a lower RAM clock.

 

[TheHunter]

I used this on my old Q9450 @ 3.6Ghz (455mhz fsb), basically anything over 445mhz fsb needs a lot of tweaking, especially GTLRs.

cpu multi: 8x
fsb 455mhz (3.640ghz, ram 1092mhz)
PCIx: 100mhz
mem. enhance:turbo
ram multi: 2.40b
5-5-5-15 (or 18 for stability)

adv timings:
tRRD: 5
tWTR: 6
tWR: 10
act to act refresh tRFC: 83
tRTP: 8
Performance Level:7 (8 for stability, Static tRead?)
Phase adjust - auto
cmd. rate: 2t

skew, driving
800mv
cpu skew- 50ps
mch skew- normal

ram:2.25v
pcix: +0.10v
mch: +0.300v (can do 0.275v but i noticed slower speed in aida64 mem. bandwidth)
fsb: +0.15v (or +0.20v with llc off, but i think you wont need that much with dual core)
*GTLR1: -3%
*GTLR2: normal
LLC- auto
cpuv: 1.343750v (real in windows 1.329v or with LLC off ~ 1.3750v, idle ~1.339v)

I helped with this guide:
http://home.exetel.com.au/seasalt/Overclock.html