RAM on divider - Big perfromance hit

Conrad11

Conrad11

Conrad11

Soldato
Joined
12 Jun 2005
Posts
5,361
Hi there,

I currently have my ram on diver 1:2 as my motherboard will only let it run at that or 1:1 and my FSB is 300, so my ram is running at DDR300 - will that be a big performance hit I am on a AMD system.

Would it help if i lowered some of the timmings?

Thanks.
 
Lowering the timings will make a big improvement. AMD systems don't take that much of a hit when running on a divider on tight timings.
 
Thanks - I only ask as people say running on a divider won't make much of a perfromance hit, but surely if its on the lowest divider...?

Does everyone with an AMD system just lower the divider and the timmings then?

Currently i have managed to get the timmings down to 2-2-2-3 (i think - memtest running now). I don't think the ram will run the CAS at 1.5.

EDIT:

My RAM is corsair twinx 3200cpt2 2x1gb (something like that anyways)
 
Well you'll certainly take a bigger hit than if you were able to run 2-2-2-3 on a higher divider. But it was pretty normal for people to clock their Opterons to 300mhz and have 2gb running at no more than DDR400 with 3-3-3-8 timings.

Sheer mhz aren't that important with the AMD DDR systems due to their memory controller being on-die.
 
I think at DDR400 my ram is default 2-3-3-6 (something like that anyways)

But for some reason - whenever I have my ram on a higher divider (regardless of the timmings) my system isn't stable.

- If my RAM passes memtest - does that mean that stability issues are not (or less likely) to be cause by the RAM.
 
Well MemTest finds out what your memory can do without the extra stresses of running an operating system. It's more of a first-stop when memory overclocking rather than a definitive test. It's common for memory to pass an all-night MemTest, but to be the cause of an Orthos blend test failure once in Windows.

Which board do you have and what are the available memory dividers?
 
Well you've got a pretty damn good board there. Tbh I never had the chance to play with one, but I can't see any reason why you can't be stable on the 133 divider. Other than DFI's being famous for being quirky as hell with some memory.

*edit*

You could have a read through these posts on the Corsair support forum. Other than that maybe Yellowbeard will be able to help you the next time he's around here.
 
Last edited:
Cob said:
Well you've got a pretty damn good board there. Tbh I never had the chance to play with one, but I can't see any reason why you can't be stable on the 133 divider. Other than DFI's being famous for being quirky as hell with some memory.

*edit*

You could have a read through these posts on the Corsair support forum. Other than that maybe Yellowbeard will be able to help you the next time he's around here.
Click to expand...

Thanks for your help Will add to reading list.

currently running DDR300 with timmings 2-2-2-2. I am doing an orthos blend test atm.
 
Cob said:
It's common for memory to pass an all-night MemTest, but to be the cause of an Orthos blend test failure once in Windows
Click to expand...
That's not quite my view, I think if the memory can run for hours and hours in Memtestx86 then that means the memory is good but if the same set-up then goes and fails something like Prime *Blend* or *Large FFTs* its either something to do with the CPU (if its overclocked) or it means the onchip memory-controller is spazzing out.

Memtest doesn't seem to pick up northbridge errors on INTEL systems or the on-chip memory controller in A64 systems?

On an INTEL system its possible to adjust the NBvCore to correct these errors but the A64 on-chip memory controller is still a bit of a mystery to me? as in I'm not sure how to *boost* its power?

Back to the orginal question I have set most of the A64 systems I have built to run on a memory divider, raw Memory MHz doesn't seem to give a great boost on A64 instead its Processor speed, memory timings including the all-importtant *1t* setting.
 
Last edited:
Hi,

Thanks for the reply. I am runing Orthos blend test at the moment (been running for about 1.5 hours). Seems to be going ok - temperatures are ok..etc. I am running just to check the overall stability of my System.

Current settings are:

CPU = 300 * 8.5 = 2550Mhz
RAM = DDR300 (1:2 Ratio Divider) 2-2-2-2-1T
HT Link = 300 * 3 = 900Mhz

How long do you usually run Orthos for to know its totally stable?

Thanks.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom