E5300 DDR2 Underclocked RAM = Instability?

Soldato
Joined
26 Jun 2009
Posts
3,023
Location
Sheffield
Hey there,

I've been setting myself up for a spot of overclocking on these components:

Intel E5300
Gigabyte G41M-ES2L
1x1gb Kingston 667mhz cheapo rubbish ram.

Now the wierd thing is, I set everything up manually, all stock settings for now, checked it was stable, and then it was time to start overclocking. My first rule is always change 1 thing at a time, so first thing was to drop the memory multiplier from 3.33 to 2.66.

This meant that my 667mhz ram would be running at 533mhz, but it wouldn't even boot? But if I leave the multi at 3.33 with an FSB of 210, running the ram at 700mhz, it does boot?

Any ideas on what the problem may be? Just bought 2x1gb of 1066mhz DDR2 on MM, so when they arrive I should be able to run the default multi and overclock it on the FSB, but I was just wondering if anyone has had any experience of this board and could offer any idea as to why changing the multi would have made the ram unstable?
 
Hello Muel :)

My first rule is always change 1 thing at a time
I set everything up manually
I'm confused? . . . your saying you only change one thing at a time yet in the next breath you have said you changed everything to manual? :confused:

so first thing was to drop the memory multiplier from 3.33 to 2.66
  • 1x1gb Kingston 667mhz
With that chip running stock and that memory I don't think there is a defined strap and associated memory divider for anything under DDR2-667?

Intel® Pentium® E5300
  • 13 x 200 = 2600MHz & Memory [2:1] DDR2-800
  • 13 x 200 = 2600MHz & Memory [5:3] DDR2-667
I suppose your thinking you should be able to run the stick at 266MHz (DDR2-533) but I was never able to get that working myself in the past on several different boards . . . would be interesting to see a screenshot of CPU-z showing the [SPD] tab? . . . . if you change the FSB to 266MHz (1066MHz System Bus) you may be able to change the state of play? :cool:
 
Can I just ask why you are wanting to under-clock your memory in the first place?
Well I guess its so he can raise the FSB without overclocking his "667mhz cheapo rubbish ram" :D

Just bought 2x1gb of 1066mhz DDR2 on MM, so when they arrive I should be able to run the default multi and overclock it on the FSB
Intel® Pentium® E5300
  • 13 x 266 = 3458MHz & Memory [2:1] DDR2-1066
  • 13 x 320 = 4157MHz & Memory [5:3] DDR2-1066
The problem is likely to be, that mobo will only treat that processor as a "native" 200MHz-FSB chip and will restrict it to basic memory multipliers, the only way I know to get around this is to perform the "BSEL Mod" to change it to a native 266MHz-FSB chip . . .

I found the newer Intel® Pentium® E6xx series chips which are "native" 266MHz-FSB (1066MHz System Bus) allowed much more flexibility when it came to overclocking and memory options!
 
When you change the strap downwards you are in fact tightening the latencies - so 333 or 400 would be much looser and therefore equate into a higher overclock. But a corresponding decrease in board efficiency. When we were benching SuperPi with the Asus P5W boards and later the P965 boards it was very handy to have a tool that allowed a strap selection. I seem to recall that Felix made one such tool. But I could have that wrong and giving credit to the wrong guy. Currently there is a SPi tweaking tool made by a German benching forum member that does include a option to select strap.

If I have misunderstood what you are asking my apologies.
 
Sorry guys, forgot about this. I would reply to you all but I don't have time I'm afraid! I solved it anyway. I got 2x1gb of 1066mhz ram, and upped the FSB to 266mhz which I think Wayne recommended in one of his posts?

Anyway, it's running lovely and stable at 3.4ghz using stock voltages now.

Wayne, when I said I changed 1 thing at a time, but set everything to manual, I mean I inserted the stock settings manually all at once, made sure it was stable, then started changing one thing at once. :)

Again, thanks for all your help and advice.
 
Back
Top Bottom