Memory issues with overclocking Sempron 3400+

[ruffryder201]

Hi folks

 

Wondering whether anyone can help me with this one (I tried looking and hoped to find a thread in this forum about something similar but to no avail):

 

Basically I recently upgraded to a Sempron 3400+ with my 512MB PC4400 Corsair memory and all was still good. Then I felt 512MB of XMS PC4400 is not enough for games, benchmarks and stuff so I bought another identical stick of Corsair PC4400.

 

With one stick of RAM I can achieve a variety of combinations of overclocks up to and including a FSB of 275MHz but with the two sticks of memory using slots 1 & 2 (the two closest to the CPU) FSB is limited to 230ish MHz, if I use slots 1 & 3 memtest is only happy up to 260MHz (even then I get the odd error in this test). It will boot at 270MHz but will blue screen within half a minute.

 

Does anyone know how to stabilise 2 sticks of memory in this situation? I haven't tried every combination of memory slots or messing around with timings in the BIOS yet. Which one is the option for selecting a timing of 1T or 2T by the way?

 

Thanks in advance for any help.

[steelballzz]

Which one is the option for selecting a timing of 1T or 2T by the way?

 

.

 

 

hi

 

CPC

disable 2T

enable 1T

 

i had similar problem with my athlon 3400 hammer 754...crappy memory controler

try a sempron 3100 or a 2800

cheappo and good clockers

 

[Guest Blooz1]

I think you should start by testing each stick individually to determine what each one will do by itself, especially since they were purchased seperately. Even if the chips are the same brand and model, they're from different manufacturing batches, and one might not run as fast as the other.

 

Also, if you think about it, 2x512 will stress your mem controller a lot more than a single stick. A single stick will usually clock higher than 2 sticks in the same board with the same CPU.

[ruffryder201]

Hi guys

 

Thanks for the posts.

 

I do not think the Sempron 3400+ is rubbish - it's 90nm and based on the Palermo core. I got rid of my Sempron 3100+ as it was a Paris core and can not overclock as high as my 3400+.

 

Thanks for letting me know how to enable the 1T/2T option. Just how much of a difference does this setting make?

 

I'll fiddle around with my memory some more tonight and will let you know how I get on.

[CPDMF]

Are both sticks of your RAM the same revision? Even though they are the same stick they could come from different batches of chips and PCB's.

[ruffryder201]

Hi

 

Although the font of the writing on the heatsink is slightly different to the other yes the version and other details appear to be the same.

 

I do not think the stress on the memory controller is the limiting factor here. If it were then I'm sure it would do what my Paris core Sempron 3100+ would do - restart the PC even with 1 stick beyond 265MHz FSB. In this case though I'm getting a blue and memtest is coming up with errors all over the place.

[grease monky]

slots 1 & 3. satas in 3&4.tcl 2.5 And what are your timings?

[ruffryder201]

Yes I am using memory slots 1 and 3 at the moment and sata port 3 (as 1 &2 limit me to 240MHz) and timings have been slightly loosened from 2.5-4-4-8 to 3-4-4-8.

 

Both sticks of RAM are version 1.1 by the way.

[ruffryder201]

Hi people

 

Thought I would let you guys know that by inserting my memory into slots 2 & 3 the RAM is now working as good as gold!

 

I'm sure when I had just the one stick of PC4400 I couldn't take it over 275MHz (I'm not sure) but now it will go to 293MHz at 2.5-4-4-8 without an error on memtest. Bizarre but just goes to show some PC problems can only be solved by trial and error rather than trying to explain why it's gone wrong and trying to figure a solution that way.

 

Mind you having said that I cannot run the memory at 1T as when I tried I had to reset the BIOS whatever you call it (you know the switch you do with the jumper). Perhaps trying to do 1T at 270MHz FSB might have had something to do with it.

 

As nice as having 1T would be I'm pretty content at what's been achieved.

[Guest Blooz1]

Nice!

 

Glad you worked it out.

[ruffryder201]

Damn it! I think I've spoken too soon! The hell is the matter with it now!

 

Using a multiplier of 8 FSB will go all the way to 290+MHz. On a multiplier of 10 FSB will go to 260MHz before CPU calls it quits. This is all achievable with memtest showing no errors. Go for a multiplier of 9 however, and memtest throws errors left, right and centre even with FSB as low as 250MHz. It's as if the memory is allergic to the CPU's mulplier of 9.

 

Is this bizarre to you guys or other people experienced such weird allergic reactions?

 

Quite a shame really as I was hoping it would cope with 290x9 (2610MHz) never mind I'll settle for 261x10 (2610MHz) for now and see what else I can do.

[smolt]

Something to think about.If your useing a multiplier of 8 or 9.That puts you useing a divider of 166 with lose timing. better off useing 150 and tighter timmings and t1. The rule of thum is if you have to use a divider use one with the tightest timmings as possable