New Posts  All Forums:Forum Nav:

Mixed RAM Question

post #1 of 16
8/26/05 at 11:42am
Thread Starter 
I have a stick of 1GB Kingston comming in today. After I install and memtest and all that can I use it with one of my 256MB sticks?

I know some motherboards support this. Some don't. Does the 9300?
Reply
Reply
post #2 of 16
8/26/05 at 1:12pm
Yeah you can use both
Reply
Reply
post #3 of 16
8/26/05 at 1:20pm
I'm running a 1GB(crucial) and 256MB(Samsung) stick in my XPS 2 (same machine pretty much), no problems at all.

John
Reply
Reply
post #4 of 16
8/26/05 at 1:26pm
You can run both sticks with no performance hit. This is because RAM runs at the exact same speed as the FSB, so dual channel offers no performance improvement.
Reply
Reply
post #5 of 16
8/26/05 at 1:29pm
The ram doesn't run at the same speed of the FSB, the 533Mhz Dothans only have a 133Mhz FSB so the memory is running at twice the speed - if you check the memory information you can see the divider is at 1:2. For the memory and FSB to run in sync for 533Mhz ram you're looking at the 1066Mhz Pentium4's

The reason the Pentium-m's don't benefit from dual channel is simply because of the shorter pipeline and relatively low clockspeeds, the architecture doesn't rely on bandwidth like the Pentium4 does. Also, you can run dual channel with two mismatched memory sticks (as mine is currently) however according to the chipset documents it's a non-interleaved mode,

John
Reply
Reply
post #6 of 16
8/26/05 at 2:00pm
John, I was talking effective speeds, which still accounts for some performance in sub-system communication. And the P-M's are memory bandwidth deprived. When one puts a P-M chip onto the Asus desktop board using the adapter, the extra memory bandwidth does substantially increase performance. I'll look for the review, but now that we're hitting 2.26Ghz stock, memory bandwidth does mean something.
EDIT: Couldn't find the review, but here's the thread over at xtremesystems.org
Reply
Reply
post #7 of 16
8/26/05 at 2:13pm
Even at effective speeds, the memory does not run the same speed as the FSB, as I said if you check the divider you'll see it's running 1:2, not 1:1.



If the Pentium-m was bandwidth deprived, then doubling it through dual channel memory would make a large difference, but it doesn't.

John
Reply
Reply
post #8 of 16
8/26/05 at 2:18pm
As I said, check out the thread above. And indeed the divider is 1:2, but I think that brings up another interesting point: the low external clock for the entire system (133) is also a performance hinderer. Anyways, check the thread and see the improvement, these Pentium M's have a ton of potential.
Reply
Reply
post #9 of 16
8/26/05 at 3:40pm
Of course clocking any processor higher is going to make it go faster, I know full well of the Pentium-m and its capabilities, I've been toting them from the start. Your comments are just so general, of course a low FSB is going to mean lower performance but just look at the P4 - massive bandwidth, huge FSB and where does that leave it? You can't just magically change the aspects of a processor architecture which you want to be better, there are tradeoffs to be made.

John
Reply
Reply
post #10 of 16
8/26/05 at 4:33pm
Thread Starter 
Question: How the hell do I get the screws off the door if everything I have tried failed?

I've tried every screwdriver, glasses screwdriver, butter knife, sharp knife, tweezers, and anything else that looked like it would work in there.
Reply
Reply
post #11 of 16
8/26/05 at 4:37pm
Well, granted that there's production cost and architecture revising involved, but it could be quite possible for the Pentium-M within a year to ahve a 200MHz external clock and maybe DDR2-667 RAM. I wasn't referring to the insanely overclocked ones either, just look at the benefits at stock and the performance. A higher FSB doesn't necessary mean high latency and a longer pipeline as long as long as the architecture is revamped correctly. I never insinuated that you weren't aware of the Pentium M's capabilities, I was just merely providing some evidence for my opinion.
Reply
Reply
post #12 of 16
8/27/05 at 4:50pm
Thread Starter 
Well long story short I stripped my screws, but luckly not too much.

The RAM is in and under memtest right now. I kept the other 256MB stick in there, but I figure that is fine. I memtested them a while ago with no errors anyway. So any errors must be the new RAM.

The door is screwed on again, but not super tight like it was. Just enough to secure it down. FOr now on if I ever have to open that door again I know to use the proper screwdriver, and not anything else.
Reply
Reply
post #13 of 16
8/27/05 at 5:18pm
Do yourself a favor and spend $5 to get a nice screw driver set to work on your laptop, I somewhat stripped some screws too before I got mine.
Reply
Reply
post #14 of 16
8/27/05 at 10:49pm
Thread Starter 
I plan on getting the right screwdriver for sure.

Anyway rather than making a new thread I ask...

How long is long enough on Memtest?

I've been going over 6 hours now with no errors. Is it safe to assume it's fine? I just may let it run overnight, but by now if it was to have errors it would have had them, right?
Reply
Reply
post #15 of 16
8/27/05 at 11:12pm
Just use them in tandem. It will work fine
Reply
Reply
post #16 of 16
8/27/05 at 11:44pm
6 hours is more than sufficient, you're good to go.
Reply
Reply
New Posts  All Forums:Forum Nav:
  Return Home