DRAM Frequency - how do i change it?

Hi again,

It gets even more interesting!

I've belatedly woken-up to the fact that I did in fact buy not one but two new Transcend PC3200 RAM modules. Because (as already explained) I didn't think it mattered to mix modules with only a single-channel memory controller installed (Sempron 3400+), I just added one of the new modules to the Kingston module already in situ. The other one I put into my HTPC (replacing an old PC2700 module, causing its RAM frequency to increase to 400 MHz).

Realising this as I say, I've now swapped the second Transcend with the Kingston, thereby pairing the two identically-spec'd Transcends (shipped in the same shipment from the same supplier) in the one PC - my workstation. I took it for granted that this would at last give me what I was aiming for all along - viz. 1024 MB running at 400 MHz.

Did it? Did it hell! 333MHz

Can anyone explain this?

Edit
And, yes, I have (now) tried each module individually and they each run at 400 MHz. I've also set the timings to 3-3-3-8.

 

Anyone...?

 

I've been out of pocket for a couple of days but I'm also at a loss here. The behavior is very similar to what happens when one trys to run 4 gigs of PC3200 memory in XP - it throttles back to 333MHz and I'm not aware of any fix for that one either.

My first inclination in light of this new revelation re: two identical sticks of Transcend PC3200 would be to focus on potential BIOS updates (check mobo mfg BIOS revision history) to determine if there is any evidence of a common problem in this regard or any reference to a fix. The next place I might look would be FAQ's at AMD as it relates to this processor and I'll assume this is a 754 Sempron (?). And lastly, I'd suggest you try to get Transcend's tech support involved in this issue. Again, I'm at a loss.

One quick story (of many) to share re: memory. 3-4 years ago I had someone request a machine identical to one of mine that he saw. DFI LanParty mobo, 462 Athlon 2600+ mobile OC'd to 2400MHz actual and 1024MB of Corsair XMS (2x512). I could not get the exact same memory (2-2-2-5 with platinum heatsinks was out of stock at New Egg) so I bought XMS Pro with the led heatsinks (also 2-2-2-5 with identical specs) which was more expensive. Couldn't even get the machine to boot at 12x200. I then tried XMS 2-3-2-5 with Platinum heatsinks - it ran but with lots of errors. I was finally able to get 2-2-2-5 Platinum XMS and his machine is still running flawlessly today. At the time, all of that memory was close to $300 a pop. All very high end and all from the same manufacturer with very close or identical specs - while this doesn't help you with your problem, perhaps there's some comfort in knowing just how elusive a solution to some of these issues can be.

 

Hi kosketus. Memory modules of the same general speed (PC3200 a.k.a. DDR400) should be able to be freely mixed on any PC3200-capable motherboard. Some BIOSs are not written correctly and you may have to go in and manually set the memory frequencies and timings.

CPU-Z has all the info you need to manually set the timings. You must use the highest of each setting between the 2 modules. In this case the proper setting is 3.0-3-3-8 since both modules support operation at 200/400DDR MHz.

CPU-Z and Everest are simply reading the SPD (Serial Programmable Device) PROM settings right from of each memory module. The BIOS is supposed to do this automatically, but it may not be doing this right.

Perhaps there has been a BIOS update issued from AOpen that addresses this problem. Maybe not. Go look for one.

 

For the benefit of all readers, this is a little misleading. Although this generalization can't be tagged as theoretically incorrect, I would strongly advise against mixing memory. A disproportionate number of hardware problems can be attributed directly to memory and that certainly includes mismatched modules of the same general speed in MHz.

 

Having had all possible help here in this Forum - and very much appreciated, may I say - and having acted on it, it seems to me that this has to be a BIOS issue. I already upgraded to the most recent BIOS (there has been no BIOS update since September 2005).

As I mentioned at the start, my initial attempt to get help from AOpen fell on deaf ears - not even an acknowledgment (maybe my message got lost in cyberspace...?). I've now sent them a follow-up message making clear what I think of their support, and repeating my request for their help in resolving this performance issue. I'll report any developments.

 

Just reflecting on this - which I didn't question when I first read it, reflecting as it did the experience I myself have been having.

How many thousands (or hundreds of thousands?.. millions?..) of people - I ask myself - have not bought their RAM all in one go? How many of those, not having dual-channel memory, have not carefully matched any subsequently-bought modules with the first one, in the fond belief that it wasn't detrimental not to do so? Can Rockster2u's critique really be justified, because if so its implications are pretty shattering?

 

Well I've heard nothing at all from AOpen Tech Support. Total silence (at the second time of asking).

I shall certainly never be buying any more AOpen products after this experience.

 

Hi again!

Just to complete the story. My PC was becoming more and more laboured so I finally steeled myself to format the C: drive and reinstall WinXP. After doing so, Product Activation was accomplished painlessly over the internet.

But things continued to go rapidly from bad to worse 'til the PC became unusable. I tried swapping the PSU: no better. Then I ran Windows Memory Diagnostic, which soon revealed that one of my two Transcend memory modules was useless.

So I now believe that the problem I began this thread with must have been due all along to a memory module which was faulty, though at that stage not so faulty that it couldn't run, to spec, when on its own. However, it must have been deteriorating even then and that process seems to have rapidly speeded-up in recent days - to the point where the whole computer was being affected.

For some reason I can't fathom, after I removed the faulty module I couldn't boot into WinXP any more. Having tried everything I could think of and got nowhere, I gritted my teeth and formatted/reinstalled AGAIN.

Now all seems well except - and this really p----s me off - Windows Product Activation now won't activate my copy of XP any more, saying that it has already done so the maximum allowed number of times