Memory speed issue on new motherboard

Zantor · Post by **Zantor** » Mon Jun 01, 2009 12:07 am

I upgraded from my EPoX mobo with a 2.2 GHz AMD Sempron to an ASRock motherboard with a 2.8 GHz Pentium 4.

When I got windows up on its feet after installing Windows XP after getting rid of Windows 7 (I got to where I couldn't stand it), I checked to make sure the processor, memory, and AGP card were working to their fullest. Sure enough, the CPU and graphics card were, but not the RAM.

I have an ASRock P4i65G socket 478 motherboard. The memory interface is dual channel and rated for DDR 400 MHz (PC3200). Here's the thing, though, and it's quite annoying. Whether I set the speed and/or timings to manual or automatic, it only runs at 266 MHz; it's as if the memory controller completely ignores the settings I set in the BIOS. One possibility I have considered is that the RAM is undervolted; if SiSoftware Sandra 2009 is correct, the voltage available for RAM is 2.5V, a little too low for the 2.6V RAM I have in my machine right now. If this is not correct, then I must look elsewhere for the root of the problem. The other possibility, which I seem to have ruled out, is incompatible timings.

Searches on Google for a solution are inconclusive. I am seeking advice or what not, but I will not purchase a new machine; I want to make what I have work. I am willing to sell what I have and get a new set, or trade for a twin set of RAM modules if I need to.

Duper · Post by **Duper** » Mon Jun 01, 2009 12:44 am

update your bios.

Zantor · Post by **Zantor** » Mon Jun 01, 2009 10:16 am

Have you updated BIOS before? Do you prefer from DOS or Windows?

Krom · Post by **Krom** » Mon Jun 01, 2009 10:39 am

Always update your BIOS from a pure DOS boot disk.

Zantor · Post by **Zantor** » Mon Jun 01, 2009 12:42 pm

ASRock provides a DOS flash utility that boots straight from the disk. Should I use it or should I get the files and make them work from a DOS 6.22 boot disk?

Another question. Would win98SE DOS boot disk work?

Duper · Post by **Duper** » Mon Jun 01, 2009 6:46 pm

hmm wait. i just reread your post.

the boot disk. It won't matter. You just need to bypass the windows load and provide an environment to run the update software without it interacting with the BIOS. (don't forget to change the boot order of your drives)

Reading over your system stats though, I think says a bit more. Your system probably won't run Windows 7 well. Even though it's not the hog that Vista is, it's still a hog compared to XP. I'm lacking in my hardware savvy these days, but isn't that a single core processor? If so, you might want to consider upgrading to a multicore processor. That's still low end by todays standards.
It says: \"OS - Microsoft® Windows® 98SE / ME / 2000 / XP compliant\"
It doesn't matter what you overclock stuff to, this is a \"slow board\".

Also, what RAM do you have in the board? and how much? Are you sure that all the sticks are working properly; that there are no bad sectors on the sticks? Sandra won't tell you that. go get Memtest.

Zantor · Post by **Zantor** » Mon Jun 01, 2009 8:33 pm

Duper, you misunderstand regarding OSes. I WAS running Windows 7 RC on my AMD board and it ran fine even though it just met the memory requirements. I have switched BACK to windows XP after 3 weeks of dealing with windows 7.

The processor is an Intel Pentium 4 2.8 GHz single core processor with HyperThreading technology (Northwood core), which uses a second pseudoprocessor/logical processor. It's a technology Intel developed to increase performance in multithreaded applications; the results can be application-dependent, but can make a significant difference. I noticed it often when compressing and decompressing files, and programs that are heavily multithreaded.

I have two DIMMs of PC3200/DDR-400 RAM in dual channel configuration. I have had both modules for over a year and have seen extensive and intensive use, and I have received no consistent memory errors that would suggest a bad memory cell.

----------------
Now playing: Basshunter - Angel In The Night
via FoxyTunes

Duper · Post by **Duper** » Mon Jun 01, 2009 8:43 pm

ahh, my bad. thanks. Still, check the memory and make sure one or both sticks haven't gone bad (I assume you have two)Particularly if nothing else has changed.

Also, see if there is a mother board driver update. Sometimes i would get speed bumps with new VIA driver updates. I can only assume the same would be true with Intel boards. *shrug*

Zantor · Post by **Zantor** » Mon Jun 01, 2009 8:49 pm

I installed the chipset INF update that ASRock provided. I am looking into a newer driver at Intel's website.

Update: I found a newer INF update (made in 2008 rather than ASRock's in 2005) and will be installing it to see if it helps.

Krom · Post by **Krom** » Mon Jun 01, 2009 9:20 pm

What bus speed is that P4 running at? Its possible it is one of the 533 MHz bus chips, which operates at 133x21. So if the memory is running in sync it will operate at 266 MHz. Running the memory in sync like that will probably already give you the best performance compared to running it out of sync to reach 400 MHz (causes big latency penalties).

The only way you could get the best performance out of 400 MHz memory is if you got one of the 800 MHz bus chips, and I don't think too many of those were made for socket 478.

If there is an option in BIOS to run the memory at 400 MHz as opposed to 266 in sync, it would be a 3:2 ratio or 1.5 to 1, depending on how BIOS lists it.

Zantor · Post by **Zantor** » Mon Jun 01, 2009 10:18 pm

The BIOS update was a success. If I let the BIOS determine memory speed automatically, it'll go to 160 MHz; it's odd that it won't do 166.

I have successfully manually set it to 200 MHz and CPU-Z reads it running at 200 MHz, doubled to 400. I am going to put the second stick BACK in, and see if they will coexist. If so, then I have achieved my objective.

Edit: I have inserted the second stick of RAM and they are staying at 200 MHz happily at 3-3-3-8 timing.

I have achieved my desired objective. Thank you all for your help and advice.