Poor Memory Performance in PC Mark

If you have a quick look here: http://service.futuremark.com/compare?pcm=1267300 you and look at the memory benchmarks, you will see on the 384kb tests, my computer sucks! I have seen computers getting very similar memory benchmarks to me except in the 384kb tests where they are about 8 times faster.

Ideas anyone?!

I don't know what are you talking about. AMD systems are not exactly spendid in memory performance so your mark looks fine with me. If you want better increase the amount of memory and use dual channel, say 2x512 or at least 2x256.

What sucks there is your hard drive score. 718 is rather lowly even for the 2 MB cache PATA drive. That is certainly something you can improve. Raptors in a RAID 0 should be doing well over 2,000, even over 2,500 with proper partitioning. My IDE RAID 0 does about 2,050.
Install the newest chipset drivers, enable system bus management/controller, defragment, and partition that RAID array properly.

Another thing - many people are taking benches while using all sorts of background junk applications (messengers, sound applications, antivirus, "real" audio, and all sorts of similar useful and useless crap). That is of course reducing the memory performance as well.

Hmmm, well the reason I bring up the "poor" benchmark was because I was looking at some other athlon systems on the futuremark site and they had much higher 384kb memory results, so I thought my rig was lacking in some way, but if you think that is normal then it sounds like something funny was happening on the other peoples computers as they were scoring over 8 times higher than I was.

Going on to the HD, I know 700 is low, but its some lame 5400 rpm hd, its not exactly the fastest in the world, that was benchmarked without my raid array, and hence the low score. Also, you mentioned that if partitioned correctly you can get a much higher score, do you mean the stripe size for the RAID? I have it set to 16k at the moment and score about 2100 - but I think its being limited by the controlers bandwidth, as it has to interact with the southbridge which only allows 133mb/s - however I'm only achieving 100mb/s so there must be some overhead or I have got a bad striping size!
What do you think?

5400, I was under impression that is Raptor. For 5400 that is fine. Partitioning only as partitioning the volume, that significantly increases performance and decreases fragmentation.

About memory - your memory score is over 5,000 on 384 MB. That seems very OK to me. I don't know what scores can be much more than that. I get over 11,000 using 1 gig in Intel dual channel with PC4200, which is as good as you could hope for in my setup. I'd like to see who has 8x5,000 = 40,000. That is maybe on a static ram or something, certainly not 384 MB DDR.

Ah I'm sorry, I think I have confused you a little bit, I was refering to the actual benchmark for 384k chunks, rather than the score as a whole, basically what I'm saying is that my system seems to benchmark - in comparison to some Athlon systems the 384k chunks quite low, as I said some people were getting over 8 times the performance in that specific test... I was just wondering how they can manage such a feat, I know my system is pretty sound other than that though - when benchmarking with the raid array anyway.

I don't read much the futuremark database but it seems to me your system is doing as expected, except for that hard drive.
I have one older system on Giga-Byte 7DXR+ with two 20 gig 5400 hard drives in Raid 0 array and I get about 1,300, this is well partitioned system. That is just a little shy of a single Raptor. Meaning with a Raid 0 and 5400 hard drives you can still get a nice performance. Single 5400 drive would score about as much as your does, laptop drives probably 200 less.
So there is a big room for improvement. Athlon XP do not score as good as Intel on memory but they still work very, very good in the "real world," so I wouldn't worry about that score at all.