if I can get my unit up and stable, I have every intention of attempting some memory upgrades on this. It appears that it's a normal 256mb DDR. Anyone have any further info on this? Timing>? Unbuffered?
once it's running it will be a simple matter of trying some different chips. It will either boot or not. we'll see.
I experimented a bit with my Thecus N5200's ram memory, but didn't do any testing to see whether I gained any performance. I had success with two 512MB SIMM modules and failure with two 1GB SIMM modules. All were non-ECC, unbuffered and DDR400 / PC3200 or better. I don't know what is really needed. See my "Compatible DDR Modules" post above for more details.
I used IOZone as suggested in the TomsHardware link you sited, to test the default 256MB DIMM versus a 512MB DIMM. I tested with my 95% full, but not fragmented, RAID 5 - 5 x Seagate 400GB. I didn't test with ethernet jumbo frames (N5200 offers 4K, 8K, 12K & 16K).
My test PC plugged directly into the N5200BR switch: Athlon 64 X2 4400+; Nforce4 chipset; Marvell Yukon Gbit ethernet; w/1 GB RAM (I robbed two DIMMS to test with one) on a clean WinXP Pro SP2 w/patches.
Ram tested on Thecus N5200BR, besides the default 256MB module, was a Kingston HyperX PC3200/DDR400 detailed on a previous post.
I did the "new" Tom's hardware recommended Read Performance tests with 64KB record sizes against files 32KB - 1MB. I ran the 256MB one twice because it spiked a bit funny. You can graph this yourself.
Really, I'd say there is nothing compelling about upping the ram, at least not for a RAID 5 setup like mine.
Another performance test that might have more meaning to some. I backed-up my N5200 to an Athlon 64 3200+, Nforce3 mb, nforce Gb; w/4xWD 400GB to onboard SiL SATA RAID 0 (software RAID) under WinXP. I think the write speed was my bottleneck. 1243.59GB (159898 files) copied in 16:41 hrs is about 20.7 MB/s