How can you push the FSB to 390 without killing your hard drive and other PCI/AGP-based components? With FSB 390 Mhz your PCI bus must run at 49Mhz!!

However, I see that the PCIEX bus clock can be set, does that mean that on this Shuttle overclocking the FSB doesn't overclock the PCI express (and hopefully also PCI?!) bus, and therefore your hdd and sound cards are safe?

One question you stated in your review that you did the tests with jumbo frames disabled does that mean that the shuttle supports jumbo frames or that some of the other boards were jumbo frame capable.

I have bought two SFF systems, one from Biostar and one Shuttle. The Biostar one is still going after 3 years, the Shuttle was a total waste. Perhaps I just got a bad one.

Is the market for these things healthy (yet)? I always thought of Shuttle's XPCs as a fag/niche market.

GMA 3100 in G33 is NOT the same as a GMA X3000. Not even close. The 3d core is, in fact, pretty much identical to the one found in i945 (gma 950), though it has a much updated name...

I'm glad they stuck with the G5 case, its so much better than the G6 design.

Looks like intels overclocking on G series chipsets is the same with 33 and the 965 boards.

and since all of Shuttle's previous XPCs have been smart enough to disable their IGPs when discrete graphics cards are installed, we expected the SG33G5 to do the same.

not true, actually. This same jumper switch has to occur in their SD11G5.

I saw someone mention you were using CAS 4 timings. Did you keep those for the overclocking portion, or loosen them? If not, if this one is like other recent Intel Shuttles, it means you were probably running DDR2-975 (the 667 is 5:4, so 390x5/4=487.5x2=975) at CAS4, but the RAM is only rated for CAS5 at DDR2-1066. Obviously that doesn't necessarily mean the RAM's the limiting factor here, but it's cutting it close.

Just re-read your SD39P2 review from earlier. In cases like this review, do you ever go back and see if the issues you highlighted that could be fixed with a BIOS update ever were?

So can you run the RHYBTH (whatever it is) demo or not. In the power section you said you couldn't, but then the next paragraph said that that's what you used to test for o/c'ing stability. Next up, you say that prime95 errored out, which you didn't mention at all. I'm assuming that you meant to say that you tested stability with prime?

I thought the SS51 was the first g-series. I've had that for over 5 years... still my primary computer. Maxed out with a 2.8 p4 and a ATi x800 PE for video.

Actually if anything this review left me impressed with Gigabyte's p35 motherboard more than anything, really.

Gave a quick read through. Very good article, as always. Shuttle made some weird design choices, like the graphics jumpers and the PSU fan, but everything else seems in place.

One thing: in the conclusion, did you mean to say that the PSU fan was easy, or not easy, to replace? You wrote that it was easy.

First post!