shuttleteam, neither your comment nor your screenshot are science, and "the Netgear just flat out owns the Intel!" is an interesting analysis, but I can't find how you're supporting it. I appreciate your enthusiasm, but please try to be as constructive as possible.
I admit that I'm a BSD & Linux person mostly, and the Netgear 302 delivers ZERO Mb/s on my systems, not being supported at all, so that's a big "thanks for playing, guys" from me. The Pro/1000MT is very well supported in contrast, including there being a Linux driver written by Intel. The Intel gig boards have a similar architecture and it's likely that Intel gig boards made several years from now will still have a similar architecture, while Netgear uses the chip of the month. From a perspective of being well supported and Just Working, Intel simply wins. I don't know about you all, but I like it when hardware Just Works.
In my personal lab testing, the Nat Semi chipset (which powers many low end gig boards) was a poor performer and didn't even sustain 100Mb/s on a gig link (my 21140 boards can do that. NOTE - I've been told that there were driver performance improvements in Linux and BSD that have since helped a lot), the Intel Pro/1000MT (Intel's chipset) were solid, and the 3c996BT (BCM5701) was neck and neck with the Intel until you hit max jumbo MTU, where the 3Com pulled ahead. The 3Com boards are substantially more expensive than the Intel (more than double the cost), and so their price/performance just isn't as attractive. The Intel boards can be had in the $41-$46 range OEM, or $35 through Intel's eval program, which would be great price for an Intel 10/100 board IMO, and you're getting a gig board. They seem to me to be the current best choice.
The Netgear 302 board has a new chipset that I don't know. It's NOT supported by the OSs I really use, so I have no useful performance data and no driver source to read (Windows performance is a non issue, not to mention an oxymoron). My gut, however, is to say this: designing a gigE chipset that performs really well is not an easy task. Most of the no-name NIC controller vendors who have designed 10/100 chips have designed chips that are pretty lousy (and some can hardly handle 100Mb/s), so any of those vendors designing a gig chip strikes me as likely to result in a chip that's just not as good as the other players. Certainly that's the case with the NatSemi chips - I know of several important hardware things they screwed up that directly impacts performance. In contrast, Intel's been doing Ethernet controllers for a VERY long time and their 82557/8/9 10/100 chip is known to have very good 100Mb/s performance. And the BCM5701 chip on the 3Com board is a third generation version of the Alteon Tigon core, which was an early gigE NIC chipset with very good performance (for a long, long time, the Tigon II - found on the 3c985SX and the Netgear GA620/620T - was the gigE NIC core to get for PCs).
skyking, today, do you need gig at home? No. But if you're buying NICs today to build systems today, it is worthwhile to think about what you're going to want to have a few years from now. My personal philosophy is that I try to buy stuff that I think is going to last, technologically and reliability wise. I've got a closet full of obosolete gear, and I'd like to be able to get as much out of boards as possible before they take the death march into the closet. I would not buy a 10/100 NIC anymore unless I was super ultra cost sensitive for a job. The cost of a gig NIC is just soooo cheap now, and it has a longer obsolescence lifetime. And in a few years, when you buy your new Athlon 64 and want to connect it fast to your current PC, you'll be able to make better use of that gig link.
And yeah, I want the coolest home network I can get! Now, if only I could figure out how to get some 10Gb/s Ethernet gear into my basement. I'll put that next to the 40lambda x OC192 SONET ring
