I'm late to this party, but my brother and I were looking through this review, and had a few questions:
1. Is the DIII logo on that fan standard with any manufacturer? Or is this just a reference? I couldn't find mention of the brand of card. If it's reference, would be nice to know it off the bat.
2. Card availability is barely slim now, and it's 3 weeks later. It would be great, although I know hard to do, if you could mention availability on day of release, and back that up with links/evidence. I think when reviewing "released" cards attention should be made to "available now" features, and I mean outside of OEM's like Dell and Alienware. Afterall, your customers are mostly do-it-yerself-ers. OK I see the pricegrabber thing at the end of the article.

Cool review, great card.

Anyone have the power consumption figures for the 6600GT? I called Albatron and couldn't get them, and they aren't on Nvidia's website, no estimates even.

Well, that makes my mind up...

I was going to get a PCI-E board once I got the money, but I'm going to go AGP...

Now, do I get the $300 P-M board, or the $100-150 A64 board (with $200 processor on either - the P-M overclocks like mad, FWIW)?

Is the video hardware acceleration/decoding broken on the 6600 as it is on the 6800 line?

Other than the eVGA LE cards that were on newegg for 2 days, has there been another 6800 "vanilla" with 256? I haven't seen one. And calling that eVGA card a "vanilla" is a stretch.

edit: should be reply to #19.

Where are the HL2 Benchies!?!

As is evident from the benchmarks the GeForce 6600GT AGP puts all of the previous generation graphic cards to shame.

He he he. Looks like Sander Sassen uses the same words I do you bunch of little girlies!

Taken from the Hardware Analysis link of Wednesday's Shortbread

But in all seriousness:
6600gt = 500/900 (128bit)
9800p = 380/680 (256bit)

On a clock for clock the 9800p does very well, other than in Doom3. The 6600's ability to pull ahead with 8GB/s less membory bandwidth is astonishing. I'd like to se a comparison of these at the same clock speeds - again, just for acedemic purposes.

to say the 9800pro has shame for being beat by the midrange of a next generation gpu is stupid... that's what is supposed to happen.

Dear TR. What am I doing. Here I sit, with an unplayed validated steam copy of HL2 and I am reading your 6600GT AGP review. Is this normal and should I be gaming instead? BTW, enjoyed the review as usual.

what's the power consumption like on this thing?? will the power supply in my shuttle SB61g2 blow up with a card like this?

OMG, once i get my shuttle, im so gettin this card.....

When can we buy it?

It lists your 6800 as having 256 MB of ram. Is this right? Did you just downclock one of your 6800GT's or something?

the first one to ask why the test wasn't done on an Athlon64 gets shot :)

our 9800 pros are hanging their heads in shame :( Still, that cards going to last me until I move to Athlon64 with PCI-E, so this release is of mere acedemic interest to me.

Edit: great review, would still like to see a minimum fps number posted somewhere.

Cool review of a promising looking card. Couple comments.

1) No overclocking tests? ;)

2) It would be nice to see some CPU scaling tests when reviewing low-mid range cards. I can imagine a lot of people buying this card to replace aging GF4Ti or Radeon9700 series cards and dont have top line CPUs. It would be nice to see how they would fare with this card and a few newer games (ie Doom, HL2, Farcry)

hmmm, a bit of an upgrade to my 9800 Pro...

Buy this, or wait and get 6800 NU?

decisions, decisions..

Great Review, BTW, as usual

Hmm. The Beyond3D and ExtremeTech reviews don't have the same 128MB PCI-E issue. Configuration / test rig problem maybe?

To everybody who said I was stupid for buying a AGP motherboard now instead of waiting for PCI express: HA HA.

I have one other possible reason for the behavior other than the guesses in the article. It could be that PCIe is having problems with latency inside the north bridge of the chipset. It is a new tech, and the first gen may not be doing the job very well. It probably takes some very different techniques to translate data into the thin serial style of PCIe instead of the wide parallel of the old PCI bus. Another related thought is that there could be some problems involving conflicting data requests for PCIe and PCI. Future motherboards that remove PCI entirely may be better.

Now those are some pretty strange numbers comparing PEG and AGP versions of the same core. Maybe nVidia can shed some light?

here we go..again...

fp.