14 Comments(s). 1 Pages(s). Showing page 1. [ 1 ]

   #9. Posted at 01:23 PM on Apr 6th 2006 Edit   Reply

If I was a 3D gamer, there is no way I'd buy ATI right now. (Sorry, ATI.) Their designs may be a bit faster at higher price points, but they still crank out the heat, even when idle. For the midrange, the X1700 might be able to compete with the 7600GT if they don't screw it up like the X700. Either that, or wait for X1800GTO prices to fall and be comparable.

The bottom line is, unless you plan on playing some really graphiaclly intense games at high resolutions, the 7900GT is enough for you.

If you're creating an HTPC, the X1600 (or the new X1700) should do you well if you get one with a passive cooler or put a Zalman VF-700 on it., assuming your case can take the heat and you can take the noise. If not, 7600GS and purevideo are the way to go.
collapse

   #1. Posted at 10:43 AM on Apr 6th 2006 Edit   Reply

32-bit memory bus option? WHY? Does anyone even still make memory they could use on a video card with a 32-bit memory bus?
collapse

   #3. Posted at 11:11 AM on Apr 6th 2006 Edit   Reply

Nice, ATi appears to be catching up to the 6600GT just in time for nVidia to leave them behind again.

Quiet die-shrinks on their hardware do tend to help alot. I've got an RV351 Radeon 9600 Pro that OC's like gang bustas, running at 550MHz.
collapse

   #5. Posted at 11:58 AM on Apr 6th 2006 Edit   Reply

I don't think ATI is playing catch-up mode, since after all, they do have the fastest single card out there. Also, don't you think 48 pixel shaders is gonna take up some die space? The x1900 has a lot more hardware under their roof than a 7900, so of course their die is gonna be bigger.
collapse

   #4. Posted at 11:42 AM on Apr 6th 2006 Edit   Reply

ATi flailing around with GPU process shrinks and alternate vendors...not surprising. In catch-up mode again.... The R580 (X1900...) is 353 sq. mm while the G71 (7900..) is 192 sq.mm, both on the same 90nm process. The R580 consumes about 40% more power than the G71 (at the default clock-speeds for both parts). The X1xxx-series 'unified' GPU architecture, so glorified by ATi Marketing, is very silicon-inefficient. So a process-shrink of the X19xx family is the only short-term viable solution. The 80nm "half-node" process at TSMC has the same design rules as the 90nm, so a shrink is pretty straightforward and will get the per-unit cost of the GPUs down. However, no problem for nVidia also transferring their 79/76xx GPUs to TSMC 80nm, if/when they feel their cost-advantages have been uncomfortably eroded -- probably already have pre-production silicon.... And they will drop into current board designs, of course.
collapse

   #2. Posted at 10:53 AM on Apr 6th 2006 Edit   Reply

not very compelling in the midrange.
collapse
14 Comments(s). 1 Pages(s). Showing page 1. [ 1 ]
 
Name/Password: / Remember
Reply to:
[click to clear]

[RED] [GREEN]
[BOLD]
[ITALIC] [STRIKE]
[UNDERLINE]

Notice: All posts should abide by the rules, please.
Note: Ctrl-Enter submits the post. (In IE)
DThread keys: Click on a reply to position the blue bar. 'A'/'Z' move it up/down.
Jazztags: (they MUST be closed)
    r{ red }r     g{ green }g     /[ italic ]/     *[ bold ]*
    _[ underline ]_     -[ strike ]-     s[ sample ]s     o[ spoiler ]o  q[ (QUOTE) ]q