First looks at GeForce 6800Go

[kemosabe]

http://www.anandtech.com/video/showdoc.aspx?i=2268
http://www.tomshardware.com/mobile/20041108/index.html
http://www.extremetech.com/article2/0,1558,1720128,00.asp
http://www.hexus.net/content/reviews/review.php?dXJsX3Jldmlld19JRD05MDU=

NVDA looking more solid this time around, but I guess once again it's going to come down to whether they're able to control heat/power issues well enough to get the clocks up to the target specs.

Wild card: M28 with 16 pipes? Eh, Mr. Baumann?

 

[Dave Baumann]

Well, when have ATI had "specific" notebook parts (i.e. ones that weren't from the desktop line) recently, and how many 12 pipe specific ASIC's are there in the desktop ASIC line?

 

[kemosabe]

Well, when have ATI had "specific" notebook parts (i.e. ones that weren't from the desktop line) recently, and how many 12 pipe specific ASIC's are there in the desktop ASIC line?

Ah, Dave, is there ever a non-convoluted answer to these simple queries? I really don't know where you're trying to lead me.

 

[PiNkY]

~ their notebook parts have hitherto been closely related to their desktop parts & r420 has 16 pipes

 

[kemosabe]

~ their notebook parts have hitherto been closely related to their desktop parts & r420 has 16 pipes

No, up until now the high-end notebook parts have been based on the mid-range desktop ASICs. Mobility 9600/9700 only have 4 pipes whereas R350/360 have eight. Even Mobility 9800 has half the pipes of X800, so there has been no precedent of pipeline equivalence between desktop and notebook. Obviously with M28 and NV41M things are changing, and if that means that each desktop part will have a mobile counterpart, then a 16-pipe M28 is to be expected. I'm just not sure that's what Dave is trying to communicate. :?

 

[kemosabe]

With a little help from friends on another board, I believe Dave's comment finally makes sense. Unless I'm mistaken, he is saying that the notebook ASIC will essentially be the same as the desktop R420, i.e. 16 available pipes out of which 2, 3, or all 4 quads will be activated. Hence Mobility 9800 (8 pipes), Mobility X800 Pro (12 pipes) and Mobility X800XT (16 pipes). I take it Mobility 9800 is the same as Dell's X800SE part (two defective quads). Feel free to say I'm out to lunch, Dave.

 

[Bjorn]

http://www.anandtech.com/video/showdoc.aspx?i=2268

300/300 (NV) vs 400/400 (ATI). There will suppsedly a 450/600 of the 6800 and 450/?? version of the M28. Seems to match the 6600 vs X700. Low end clocked much lower then Ati's corresponding card. Will be interesting to see the power consumption of these cards at their different speeds.

 

[jvd]

man looking at anands benchmarks both of those cards are great.

THough the ati card is faster in most of the games and sometimes by a huge amount.


Only question is if these two cards are going to be priced the same. If the nvidia part is priced lower it will be very attractive

 

[Bjorn]

Only question is if these two cards are going to be priced the same. If the nvidia part is priced lower it will be very attractive

The 300/300 version will probably have to be a bit cheaper then the 400/400 M28. Just like the vanilla 6600 vs X700 Pro. The 450 MHz version should be a really good DTR chip though.

 

[ChrisRay]

Those results are pretty impressive,.

 

[DemoCoder]

What is it about the Battlefield engine that makes it so ATI specific? 1/2 the performance? It's a DX8 game.

 

[Jawed]

What is it about the Battlefield engine that makes it so ATI specific? 1/2 the performance? It's a DX8 game.

Multi-texturing?

Jawed

 

[DemoCoder]

I just can't believe Battlefield 1942 only runs at ~30fps. It didn't run that slow on GF4-4600s.

 

[mboeller]

http://www.anandtech.com/video/showdoc.aspx?i=2268

300/300 (NV) vs 400/400 (ATI). There will suppsedly a 450/600 of the 6800 and 450/?? version of the M28. Seems to match the 6600 vs X700. Low end clocked much lower then Ati's corresponding card. Will be interesting to see the power consumption of these cards at their different speeds.

According to this german website the 6800Go @ 450 / 600 MHz will consume around 60 Watt!

 

[Bjorn]

According to this german website the 6800Go @ 450 / 600 MHz will consume around 60 Watt!

Pretty much guarantees that it will be a DTR chip only. Though it will be interesting to see how much the 300/300 6800 uses compared to a 400/400 M28. And if the notebook's with the M28 will go as high as 400/400.

 

[mczak]

http://www.anandtech.com/video/showdoc.aspx?i=2268

300/300 (NV) vs 400/400 (ATI). There will suppsedly a 450/600 of the 6800 and 450/?? version of the M28. Seems to match the 6600 vs X700. Low end clocked much lower then Ati's corresponding card. Will be interesting to see the power consumption of these cards at their different speeds.

According to this german website the 6800Go @ 450 / 600 MHz will consume around 60 Watt!

And according to this website, http://www.chip.de/news/c_news_12484551.html a 250/275 GF6800Go still uses 35W!

 

[nobody]

http://www.anandtech.com/video/showdoc.aspx?i=2268

300/300 (NV) vs 400/400 (ATI). There will suppsedly a 450/600 of the 6800 and 450/?? version of the M28. Seems to match the 6600 vs X700. Low end clocked much lower then Ati's corresponding card. Will be interesting to see the power consumption of these cards at their different speeds.

According to this german website the 6800Go @ 450 / 600 MHz will consume around 60 Watt!

And according to this website, http://www.chip.de/news/c_news_12484551.html a 250/275 GF6800Go still uses 35W!

And i wouldn't trust one of these websites, they both have written to much BS in the past.

 

[Pete]

man looking at anands benchmarks both of those cards are great.

THough the ati card is faster in most of the games and sometimes by a huge amount.

Only question is if these two cards are going to be priced the same. If the nvidia part is priced lower it will be very attractive

The 6800go acquits itself very well, considering its much lower clocks (M28 is clocked >33% higher!). It's only really behind in Far Cry and SW:BF, and Far Cry should become much closer once Anand tests with the 1.3 patch.

But I agree that price, not clocks, is everything. If ATi can deliver the M28 at the same price as the NV41M, then those numbers show ATi in a slightly more positive light.

I just can't believe Battlefield 1942 only runs at ~30fps. It didn't run that slow on GF4-4600s.

You mean Star Wars: Battlefront? BF:Vietnam hits 100+fps on both at AT.

BTW, I'm curious why every web site isn't posting at least some power draw figures. Kill-a-Watt meters are $35, and it should be relatively straightforward to measure AC power draw in idle and load tests and publish the difference.

 

[mczak]

And i wouldn't trust one of these websites, they both have written to much BS in the past.

It's certainly true that these are not reliable sources. However, the fact that some review units apparently had clocks below nvidia's suggested clocks (for instance the sager (a 12-pound "notebook") of tomshardware and the Voodoo Envy of extremetech) could indicate that power usage may indeed be a little excessive.

 

[kemosabe]

According to Brandon Bell:

ATI has also developed a 16-pipe version of M28, but this product will likely never see the light of day.

:?