g70?
- Thread starter kingmike
- Start date
I know this is a prototype but there seems to be alot of heat. Is that normal for prototypes (I know that explosions are possible in early development but at this stage?...). The mention how how many watts to drive this beast is a little concerning, hopefully they can refine this along with a die shrink.
Anyone else have thoughts?
Anyone else have thoughts?
I'm suspicious of the marking on the package.
I'm not sure what they're doing in w/Korea markings. The Fabs are in Taiwan, and most of the package houses are there or in Hong Kong.
But assuming they're real, don't worry too much about the damage to the board. The upper right hand number (0505a1) suggests to me that it was packaged on May 5th, and is of first silicon (of the first tapeout w/no metal revisions).
This would suggest that this is one of the first chips back from the fab, and likely there for a physical mockup and not a functional one anyways (or anymore). Its not too surprising to kill a few of the first batch.
I'm not sure what they're doing in w/Korea markings. The Fabs are in Taiwan, and most of the package houses are there or in Hong Kong.
But assuming they're real, don't worry too much about the damage to the board. The upper right hand number (0505a1) suggests to me that it was packaged on May 5th, and is of first silicon (of the first tapeout w/no metal revisions).
This would suggest that this is one of the first chips back from the fab, and likely there for a physical mockup and not a functional one anyways (or anymore). Its not too surprising to kill a few of the first batch.
ATI report their revs as "A<Silicon Revision><Metal Revision>", so the A13 on R420 would suggest first silicon but then went through two metal revs subsequent to the intitial version. Over the past few years virtually all NVIDIA's parts have been reported as just "A1" so I actually think they are numbering the revs the same as ATI, but missing the metal rev number off the end (i.e. we can only see that this is first silicon, not the number or metal revs).
Ah, that sounds plausible.
I'm sure other companies can be different, but A1 is our first silicon. A2 is our first metal rev. B would be a full layer rev. We also have a prefix for what fab it came from. (T for TSMC, U for UMC, etc.)
FWIW, here's a 6600GT chip, with A2 on it.
www.ixbt.com/video2/images/nv43/6600gt-chip.jpg
A NF3 with A6
http://pc.watch.impress.co.jp/docs/2003/1022/sindan01_12.jpg
HSI with A4
http://prohardver.hu/html/vga/nv43_agp/cardz/hsi.jpg
GF2MX with B2
http://www.technoyard.com/hardware/video_cards/nvidia_ref_MX/chip_lg.jpg
www.ixbt.com/video2/images/nv43/6600gt-chip.jpg
A NF3 with A6
http://pc.watch.impress.co.jp/docs/2003/1022/sindan01_12.jpg
HSI with A4
http://prohardver.hu/html/vga/nv43_agp/cardz/hsi.jpg
GF2MX with B2
http://www.technoyard.com/hardware/video_cards/nvidia_ref_MX/chip_lg.jpg
If this is a 2005 week 5 board it's definitely got to be an early prototype. They probably overclocked it to see its potential and I can imagine that they got close to 150 Watt. But many months have passed now so the power draw and length of the card could have decreased significally. Let's wait and see...Khronus said:
Oh yeah, on that 150W thing: The resoning its based on is simply rediculous, by the same means both X850 XT and 6800 GT would also use 150W. The presence of the 6-pin power supply line simply shows that the design is at least close enough to the 75W limit of your everyday PEG slot to warrant the use of an external source.
According to the recently leaked nVidia document, the second 7800 GTX in an SLi system requirers an additional 8A on the PSU's 12V line (26A vs. 34A), that's not even 100W. (still a lot, though)
Anyways, going by the history of nVidia's PSU recommendations, this figure probably contains some healthy safety margin, I guess we'll have to wait for xbit to measure its real world thirst.
According to the recently leaked nVidia document, the second 7800 GTX in an SLi system requirers an additional 8A on the PSU's 12V line (26A vs. 34A), that's not even 100W. (still a lot, though)
Anyways, going by the history of nVidia's PSU recommendations, this figure probably contains some healthy safety margin, I guess we'll have to wait for xbit to measure its real world thirst.
Similar threads
