• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

R600 and G80 delayed.

D

dnottis

Supreme [H]ardness
Joined
Aug 14, 2001
Messages
4,768
Looks like the G80 officially got kicked back...

http://theinq.net/default.aspx?article=34479

"We all expected this chip in October or November but it looks like the chip might be delayed even further."

Oh well, maybe by February or March we'll see the R600 and G80 actually emerge. Sucks for all those folks waiting on a GPU in Sept / Oct though. I guess it's officially time to jump on your 7950 GX2 or X19xx video cards seeing as though it will easily be another 3-6 months.
 
The Inquirer isn't the most reliable source, but a delay in next gen video cards is certainly possible. Personally, I'd rather see them delayed if it means they perform better.
 
Since Vista seems to be delayed to march, it's a genuine possibility the videocards is following this.
 
How can they be delayed when their release dates hadn't been announced yet?
 
how can that vr-zone link be correct i thought it was stated that g80 would not be unified? close to 700M transistors!?
 
-=JacKaL=- said:
how can that vr-zone link be correct i thought it was stated that g80 would not be unified? close to 700M transistors!?
Click to expand...
Moores Law still holds true!

Moore's Law is the empirical observation that the transistor density of integrated circuits, with respect to minimum component cost, doubles every 24 months.


7900GTX = 278 Million Transistors
8XXX = 700 Million Transistors Sounds about right.

Also note that the G80 is the first set of cards of it's kind to support DX10 and a lot of new features.
 
Also note that the G80 is the first set of cards of it's kind to support DX10 and a lot of new features.
Click to expand...

Ther's speculation that the G80 is only partial DX10. Think of it as a DX9 card that can do DX10 as opposed to a DX10 card that can do DX9. This is only speculation.
 
I tend to agree with those who say it is delayed because of Vista. If they launch before vista is released then no one is going to be able to use all that new powerful hardware, plus they jst released the 7950....and if they launch their new stuff in 2 months then they wont fully achieve profits on what they just released
 
Wow, when I first saw a blurb about that 384-bit memory interface, and 768MB ram, I thought it was a laught. But honestly, oince you get above 256-bit (4x64) busses, the next step to 512 is mighty expensive in terms of PCB cost.

384-bit is a nice compromise, although the required memory sizes will be a bit confusing for people.
 
The cards are going to be released around holiday season...just wait and see...wait and see :cool:
 
defaultluser said:
Wow, when I first saw a blurb about that 384-bit memory interface, and 768MB ram, I thought it was a laught. But honestly, oince you get above 256-bit (4x64) busses, the next step to 512 is mighty expensive in terms of PCB cost.

384-bit is a nice compromise, although the required memory sizes will be a bit confusing for people.
Click to expand...
Well, the X1800 and X1900 cards technically have 512-bit memory interfaces, since there's two 256-bit buses going in opposite directions. I doubt there'd be much difference in PCB cost if that were a single bi-directional 512-bit bus instead.
 
Zero82z said:
Well, the X1800 and X1900 cards technically have 512-bit memory interfaces, since there's two 256-bit buses going in opposite directions. I doubt there'd be much difference in PCB cost if that were a single bi-directional 512-bit bus instead.
Click to expand...

Are you really sure about that?? ;)

It's an internal 512-bit bus...
mem is still 256 isn't it?
 
The memory controller works as a 512 bit bus, 256 bi-directional (seperate paths), the paths to the memory is still 256.
 
We will definitely see the G80 by November, I see it in the crystal ball which has remained dark and without any prophetic visions for some time. Today it woke me in the morning and when I looked Voila. :D

I imagine several cards coming too. Such as 8200s 8600 and even the 8800s. ;)
 
Trimlock said:
The memory controller works as a 512 bit bus, 256 bi-directional (seperate paths), the paths to the memory is still 256.
Click to expand...
Technically it's two separate uni-directional buses going in opposite directions. However, my point was that the X1800 and X1900 memory bus implementation would be the same physical size as a single 512-bit wide bi-directional bus. Therefore, to implement one wouldn't cost any more than current boards to produce.
 
Lord_Exodia said:
Um... No offense, but what does the RD600 motherboard chipset have to with with this thread? The thread is about the G80 and R600 The RD600 is a motherboard chipset
Click to expand...



haha yea we already got that out there ;) (refresh the thread and check a page back)
 
dwilson got me all excited but then i realized it was their mobo chipset and then i was sad again :(
 
Zero82z said:
Technically it's two separate uni-directional buses going in opposite directions. However, my point was that the X1800 and X1900 memory bus implementation would be the same physical size as a single 512-bit wide bi-directional bus. Therefore, to implement one wouldn't cost any more than current boards to produce.
Click to expand...

No, that's not true. The 512-bit "ring bus" is internal to the chip. Think of it as a highly-optimized take on a crossbar memory controller, with the crossbar logic spread out across the chip to prevent hotspots.

The "ring bus" interfaces with external memory at 8 points, and each interface is 32 bits wide (8x32). This is contrary to ATI and Nvidia's previous crossbar design, which were (are) 4x64.

http://www.beyond3d.com/reviews/ati/r520/index.php?p=05

This is a lot like the "256-bit" DualBus on the Matrox G400, which was really just a full-duplex 128-bit interface inside the chip to a 128-bit wide memory controller (the reason it had such good 32-bit performance). The full-duplex capability means more efficient use of the memory bandwidth available, but it cannot increase the maximum bandwidth.
 
Zero82z said:
Technically it's two separate uni-directional buses going in opposite directions. However, my point was that the X1800 and X1900 memory bus implementation would be the same physical size as a single 512-bit wide bi-directional bus. Therefore, to implement one wouldn't cost any more than current boards to produce.
Click to expand...

Thats true, we'll see if the 384 bus (if its true) would contribute to better performance with AA or not, I still think the 256 is enough especially with GDDR4 being introduced to us we are overcoming any problems with memory bandwidth with out having to increase the bit rate

At this point its just exciting to get the actual cards, for once I'm really excited to get either or cards and can't wait to see if what ATI is bringing to the table really kicks ass or flops
 
Am I the only one that doesn't mind if these cards aren't released until we can actually use them? Why rush them out the door when DX10 software doesn't even exist when they can use the extra few months to optimize power consumption and iron out bugs?
 
Lamont Cranston said:
I tend to agree with those who say it is delayed because of Vista. If they launch before vista is released then no one is going to be able to use all that new powerful hardware, plus they jst released the 7950....and if they launch their new stuff in 2 months then they wont fully achieve profits on what they just released
Click to expand...

In Vista with dx10, it'll be a dx10 card. In XP it'll be a dx9 card. On Linux it'll be an OpenGL card.

There's absolutely no reason next gen cards need to wait for Vista. I don't know why everyone assumes this.
 
Trimlock said:
Thats true, we'll see if the 384 bus (if its true) would contribute to better performance with AA or not, I still think the 256 is enough especially with GDDR4 being introduced to us we are overcoming any problems with memory bandwidth with out having to increase the bit rate

At this point its just exciting to get the actual cards, for once I'm really excited to get either or cards and can't wait to see if what ATI is bringing to the table really kicks ass or flops
Click to expand...

Of the specs posted, the 384bit memory interface seems the most realistic. GDDR4 was more than enough for this generation, but if you double the horsepower next gen, you're going to need a hell of a lot more memory bandwidth. I don't think they would have gone to 384bit if they didn't think it was going to get them some performance.
 
PRIME1 said:
Thanks I needed a laugh.

The Inq being an official source :D

I don't remember EVER seeing an official release date for the R600 or the G80 :rolleyes:

The Inq is one step away from telling us all about the exploits of Lindsay Lohan and space aliens.
Click to expand...

Absolutely. The Inq tries to make itself out to be accurate. What they actually do is write a story about nearly every possible outcome of every rumor. Then when one of the stories inevitably comes true, they say "look, we were right", and they ignore the 50 other stories where they were embarrassingly inaccurate.

The Inq is nothing more than entertainment.
 
Rabid Badger said:
Am I the only one that doesn't mind if these cards aren't released until we can actually use them? Why rush them out the door when DX10 software doesn't even exist when they can use the extra few months to optimize power consumption and iron out bugs?
Click to expand...

/sigh

this is posted yet again, these cards CAN BE USED with XP, there is no bugs needed to be ironed out, power consumption will go up due to the increase in transitor count, theres no reason to wait for vista for these cards to be released!

These cards are going to kick ass and take names, why you would need vista for that to happen is beyond me, not to mention when vista is shipped we will still have to wait for the D3D10 patch!

Of the specs posted, the 384bit memory interface seems the most realistic. GDDR4 was more than enough for this generation, but if you double the horsepower next gen, you're going to need a hell of a lot more memory bandwidth. I don't think they would have gone to 384bit if they didn't think it was going to get them some performance.
Click to expand...

we'll see, the GPU's act very different with different memory types, like the R580 getting a decent increase with just adding GDDR4 onto the PCB, not like we were bottle necked with memory bandwidth to begin with, alot of people are suprised we saw any noticable increases with it, how ever with all these rumored spec's and resource allocation rumors the 384bit bus might just be the blessing we wanted (bring back quincux! :D )
 
Trimlock said:
/sigh

this is posted yet again, these cards CAN BE USED with XP, there is no bugs needed to be ironed out, power consumption will go up due to the increase in transitor count, theres no reason to wait for vista for these cards to be released!

These cards are going to kick ass and take names, why you would need vista for that to happen is beyond me, not to mention when vista is shipped we will still have to wait for the D3D10 patch!



we'll see, the GPU's act very different with different memory types, like the R580 getting a decent increase with just adding GDDR4 onto the PCB, not like we were bottle necked with memory bandwidth to begin with, alot of people are suprised we saw any noticable increases with it, how ever with all these rumored spec's and resource allocation rumors the 384bit bus might just be the blessing we wanted (bring back quincux! :D )
Click to expand...

I'm pretty sure Vista RC1 has DX10 already. According to dxdiag it does anyways.
 
defaultluser said:
No, that's not true. The 512-bit "ring bus" is internal to the chip. Think of it as a highly-optimized take on a crossbar memory controller, with the crossbar logic spread out across the chip to prevent hotspots.

The "ring bus" interfaces with external memory at 8 points, and each interface is 32 bits wide (8x32). This is contrary to ATI and Nvidia's previous crossbar design, which were (are) 4x64.

http://www.beyond3d.com/reviews/ati/r520/index.php?p=05

This is a lot like the "256-bit" DualBus on the Matrox G400, which was really just a full-duplex 128-bit interface inside the chip to a 128-bit wide memory controller (the reason it had such good 32-bit performance). The full-duplex capability means more efficient use of the memory bandwidth available, but it cannot increase the maximum bandwidth.
Click to expand...
Excellent brain-food here, thanks for the explanation.
 
I just wish one or both of the these companies would just come out with an announcement. I am one of those waiting patiently for the Direct3D 10 cards (stop calling it DirectX10!) to build my next system, and the rumor mill is driving me nuts!

AFAIK, Crysis and Bioshock will both take advantage of D3D10 for teh eye candy, and Crysis is due out by years end.

They just gotta get it out there before Christmas! C'mon guys, you don't wanna miss the holiday revenue! :eek:
 
Combover said:
I am one of those waiting patiently for the Direct3D 10 cards (stop calling it DirectX10!)
Click to expand...
We will never stop calling it DX10 for the following reasons: D3D10 takes one more letter to type than DX10, and you have to un-shift to type the 3, then re-shift to type the second D--thus it's simply less efficient to bother with D3D10. Put another way, the lazy solution wins every time.

Also, who cares if the upgrades only apply to the D3D API? That's the only part of DirectX that's been of interest to gamers for quite some time now. I'm sure there will be many nice improvements in the other APIs some day, but that's the one that offers the advancements that we notice/care about. In short, 99% of the threads around here that discuss DX9 or DX8 only apply to D3D, so why do we have to get a corn-cobb up our butts about it now, all of a sudden?

Combover said:
comb
Click to expand...
If you want to talk about a real breaking of a real rule, signing your posts in the body of the message is a violation of [H]ardforum rule #25, see it here:

http://www.hardforum.com/showthread.php?t=760666
 
Commander Suzdal said:
If you want to talk about a real breaking of a real rule, signing your posts in the body of the message is a violation of [H]ardforum rule #25, see it here:

http://www.hardforum.com/showthread.php?t=760666
Click to expand...

I didn't speak of breaking any rules, but thx for pointing that one out. Original post edited.

^ Nvidia and ATI would be crazy to hold off on a graphics card release to wait for Vista. We know these cards will run DX9 and prior games, and I would expect they would peform a bit better than the current round of cards.
 
If Nvidia and ATI release new video cards by the end of the year wont there most likely be a Refresh of New Cards by the time Vista does Drop, It most likely be better to actually buy a video card around the time of the Real Vista launch.
 
by the time Vista does launch, both companies should have refreshes around the corner.
 
You must log in or register to reply here.
Back
Top