Sure it was nine G80 SKU's, and not G8x's? Being spoiled for choice is one thing, but nine derivatives of essentially the same thing is just nonsense.
Sure it was nine G80 SKU's, and not G8x's? Being spoiled for choice is one thing, but nine derivatives of essentially the same thing is just nonsense.
Originally posted by: tuteja1986
lol newbie ;( ... it takes atleast few months to counter attack ... i say G81 will come out in April earliest. Also Nvidia wouldn't need to worry about R600 when they have all noobs believed Nvidia is the best and ATI knows crap all. A noob like you Giffen
if you ask me, nvidia are not just and have not just been sat on their arses.
i personally think they havent played all their cards, for me G80 doesnt seem like the full thing. why release your absolute best hand, when the competition can be beaten easily with a worse (relatively) hand and be beaten for some months?
its awesome, no doubt, but i think theres a little more to come. remember that G80 is manufactured on a tried and tested 90nm. the move to 65nm surely isnt far off and when it happens, i think were gonna see some nice clock increases. (if ati is aiming 700-800mhz, no reason why nvidia cant aim for the same, and they will of had time to tweak and slim an already impressive design)
i heard a rumour that G80 was infact designed with a 512bit memory in mind, but that it wasnt fully utilized because of cost (cost to implement, cost of it not working reliably). 384bit does look odd, as generally things tend to just go to another power of 2.
personally i dont think we'll see a refreshed g80 with a 512bit memory bus (dont think they would shrink to 65nm and increase bus width) thought id like to be wrong. id envisage a die shrink and copious clock boost in a effort to combat R600.
well the way i understand it is that ROP's arent fully utilized most of the time.
i remember the 6800's had 16rops to 16 pipes, but for the 6600 nvidia used 4 rops to 8 pipes and it still mopped the floor with ATi's offerings
if pixels are gonna spend along time in the pixel pipeline, then your ROPs are just sat doing nothing.
if you cook up some good scheduling/allocation thing, then you could get away with it?
something like that, its not worth it to have 1:1 ratio on rops to shaders, most of the rops just wont get used.
Originally posted by: apoppin
there is already a Part Two:
in part
http://www.theinquirer.net/default.aspx?article=35708
January 30th!
:Q
edit:
HIGHly doubtful
let me guess .. Feb or March ... DAAMit will lay low till then
i dont believe that lol
still i wanna see the cooler..... i love a good meaty air cooler. especially well engineered ones. hopefully its lots of copper and heat pipes, it'll look like a work of art!
which is something i just dont get
Ati's mobile parts often have very good heat/power figures....yet none of this seems to filter up to the desktop cards. though again ati's high end mobile parts are that scarce that no ones tested one (x1800 mobility radeon?)
and another thing
why design the cooler to be bigger than the PCB?
if the cooler is gonna dwarf the PCB why limit yourself? why not just make the PCB bigger (match the HSF size) and give your self more breathing room?
its a bit daft to be saying hey look we crammed all this ace stuff onto a PCB the same size as our old PCB's...... but oh yeah it still wont fit in alot of cases coz the HSF is just mahoosive.
seems like pointless exercise for me lol
read the bloody article man
64 shaders yes, but 4 x more work they can do! (if the INQ is correct of course, which remains to be seen)
you havent been able to compare pipes to pipes for ages now.... its just not that simple anymore
Hmm... I want to get a DX10 card for my rig. I have a 7900GTO that is eligible for step up for 50-some more days. I think I'm gonna get a 8800GTX at the end of my step-up, then sell it to my roommate and get a R600 He's a nvidia fanboy so I shouldn't have that much of a problem
He's a nvidia fanboy so I shouldn't have that much of a problem 
5150Joker
Diamond Member
Hmm R600 monster + Vista/DX10 official + Crysis = happy 5150 Joker. If the rumors INQ is floating around about its PCB being shorter than the 8800 GTX and only consuming 200-220W then this thing will be a winner. No point in owning a DX10 card right now though :\
Well theoretically they can do 4 times the work. In practice what does it really work out to? 2 times the work at best is my guess. Maybe hit 2.5-3 times the work done each cycle after a year of optimizations.
Edit- it will be very close as always. Both these companies know exactly what the other is coming out with for each wave of cards. When they can get it to market is a little less known by each...
It all depends on the shader code. If a relatively simple fragment program writes to all 4 components of the vector (RGBA) then I see no reason why the vec4 shader would not approach the theoretical 4x performance. However, if some color components have a dependency on the other color components or if not all components are being used, then you could see a drop in efficiency. A lot of this also depends on the shader compiler and how well it optimizes the code.
apoppin
Lifer
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
Yea, G8x deriavtives. My bad.
Edit - Heres the link. Seems to say G80 derivatives..
Link
And heres the one with the 5 G80 derivatives.
Link 2
G80-400 (This could be the 8800 ultra)
G80-200 (Possibly the 8800GT)
G80-600 (mid range e.g 8600?)
G80-875 and G80-850 (low end e.g 8300?)
8800GTX (G80-300)
8800GTS (G80-100)
Should post this in the G80 thread.
lol the PCB shorter point is maybe a little moot, considering INQ also said that the cooler is bigger than the card. though it may only be like 10mm longer or something, but we dont know for sure.
Intresting times. My wallet is not the fattest around so I have plenty of time to ponder which one of these chernobyls to get
My only consern are the drivers. Seems to be the trend to release hardware as fast as possible and drivers are lagging behind badly. Seen plenty of people complain about G80 drivers on other forums so it isnt ATI only thing anymore.
My only consern are the drivers. Seems to be the trend to release hardware as fast as possible and drivers are lagging behind badly. Seen plenty of people complain about G80 drivers on other forums so it isnt ATI only thing anymore.
I expect the R600 to have the performance edge over nvidia, even if they have to run a monster cooler and have it suck some silly amount of watts (/me still remembers the massive uproar over the 6800's power requirements and now look at us). That said ATI releasing a card and then actually being able to buy it are different things.
I expect nvidia will be more interested in selling lower end 8 series cards as that is where a lot more of the money is, and where ATI won't be able to compete yet.
Personally I'll wait for the smaller, cooler second gen cards to come out. By then we'll have some apps that really need the performance provided (for something more then yet higher levels of AA, and AF), and hopefully the prices will have settled down to more reasonable levels.
I expect nvidia will be more interested in selling lower end 8 series cards as that is where a lot more of the money is, and where ATI won't be able to compete yet.
Personally I'll wait for the smaller, cooler second gen cards to come out. By then we'll have some apps that really need the performance provided (for something more then yet higher levels of AA, and AF), and hopefully the prices will have settled down to more reasonable levels.
Originally posted by: apoppin
that is pretty sexist
say that to your GF
:Q
Why, yes... You are right
I decided to tell my wife that... She agrees!
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter