If I wanted to quote minimum mwave had the 470 for $330 when I replied earlier but I was simply quoting what was stock at NE at the time...Quote:
Originally Posted by Manicdan
Printable View
If I wanted to quote minimum mwave had the 470 for $330 when I replied earlier but I was simply quoting what was stock at NE at the time...Quote:
Originally Posted by Manicdan
Maybe I'm looking at the 460 in the wrong light, but i don't see it'll be a great buy compared to the competition with all things considered, especially if ATI cuts prices to counter it :shrug:
Bah, this gen of cards hasn't been very exciting IMO. Nvidia's broken Fermi cards and ATI's more of the same 5000 series.
Bring on ATI 6000 vs. GTX500! :)
There's a big drop in cores but core/shader clocks are unchanged, unlike with 480 to 470.Quote:
Originally Posted by Olivon
GTX 480 has 24% more FLOPS than GTX 470, while GTX 470 has 27% more FLOPS than GTX 465.
Incorrect in your mind maybe, newegg and pricewatch both show an average of $60 more for the 470 for the brands I would buy.Quote:
Originally Posted by highoctane
Nice try though...
:rolleyes:
Aside from the fact that you are overlooking the fact that the GTX470 is a very good overclocker, you show Nvidia! Teach those meanies a lesson and buy that 5850! :rofl:Quote:
Originally Posted by rogueagent6
Also, for future reference please don't try and suggest that the cost difference in energy of running a 5850 vs. a GTX470 would be anything more than a drop in the ocean for anybody that can afford a 470 or 5850 and the corresponding hardware to utilize them fully.
And the 5850 isn't a good clocker? :rolleyes:Quote:
Originally Posted by ElSel10
Are you going to pay my power bill for me? Thought not. When one card offers no tangible increase in performance over what I currently have, and uses more power, plus was 7 months late to market, why should I buy it?Quote:
Originally Posted by ElSel10
This.Quote:
Originally Posted by rogueagent6
If you have a 5850, keep it. If you are buying something new, get the GTX470, if the price difference is low.
At least that's what I would do if I was Jesus.
Here in Portugal, cheapest HD5850 is 240€, cheapest GTX470 is 330€, so a 90€ difference is quite too much to choose the 470.
That's a big if.Quote:
Originally Posted by Luka_Aveiro
http://www.anandtech.com/show/2947 <- reviewQuote:
Originally Posted by ElSel10
175W declared power consumption
From the article:
"Given the 175W load power, it shouldn’t come as a surprise that AMD and its partners are doing some recycling on board designs. The launch 5830s will be using the 5870’s board due to the similar power usage of the two cards. This is something that AMD says may change in the future if vendors want to do their own boards."
Have i said otherwise ?Quote:
Originally Posted by -Sweeper_
i havnt sean anything 5830 that used a 5870 pcb just the POS vaporX/xfx non refrence 5850 and the asus CU 5850 pcbs.
NV needs a new smaller chip to get what they want thats the only way, the gf100 is crap for gaming or desktop use its only good for benching, cuda and the extreme enthusiast. they need to drop the power down so that means that they need a smaller part not a gimped part. they have to stop the chipzall and chipzooki strategy they need upper midsized parts they cant compete by cutting bad dies down, it limmits the clocks to much as u dont get any less power consumption by having clusters dissabled
oh the 5850 overclocks, but it doesn't scale as well as the gf100 cards. the gtx470 and gtx480 respond really well to a boost in mhz, cypress must be held back by something else.Quote:
Originally Posted by rogueagent6
overclock.net member did some home benching
if you don't like the power consumption of the gf100 cards, that's fine don't buy one, but there is no need to constantly remind people who don't care which card is less thirsty. it's like saying "my corvette is better than your viper because it gets better gas milage" :shrug:
i thought that this is on NV wanting another 8800gt. they need to make it so it works in normal computers and not custom ones like we here all have. u cant just stick a gf100 into a store bought computer like u can with a g92 or the 5770/5850 and if u are targeting that $200-300 range u need the lower consumption. that means that they need a new chip, if they do get a new chip thats smaller on that architecture so they demonstrate that scalability that they were saying the fermi was all about and by cutting the die and pad size they should be able to do 1/3 the shaders with half the power as tsmc 40nm needs alot of extra power as u make the die bigger (its not linear.)
Reread that quote you posted. They are using the 5870 board due to similar power usage of the two cards.Quote:
Originally Posted by RaV[666]
The 5870 board is a result of the GPU on the 5830 having similar power usage to the 5870, which implies 5830 GPUs are poorer parts that can't make the cut for a 5870 or 5850, and the higher power usage is a result from poorer grade silicon and necessitates the more expensive 5870 PCB rather than that of the 5850.
:up: :ROTF:Quote:
Originally Posted by Luka_Aveiro
i like your humor :D
i think you mean 2/3? :DQuote:
Originally Posted by zanzabar
and a dual 465 card... how?
if the single 465 has a tdp almost identical with that of the 470 but is around 33% slower... that doesnt make any sense... then they could just make a dual 470... same power but 33% more performance...
i'm not sure i'd put ANY of those into a store bought computer, seeing as most store-boughts have a 250-350watt psu on board...Quote:
Originally Posted by zanzabar
also, power consumption increases because of a lot of things... die size being just one. fermi's "power problem" is leakage... but that leakage is also the source of it's overclocking prowess. B1 will change many things...
leakage isnt really the power problem, they have to run way more power through the core based on die/pad size due to bad tracing and power piping on the 40nm node from tsmc parts, u even have to run 2 power trenches with tsmc 40nm per set of transistors, so when u have a product with the performance that a market needs and its the point were your core sales are going to come from u should accommodate them. look at the g92 they have products for the desktop that go less than 65W and some that are over 150W. ati has the 5770 and thats only 90W, most newer computers with a quad core seam to have a 450-500W psu as thats the cheapest that u can make right now and HP, dell and acer are starting to use 1 psu for the entire non gaming desktop line so the capacity is there. the gf100 as of now with the 470 and an i7 both at stock u can get over 600W with total system draw thats just to much for the normal end user to stick into a computer.
im in no way saying that the power consumption makes the 470 or 480 a bad card they are fine for the market that they are for, sure i would like to see less power draw and higher clocks like every1 else, but when u want to take upper mid performance and give it to the masses as the g92 did and thats what seamed to make it popular was that u could shove 1 in almost anything and if u wanted more u could get a higher clocked one or a second one. with the 460 or 465 they are just doing what ati did with the 5830 and unless u want to oc the card and upgrade frequently its a terrible card since its not a smaller die but only has half the shaders/rops it uses the same power as the higher end and is not vary fast, this is the same the only way to cut consumption to make it work in this application is to cut clocks, cutting clocks leads to bad performance per watt and like was said the gf100 likes clocks. so OEMs wont use it as it will cost them more and make the computer louder.
so what they need to make it like the g92, is to cut power and die size so u can use it without a new psu on normal systems (not high end not bargain) and keep the clocks up for performance, it seams like a new die would accomplish that. NV had said that the gf100 was scalable, that means that they can make new parts from it not that they can disable clusters so they still leach power.
What they need to do is have something with 406 or 384 shaders with 750mhz clocks. The low clocks are really hurting the 470.
While they are at it, get the card at the 300 dollar price point. Such a card would be slightly faster than the 5870. However it might need a decent cooler or a revision to make feasible at the 300 dollar price point. I wouldn't say it would have a 8800gt effect but it would atleast be the best bang for your buck videocard currently available.
yeah good point zanzabar...
what happened to fermis amazing scalability?
fermi has taped out how many months ago? 7? 8? and there still isnt a cut down fermi gpu...
and thats supposed to be a very flexible and easily scalable design?
ati had rv870 and rv840 launch within weeks, and a complete lineup from top to bottom within, what, 3 months?
its going to have been almost 1 year by the time gf104 is widely available in retail!!! 1 year! and thats not scaling at all, gf104 will probably replace gf100... nvidia will offer single gf104 for mainstream and dual gf104 for highend...
yeah, thats a very flexible strategy alright nvidia... :D
tbh i wish theyd have done this from the start... id have loved a 384sp gf100 for 250$ and they could have sold dual 384sp cards instead of 480s, which would have been faster, run cooler and cost them a lot less than 480s...
i really hope nvidia will finally get over their arrogance and revise their strategy...
the current market situation sucks, without any real competition and all... :/
So with only 352 CC's can we expect better than 5770?...Quote:
Originally Posted by Olivon
This is VERY disappointing as I was expecting 416 CC's and would have been simply disappointed with 384, but 352!... :confused:
Perhaps they literally mean the GTX465.5 will replace the 8800's (9800, GTS250) and will be the new 8800 effect!
Hmmm would be time for a Hitler video if they wern't banned...
Edit: Lol! Perhaps this is how Nv got >20% Yields
To be entirely fair, supposedly Cypress was finished quite early. Had they launched the 5870 and 5970 as soon as the designs were finished and the chips were ready, the lead time for the cut down parts would have been a bit longer.Quote:
Originally Posted by saaya
At this moment, all the rumor about nvidia is as stink as junk food until it actually materialize.
yeah... still...Quote:
Originally Posted by Oberon
1 month after the launch ati had 2 physical chips in the market offering at least 4 products... its been 1 month since fermi was finally launched and there is one chip out with 2 products...
its time that nvidia realizes that its just not a good idea to have one big chip at a time and recycle the last generations chip for the mainstrem and entry level segments... i dont believe that 40nm is a one off... i really dont think we will go back to the yields and 2 year process cycle weve seen so far...
Nvidia's Geforce GTX 460 gets listed - Fudzilla
Hope that's not real prices, 280-300€ :down:
All Fermi products are priced incredibly high in Europe right now.
http://www.fudzilla.com/content/view/18777/1/Quote:
Some of European e-tailers/retailers have jumped the gun and have listed Nvidia's upcoming Geforce GTX 460 card. Unfortunately, the specifications are scarce at the moment and neither one of those shops have the precise clocks, or any other information for that matter.
The only info regarding the specs of the card is that it supports DirectX 11 and that it has 1024MB of memory. The good side is that we now have the price of the card, or at least an average price that you should see in Europe once this card shows up. The listed GTX 460 card hovers around €300, and while some shops are listing it for as low as €284 others go as high as €332,36.
This price puts the upcoming GTX 460 in direct competition with the HD 5850, and all we need now is to see if the performance is on level with this price. The stores that are listing these cards will certainly get an angry call from Nvidia reps and in that case we have a screenshot ready, as those things makes us quite happy.
Some shops are bold enough to say that the card will be available in next 4 to 7 work days while other are more realistic with availability date set at 10 to 20 work days.