maybe I should run the benchmark to confirm such fact... 2.8 fps looks like a too low number
Printable View
maybe I should run the benchmark to confirm such fact... 2.8 fps looks like a too low number
Something does not add up on this slide. 5870 CF getting less FPS then single card? Driver bug perhaps?
I am busy now, but I will attempt to have a benchie done by tomorrow morning on 5970 under several resolutions, heaps of those numbers look odd
Sounds like newegg is vey greedy. Shame on them...
Having shadows on etc is the different in most cases of being killed by the guy behind you or killing him. I like to play with max details for this reason and like I said if I wanted anything less I would play games on a console. To get the full experience of any game at 2560x1600 you need a dual gpu configuration and anything less is unacceptable.
No it has not been reviewed yet. Most review sites are too chicken to touch water-cooled cards. It does look attractive at 850 clock at 1700 shader.
You can buy it off evga's site though
http://www.evga.com/products/prodlist.asp
ASUS GTX 580 SLI Review
http://www.overclock3d.net/reviews/g...usive_review/1
Is Hardwareheaven.com credible?
At the bottom of the page they post videos of the card's fans noise.
The GTX580 and GTX580 SLI videos are EXACTLY the same. :confused:
Load both of them up first, then try and start them at almost exactly the same time(i did very close:D) You will see they are the exact same video. Someone ought to tell them they made a booboo.
I've read the language of the article and it does have that alluring NV bias to it.
Hope they fix those vids before someone cries foul. ;)
I don't know if they can be trust, the only problem i have see was for F1, they was not able to run in CFX .... but here my result with CFX in F1 2010 with 10.10a+ CAP ( old bench screenshot, without 10.10c and new cap ) ... all maxed in game settings: 16xAF + 8xMSAA - DX11 ultra settings .... 1920x1080
They call that a driver problem ? they don't even know how to set it, or they have not use last drivers ... (I7@4ghz+ anyway)
http://img179.imageshack.us/img179/6/f12010bench.png
It is posible some one get burned by oc :shakes: :rofl:
If you see unreal performance drops, lags and weard behavior of your pc. It may be overclocked too hie.
Anyway i call that overclock data loss. It reparing data with more cycles then usial. Maybe pci-e voltages rise will help maybe gpu voltages,maybe graphic memory :shrug:
Off topic:
I know you ppl you havent play real game that slelge hammers fpu math coprosessor 100% and slight oc it makes crash of game.(is not abaut that game is bad written)
Anyways i wrote this: Tips for quality gaming + Networks And Windows Tips
for vets(old players)in game > Thats makes me master too not just admin :D :p:
http://www.bz2maps.com/phpBB/viewtopic.php?t=1597
Does a 15% performance update justifies upping the generation's number by a single digit
Either the evolution of GPU performance has *really* slowed lately, or we're dealing with fraud here (not a first for nVidia, I'm sure).
How come they can get away with it? Computer Hardware is a multi-billion industry, such tricks can make great difference (the card's real name is GTX 485)
my bad, i stand corrected!
but no, he wasnt the only one that said availability will suck... several others have mentioned that theres supposedly a small batch of cards for launch and thats it... supply after that will be bad or non-existant.
in the end, who cares though, cards ARE available right now, your right... and thats what matters... whoever wants one can get it. thats good news, and really well done by nvidia, gotta give them that :toast:
i didnt expect this... they really did a 180 after their fermi pre-pre-pre launch bs and delays...
thx! :toast:
maybe gf100 had dividers in place to adjust the clocks of individual segments of the chip? and before gf100 launched they had already figured out which ratios work best and in gf110 they cemented the ratios and could remove buffers and dividers?
and maybe the reworked power planes allowed them to move things around a bit, avoiding hotspots which in turn allowed them to pack transistors more tightly?
too bad we will never know... this is the kind of stuff id love to know, and its a shame that by the time its not sensitive information anymore, its just forgotten instead of made public... would be such a nice read...
sweet, thx! :toast:
too bad they didnt run more tests... too bad theres no way to limit a 580 to 480sps, looks like we will have to wait for a 570 for that... clock for clock a 580 is around 8% faster than a 480? and it has more sps which would result in a 6-8% boost, so... it looks like there is close to 0 ipc boost?
the boost seems to be sps and clocks only, right?
hmmm but that wouldnt explain the bittech numbers... they see huge gains at 2560x1600 of the 580 over the 480...
how can a 10% clock boost and 7% more sps result in some cases in a performance boost of more than double those theoretical gains combined?
so either gf110 is tweaked for 2560x1600 and/or aa, or maybe bittech used older 480 numbers of an older driver and compared it to a 580 with the latest driver? but i cant imagine them doing that...
hmmm
Product naming is pure marketing and both companies are doing what every one else would do (this time that is).
Both HD68xx and GTX580 have new bits introduced to them. Just to name few:
GTX580 - better culling, full speed FP16, other minor tweaks.
HD68xx - new UVD, improved display output capabilities, improved tesselation performance, other minor tweaks.
Is that enough for new generation moniker? I think it is, at least in current realities of I.T.
if we had a poll, i think we would see a majority choose 485 over 580
however since there is nothing really wrong with picking a new number, ofcoarse marketing would suggest they go with a new number to remove any implications with old issues (lack of cores, high tdp)
If they named it 485 what would they name all the other incoming GF11x parts?
so theres plans for more than just a 570 in the upcoming months till 28nm?
when is a 560 coming?