r600 was a total replacement it not necessary for company to say we are replacing important thing is what they do
r600 was a total replacement it not necessary for company to say we are replacing important thing is what they do
When i'm being paid i always do my job through.
I don't think the R600 is the right card to make this case, as ATI made two low end parts and a few laptop parts. There was around a six month gap between the 2900 and the 3800, so I'd still count that as a refresh. ATI might have pushed the 3k series in early, but they didn't dump and burn the 2k series to do so.Originally Posted by kromosto
anyone want to take a wild guess as to when tsmc will have their 28nm process ready for nvidia's shrinked fermi ?????
ill start and say Q4 2011
Q4 2010 or Q1 2011. Risk production very early Q4.
From what we have been hearing, TSMC's 28nm process is quite mature already (~60% yields @ LP and ~25% @ HKMG / HP) considering the resources that were dedicated to it when their 40nm process tanked.
Last edited by SKYMTL; 03-19-2010 at 07:33 AM.
I said it should not that it will. See the difference? Yeah, read between the lines smart guy.
ok my bad i didnt mean r600 i mean x1800old days and i am an old man just sorry about my memory
When i'm being paid i always do my job through.
Tsk, don't encourage goldentiger, he'll post his list again, he's done it when nobody even asked. I'm sure the people with cards that died on that last Nv driver agree with you, quality drivers! meh
allright... and at what point in the production did we start to hear about 40nm problem at tsmc ????
Over 6 months before the 5xxx series ever launched, first general indicators could be considered the weak supply of 4770's.
Work Rig: Asus x58 P6T Deluxe, i7 950 24x166 1.275v, BIX2/GTZ/D5
3x2048 GSkill pi Black DDR3 1600, Quadro 600
PCPower & Cooling Silencer 750, CM Stacker 810
Game Rig: Asus x58 P6T, i7 970 24x160 1.2v HT on, TRUE120
3x4096 GSkill DDR3 1600, PNY 660ti
PCPower & Cooling Silencer 750, CM Stacker 830
AMD Rig: Biostar TA790GX A2+, x4 940 16x200, stock hsf
2x2gb Patriot DDR2 800, PowerColor 4850
Corsair VX450
ATI launched their HD 4770 in April 2009. The first rumors of low yields from what I can see in Google News were from December / January 2008.
I had nearly forgotten that as well, because the 1900 was really strong.ok my bad i didnt mean r600 i mean x1800
If nvidia can fix the problems then this card will be a must have, if a refresh has the raw performance with all the visual enhancements it offers via transparent AA ect I'd sleep in a tent to get one early.
If those memory clock speeds are true that is quite interesting. I take 1 of 2 things from this. A) GF100s architecture doesn't require high bandwidth or B) Nvidia epic failed with the memory controller.
Hopefully it is the former. If true, this is the fist time in a while where Nvidias flagship didn't have rediculous memory bandwidth in comparision to ATIs line ( 2900 being the exception with its 512bit bus ) Hell the GTX 280 has more memory bandwidth than the GTX 470 if this info is to be believe but again as I mentioned perhaps this isn't a big issue in light of the architecture changes. Interesting none the less...
As far as core and shader clock, 1400 seems reasonable. Those numbers would put the 480 at roughly 2x the GTX 285... at least on paper (doesn't this sound familiar lol... ; eg 240 - 480 CC, similar clock speeds ect )
Feedanator 7.0
CASE:R5|PSU:850G2|CPU:i7 6850K|MB:x99 Ultra|RAM:8x4 2666|GPU:980TI|SSD:BPX256/Evo500|SOUND:2i4/HS8
LCD:XB271HU|OS:Win10|INPUT:G900/K70 |HS/F:H115i
At least they priced them with some sense, and this isnīt very common in the green side.
BTW Would be great to know how many ROPs and TMUs they finally have.
Going on a 480 SP GTX480, it would have 48 ROPs and 60 TMUs.
For the 448 SP GTX470, it would have 48 ROPs and 56 TMUs.
512 SPs would yield 64 TMUs and 48 ROPs.
What a pity that I only have a 1000w psu ffs
If the GTX 470 has a cut down memory interface, how would it retain the same number of ROPs?
Remember, on the GF100 architecture, ROPs, the memory controllers and L2 cache need to scale in a linear fashion to one another. ROPs come in groups of 8, L2 cache in blocks of 128KB and memory controllers in 64-bit increments.
Problem is that you're talking strictly shader power - in terms of TMUs and ROPs, theyre the same or lower than the GTX 285...
But that can mean anything... what structure and die size are the test chips at? Obviously that would be excellent for a 300+mm2 complex test structure. Not so hot for a 50mm2 dram test structure
have my single 5xxx now for about 4-5 months playing latest games, never had ANY driver issue and i always update after a few days of release to the latest version, there are always a few bugs you find those for ALL vendors.
what are people always complaining about on single card setup, sure it isn't the person behind the keyboard and mouse... I haven't posted anything when not asked about it. Seriously, you don't need to lie to try to bolster some imaginary position. I posted it exactly ONCE... and the thread I was having a conversation with someone in was locked, so I posted it into the next section of the Fermi thread. That is not "posting again, doing it when no one asks" repeatedly as you imply. Long story short, I tried two different 5870's with two different setups, each in each one, no dice: lots of rendering errors, cruddy control panel, many glitches/annoyances, etc.
No need to be insulting toward people... it's not the "person behind the keyboard and mouse". Driver issues don't affect EVERY SINGLE LAST PERSON who has a product always, as anyone who works with computers on more than an MS Word level knows. I assume since you're on these forums you already knew that, of course.
Last edited by GoldenTiger; 03-19-2010 at 09:05 AM.
Oh, I agree. But with risk production still ~5 months away and volume production 8-9 months away, 60-65% on a DRAM test structure and 25% on a HP part is encouraging all things considered.But that can mean anything... what structure and die size are the test chips at? Obviously that would be excellent for a 300+mm2 complex test structure. Not so hot for a 50mm2 dram test structure
Last edited by SKYMTL; 03-19-2010 at 09:02 AM.
the saying "epic fail" is getting annoying already, next time i see epic fail im going to break my keyboard on my face LOLLLL.........
on a serious note, seeing how the memory will make minimal performance increase for most people, im sure nvidia chose lower frequency gddr5 to keep cost down, thats my guess.
so the 60% yield are on a dram test chip and not some semi complicated process???? we might have to expect some more problem then .....
I would expect 4000 (quad) for the top model.
ok not 58xx driver problems againits fermi thread so lets stick to topic i am really excited and don't want to hear any ati related cr_p
XFX GTX480 & GTX 470 pictured via Hardware Canucks
http://www.hardwarecanucks.com/news/...ards-pictured/
Bookmarks