The bolded part doesn't fit with the leaked pics of Cayman card equipped with 8 GDDR5 chips on the front side. :shrug:
Printable View
The blurred ROP part could be anything, but 32 or 64 are about the only numbers that could make sense.
Funny how all slides leaked to the internet except one all uses codenames, not product name.
I realize there aren't 100% answers yet, but I've been following this thread since the beginning, and I just keep getting more confused... :shakes:
First BARTS was the Mid Range replacement for the 5770 (6770 for <$200) approching 5850/5870 performance, to be released in Mid October.
Then there are conflicting reports saying theres a delay on it's release until November...
Personally I'm hoping the delay is on the Caymen parts (5900 performance?)...
What really sucks is the fact that I've been holding on to my 4870 for quite some time.
I'm not a huge gamer, but this card is getting a little long in the tooth, and I'm ready to upgrade my GPU!
I'll hold onto this card for a few more weeks, but if anyone has any legit info I'm all ears... ;)
This is an insider info request (within NDA of course), is it worth waiting for?
Thanks........ Dave
it will be worth the wait, besides it will be worth it to wait because price drops for the HD5800 series cards will start....especially for black friday im sure
Educate the ignorant, what do ROPs have to do with RAM chips? Are the ROPs tied to memory controllers? If so, how many ROPs per 64-bit controller? Are TMUs tied to memory controller(and that way to ROPs)?
I know I should read those B3D architecture analysis articles some day...
Anyway, it seems clear from the pic that there are 32 ROPs in the 3 RPE part.
So will Cayman be launched as well this year? Or just Barts?
After doing a bit of noobish image forensics I'd be more inclined to guess:
1 RPE 32 TU 16 ROP 640 SP for Caicos?
2 RPE 64 TU 32 ROP 1280 SP for Barts.
3 RPE 96 TU 32 ROP 1920 SP for Cayman
http://img829.imageshack.us/img829/3...tial237549.jpg
Yes bro AFAIK since RV 770, the external buswidth is tied beetween the ROPs and memory controllers, beetween 4 or 8 ROPs per 64 bit MC. 8 VRAM chips on the same side mean 256 bit buswidth config (32 bit per chip), and 256 bit means 4 MC, my math concludes that the ROPs number should be 32 (16 is out of question for Cayman, though at first i thought perhaps Bart would be a 16 ROP chip, alas it turns out to be a 32 ROP chip too).
TMUs are tied to the SIMD array IIRC (that's why B3D forum is questioning the 64 TMUs implied in Bart, since that would mean a 16 SIMD array with 80 SPs per array, creating a non power of "2" SIMD array with all of its chaotic implications). CMIIW
AFAIK Cayman will be launched this year too (November launch, December real availability ?). Antilles is still questionable (no leak on the final product so far), while the lower end chips are slated for next year.
holly hell!!!! if that is not a fake then cayman is some kind of monster.... lol now about that fermi mid-life kicker.... it better be some kind of "double dragon flying kick" to keep up with this baby
As was noted before, why would they want to hide the GPU-Z version number? Perhaps to hide the fact that they edited the screenshot of a version that is known not to be able to display the HD6800 properly like this?
we're so close to release date... thb I would not think that to be fake, numbers seem pretty close to stipulations
WTH..... 2gb :shocked:
this is fake
Sadly, I missed that.
The rest of his entries are mostly about attacking a site, that is run by probably the most fanATIc guy, you have ever seen. He's like a Charlie, but worse. If he posts something new, I can translate other stuff in case he posts more.
To make it look real?
We all know, that GPU-Z is mostly a database. He might have access to new HW, but why risk a good relationship with somebody and leak it? Doesn't make sense.
or fack picture (GPU-Z) :DQuote:
or a kick-ass card
48 rops, 2GB card...
wut
maybe just a bad dream for nvidia fanbags
and maybe just a very bad dream for ATI fan... When GF110 come and kick HD 6800 ass :D:D
whatever it is, new competitive cards from both companies are great for everyone ;)
however, i don't think nvidia will have their cards ready before 2011.
if cayman really features the specs we're talking about here, it'll be more powerful than i'd have expected. let's hope it's true.
that's exactly my point, we might even see another de-ja-vu from what happened between 2009-2010
nobody know what nvidia hide for HD 6000, I think nvidia have surprising for HD 6000 Because everyone Including AMD Surprised from nvidia When Release GF104 and Produced GTX 460 Which is the best deal you Can get it especially 2 GTX 460 SLI and Embarrassed More than card form AMD . nvidia Able to do new core (like GF104) In a short period of time so do not Exclude that nvidia working (now) on new core (GF110), may be will come After a month 2 month 3 month We do not know yet but it will come.Quote:
Only in woodscrewmark I assume. I do not expect big ass kicking from them, not until 25nm, even then it's just a dream.
I agree with you :)Quote:
whatever it is, new competitive cards from both companies are great for everyone
however,
The 104 is nice because nvidia sell it dirt cheap, for the amount of silicon it uses it isn't that impressive. But now nvidia have put a midrange chip that big into the market, it seems AMD have no problem making a bigger midrange chip too.
I think nvidia can make competing chip even if AMD go bigger, but the bigger AMD goes the harder it'll be for nvidia to beat them, so unless they can increase their transistor density it probably wouldn't be too smart to push out another 3 billion transistor GPU
Indeed it will come, after the 6800 series, you see they took on the 5K with the 460 and the 6K will have an answer for that.
They could very well be faster than the 6K but I wouldn't expect a larger difference than what the 480 had over the 5870 and comes the holidays and the Chinese new year so it will be the end of Q1 when it can reach the shelves unless they do it before xmas, but it's way too silent for that to happen.
How can 48 ROPs work with 256 bit buswidth ? the comeback of a ringbus ?? :confused:
AMD to debut Radeon HD 6000 series on October 19
http://www.digitimes.com/news/a20101001PD208.htmlQuote:
Monica Chen, Taipei; Joseph Tsai, DIGITIMES [Friday 1 October 2010]
AMD has decided to officially launch its Radeon HD 6000-series on October 19, a week later than the original schedule of October 12, with the Radeon HD 6870 possibly the first for launch.......
gtx460's were sold fairly closely at a loss, it was initially built to compete with Junnper, however it could rival low end cypress. Lets hope they can get their eggs in one basket and avoid releasing yet another half baked product... it really hurts us buyers when this don't happen
**EDIT**
there i also a big hole in that statement 1 to 3 months to "it will come" can mean the difference of (first mover advantage, pricing advantage, marketing advantage), when talking about releases its not only the specs what count, but how the whole exercises is carried out.
GF100 was superior to cypress in performance (ALONE) however pricing and the release time made it a not so desirable product. If your statement is right (which I doubt) and GF110 comes out (whenever) they better do it before Christmas, otherwise u might see them releasing Q1 2010.
methinks it depends on performance, availability and MSRP
HD 5850 and HD 5870 prices went up after launched
GTX 460 786MB and 1GB prices went down after launched
You can get a Galaxy GTX460 1GB GC for $159 (after MIR) now
How does this free upgrade to Cayman work?
Even if i think the GTX460 is a good card, and specially a good deal price/perf for a middle range card.., i believe it was a nice surprise after the GTX480 or 470, and some of the disappointements thoses 2 cards have bring ( power ask, heat, price, relative performance gain ) ... but you forget the GTX460 is not the flagship, the die size is big, power ask for a so middle range card is high, then most cards come highly OC,(remember the launch, 10-12 cards OC in each review ) ... the base 1gb is not so impressive, but the card is cheap ... when you see the card have +- the same die size of a 5870 and ask more power.... It cost nearly the same price to Nvidia to make a GTX460 to AMD for make a 5870.... AMD can cut the price of 5870 to 300 dollars and they will still make more money of Nvidia with the GTX460.
It's a nice middle range card, with good performance and sell really cheap, but it's not as Nvidia have make a refresh of the GTX480 with it..
Nvidia have put all this team @work for release a new core, but i really don't see them redesign completely the GF100 in a so short time, not for 2010... and i don't believe before march April 2011 ( will be fair +-1year after Fermi launch ) But who know, we can have a good surprise.
Edit and for the tips: i have build 3 computers for friends and my brother lately, where i have install in each a GTX460 .... price/perfs they have a really nice ratio ( outside some drivers problem ), specially if you don't want put the price of a 5870 or a GTX480. ( and they don't wanted to put more money for a gpu )
I think i should have stop read what write Digi a long times ago, it look like they jump from a board to another each time something surface on the net, specially with their 6K story. It's not so the infos they write, but mostly how they write it..
It sounds to me like ATI is trying to find a leak and plug it. It looks like they are sending out different info to different channels in order to narrow down where leaks are coming from. Hence why there is conflicting "info" all over the place. This step was long overdue IMO.
As with all launches, you should take every bit of pre-release rumor with a massive ship full of salt.
Well offcourse... nobdoy should never believe anything before see it .. the 1st rules stay the best
SKYMTL> i agree, Ati send fakes, since RV770. They don't want the nda to be broken months before the offcial release.
Good idea, but for us it's a bit sad, we love leaks ^^
So digitimes now changes their previous story and tells us 68xx is scheduled to launch in 16 days? Basically a week "delay" from Oct 12th.
When does NDA expire?
around mid october ... probably around the 12th or the 18th
..holy crap! 34 pages!? wow, one can only wonder (or read), what has been said in those pages...
:shocked:
http://vr-zone.com/articles/amd-rade...le-/9999.html#Quote:
Delays and shortages; Radeon HD 6850 and HD 6870 expected to come first
Whilst many are predicting HD 6000 Series cards to hit retail in late October or early November, our close sources have suggested availability only in November. Apparently, the distributors have not been given pricing of these new cards, nor have they been invited to place orders till date. It takes some time for paperwork to be filed and products to arrive.
In what might surprise many, one source is telling us that AMD will skip the HD 6700 GPUs for now and have the HD 6800 GPUs take the lead owing to delays in production. Expect very limited availability and if the yield does not match expectations, we might see a price increase in the couple of weeks after the HD 6800 GPUs hit retail....
buy one and then sell later on???
didn't Movieman say something regarding this?
ahh good ol chiphell, witha piccie of a HD6X70... man its only weeks away and the best chiphell can come up with is a top view snappy, which looks almost like a reference 5000 series (I have noticed the back plate is different in the piccie though)
I really need a new VC, this 4870 is getting a little old....
I'm just hoping the price is gonna be in the 460GTX range. (<$200). :p:
I have money burning a hole in my pocket, hope it's worth the wait!
NDA's....... (New Doodle Agendas???)
man NDA's have gotten a lot tighter in this industry, specially taking into account how they're making part of their market strategy... its mega-nuts
nda = non disclosure agreement ... you know .. legal papers that people have to sign so if they break the disclosure part they could get sued .... because if the competitior knows about the architecture design of the next big gpu comming in 6 months ahead ... could make small aleration to allow more aggresive competition... and no ... dont say competition is good and it should be a base to forget all about nda's .... competition is good ... but people signing nda's have to respect them ....
... lol I was being sarcastic, I know exactly how the whole NDA doco binds individuals... I just wanted to spice up the acronym just a touch
Yea I wonder how exactly the NDAs are formed these days, looks like chiphell have had a card for several weeks already but they are only able to post more info little by little and this close to launch we only got this low quality topview pic etc. :p
Like does the NDA have specified what info can be revealed at what date or something. xD
25 sept: Rumors about specs OK to post.
6 oct: low quality pics = OK
12 oct: high quality pics = OK
etc.
It's clear that AMD uses the NDA part of their strategy/marketing IMO to create more interest. :p: On the paper this Barts doesn't sound like any spectacular, yet they've managed to make it seem like it is (not saying it won't be good cuz it'll prolly be better than any other product at the time from a price/performance POV) but it's still on 40nm and pretty much a stop gap before the real stuff gets released on 28nm.
maybe chiphell doesnt have anything signed and they got their info from someone else that signed an nda but they cant show more then a 2d layout of the card in fear of revealing details that could make them known who they are etc....
I totally agree.
This leak can only do good things because they are doing more harm to nvidia than harm to themselves because NV has alot more cards to sell. The hot card right now is an NV card and throwing smoke over it, only preserves their marketshare and it appears to be working. AMD probably leaked this on purpose because they seem so timed and the leaks themselves are directed(such as barts leaks first). A massive score to get peoples attention, etc.
how does a leak look good for a company more then harm them????
That depends what leaks. so far we have performance estimates and looks. but not price leaks.
and price is the only response nvidia can give right now.
Seriously, how much harm do you think leaks like that are if they are like within a month to official launch, if stuff had been leaked a month or two ago it would be much of a concern and I'm pretty sure nvidia knows more than we think anyway. Leaking stuff at the right time before a launch, ie not too early could actually be beneficial from a marketing standpoint to create more interest or hype to increase sales at launch and positive launch sales is quite important.
that pic is a total fake plus its photochopped
Umm, you get the meaning of the word wrong...
http://www.thefreedictionary.com/fake
fake
adj.
Having a false or misleading appearance; fraudulent.
n.
1. One that is not authentic or genuine; a sham.
stop wasting this thread space, there needs to be plenty of room for more rumors thanks :p
Sorry!
I heard from my friend that his sister's husband (works for AMD) was flying to India soon in preparation for the launch of the 6k series!
While this doesn't really tell us anything, it is actually true.
Still 1GB of VRAM. :( Hope 2GB parts will not take 6 months to be released.
Thanks again cold2010 :up:
Google translate not work very well ... The guy is really talking about 5830 performance for HD6850 ? :confused:
Original thread :
http://www.xfastest.com/viewthread.php?tid=51709
barts pro?
223,CAYMAN GL XT (6700),NI CAYMAN
224,CAYMAN GL XT (6701),NI CAYMAN
225,CAYMAN GL XT (6702),NI CAYMAN
226,CAYMAN GL XT (6703),NI CAYMAN
227,CAYMAN GL PRO (6704),NI CAYMAN
228,CAYMAN GL PRO (6705),NI CAYMAN
229,CAYMAN GL (6706),NI CAYMAN
230,CAYMAN GL LE (6707),NI CAYMAN
231,CAYMAN GL (6708),NI CAYMAN
232,CAYMAN GL (6709),NI CAYMAN
233,CAYMAN XT (6718),NI CAYMAN
234,CAYMAN PRO (6719),NI CAYMAN
235,ANTILLES PRO (671C),NI CAYMAN
236,ANTILLES XT (671D),NI CAYMAN
237,BLACKCOMB XT/PRO (6720),NI BLACKCOMB
238,BLACKCOMB LP (6721),NI BLACKCOMB
239,BLACKCOMB XT/PRO Gemini (6724),NI BLACKCOMB
240,BLACKCOMB LP Gemini (6725),NI BLACKCOMB
241,BARTS GL XT (6728),NI BARTS
242,BARTS GL PRO (6729),NI BARTS
243,BARTS XT (6738),NI BARTS
244,BARTS PRO (6739),NI BARTS
245,WHISTLER XT (6740),NI WHISTLER
246,WHISTLER PRO/LP (6741),NI WHISTLER
247,WHISTLER XT/PRO Gemini (6744),NI WHISTLER
248,WHISTLER LP Gemini (6745),NI WHISTLER
249,ONEGA (6750),NI TURKS
250,TURKS XT (6758),NI TURKS
251,TURKS PRO (6759),NI TURKS
252,SEYMOUR XT/PRO (6760),NI SEYMOUR
253,SEYMOUR LP (6761),NI SEYMOUR
254,SEYMOUR XT/PRO Gemini (6764),NI SEYMOUR
255,SEYMOUR LP Gemini (6765),NI SEYMOUR
256,CAICOS GL PRO (6768),NI CAICOS
257,CASPIAN PRO (6770),NI CAICOS
258,CAICOS PRO (6779),NI CAICOS
6739 = BARTS PRO IT SEEMS
-http://ht4u.net/news/22583_catalyst-108-treiber_verspricht_mehr_performance_und_verraet_we itere_codenamen/
According to the device ID it's a Barts Pro. And 5830 performance wouldn't be that bad if it was called 6700, but changing it to 6800 still doesn't make much sense.
more is better .... we need moarr vram
Every couple of generations, there is always a flagship card to skip.
For example, the 3870's.... or for nVidia, the GTX280's... or even 480's...
The reasons differ. Price, competition, etc.
Will the 6K be a flagship to skip? I'm holding onto my 4870x2, saw no point in going to 5870/5890 (oops mean 5970), and this thread is becoming increasingly depressing as to what the 6xxx flagship will bring to the table.
So, Bart chips as the base of HD 68xx cards seem to be true. Well, it's rather unpleasant finding for me, but still just half of the story. Gotta know the pricing of these babies to really judge it whether it's just a rather confusing & quite dissapointing resegmenting but still a great value product, or a total bust as AMD cash cow consumer conning product.
Maybe the reign of most powerful card will still reside with the almighty 5970.... oh!... I meant 5890....:rofl::rofl:
Wrong.
Just an example:
http://www.thg.ru/graphic/amd_radeon...s_2gb_8xaa.png
More detailed research here.
TL;DR: If you game at 2560x1600 or use Eyefinity it is a good idea to have over 1GB of VRAM.
Hopefully that's 68*50*, or the rumour about the renam is true. I will buy a new radeon if it has 2GB of ram and at least ~40% increase in eyefinity performance. Otherwise I won't.
Edit: barts pro. Should have refreshed my screen. Shame on ATI if true for the annoying, misleading, and uneeded name change. Unless Barts outperforms cypress that is. That would be awesome.
Those pics confirmed the renames, though by definition it's not a rename because the product hasn't even been released :ROTF:
I don't understand why people are so mislead about VRAM. Every time we cross over to a new density the same thing happens. I saw this when cards went from 64 to 128, 128 to 256, 256 to 512, 512 to 1024, and (now soon) 1024 to 2048. Someone will review a card with more memory and act surprised that it isn't any faster. Like with your computer's RAM, more only matters if you don't already have enough. If you're paging, things slow way down. If you're not (ie: your content all fits into VRAM already), you aren't going to magically gain anything. It's a bit of a catch-22 as well. Developers don't make games with larger texture sets until there are cards with more VRAM getting to be mainstream. People don't often buy the more expensive cards until they see a benefit. The only things saving us are time (and its implications like DRAM getting cheaper regardless so eventually it isn't expensive to double the VRAM) and fancy extras like AA eating up memory space and bandwidth even if the game's assets don't by themselves.
To quote from the article:
I didn't state that 2GB would be useless with massive amounts of AA combined with 3 monitors.Quote:
I picked a couple of the games that were running plenty-fast in the previous CrossFire chart and added 8xAA to them
However, looking at the 5760x1080 single card section, I still only see a difference of 1 - 3 frames.
So yes, if you want to run 5760x1080 with 8xAA then 2GB will be worth it for you. But realistically, the number of people who even run eyefinity is miniscule, let alone those who think that 8xAA is necessary. Everyone else will not notice a difference between 1GB and 2GB VRAM, so I think most of them would prefer having 1GB of VRAM and lower costs on the cards.