I'm likin that case shot...it's whetting the appetite
Printable View
I'm likin that case shot...it's whetting the appetite
Gah, I just want my 5870 already!
It was kind of obvious when AMD announced that it would launch prior to W7. It was either september or october and if it was to be september it had to be a paper launch considering the 40nm problems at TSMC. So maybe it's just what I was assuming considering all facts.
The 4870's have dropped so much in price already, that it's a rather good deal for now. Take for example, a Sapphire 4870 1GB is only $140 on Newegg.com and cannot go much lower even after ATI releases 5870 cards.
However, 4890's have a lot more room to drop in price, so if you do not want to be disappointed, it's better to just wait 2 months or get a 4870 1GB now.
There goes my expert wallet advice!
MSI 4890 for $200 @ZZF - $30MIR - $30ebillme rebate
Was as low as $126 after all those rebates plus a $5 cashback from bing.com
http://forums.anandtech.com/messagev...&enterthread=y
Ughhh.. ebillme is not exactly a "regular" rebate.. lots of people do not want to go through the trouble, and once used, it's no longer valid.
Plus I would stay away from ZZF since it is quickly going down the hill as of late. Remember the crash of MonarchComputers.com? Thousands of customers were left out in the cold.
I'll let the numbers do the talking: http://www.resellerratings.com/store/ZipZoomFly
In the last 6 months, it earned a rating of 6.5 out of 10, which is much worse than its lifetime rating. If you read up on [H] forums, you'll see how much bad rap ZZF has gotten. Usually, their "awesome" deals are the crapper deals.
some news on the card...
"from the way they have it setup, expect the minimum ROP layout to be 8. The better part being 32 (x2 with the X2 chip). The SPs are unchanged in layout from the looks of it.
8-8-8-8 is the layout I imagine they will use and just chop off 8 for each market segment they intend to engage in."
Thanks to "Flyordie" at TPU for that info. so double what we have atm ?? no extra sp's :( hmm wonder what this will do...
(here's hoping unrealistically that double rops = double performance haha)
I'd be quite shocked if the SP count didn't increase. Sure they can increase the efficiency of the shaders but not to the extent it will blow the current offerings out of the water. It would be nothing more than a shrink with some tweaks if they didn't up the shader count. I'm sure they've done a bit to increase the texture performance as well but I doubt they'd do this at the cost of the shaders ( AMDs last few generations have been weaker texture wise than Nvidia's by far but they seem to make up for it in shader power ) Now I recall reading some info about AMD saying we'd understand their design choices when we see the cards and that we might be shocked by the cards ROP layout so we shall see.
Regardless I'll be surprised if the cards disappoint. The real question remains will Nvidia keep up and if so will they surpass AMD by any tangible amount.
IN a recent interview with Richard Huddy he stated that one of the most reworked parts of the chip is the complexity of the scheduler but the whole chip is much more complex than previous generations.
Link in German(can't go to babblefish at work)
BTW- Can we get a link to the ROP post?
Interesting article... here's the Google translation, I can't wait to see the real deal:
Quote:
Preview on PC Games Hardware 09/2009
DirectX 11: AMD and Nvidia to reveal details of technology and hardware - Preview on PC Games Hardware 09/2009
PC Games Hardware has detailed with AMD and Nvidia on DirectX 11 and the expectations of the Radeon and Geforce-makers to the new API, and Windows 7 talk.
DirectX 11 Nvidia Interview Tamasi Huddy AMD PCGH 09 2009 a [Source: see Image Gallery] Tony Tamasi (Nvidia) and Richard Huddy (AMD) to reveal what they and their companies from DirectX 11 and promise to some extent, what priorities and improvements in the hardware included.
We offer a preliminary kind of appetizers schonmal a few interesting excerpts from our interviews with the two representatives from AMD or Nvidia. On the PC Games Hardware versions with Issue-DVD, ie DVD, Extended and Premium, see also an over 40 minute video in which we view AMD's Richard Huddy of things out on DirectX 11.
Nvidia Tony Tamasi [Source: see Image Gallery] Tony Tamasi, Nvidia
Question:What could be hard-wired units in the future most likely on the shader will be executed?
Answer: Of course it is possible to any kind of graphic calculation of programmable units to execute. The really interesting question, however: you should - both in terms of performance as well as on the power? Since Intel despite hype and high expectations Larrabee has not yet delivered, is the last word here is not yet spoken. Intel has already indicated that we expected today, with current high-end GPUs to be able to compete.
If they are really in a year's supply, they are already quite behind the GPU performance development behind. According to the photos, which Intel has shown, it looks as would both the chip size and power probably also significantly higher than current GPUs. It could be concluded that this approach has significant disadvantages in the cost and power brings with it - quite apart from the previously unanswered question about the performance.
AMD's Richard Huddy [Source:see Photo Gallery] Richard Huddy, AMD
Question: Do you have an existing design as a base made from scratch or start? What parts needed the most conversions?
Answer: For Evergreen, we pretty much have to break new ground, but it is in some essential points of our current architecture derived DX10.1. The efficiency of the HD-4000-series makes it a perfect starting point, but DX11 requires a lot of fine-tuning to the specifications, which are then at certain points had to be added.
In addition, the Tessellator and thus also of Hull and domain shader fairly profound changes to the scheduler requires. Overall I would say that our main-Scheduler, which work on the ALUS distributed resources and the various steps assigns the component is, what the biggest overhaul required.
Overall, he is more than twice as complex as in previous generations. If you see the final specifications, you will find that we have other pieces of hardware have significantly inflated, among other things, simply because we now can produce more complex chips.
What does complexity tell me about performance.Older games see nice gains with more rop's and tmu's,newer games are shader bound.:yepp:
Complexity means that RV870 will have over 2 billion transistors in comparison to the 950 million on rv770.
http://www.forum-3dcenter.org/vbulle...postcount=1884Quote:
September 17 in San Francisco issued a formal version of a full range of evergreen.
AMD DX11 graphics chips into high-end, performance, mainstream, entry-四档, high-end, code-named Cypress, the performance of code
Codenamed Redwood, are the mainstream of the two code-named Juniper and Cedar, entry, code-named Hemlock.
And flagship product Cypress uses the MCM design, is composed of 2 Redwood, Redwood chip size 300 square millimeters,
Juniper chip size of 181 mm2, Cedar die size 120 mm2. Currently, AMD DX11 performance and mainstream Products has proceeded smoothly, but the flagship product in the chip package Cypress encountered a little problem, Cypress
May be deferred to the AMD DX11 performance, after the mainstream market
Cedar = 120mm^2
Juniper = 181mm^2
Redwood = 300mm^2
Cypress = 2 x Redwood (MCM?)
Translation : 5870x2 is delayed :down:
"The SPs are unchanged in layout from the looks of it." + "Redwood chip size 300 square millimeters" + 40nm = I don't understand :confused:
Not really. Usually the X2 comes out later.
Also there have been rumors of problems with it all along, mostly due to packaging costs/complexity.
Only about 2 years ago, was first being rumored a little before the RV670 launched in late 2007 which amusingly enough was about the same time that R800 should have been in development for about a year.
That the SIMDs are still setup similar enough to RV770, though I don't believe that particular rumor.
The title doesn't seem right...
Now that's more like it.Quote:
AMD DX11 RV8xx Card Exposed Intentionally (aka marketing)
If that 300mm˛ @ 40nm is anywhere near reality it sounds very good. That would mean some truly serious expansion of processing resources. Like, literally double everything, but since that wouldn't make much sense, it could be *just* doubled SPs and TMUs, or more.
Much like RV670 -> RV770.
TPU link
He had stated 3 days earlier that he knew something but he couldn't say anything yet. I reminded him 3 days later and that's when the comment came up.
I don't think he is too sure on the SP setup since the ROP setup is what he said was the only information he had.
Thank you for the link.
Honestly I cannot take anything he says serious after this post.
He sounds decently knowledgeable on your basic computer/techy/OCing info but seriously doubt he has any real connections/sources.