Thursday, September 19th 2013
AMD "Hawaii" R9 290X GPU Specifications Revealed
Here are the first set of specifications for AMD's next high-end GPU silicon, on which the company will no doubt carve out several SKUs from. Codenamed "Hawaii," and slated for unveiling on the 26th in, well, Hawaii, the 28 nm chip is what AMD will take NVIDIA's GK110 silicon head-on with. It is based on AMD's second-generation Graphics CoreNext micro-architecture.
With an estimated die-area of 430 mm² (18% bigger than "Tahiti,") the chip physically features 2,816 stream processors (SPs) spread across 44 clusters with 64 SPs each (a 37.5% increase over "Tahiti"). The chip features four independent raster engines, compared to two independent ones on "Tahiti." This could translate into double the geometry processing muscle as "Tahiti," with four independent tessellation units. The memory interface of the chip is expected to be 384-bit wide, based on the GDDR5 specification. Given the way TMUs are arranged on chips based on this architecture, one can deduce 176 TMUs on the chip. The ROP count could be 32 or 48. The chip will feature hardware support for DirectX 11.2, including the much hyped shared resources (mega-texture) feature.
Source:
3DCenter.org
With an estimated die-area of 430 mm² (18% bigger than "Tahiti,") the chip physically features 2,816 stream processors (SPs) spread across 44 clusters with 64 SPs each (a 37.5% increase over "Tahiti"). The chip features four independent raster engines, compared to two independent ones on "Tahiti." This could translate into double the geometry processing muscle as "Tahiti," with four independent tessellation units. The memory interface of the chip is expected to be 384-bit wide, based on the GDDR5 specification. Given the way TMUs are arranged on chips based on this architecture, one can deduce 176 TMUs on the chip. The ROP count could be 32 or 48. The chip will feature hardware support for DirectX 11.2, including the much hyped shared resources (mega-texture) feature.
97 Comments on AMD "Hawaii" R9 290X GPU Specifications Revealed
BTW: Those aggregated percentage "Overall analysis" charts...might pay to take them with a grain of salt. According to the chart,
this supposed 290X has a 22.3% gain over the Titan for the Valley bench w/AA, yet the results chart has the AMD behind the Titan in both AA tests (along with both of the other benches) ???
*goes in hiding*
1920 x 1080 -> 2560 x 1600 . Per card. Again: compared to itself from 1920 to 2560 resolution performance.
EDIT: no sorry, that's their second set of 3 charts. The first (which you copied) is about the performance decrement between using no AA or AA. Per card. Compared to itself.
They say: "As you can see, the Titan and the 780 drop 33-34% in performance (when applying AA), whereas ........... Yes. Yes ....... " ...the R9 290XT would drop 29.9% performance according to their chart.
yeye i know it's a rumor
Kind of seems like an odd metric to use given the non-stock clocks, and the probabilities of the user of a $600-1000 card using no AA in their in-game settings.
Anyhow the performance difference between OC'ed AMD part, the stock 780, and stock Titan using the highest game i.q. benchmarks comes out to be 1.7% and 4% over Titan at 1920 and 2560 respectively, and 7.8% and 11.6% over the GTX 780.
Not overly conclusive without knowing what the reference core/boost/memory clocks are on the AMD part
There is also the no small matters of the card being benched obviously being a development/validation board ( shitty generic blower fan, jumpers on the PCB, relatively Spartan layout) which is in the AMD designs favour. The other matter is that the 2816 core version is supposed to reserved for FirePro boards while the Radeon part would have 2560 enabled- so which of the two is being benchmarked?
that would be disappointing
Which is why I noted that only a certain amount can be taken from the available information- i.e.
1. If the AMD card has a dynamic boost similar to Nvidia's then it would stand to reason that a better cooler would produce better results than the basic shroud of the engineering sample.
So, depending on the shipping clocks, and whether the card ships with a dynamic boost, the card could be faster...or slower than these benchmarks. Hardly definitive. The only certainty I think is that the shipping reference card overclocks better than 1020/5000 - at least I would hope so given what the GTX 780 and Titan can achieve with a larger die.
2. Maximum boost in benchmarks of the reference part will determine how it fares in reviews. Will an OC'ed card beating a more expensive card when overclocked is laudable. It is neither unusual, nor the basis for most comparisons. Case in point being that an overclocked HD 7950 or GTX 670 would also best a stock HD 7970 or GTX 680. While the benchmarks show the ability it is taken as read that the 7970/680 will also overclock in turn to restore the balance of power in most cases. A more recent example would be an OC'ed 780 beating a stock Titan - all kudos to the Jetstream, but no one is suggesting that turning up the frequency of the Titan will allow the 780 to retain its lead.
Anyway I can't wait on 25th!
The argument started when i said after overtaking GTX 780 there is not much left to overtake Titan - You yourself told me that Titan is 8% faster hence my argument was valid, i.e - there is not much left to overtake once you are done beating GTX 780.Seriously ? Tell me whose the troll now ? - I gave you a link of a review of overclocked GTX 780 which beats Titan that means if AMD's flagship card is faster than GTX 780 it will easily be able to beat TITAN - hence forth i was showing you the valid point but you are smoking some good trolling stuff here. - (Link for those who are lazy like me-
www.techpowerup.com/reviews/EVGA/GTX_780_SC_ACX_Cooler/26.html )I don't think i need to explain again...
Case in point: AMD card. Unigine Valley 1.0 bench at 4xAA (1920x1080) 70.1 fps
GTX 780 overclocked (on air cooling). Unigine Valley 1.0 bench at 8xAA (1920x1080) 84.3 fps
EDIT: Oh and please don't start with pathetic attempts of show casing specific results that shows Titan is faster than AMD's flagship and in other GTX 780 has even higher scores. It just makes you look more sad, cause both the results were produced using different PC configs.knock knock - Anybody home ? I have constantly said if AMD's card is faster than GTX 780 it will have no trouble beating TITAN - hence i gave the reference of Overclocked GTX 780.
I see you are trying very hard to spin this, but it is clear now whose the troll here.