AMD States that the 290/290X have hardware frame pacing, so that should make CF work in all setups, and not be reliant on driver support.
ocUK hinted that the 290 is a smidgen slower than the GTX 780 but very much faster than the GTX 770 - and will be much cheaper than the 780. So if AMD has any sense, i'd expect the 290 non X to occupy a price spot of 450$-500$ which would be fantastic IMHO. Overclocking would definitely play a role in the overall value and the hints there are, again, that it OC's and scales very well. Let's just hope that vendors don't freakin' voltage lock their cards......
So if these rumors are true the 290 non X will definitely be the king in AMD's lineup in terms of performance per dollar (along with 280X). Even better, assuming CF-eyefinity is fixed, that could make the 290 a heck of a deal for crossfire. I'm optimistic about frametimes with the 290, because the cat 13.11 drivers improve single card frametimes on the 280X to be better than the GTX 770 per PCPer's benchmarks....hopefully AMD has wisened up about software. IMHO, for AMD, it's all about software quality - that is what they need to work/focus on the most. NV has been better than AMD in this area for some time and i'd like for AMD to match NV in that respect.
Won't that affect OCing? All those transistors in such a small die? These might not be stellar oc'ersHawaii transistor density is certainly impressive, and for all we know could be even higher as all AMD has given us is "more than 6 billion transistors"
The 7970 GE already had the better of the 770 in frametimes long before 13.11sI'm optimistic about frametimes with the 290, because the cat 13.11 drivers improve single card frametimes on the 280X to be better than the GTX 770 per PCPer's benchmarks....
Won't that affect OCing? All those transistors in such a small die? These might not be stellar oc'ers
Now, that is a misrepresentation of reality. AMD never stated that the 290/290X have hardware frame pacing - some folks assumed that because crossfire is done over PCI Express that it has "hardware" frame metering. This is not the case. In fact, all "frame pacing" is done through software.
SKYMTL of hardwarecanucks had a rather verbose discussion on this a few weeks ago, but the TL'DR is that since displaying pixels on the screen is always done on a driver level, all frame pacing must be done through software. There is no "hardware" frame pacing. Pixel manipulation and data management that is received over the PCI-E bus is always handled by software. Hardware frame pacing is a myth, the fact of the matter is that NV has had software do frame metering for some time while AMD only recently decided to catch up. And still has not implemented it for eyefinity CF.
Again, for AMD is all about software. I think the 290 and 290X will be awesome cards but let's not kid ourselves about hardware frame pacing. AMD needs to get their software on par with nvidia, and I think AMD realizes this these days. IMHO, they should spend slightly less on game bundles and throw more money at their software developers. That software pays off - software is why NV is so loved by their users; the best hardware in the world is useless without high quality software to back it up, period. I do re-iterate that I think AMD realizes this now, so I don't expect AMD to replicate the awful launch software situation that the 7970 had. I don't think they will make that same mistake twice.
Generally higher transistor density increases performance at the cost of power consumption.
So even it doesn't clock as high it is quite possible it will scale better.
Look at Pitcairn - they have a higher density than Tahitis and they OC pretty well.
I've always wondered why AMD was able to attain so much higher transistor density on an equivalent manufacturing process over Nvidia.
AMD must have re-designed the GPU from scratch not just widened it. How can you have 2x the ROPs, 512-bit memory controller, nearly 40% more TMUs/steam processors but die size goes up only 16% from Tahiti XT? Something is definitely smaller inside Hawaii than Tahiti. The memory controller is presumably smaller than 384-bit one in Tahiti, but what else?
May wind up with a $500 card you can overclock to 290X stock performance, which would be really nice considering now your only options are $650 or $1000 for a card that fast.
Unleash the Kraken!
Either the TMU number is wrong in that slide for the R9 290 or they decoupled them from the CUs.
ps: if those prices are correct, does anyone really believe that NV will lower GTX780 and TITAN prices ??? I DONT.
Wow, thanks for that. I was not aware that GK110 had higher Xstor density than Tahiti. And yes, that is quite a jump to Hawaii.
To put things in further prespective, Pitcairn clocks in at 13,2m trans/sqmm and GK104 at 12,0 trans/sqmm.
Hawaii transistor density is certainly impressive, and for all we know could be even higher as all AMD has given us is "more than 6 billion transistors"
I've heard about the same rumor, but can't assign as to who said so. I have yet to find documents attesting that 28nm hpm has higher sram densities.Perhaps this is 28nm hpm process?
http://www.chipworks.com/en/technic...og/a-review-of-tsmc-28-nm-process-technology/The fourth and final 28 nm process offered by TSMC is the HPM technology. This process is targeted at mobile applications and apparently will support both high performance transistors and low power transistors on the same die, thus enabling higher performance mobile devices, while continuing to improve power performance, which is critical in battery powered gadgets. TSMC claims the technology can provide better speed than 28 nm HP while giving similar leakage power to 28 nm LP. The wide performance/leakage coverage apparently makes 28 nm HPM ideal for applications from networking, tablet, to mobile consumer products. The HPM process has not yet been seen in the market yet, but Chipworks expects to see an example of the technology soon in a mobile device from a leading manufacture sometime soon possibly a future variant of the Apple A6.
If R9 290/X put pricing pressure on the 780s, if I was a Titan owner I would be putting up my card for sale ASAP before its resale value tanks completely once R9 290X is released. Most already don't believe it's worth $1,000 and if R9 290 $499 can overclock to trade blows with Titan OC, oh boy...
We also don't know how many transistors Hawaii has, just that it's greater than 6 billion and less than 7. Everyone's assuming it's pretty close to 6.0 billion because even that number is already way above what we were previously expecting. That said, there's also the possibility that Hawaii is even more dense than that. Anything higher than 6.5 billion would be truly surprising, but I wouldn't count out the 6.1-6.4 billion range.
Tahiti was AMD's first gpu on 28nm and they went with very conservative densities. As HurleyBird says, Pitcairn puts quite some more on the table.
I've heard about the same rumor, but can't assign as to who said so. I have yet to find documents attesting that 28nm hpm has higher sram densities.
http://www.chipworks.com/en/technic...og/a-review-of-tsmc-28-nm-process-technology/
Is RS really comparing non reference to stock again?
/mindblown
I don't see why Nvidia feels the need to lower prices much, if at all. It's been proven time and again that loyal NV customers will buy Nvidia no matter what. Sure some will get frustrated and switch, but look at the market share #'s over the years it takes a colossal blunder by Nvidia for AMD to gain any meaningful share. And even when they do, Nvidia inevitably gets their ducks in a row and the numbers shift back again.
Yeap.You think the 290 pro should have 160TMU?