Friday, July 1st 2016
NVIDIA GeForce GTX 1060 Reference Board Design and Clocks Confirmed
A leaked slide from NVIDIA press-deck for the imminent launch of the GeForce GTX 1060 confirmed the reference board design, which first surfaced in Hong Kong. The slide also reveals clock speeds, and other key specs of the card. While it doesn't reveal the GPU nominal clocks, it mentions that the GPU Boost frequency will be set as high as 1.70 GHz. The memory is clocked at 8 Gbps, which over the GPU's 192-bit GDDR5 interface, puts out 192 GB/s of memory bandwidth.
The chip features 1,280 CUDA cores based on the "Pascal" architecture. The card draws power from a single 6-pin PCIe power connector, its TDP is rated even lower than that of the AMD Radeon RX 480, at 120W (vs. 150W of the RX 480). NVIDIA has been making huge (and successful) performance claims for its "Pascal" GPUs so far. The GTX 1060 is claimed to be faster than the GeForce GTX 980 from the previous generation, and "much faster" than the RX 480, which means that NVIDIA intends to price this card competitively to the RX 480.
Source:
VideoCardz
The chip features 1,280 CUDA cores based on the "Pascal" architecture. The card draws power from a single 6-pin PCIe power connector, its TDP is rated even lower than that of the AMD Radeon RX 480, at 120W (vs. 150W of the RX 480). NVIDIA has been making huge (and successful) performance claims for its "Pascal" GPUs so far. The GTX 1060 is claimed to be faster than the GeForce GTX 980 from the previous generation, and "much faster" than the RX 480, which means that NVIDIA intends to price this card competitively to the RX 480.
117 Comments on NVIDIA GeForce GTX 1060 Reference Board Design and Clocks Confirmed
Also, I may not like the pricing but I'm not about to buy an inferior card to spite Nvidia. I won't go back to dual cards so at 1440p I want the fastest I can get however, the 1080 isn't a big enough jump and we all know a 1080ti is waiting in the wings. We just have to hope Vega can cause an upset. No dude, shit graph. By making the origin zero you have reduced scaling bias but by making the x-axis go to 4 you increase the scale bias. If you want to make it neutral, scale starts at 0 and ends at 2.
The demand for GP100 in Teslas is enormous at this point, and Nvidia is earning more per wafer on these so it's going to tie up a lot of the wafer production volume for several months to come. Combine this with higher demand than previous generations and you'll get a shortage. There will be a steady stream of GP106s and GP104s coming, but customers might have to wait a few weeks. Nvidia is planning to release more products based on these in the coming two months, including mobile products based on both GP104 or GP106 for the "back to school" season.
Edit: BTW: GTX 960 has 1024 cores vs 2048 cores in GTX 980.</math-police>
Much faster. Wow:
And yes, I did post it twice. I have reasons... maybe.
We know, for a fact that, RX 480 is held way back because of the 150w threshold AMD put on the reference design. Reference is 1266 MHz with AIBs reporting 1460-1600 MHz (15-26% faster) doable with better cooling and more power. Add that knowledge to NVIDIA's graph, they look well-matched. Assuming everything goes well, that should mean Vega is also well-matched to GTX 1070/1080--but that's clueless as to how HBM2 plays into it.
The purpose is to make u buy this and that...is called MARKETING.
:))))
and i totally agree on the front of the 1080ti, there is no point in buying any of the lesser cards if you can afford that price range.
This was done before any significant amount of stock was produced so as to trigger a response from the high end enthusiast market, with the understanding of AMD's plan to release Polaris prior to Vega.
It was purely a canard to seem to release these cards before they actually were able to, to try to lock up the market. It was a success as most buyers are still waiting for pre-orders and stock to ship.
TBH, I have a bad feeling full stop about AMD and Nvidia with future cards. I have so many 'wishes' for Vega but I don't think that will be anything other than 'reduced expectations met'. I hope I'm wrong but the node shrinks are giving less room to play with on voltages and thermals.
It's good to see the 970's and 980's dropping further though. They obviously had way too much stock on hand going into a new product launch, even with stopping production on 970's and 980's 2 months ago.
Guess we have to wait for something...Wish it would come sooner. See, that will be interesting because if vendors can make the card clock it should provide some great performance. It's all going to depend then how the 1060 compares clocked to its higher point.