TweakTown NewsRefine News by Category:
NVIDIA might have just launched its next-gen Titan X graphics card, but thinking into 2017 - what will NVIDIA do since it has so many high-end, enthusiast graphics cards out on its Pascal architecture? Volta, baby.
According to the latest rumors, NVIDIA will launch its Volta architecture in May 2017 at its own GPU Technology Conference that is being held May 8-11. NVIDIA's current roadmap has Volta ready for 2018, but the new rumors suggest NVIDIA is preparing Volta for a big unveil in the second half of 2017 - over six months ahead of schedule.
Volta was meant to arrive on the 10nm process, but according to Fudzilla, it will use the 16nm FinFET process - the same node that the new GeForce GTX 1080, GeForce GTX 1070 and GeForce GTX 1060 are all on. Volta will use stacked DRAM, so we should expect it to be quite the monster when it comes to performance. Fudzilla adds that the performance per watt is "expected to increase tremendously", which is exciting to hear.
With the release of the Radeon RX 480, we have seen a newly-focused AMD that isn't trying to compete in a market that it can't compete in yet against NVIDIA. The mainstream market is where the money and consumers are, and this market is thriving.
Well, now we have leaks on SAPPHIRE's upcoming Radeon RX 470 and RX 460 graphics cards, led by the SAPPHIRE Radeon RX 470 Platinum Edition. Sapphire's take on it looks like AMD's reference card, and that's not a bad thing - but the price will be what gamers will jump at - with a rumored price of just $179 USD.
SAPPHIRE's Radeon RX 470 Platinum Edition PCB features a 4+1-phase design, whereas the Radeon RX 480 had a 6+1-phase design. One single 6-pin power connector is required, keeping PSU requirements nice and low.
NVIDIA has just announced its new monster graphics card, the new Titan X. NVIDIA's new Titan X is based on the Pascal architecture, featuring a pretty damn good increase in specs over the already fast GeForce GTX 1080.
The new Titan X has 3584 CUDA cores compared to the 2560 CUDA cores on the GTX 1080, while we have 12GB of GDDR5X on a 384-bit memory bus, compared to the 256-bit memory bus on the GTX 1080. The new Titan X uses the GP102 which is based on the 16nm FinFET process, while GP104 powers the GTX 1080 and GTX 1070.
Power consumption wise, we're looking at a 250W TDP - up from the 180W on the GTX 1080, while we have 12 billion transistors on the new Titan X, compared to the 7.2 billion found on the GP104-based GTX 1080. The memory bandwidth has been increased thanks to the 384-bit memory bus, where we have a huge 480GB/sec - very close to the 512GB/sec offered by HBM1 on the AMD Radeon R9 Fury X.
AMD impressed the world with its Radeon RX 480, priced at $199/$239 for its 4GB/8GB versions - and still coming out on top when it comes to performance for your dollar against NVIDIA's new GeForce GTX 1060. Well, the AIB partner cards are now rolling out with the devilish PowerColor Red Devil RX 480 8GB announced.
PowerColor has priced its new Red Devil RX 480 8GB at $269 in the US, with its GPU clocked at up to 1.33GHz and the 8GB of GDDR5 clocked at 2GHz. There's a single 8-pin PCIe power connector, with 6+1-phase power with each phase providing 25W of power compared to 22.5W on other boards.
PowerColor has deployed its latest fan technology dubbed Double Blase III, which "increases airflows and prevents dust deposition on the fans". Double Blase III is made with triple 80mm 2-ball bearing fans, and 2pcs of 8mm and another 2pcs of 6mm nickel-plated heat pipe for the best heat dissipation.
AMD has been ahead of the game when it comes to Asynchronous Compute support in their GPU architectures, and now id Software is throwing their weight behind Async Compute when it comes to their idTech6 graphics engine.
id Software's Billy Khan talked with Eurogamer recently, where he said that other developers will begin to see the benefits of Asynchronous Compute, and will take better advantage of it. Khan said: "Doom is already a clear example where async compute, when used properly, can make drastic enhancements to the performance and look of a game. Going forward, compute and async compute will be even more extensively used for idTech6. It is almost certain that more developers will take advantage of compute and async compute as they discover how to effectively use it in their games".
The developer also said that other game developers should jump right into Vulkan as well, with id Software's Axel Gnetting adding: "Vulkan actually has pretty decent tools support with RenderDoc already and the debugging layers are really useful by now. The big benefit of Vulkan is that shader compiler, debug layers and RenderDoc are all open source. Additionally, it has full support for Windows 7, so there is no downside in OS support either compared to DX12".
Tiago Sousa of id Software chimed in, saying: "From a different perspective, I think it will be interesting to see the result of a game entirely taking advantage by design of any of the new APIs - since no game has yet. I'm expecting to see a relatively big jump in the amount of geometry detail on-screen with things like dynamic shadows. One other aspect that is overlooked is that the lower CPU overhead will allow art teams to work more efficiently - I'm predicting a welcome productivity boost on that side".
TSMC has been hitting its stride this year, with the delivery of the 16nm FinFET process for NVIDIA's current GeForce GTX 1080, GTX 1070 and the soon-to-be-released GTX 1060. But what about the future?
Samsung has been pushing its Extreme Ultra-Violet Lithography (EUV) for the 7nm node, but TSMC isn't far behind, with the Taiwanese giant set to start 7nm node trials by the end of 2017. TSMC's co-CEO, Mark Liu, said that the company plans to push into EUV for 5nm by the end of 2020. Liu added that the company will simplify the process, improve density and cut down on the overall costs of the new node.
TSMC has also said that it has made more than one improvement to its existing EUV infrastructure, while it claims to have implemented 125W EUV source into its ASML NXE:3350 equipment. TSMC also says it now has in-house EUV technologies for inspection, masking and repair for lithography. When it comes to 7nm, TSMC has hit integration for scanners, masks and photoresist, while they'll be adding two more EUV scanners in 2017.
XFX will soon launch its new Radeon RX 480 graphics card, which has been spotted online courtesy of Quasarzone. The new card features an awesome looking XFX logo at the top and a dual-fan cooler.
We will see a single 8-pin PCIe power connector on the XFX Radeon RX 480, while it will rock a factory overclocked GPU. There are four large copper heat pipes keeping the Polaris 10 GPU nice and cool, while we have a rectangular thermal pad solution that looks like it's keeping the VRMs cool. The reference AMD Radeon RX 480 was a great card, but the XFX variant should be an interesting take on the RX 480.
We're not far away from the launch of NVIDIA's GeForce GTX 1060, and we're already hearing about the mobile variant, which is reportedly not going to have much of a performance difference between the desktop and notebook parts.
Polish website Purepc is reporting that the only difference between the GeForce GTX 1060 on the notebook and desktop is the clock speeds, where they said: "The only difference between the GeForce GTX 1060, for laptops and desktop version, will be reduced core speed. 1405 MHz instead of 1506 MHz in the basic mode, and 1671 MHz instead of 1709 MHz GPU Boost 3.0".
If this turns out to be true, the GTX 1060 will usher in a new performance standard for gaming notebooks, where 1080p 60FPS shouldn't a problem. Now we need to know the pricing on GTX 1060-powered gaming notebooks, which should be revealed in the very near future.
AMD has released its Polaris 10-based Radeon RX 480 into the wild, with the next-gen Vega architecture in the oven, and ready for early 2017. But now we're hearing rumors of the Vega 10 and Vega 11 GPUs, in the middle of 2016.
In the new OpenCL driver, a few new chips were discovered under 'GFX9': Greenland, Raven1X, Vega10 and Vega 11. Greenland is something that has been swinging around the rumor mill for a while now, a new GPU that will reportedly rock 4096 stream processors, and a new SOC v15 architecture.
Vega on the other hand is a "high-end architecture for high-end gamers" according to AMD, and has an early 2017 release window. Vega will be the first GPU to utilize the faster HBM2 memory standard, which is something NVIDIA is using on its professional side of things on the Tesla P100 graphics card. Vega will be the first GPU to utilize the faster HBM2 memory standard, which is something NVIDIA is using on its professional side of things on the Tesla P100 graphics card.
Colorful has just announced it has four custom GeForce GTX 1060 graphics cards on the way, ready to cover gamers in all segments of the mid-range market.
Colorful is aiming its new iGame series of GTX 1060-based cards at gamers who don't want to throw their wallets at the prices of the current GeForce GTX 1070 and GTX 1080, while offering great performance for 1080p 60FPS gameplay. Colorful's new iGame GTX 1060 series is led by the iGameGTX1060 X-TOP-6G, which has base/boost clocks of 1620/1847MHz, respectively. It will also feature 5+2-phase IPP power.
The second card in Colorful's iGame GTX 1060 arsenal is the iGameGTX1060 S-TOP-6G, which will have 1594/1809MHz base/boost clocks, respectively. The iGameGTX1060 S-TOP-6G will feature 4+1-phase IPP power. Under that, we have the iGameGTX1060 U-TOP-6G which will have the same 1594/1809MHz base/boost clocks, respectively - while it will amp up the power to a 5+2-phase design.