TweakTown NewsRefine News by Category:
MSI might be working on two new GeForce GTX 970s, but the new EVGA card might having everyone crying into their wallets. K|NGP|N, the crazy overclocker himself, is working with EVGA on a new GeForce GTX 980 Classified.
We don't know much about the new card, whether it will feature the GM204 core, or the bigger and badder GM204 core. I'm hoping we see EVGA push the boundaries and use the full GM200 core, but with K|NGP|N behind the scenes tinkering around with this, we can expect big, big things from EVGA's latest card.
MSI has shown off some damn impressive new GeForce GTX 970 cards, with three awesome color schemes on offer: red, gold and green. The green version is my favorite, as it's the perfect match for the company celebrating the fact they've shipped 100 million GeForce video cards.
The company is also preparing a new GTX 970 with a black and white theme, a card codenamed GTX 970 4GD5T-OC. There's also the GTX 970 100ME, but the GTX 970 4GD5T-OC card doesn't feature a backplate, and it features cheaper components, too. Both of the cards will be made available lalter this month, with the green edition of the GTX 970 featuring a "special gift".
NVIDIA made quite the splash when it launched its new Maxwell architecture in September, but the company is reportedly preparing its new Quadro M6000 GPU which will see the visual computing market get excited, very excited.
The new NVIDIA Quadro M6000 should be based on the GM200 architecture, with a fully unleashed Maxwell chip at the heart of the card. The full GM200 would feature 3072 CUDA cores (24 SMMs, while the GM204 sports 16 SMMs). This full GM200 core also features 96 ROPs, and either 192 or 256 TMUs. The memory interface will be expanded from the 256-bit bus on the GeForce GTX 980 (based on the GM204 core) to 384-bit for the Quadro M6000.
On top of that, we should see the Quadro M6000 feature a huge 12GB of GDDR5 RAM with a total memory bandwidth of 317.4GB/sec thanks to its 384-bit memory bus. We should see NVIDIA launch the Quadro M6000 later this month, or even tonight at its CES 2015 event.
CES 2015 - If you're still searching for a GeForce GTX 970 card, Zotac will be showing their new AMP! Extreme and AMP! Edition models in Las Vegas very soon at CES 2015.
Quoted as "designed for enthusiast gamers" in their recent press release, these two new graphics additions to the Zotac family will be focused around extended features and performance capabilities.
ZOTAC will be displaying these cards in an SLI configuration, showcasing their stellar 4K gaming capabilities to the public. One system each will be displaying their AMP! Edition and AMP! Extreme Edition cards.
We only reported a couple of weeks ago that NVIDIA was rumored to launch its GeForce GTX 960 in January, but now some more information is coming through, including a price. At the time, I said that if the company launched the GTX 960 with a $199 price, it could be a hit. Well, this price could end up being true.
Various sources have confirmed that the GTX 960 will rock the GM206 core, but we could end up seeing NVIDIA use a cut down GM204 core, too. We have heard previously that the GTX 960 would feature 4GB of RAM spread across a 256-bit memory bus, but now we're hearing that the GTX 960 could feature a 128-bit bus.
Speculation is starting, where we could also see a GTX 965 or GTX 960 Ti arrive, where we could see the GTX 960 feature a 128-bit bus while the GTX 960 Ti version could throw things up to a 256-bit memory bus. We are still expecting one 6-pin PCIe connector and a total of 150W TDP, which will have mid-range systems enjoying a low-power, but high performing GPU.
Prepare for sadness: AMD and NVIDIA's upcoming next-gen GPUs could be delayed, at least the 16nm or 20nm based versions of them, according to a new report from WCCFTech.
The site is reporting that TSMC may be rolling out its 20nm technology, but TSMC is busy filling orders for Qualcomm and Apple. Flagship GPU dies are much, much larger than the likes of system-on-chips (SoCs) that Qualcomm and Apple require, so that leaves Apple and NVIDIA with a very limited supply of 20nm dies.
What does this mean for AMD and NVIDIA's future GPUs? NVIDIA is already sailing quite well with its efficient Maxwell architecture, where even on the now ageing 28nm it is pulling some seriously good numbers in both camps: performance and power efficiency. AMD is most likely the next up for a GPU refresh, but it looks like the company is going to have to side on the 28nm fence, which should mean its upcoming next-gen architecture should be quite impressive.
It looks like we're seeing the beginnings of SAPPHIRE's new GPU, which is going under the guise of 'Project NFC.' What is Project NFC? Well, NFC stands for Not From Concentrate, which is a term used in the food industry for when water hasn't been extracted from the juice of a fruit. In SAPPHIRE's tease, it is more of a pure and unaltered version, which could be a liquid cooled video card - exciting.
The company hasn't coming out and said that Project NFC is an actual liquid cooled AMD card, but SAPPHIRE did hint at it on its video on YouTube. We have possible drawings of a GPU block, and water vapor as you can see below.
The next question is, is this the Radeon R9 380X? Or would SAPPHIRE liquid cool one of its existing cards? Whatever happens, it's SAPPHIRE: which means we know it's going to be good, very good.
GIGABYTE and EVGA have been on the market with external VRM solutions for serious GPU overclocking, but it looks like ASUS is stepping into the game with its own external voltage regulator module, or VRM.
ASUS has already deployed GPUs with advanced 10-phase and 14-phase VRMs in the form of the ASUS Strix GeForce GTX 980 and the ASUS ROG Matrix Platinum GTX 980, but this is an entire new ballgame. The ASUS GPUs with the advanced VRMs still have limitations as they're consumer GPUs, with certain restrictions that stop too much power flowing through the card. The external VRM card will allow for far higher voltages being pushed through the GPU and memory, which should unlock some massive potential for extreme overclockers.
The VRM card in question features a "single 8-phase output with output voltage of up to 2.5V (with output voltage offset switches [+0.4V, +0.3V, +0.2V, +0.1V]) and current up to 500A. The card has on-board voltage control/monitoring, output current monitoring, VRM temperature monitoring, load-line calibration (0%, 60%, 80%, 100%), hotwire setting/monitoring and other features required by extreme overclockers. The board sports four six-pin PCIe (4*75W) input power connectors, which means that it can deliver up to 300W of power to the graphics board, enough power to break world's records" reports Anton Shilov from KitGuru.
According to "AMD_Chris" on various forums, AMD is working on an impressive new feature dubbed "Dynamic Frame Rate Control". DFRC would allow gamers to put a lock on the total frame rate their video card can render, which can result in a huge amount of power savings.
The feature would most likely see AMD variably adjusting the clock speeds of the cards in order to hit the desired frame rate, such as 60FPS. It might sound like V-Sync, but it's nothing like it as DFRC stops your GPU from cranking things up internally to render 100FPS, when you're only receiving as much as your monitor can put out, which is 60Hz, or 60FPS most of the time.
DFRC will underclock your GPU once you hit 60FPS (or whatever frame rate you choose), allowing the card to not pull as much power from the wall. AMD_Chris says that "the power savings were mind blowing" and we would agree, if your card is rendering 120FPS+ in a more basic game and you've got DFRC set to 60FPS, the power savings would be fairly large. We can't wait to test this new feature, that's for sure - what about you?
According to the latest rumors, the upcoming NVIDIA GeForce GTX Titan II has been benchmarked against the Radeon R9 380X and R9 390X. These are just rumors right now, as all three of these cards have not been announced yet. Even the name Titan II could change to Titan II, or Titan Extreme. AMD on the other hand could change their cards to 399X, or 400X or anything out of the norm from what to expect.
The details on the cards from the poster on Chiphell state that the Fiji XT and Bermuda XT (380X and 390X respectively), will be manufactured on Globalfoundries' 20nm process. The Titan II on the other hand will arrive as the GM200, which will supposedly come in two flavors: a cut down version with 21 MM units and a total of 2688 CUDA cores. A "full fat" version will arrive at a later date with an increased amount of SMMs and CUDA cores.
When it comes to performance, the GM200-powered GPU from NVIDIA will be 34% faster than the GeForce GTX 980. On the AMD side of things, the R9 390X will be a huge 65% faster than the R9 290X, which will feature liquid cooling that was rumored all those months ago. Performance wise, the next-gen GPUs are packing a massive punch.