When AMD and Nvidia release all of those TFLOPs numbers, it’s important to realize that those are theoretical maximums. In order of a chip to reach that number, its architecture has to be extremely efficient and powerful to use. When Pascal first launched, Nvidia released some details about what it would look like. With the release of the white paper for the architecture, there are a few additional highlights worth noting.
First off, we know that Pascal has cut the SM (Stream Multiprocessor) down from 128 FP32 cores to 64. This allows for better distribution of processing power to tasks and as each SM keeps the same amount of register files and other support hardware, throughput is increased overall. Nvidia has also tweaked the SM so datapaths are more streamlined and sharing information within the SM takes less power and hardware. The scheduler has also seen some improvements and updates to ensure the SM is constantly being fed.Cache sizes have also been increased from 3MB to 4MB and a dedicated shared memory space of 64KB per SM has been added This is lower than the 96KB per SM in Maxwell but if you consider the doubled SM count relative to a same size Maxwell chip, it’s actually an increase of 16KB per SM. Finally, Nvidia detailed the P100 interposer layout for HBM2, something for us to look forward to when HBM2 finally arrives.
For other details, be sure to check out our earlier write up on the Pascal architecture.
One of the biggest concerns about Polaris 10 has been whether or not it will be a true replacement for Fury X. With the latest leaks out, most of the information points to about 100W TDP with 2304 shaders and clock speeds around 1050Mhz. Compared to Nvidia’s Pascal GP104, this doesn’t sound very competitive, leading to concerns that Nvidia would dominate the high-end. With the release today of AMD’s more detailed roadmap, our concerns have been laid to rest.
The new official roadmap offers a bit more detail than the one AMD showed back at Capsaicin. The new one offers more detail around Polaris 10 and 11, with both chips working to replace the entire Fury and 300 series lineup. This means the top Polaris 11 chip will offer enough performance to at least match, if not exceed Fury X. This should be competitive enough against GP104. If the 2304 shader report is true, AMD has truly revamped GCN 4.0 into something that is significantly superior to GCN 1.0 while cutting power consumption at the same time.
The layout for Polaris compared to the current lineup also suggests there will be no rebrands for the 400 series. It suggests that Polaris 10 will go from about 490X to 480 while Polaris 11 will fill in 470X down to at least 460. With how well small die low power Polaris 11 has done, rebrands don’t really make any sense. Finally, Vega will drop in 2017 with HBM2 and not in late 2016 as some have hoped.
With the improvements AMD has done, I am really looking forward to what Polaris and GCN 4.0 will bring to the graphics landscape.
After revealing their next flagship Telsa earlier, Nvidia has let loose with a few more details and specifications. Based on the new Pascal architecture, the P100 will be utilizing TSMC’s latest 16nmFF+ process. As we know from the keynote, the chip will feature 15.3 billion transistors and the latest HBM2 memory. The P100 also features what Nvidia is calling the “5 miracles”.
First off, the P100 will run at an impressive 1328 MHz base clock and 1480 MHz boost. This is high for a professional Tesla card though well in line with GeForce clocks. The card won’t be using the full GP100 die with 60 SMs and 3840 CUDA cores, rather it will use a cut-down version with 56 SMs with 3584 cores. This mirrors Kepler’s launch where the cut-down Titan came before the Titan Black. In addition to the usual FP32 CUDA cores, there are also 1792 FP64 CUDA cores for Dual Precision Work. This gives a 2SP/1DP ratio, higher than anything from Kepler or Maxwell. The P100 also has 224 TMUs and massive amounts of cache and register files.
Next, we have the massive 610 mm² die on 16nmFF+. About 50% of that is FP32 CUDA cores, 25% is FP64 and rest on other parts. This means despite the massive die size, the P100 and GP100 derivatives won’t be great gamers, as games generally only use FP32 CUDA cores. There may be a GP100 variant though that swaps out the FP64 cores for FP32 ones. Even saddled with compute though, GP100 will still beat the Titan X by a good margin. TDP is a relatively tame 300W, as expected from the use of 16nm and 16GB of HBM2.
Finally, most marketing statements are hyperbole and the “5 miracles” are no exception. They are the Pascal Architecture, 16nm FinFET, CoWoS with HBM2, NVLink, and New AI Algorithms. Honestly, none of these are really that amazing on their own and have been expecting. Combining all of them in one go on such a massive chip though is pretty amazing though. While the P100 will be shipping soon, don’t expect many till Q1 2017.
This year, both AMD and Nvidia will be launching their new Polaris and Pascal based GPUs. Unfortunately, it looks like the flagship chips won’t be arriving till next year. Set to arrive in early 2017, Vega, also known as Greenland, is to be the flagship replacement for Fiji. According to information 3DCenter dug up, Vega will feature 4096 GCN shaders, the same amount as Fiji currently has.
With Polaris and Vega, there are suggestions that AMD has managed to improve GCN 4.0’s performance by 30% compared to current GCN offerings. This alone should allow a significant performance increase over the Fury X. Fiji was also limited due to the design of GCN being unoptimized for massive chips with too many shaders and if AMD has managed to fix this, Vega will perform better than expected.
Furthermore, Vega will utilize HBM2 which will finally remove the 4GB cap faced by HBM GPUs as well as reduce latency. The use of 14nm as well and other Polaris improvements will also allow for a cooler and less power hungry die. We can also expect Vega to come in at a die size similar to Hawaii rather than Fiji, with a true Fiji size successor to come later on in the process cycle.
Even when much of the excitement about VRAM is coming from HBM2, that technology isn’t quite ready for prime time yet. For now, HBM2 is still a ways away and still a premium product. To hold the line, memory vendors have come up with GDRR5X, a significantly improved version of GDDR5. In what is unquestionably good new, Micron has just started sampling their GDDR5X modules to customers, way ahead of their original summer target.
GDDR5X has been moving along quickly since JEDEC finalized the specifications back in January. It was also only last month that Micron got their first samples back from their fabs to test and validate. This means that GDDR5X was easier to implement than expected and the quality of the initial batch was good enough that there wasn’t much to change in the production process.
Micron will be offering GDRR5X in 1GB and 2GB IC’s, allowing for 8GB and 16GB VRAM GPUs on as narrow as 256bit memory buses. The biggest advantage of GDDR5X is the doubling of bandwidth from 32byte/access to 64byte/access. Combined with higher clock speeds that allow for up to 16Gbps and improved power efficiency, the new memory will be a good match for Pascal and Polaris while we wait for HBM2.
In the few days after AMD first demoed Polaris 10 to us at Capsaicin, more details about the upcoming graphics cards have been revealed. Set to be the big brother to the smaller Polaris 11, the better performing chip will drop sometime after June this year.
First off, we’re now able to bring you more information about the settings Hitman was running at during the demo. At Ultra Settings and 1440p, Polaris 10 was able to keep to a constant 60FPS, with VSync being possible. This means the minimum FPS did not drop below 60 at any point. This puts the card at least above the R9 390X and on par if no better than the Fury and Fury X. Of course, the demo was done with DX12 but the boost is only about 10% in Hitman.
Another detail we have uncovered is the maximum length of the engineering sample. Based on the Cooler Master Elite 110 case used, the maximum card length is 210mm or 8.3 inches. In comparison, the Nano is 6 inches and the Fury X 7.64 inches. Given the small size, one can expect Polaris 10 to be as power efficient as Polaris 11 and potentially be using HBM. Given that Vega will be the cards to debut HBM2, Polaris 10 may be limited to 4GB of VRAM. Finally, display connectivity is provided by 3x DP 1.3, 1x HDMI 2.0 and 1 DVI-D Dual Link though OEMs may change this come launch unless AMD locks it down.
Even though a lot of information was shared from the Capsaicin live stream, some details weren’t made known till the after party. In an interview, Radeon Technologies Group head Raja Koduri spoke in more detail about the plans AMD has for the future and the direction they see gaming and hardware heading towards.
First up of course, was the topic of the Radeon Pro Duo, AMD’s latest flagship device. Despite the hefty $1499 price tag, AMD considers the card a good value, something like a FirePro Lite, with enough power to both game and develop on it, a card for creators who game and gamers who create. If AMD does tune the drivers more to enhance the professional software support, the Pro Duo will be well worth the cash considering how much real FirePro cards cost.
Koduri also see the future of gaming being dual-GPU cards. With Crossfire and SLI, dual GPU cards were abstracted away as one on the driver level. Because of this, performance widely varies for each game and support requires more work on the driver side. For DX12 and Vulkan, the developer can now choose to implement multi-GPU support themselves and build it into the game for much greater performance. While the transition won’t fully take place till 2017-2019, AMD wants developers to start getting used to the idea and getting ready.
This holds true for VR as well as each GPU can render for each eye independently, achieving near 2x performance benefit. The benefits though are highly dependent on the game engine and how well it works with LiquidVR. Koduri notes that some engines are as easy as a few hours work while others may take months. Roy Taylor, VP at AMD was also excited about the prospect of the upcoming APIs and AMD’s forward-looking hardware finally getting more use and boosting performance. In some ways, the use of multi-GPU is similar to multi-core processors and the use of simultaneous multi-threading (SMT) to maximize performance.
Finally, we come to Polaris 10 and 11. AMD’s naming scheme is expected the change, with the numbers being chronologically based, so the next Polaris will be bigger than 11 but not necessarily a higher performance chip. AMD is planning to use Polaris 10 and 11 to hit as many price/performance and performance/watt levels as possible so we can possibly expect multiple cards to be based on each chip, meaning probably 3. This should help AMD harvest imperfect dies and help their bottom line. Last of all, Polaris may not feature HBM2 as AMD is planning to hold back till the economics make sense. That about wraps it up for Capsaicin!
Even as Polaris approaches us quickly within 3 months, the planning for its successor has long been in the works. At their Capsaicin event, AMD took off the wraps for their upcoming GPU plans with a roadmap detailing the planned releases up till 2019. In keeping with the star nomenclature that started with Polaris and ditching the islands, we will have Vega and then Navi following Polaris.
Starting off with Polaris later this year, AMD’s main selling point it seems is the 2.5x performance per watt the new GCN architecture will bring. This is no doubt due to the combination of improved hardware itself, the new 14nm LPP process and DX12 finally making use of the previously wasted hardware resources like asynchronous controllers and shaders.
Moving along, we have Vega to release in what looks to be early 2017. The biggest change it seems is the use of HBM2, replacing GDDR5(X) and HBM1 no doubt. This means we can no doubt expected all Vega releases to utilize HBM2. While this may suggest Polaris won’t be using HBM2, it could also mean that only certain Polaris chips, likely only the high-end ones, will use HBM2.
Finally, we come to Navi, which should debut in early 2018. This release will have scalability and use of next-gen memory like Hyper-Memory Cube for instance. The scalability mention suggests either the use of smaller GCN units used to build the chip to better suit the market or a new process node. For now, we are probably better off trying to figure out what Polaris will be
For the upcoming graphics generation, both Nvidia and AMD are set to use HBM2 for their upcoming Pascal and Polaris graphics cards. While Samsung has already revealed that they have kicked off mass production for HBM2, we’re getting word on competitor SK Hynix’s plans for the new memory technology. According to Golem.de, the 4GB modules will start mass production in Q2 while the 8GB variants in Q3.
With HBM2, the maximum capacity per stack jumps from 1GB in HBM1 to 8GB with HBM2. This will allow GPUs to have up to 32GB of VRAM using the 8GB modules in the Fiji/Fury configuration, 16GB using the 4GB modules and 8GB using the lowest 2GB stacks. Each stack also doubles the bandwidth from 128GBps to 256GBps. Due tot he bandwidth increase all coming from a 1Ghz clock rate boost, we can expect latency to be severely reduced.
Given that SK Hynix along with AMD were the pioneers for HBM1, it is surprising that Samsung was the first one to reach mass production for the 4GB modules. On the other hand, SK Hynix may still have locked in the 2GB stacks first, the ones more likely to be used for consumer GPUs. With 4 stacks, this will allow for 8GB of VRAM, plenty for 4K and VR GPUs, while 2 stacks will do fine with 4GB for mainstream cards. Hopefully, this means that HBM2 cards will arrive sooner rather than later.
Even as this generation’s GPUs are continuing to fly off the shelves, Nvidia is already gearing up for their Pascal launch. Despite being quieter than AMD, it looks like Nvidia will launch their Pascal cards around the same time, in 2H 2016 as AMD’s Polaris will. What’s more, 2H 2016 will see Nvidia’s flagship Pascal GPU based on TSMC’s 16nmFF+ process and utilizing HBM2. This is still a rumour right now but it does fit the time frame since 1H 2016 would be too soon and 2017 too late.
The biggest question is what does”flagship” mean exactly. Ever since GTX 680 was launched, Nvidia has been playing around with the word flagship. Traditionally, the big dies like GF110 would launch first with the smaller mainstream GF104 launching after. Kepler and Maxwell saw that switch with GK104 and GM104 launching ahead of GK110 and GM200 respectively. This suggests that the so-called “flagship” may only be GP104 and not GP100. Even if it is GP100, it may well be a cut-down version, similar to how the GTX 780 was the cut-down variant of the later GTX 780Ti. This strategy does maximize sales for Nvidia but isn’t that great for consumers.
Whatever the card is, be it GP104 or GP100, it is going to use HBM2, giving it at least 512GB/s with 8GB of VRAM but potentially much higher at 16-32GB with 1TB/s+ of bandwidth. With AMD set to launch Polaris around the same time, Q3 2016 should make for exciting times as a slew of new GPUs arrive.
With both AMD’s Polaris and Nvidia’s Pascal both launching in 2016, the GPU market is very lively this year. Both competing platforms are also set to use HBM2 which JEDEC finally released the final standard for. Within days of JEDEC’s release, Samsung has just started mass production of their HBM2 memory chips.
Starting off with 4GB, Samsung’s HBM2 uses their advanced 20nm process. Each HBM2 package stacks four 8Gb core dies on top of a buffer die at the bottom. Each of these packages will offer 256 GB/s of bandwidth, quadruple that of HBM1. For a more dramatic comparison, each 4Gb GDDR5 die only offers 1/7 the memory bandwidth and 1/2 the energy efficiency. HBM2 also has 5000 TSV (Through Silicon Via), 36x more than what GDDR5 has.
Just like HBM1 did for AMD’s Fury lineup, Samsung is expecting HBM2 to bring 95% space savings compared to GDDR5. In addition to regular HBM2, Samsung is also planning HBM2 with ECC, likely for GPGPU and enterprise work. This means not only will consumer GPUs get numerous benefits, but GPUs found in supercomputers and data centers will soon have HBM2 as well, where it is arguably more useful. Samsung is expecting to continue to ramp up production as demand increases over the year.
Ever since HBM1 was revealed and launched with Fury X, many have been looking forwards to what HBM2 would bring along in 2016. While HBM1 brought large power savings and a major boost in memory bandwidth, it was largely limited to a relatively low 4GB capacity. HBM2, however, is set to provide a boost in capacity and bandwidth by increasing the number of stackable dies. We’re now getting reports that AMD’s upcoming Polaris chips will utilize HBM2.
As a major revamp of the GCN architecture, a Polaris flagship GPU would be the natural product to debut HBM2. A flagship GPU much more powerful than current generation chips due to the new architecture and process node would likely require more memory bandwidth to feed it and a high memory capacity as it would be meant for VR and 4K gaming. Being the largest chip in the lineup, the flagship would also benefit from the major power savings, helping offset its core power consumption. The confirmation of HBM2 also suggests that we will be getting high-end Polaris chips this year.
At the same time, AMD is also confirming that they will continue to use GDDR5 and likely GDDR5X as well. At CES, AMD showed off a low powered Polaris chip using GDDR5 that was able to provide the same performance as Nvidia’s GTX 950 but with a significantly lower power consumption. With such a leap in efficiency, the HBM2 chips will likely be light years ahead of current cards in terms of efficiency if GDDR5 already shows such massive gains.
The graphics card market is full of interesting power struggles and if recent reports are true, it seems 2016 will be one of the biggest battles yet. AMD may have already put out some cards with HBM memory, and we’ve heard that Nvidia will be doing the same soon too, but don’t count GDDR memory out just yet! It seems that the upcoming GDDR5 standard is being developed by Micron, which will power mid-end graphics cards, while HBM2 will likely remain for higher end cards.
Of course, there’s some confusion here as JEDEC are already working on the GDDR5X standard, so where Micron fits in really remains to be seen, but that’s something we’ll have to wait and see. GDDR5X is said to double the bandwidth, so is GDDR6 is new standard, or just a further refinement of 5X? Either way, we can expect it to adopt a lower node, most likely starting from 20nm and working down from there, allowing for higher clocks, and lower voltages, although these kinds of improvements are the obvious targets for any increase in performance these days.
Our guess is that the revised GDDR standards will be acting as a bridge until HBM matures enough to cover a wider range of cards and budgets. Either way, 2016 is shaping up to be an exciting time in the GPU market, with new memory, new architectures, new cards and so much more on the horizon.
NVIDIA’s upcoming architecture, codenamed ‘Pascal’ is a revolutionary step towards greatly enhanced processing power in a small form factor. The advent of stackable high bandwidth memory was first introduced in AMD’s Fury line-up. While AMD was first to market with this technology, they could only utilize HBM1 which has a 4GB VRAM limitation via single GPUs and 8GB with a dual card solution. Thankfully, the second iteration allows for increased memory capacities which is vital given the VRAM utilization of many modern engines.
During NVIDIA’s GTC event in Japan, the company confirmed their flagship Pascal GPU will launch with 16GB HBM2 and feature 1TB/s total bandwidth. This quashes some early theories about NVIDIA’s professional line containing 32GB of HBM2. Obviously, these figures can increase in the future but it greatly depends on the manufacturing costs and supplies of HBM2. It’s important to remember the shortages of AMD’s Fury X line on launch, which was a result of low HBM1 yields.
Nevertheless, Pascal looks set to be an absolute powerhouse, and I cannot wait to see the performance numbers. Furthermore, NVIDIA’s pricing for the flagship model could easily exceed $1000, and perhaps reach $1500 given the pricing of the Titan X, and HBM2 costs.
We have previously reported the rumours that Nvidia was planning to use GDDR5X on their upcoming Pascal graphics cards, a rumour that not everyone bought right away. But there are quite a few reasons that we could see this happen and the newest leak seems to support this.
Last time it was a German site that leaked the GDDR5X information and this time we get news from a Russian outlet that got their hands on what looks like leaked slides from Micron of the upcoming GDDR5x memory. The site also suggests that this won’t be limited to only Nvidia, but that AMD also will want to get on board and use this type of memory on some of their graphics cards next year.
HBM, and HBM2 might be very exciting and be the next mainstream graphics memory, but it is still a costly one to produce and the production is also limited. This leaves room for the next GDDR5 standard to make its entry. GDDR5X offers double the data-rate per memory access of 64 byte/access compared to 32 byte/access of the current GDDR5 standard. Where current GDDR5 tops out at around 7Gbps, the new standard will initially offer 10-12 Gbps with a later goal to achieve 16 Gbps.
The implementation of the new memory should be relatively easy and cheap for manufacturers as the new chips will retain the same pin layout. With all this information, we can assume that HBM will be reserved for the top-tier graphics cards for the foreseeable future while GDDR5x will breath more power into mid-level and entry-level cards and allow them to perform better at the ever-increasing resolutions.
NVIDIA’s upcoming Pascal launch is hotly anticipated as the company transfers from GDDR5 to High Bandwidth Memory on a 16nm manufacturing process. The performance increase from Maxwell is predicted to be quite substantial compared to previous generation graphic card launches and it will be interesting to see NVIDIA’s first HBM-powered GPU.
Unlike AMD’s Fiji line-up, the 2nd generation HBM chips allow for increased memory capacity. Apparently, NVIDIA’s top-tier gaming chip will feature 16GB HBM2 while the compute card utilizes 32GB. This is a staggering increase from the 6GB on the GTX 980Ti or 12GB on the Titan X. The graphics card opts for a 4096-bit memory interface, DirectX 12 functionality and contains around 17 billion transistors.
The GPU’s core has been teased at GTC 2015 in Japan and looks pretty similar to earlier announcements. Clearly, with the new manufacturing process and move to HBM2, Pascal is the most important release for over a decade. HBM revision 1 has been somewhat plagued by supply chain issues but this should be resolved by the time for Pascal’s release in 2016. HBM2 is a major step up, but consumers should expect a very hefty price tag for NVIDIA’s truly next-generation chip.
Thank you TweakTown for providing us with this information.
Nvidia’s Pascal may not arrive in volume for quite in a while according to the latest report. According to BusinessKorea, both Samsung and SK Hynix are set to provide HBM2 to Nvidia for their next generation Pascal architecture. That, however, is contingent on testing to be trouble free in 2015 and volume production will only start in 2016. At the earliest, we may not see HBM2 for GPUs till at least Q2 2016 and likely later than that.
More interestingly though, is the fact that SK Hynix was supplying HBM1 to Nvidia at the same time as AMD. This means previous reports about AMD priority access may have been overblown. It also means that the GP100 we’ve been hearing about might actually ship with HBM1 given HBM2 still isn’t in production at this point.
While we have known that Samsung wasn’t planning on starting mass production of HBM2 till 2016, this is news from SK Hynix. This also means both Nvidia and AMD will have to wait for a while before they can utilize HBM2 for their Post-GCN and Pascal architectures. For those who plan on buying AMD’s HBM1 enabled cards though, it means their purchases remain relevant for that much longer.
According to multiple sources, AMD is working on 3 new GPUs as part of their 2016 lineup. Likely part of the Arctic Islands release, the 3 chips all fit the theme being called Greenland, Baffin and Ellesmere. Greenland is set to be the flagship product while Baffin and Ellesmere will probably target other segments like performance and mainstream.
While we’ve already gotten some information about what Greenland will look like, we’re also learning that the ISA (Instruction Set Architecture) is undergoing major changes. Since 2011, AMD has been working with GCN, or Graphics Core Next. There have been several iterations from the original GCN 1.0 like 1.1 which powers the R9 290/290X and GCN 1.2 which is found in the R9 285 and Fury/Fiji. With this latest ISA change, AMD is moving beyond GCN to “post-GCN” on a radically new architecture. In some ways, the time spent with GCN highlighted how AMD has a slower cycle than Nvidia but also showcased the staying power of the architectures, which is finally seeing AMD’s early invest pay off with DX12.
This new ISA is also set to net AMD massive gains in the power efficiency department. While Nvidia showed strong gains moving from Kepler to Maxwell, AMD is reporting that their new ISA will double power efficiency. This should leapfrog Maxwell and will likely be a strong competitor to Pascal. Given the use of either 14nm or 16nm FinFETs as well, we may well see even more massive power savings. Combined with up to 32GB of HBM2 which is also pretty power efficient, 2016 looks to be a pretty big year.
Both Baffin and Ellesmere are also expected to be entirely new cards as well. This will bring a much-needed refresh to the AMD lineup which largely consists of rebrands at this point. Hopefully, these new cards and Zen will bring enough to the table and turn things around for the beleaguered firm.
Thank you WCCFTech for providing us with this information
While SKHynix was the first to produce HBM for AMD, things may change next year. Samsung is leveraging their massive production base and getting into the HBM market with plans to start mass production in 2016. Given that Samsung is planning to provide 4Hi and 8Hi stacks, Samsung is skipping right past HBM1 and straight to the more useful HBM2. Samsung is marketing the use of HBM as a fast cache like eDRAM for both normal consumers but also targeted towards high-performance computing as well.
Having more suppliers for HBM is obviously a good thing for AMD as it ensures that SkHynix won’t become a bottleneck both for current HBM cards but also for their upcoming Greenland flagship in the future as well. This is also bad news as AMD reportedly has a priority agreement to get HBM from SKHynix. It was inferred at the time this meant that AMD could effectively lock out Nvidia from HBM for a while till production ramped up. With Samsung leveraging their massive fabs, it seems that Nvidia will probably be able to get all the HBM they want as well.
With HBM being in plentiful supply, Nvidia probably won’t face any issues on this front for their new Pascal GPUs. It’s important to note that while Samsung does plan to start mass production in 2016, when that ramp up will take place and how long that will take is still unknown. This means that AMD might still have an advantage, albeit now diminished. With widely available HBM2, a new process node and new architectures, the GPU landscape for 2016 looks very promising.
Thank you ComputerBase for providing us with this information
We have known for a while now that Nvidia wanted to get onto the HBM wagon too and that is pretty much a given. It’s the next generation video memory and it looks extremely promising so far. The next gen GPU should be the GP100 and it will reportedly rock somewhere between 4500 and 6000 CUDA cores and it will be coupled with HBM2 memory.
The latest news now points towards two new graphics cards: one for the consumer market and one for the server and workstations, again this is almost a logical assumption anyway. The first new GPU will use 4-Hi stacks of HBM2 memory and be aimed at the consumer market while the other business-oriented GPU will feature 8-Hi stacks of HBM2 memory. Effectively that means HBM powered graphics cards from Nvidia with up to 32GB of very fast VRAM.
Whether the new consumer model will be named the GeForce GTX 1080 is currently unknown. It could very well be that Nvidia is working on a new naming too, not only to avoid the thousand digits, but also to underline that is truly is a next generation graphics card; much like AMD has done with their Fury branding.
We’ll make sure to keep you updated as more news and leaks emerge on these new Nvidia 16nm based graphics cards.
Thank You TweakTown for providing us with this information
AMD made history earlier this month by being the first major GPU vendor to ship HBM with their top end Fury and Fury X graphics cards. Nvidia however, has been absent so far, waiting on HBM2, a more advanced version of the HBM1 shipping with Fury(X), before getting into the new tech. According to a report though, AMD is leveraging their deal with SK Hynix to get priority access to HBM2 in time for their upcoming Arctic Islands GPUs.
While HBM1 is limited to 4GB and 512GB/s, HBM2 increases those numbers significantly with up to 16/32GB of VRAM and over 1024 GB/s. Like HBM1, HBM2 is expected to be in limited supply at launch. If AMD has priority for HBM2, and the stocks are low, it may mean that Nvidia practically won’t be able to use HBM2 until the supply improves enough that AMD can’t use what is available. This might create a de facto exclusively for AMD, offering a chance for the underdog to dominate with HBM2 GPUs.
If the supply of HBM2 is limited, it could complicate things for Nvidia. Their Pascal architecture is set for 2016 and could be designed for either GDDR5 or HBM2, which vary widely in implementation. Nvidia can choose to go with GDDR5 but risk losing its lead over AMD and inability to refresh with HBM2 later on. If Nvidia does go with HBM2, supply might be heavily constrained, allowing AMD a chance to grab market share. It will be interesting to see both side’s offerings in early 2016 and the choices they make for their lineup.
Thank you WCCFTech for providing us with this information
AMD is putting the final touches on everything and preparing to launch their new Radeon R300 series graphics cards very soon, but before it even hits the market we already get information about the next generation of AMD graphics cards. The R9 300 series is set to launch in June during Computex in Taipei and it will continue to use the 28nm process as the 20nm process just isn’t viable yet for these kind of products, the costs are just too high.
But the next generation of AMD cards from the Arctic Islands series, codenamed Greenland, will be built on the 14nm FinFET technology. This means that AMD could skip the 20nm process entirely. Another detail revealed is that the Greenland card will use the second generation of HBM memory with increased bandwidth and capacity. It is expected that AMD’s 14nm FinFET process will be produced by Globalfoundries OEM.
This could mean some promising times ahead of us with more powerful GPUs that use even less power than they do today, but also heavily improved memory in both capacity and speed. I can hardly wait to see what AMD has to offer here, although we should be looking forward to the next generation R9 300 cards instead. Computex isn’t far away, so it will be an exciting summer.
AMD’s planned 2016 next-gen APUs are rumoured to be 16-core processors, based on the Zen architecture, featuring a Greenland GPU, quad-channel DDR4 support, and HBM memory, according to TweakTown.
The forthcoming APU will replace the Godaveri platform, first introduced with the Carrizo APU. AMD is releasing a series of Carrizo notebooks later this year, likely to be launched at Computex 2015 in June.
The Greenland GPU is expected to be based around the Fiji architecture, which powers AMD’s flagship graphics card, the Radeon R9 390X. Though the 390X uses HBM1, AMD is expected to switch to HBM2 for Greenland next year.