Graphics Cards - Page 232

We're only weeks away from the official unveiling of AMD's next-gen Polaris 10 GPU, with new reports from a recent event hosted by the company in Taiwan about Polaris.

The event doubled as a way of showcasing their next-gen Polaris architecture, as well as the new dual Fiji video card, the Radeon Pro Duo. The upcoming Polaris 10 has a TDP of 175W, but will "actually consume much less than that", reports WCCFTech.

The Polaris 10 reportedly hits around 4000 points in 3DMark FireStrike Ultra, compared to the R9 Fury X and GeForce GTX 980 Ti which also hit around 4000 points. All AMD will have to do is hit the $300 (or so) price point on the Polaris 10, and it would be a winner if it consumes less than 175W, and cranks out GTX 980 Ti like performance.

AMD is days away from pushing its Radeon Pro Duo into the wild, its first dual-GPU video card since the infamous Radeon R9 295X2. The press slides are now out - even before I received them (thanks, AMD...) and now we know the official specs - even though we kinda knew that from the constant flow of leaks on the Radeon Pro Duo.

The Radeon Pro Duo features 2 x Fiji XT cores built on the 28nm process, the same GPU powering the Radeon R9 Fury X. We have 4096 stream processors, so with two GPUs we have a total of 8192 stream processors on the Radeon Pro Duo. This is joined by 256 TMUs per GPU (512 TMUs in total) and 64 ROPs per GPU (128 ROPs total). This provides the dual-GPU video card with an insane 16.4 TFLOPs of performance, which is a huge jump on the 8.6 TFLOPs of compute performance from the R9 Fury X.

The GPUs are clocked at 1GHz, while the 4GB of HBM1 per GPU (8GB HBM1 total) is clocked at 500MHz, and with its 4096-bit memory bus, we have 1024GB/sec of memory bandwidth. There's a 350W TDP on the card, which is not too bad at all considering the R9 Fury X has a 275W TDP, while the price is a huge $1499. This isn't a card for gamers wanting to hit 1080p 60FPS, or even 4K gamers - this is a card for the serious, insane enthusiasts who want 4K 60FPS constant, or multi-monitor/VR gamers.

Project Quantum is one of the coolest things AMD has ever shown off, but it was never delivered to market. Project Quantum and an insane amount of power inside of its small, and super-unique chassis - but used an Intel CPU at the time, alongside the dual Fiji GPU when it was debuted at E3 2015.

But with AMD's next-gen Zen CPU architecture nearly here, the company can move away from using its competitors' faster CPUs, with one of its own. According to the latest rumors, AMD postponed the release of Project Quantum for two reasons.

First, an "internal decision to market the Radeon Duo Pro, not as a purely gaming product, but as a professional high performance solution". Second; "The lack of a high performance AMD CPU and Motherboard solution (the original Project Quantum uses an AsRock Motherboard) that could accompany the powerful dual GPU".

The last time we reported on the GPU market share, NVIDIA dominated with 82% of the dGPU market in November 2015.

During AMD's Q4 2015 earnings call, company CEO Lisa Su said: "We have clear opportunities to regain GPU share in 2016 based on the performance per watt of our new GPUs and software leadership". AMD has teased just a few hours ago that their Polaris 10 will be for the gaming PC and high-end notebook market, while the Polaris 11 will find its way into standard notebooks.

Now, for AMD to regain GPU market share, it would have to hit NVIDIA where it hurts: the mainstream video card market. This is the GeForce GTX 950/GTX 960 and GTX 970 markets - where AMD didn't do too much with the Radeon 300 series last year. The Radeon 300 series were mostly rebrands, but the new Polaris 10-based Radeon 400 series should be something completely different.

We now have a better idea of what AMD's next-gen Polaris architecture will be when it arrives on video cards next month, thanks to AMD's recent press release.

There's on specific paragraph that's worth discussing, where AMD said: "AMD demonstrated its "Polaris" 10 and 11 next-generation GPUs, with Polaris 11 targeting the notebook market and "Polaris" 10 aimed at the mainstream desktop and high-end gaming notebook segment. "Polaris" architecture-based GPUs are expected to deliver a 2x performance per watt improvement over current generation products and are designed for intensive workloads including 4K video playback and virtual reality (VR)".

But as VideoCardz notes, we need to focus on the word 'mainstream'. AMD has stamped the word 'mainstream' onto everything between casual and enthusiast segments, where back in 2014, mainstream was associated with the Radeon R7 260.

Every single day seems to bring something new to the table when it comes to next-gen GPUs, with today being no exception - in the last 14 hours or so, we've gotten our first look at the GP104.

NVIDIA's GP104 GPU is the Pascal-based consumer GeForce GPU, which will power the upcoming GeForce GTX 1080 and GTX 1070 video cards. The GP104 is made on the 16nm FinFET process, but it looks like the GP104 will power two very different video cards.

The GeForce GTX 1080 is expected to arrive with 8GB of GDDR5X, while its sibling in the GTX 1070 will reportedly sport 8GB of GDDR5 - both with 256-bit memory buses. The GTX 1080 with its 8GB of GDDR5X will have it clocked at 2GHz (8GHz effective), an increase from the 7GHz effective speeds on the current GDDR5-based offerings.

AMD is days away from the official launch of its Radeon Pro Duo, the company's dual-GPU based on the Fiji architecture and HBM1 technology. HWBattle has received their sample, and have uploaded pictures for our viewing pleasure.

The Radeon Pro Duo features two Fiji XT GPUs with 4096 stream processors each, for a total of 8192 stream processors. Each GPU features 64 ROPs and 256 TMUs, with each GPU clocked at 1GHz. Each GPU also has 4GB of HBM1, for 8GB HBM1 total. We have 16 TFLOPs of 32-bit single precision compute performance, which makes the Radeon Pro Duo the fastest video card on the market.

The 4GB of HBM1 per GPU is clocked at 500MHz, resulting in 512GB/sec over the hugely wide 4096-bit memory bus. This means that applications designs for specific workloads on the Radeon Pro Duo can hit over 1TB/sec bandwidth - which is simply put, insane. The TDP of the Radeon Pro Duo is 350W, with 3 x 8-pin PCIe power connectors required. There's 3 x DP outputs and 1 x HDMI, too. Watercooling is standard, as the Fiji XT GPU runs quite hot, but AMD's engineering department took care of that with the Radeon Pro Duo.

It looks like AMD's next-gen Polaris flagship GPU has been spotted again, with the mysterious 'C99' receiving its RRA certification. This certification puts it on a path towards being consumer ready, which is exciting. These numbers might not look like much, but the nomenclature is used for test boards. All of the boards start with the letter C, with the number proceeding the letter C being the generation of video card. For example:

This means that Polaris-based boards, alongside other GCN 4.0-based, should have the C9XXXX nomenclature. The Baffin XT-based GPU that was on Zauba recently had the C98101 codename, which tells us it's a Polaris-based product. But now the C99305, C99304, C99303, and C99302 have been spotted getting their RRA certification, back on April 8.

The RRA certification is South Korea's National Radio Research Agency, with any silicon-based electronic needing this certification before it can be a consumer product. The certification doubles as a final stage of design of the GPU, as no more changes can be made after the RRA certification has been approved.

We keep reporting on the GeForce GTX 1080 and GTX 1070, based on the GP104 GPU, but what about the successor to the insanely popular GTX 950 and GTX 960? The GP106 will be released sometime later this year, and it might be a surprisingly powerful, yet power efficient video card.

According to the new reports, GP106 cards won't require any additional PCIe power connectors, so it'll be a slot-and-forget deal. We should expect the GeForce GTX 1060 to use less than 75W of power, which will be insane given that it should deliver a nice 30-50% bump in performance over the current GM206-based GTX 960. The 16nm FinFET process at its finest.

Right now, the GTX 950 and GTX 960 are still doing incredibly well in the market, so there's no need for NVIDIA to pump out a GM206-based successor just yet. But, the GTX 970 and GTX 980 cards have been out for nearly two years now, so bring on the GTX 1070/GTX 1080!

It might be late in the GTX 980 Ti game to make that much of a splash, but Colorful Technology Company Limited, a Chinese manufacturer of some very special video cards, has just unveiled its latest creation - the iGame GTX 980 Ti KUDAN.

The iGame GTX 980 Ti KUDAN features not just a unique custom-cooler, but powerful heat sinks packing an integrated watercooling loop that will provide some insane cooling. Five heat pipes are responsible for taking heat away from the GPU, dispersing it throughout densely packed fins and the watercooling loop, which spools up when extra cooling is required.

Colorful has deployed full-time GPU thermal monitoring which provides smart control over the integrated watercooling system, which can be configured with a custom watercooling loop if you want. The iGame GTX 980 Ti KUDAN also has an Overclocking Key, which when pressed, boosts the GPU up to a huge 1304MHz Boost clock.

Soon, I'm going to have to stop reporting on these teases of next-gen video cards, because I'm just getting too excited. We're at the point now where we're talking and reporting about the cooling shroud being removed like it's another bread crumb in the Pascal trail.

Today, we have the cooling shroud removed from the Pascal-based GeForce GTX 1080/GTX 1070, with the new cooler made using die-sinking technology. Die-sinking technology is a method of shaping a form with sparks, or electric discharges. As I explained in my GTX 1080 post yesterday, these coolers provide me with the feeling of Transformers, and I don't know if I like that or not - yet.

The new pictures see the GTX 1080 cooler made from four parts, with one of the characters in the GTX 1080/1070 not placed yet, so that it can be placed as a '7' or '8' depending on the card.

After teasing a rough GPU roadmap at its Capsaicin event during the Game Developers Conference, AMD has just replaced it with an official roadmap that shows what to expect through to 2018.

This year, we know that it's all about Polaris, but in 2017 the real fun begins with the Vega architecture as AMD will be using HBM2 on their next-gen GPU architecture. In 2018, the company will succeed Vega with Navi, which teases a still unknown "NextGen Memory".

The new Polaris 10 and Polaris 11 GPUs will feature fourth-gen GCN cores, with HEVC encode/decode abilities, the it's-about-damn-time HDMI 2.0 capabilities as well as DP1.3, and is built on the exciting new 14nm FinFET process. Now, let's get into more detail.

AMD has an exciting piece of technology with its Polaris architecture, but its upcoming Radeon M400 mobility series will reportedly be filled with rebrands, except for the high-end designs, which will be based around Polaris.

VideoCardz is reporting that AMD will be launching the rebranded lineup, followed by Polaris-based mobility offerings later in the year. The new Radeon R9 M470X will be a rebranded Radeon R9 M385X, with both of these SKUs using the Bonaire XT GPU. It'll rock 896 stream processors, 56 TMUs and 16 ROPs.

Then we have the Topaz XT-based Radeon R7 M460, R7 M440, R5 M445 and R8 M445DX which will be rebranded Radeon R7 M270DX, R7 M260, R7 M265DX, and R7 M360, respectively. The Radeon R5 M430 will be based off of the R5 M330 and its Sun XT-powered GPU, while the Jet Pro will power the new Radeon R5 M430 which is jus the R7 M260DX.

With NVIDIA poised to unveil its next-gen Pascal-based GeForce cards next month, AMD can't miss out on all of the fun - with the company set to unveil its own next-gen Polaris-based video cards late next month, and early June - during Computex in Taiwan.

AMD will also reportedly be showing off their new Bristol Ridge-based APUs, which are set to hit notebooks with the latest Radeon M400 series video cards. Now, onto the Polaris news. It's pretty much fact that AMD's next-gen Polaris 10 (codenamed Ellesmere) will be unveiled, with a super-small TDP of around 100W, and will have no issues playing games at 2560x1440 at 60FPS.

The new Polaris 11 GPU is codenamed Baffin will consume even less power, with a TDP of under 50W. We heard yesterday that the Polaris 10 samples are hitting speeds of 800MHz, and right up to 1050MHz, with 2560 stream processors (but only 2304 SPs enabled on that particular sample).

AMD will be launching its first dual-GPU video card since the Radeon R9 295X2, with the new Radeon Pro Duo to be released on April 26 for $1499. Some lucky people have gotten their hands-on the Radeon Pro Duo, which is Radeon Technologies Group's first video card since it broke away from AMD last year.

The Radeon Pro Duo features 2 x Fiji XT GPUs on the 28nm process, with 4096 stream processors per GPU. We have a GPU clock speed of 1GHz, with 16 TFLOPs of FP32 performance, and 0.9 TFLOPs of FP64 compute power. Underneath, we have a 4096-bit memory bus courtesy of HBM1, with 4GB of HBM1 per GPU for 8GB HBM1 total. The 4GB HBM1 per GPU is clocked at 500MHz, with 512GB/sec of bandwidth. In total, the Radeon Pro Duo has a 350W TDP, and is liquid cooled.

But something that's quite cool, is that RTG's first video card includes a spare Fiji XT GPU. This can't be used for anything, but it's something you could use as an ornament - or you could throw it on your keychain and rock an awesome GPU around with you.

Some strange new screenshots of rather benign parts of the operating system might just show the verifiable, bona fide existence of AMD's Polaris 10 GPU. Someone decided to show off the device ID of an installed GPU and corresponding AMD Graphics Manager shots with that list the operating frequency of the elusive GPU.

VRWorld received a small tip from someone in the industry (supposedly) with some evidence of working engineering samples of the upcoming Polaris 10 GPU. We've recently learned the device ID's that are supposed to be associated to the new Radeon R9 4xxx series of cards, and this matches up perfectly. This happens to be one of them, if it's not doctored. Interestingly, the clock speed of the card is at only 1050MHz.

We know that Polaris is built on the 14nm FinFET process and should feature GDDR5X RAM. We've previously heard that HBM 1.0 might make an appearance, though it seems more likely that GDDR5 in some guise will also be included for higher VRAM capacities. Internal enhancements to GCN and the inclusion of enhanced geometry processors, a new command processor and a much better memory controller should allow for higher performance gains than we might be able to conclude otherwise.

NVIDIA blew us all away with its reveal of the Tesla P100 during GTC 2016, but there was no word on the consumer side of things. I heard whispers during the show, but nothing concrete - what we do know, however, is that both NVIDIA and AMD will reveal their new video cards before Computex (which kicks off in the last few days of May). We chatted with Hardware Canucks during GTC 2016 about all things GPUs, something you can watch below.

It was only a few days ago that we reported that the next-gen NVIDIA GeForce video cards would feature GDDR5, but it looks like NVIDIA could tap the faster GDDR5X on its upcoming GeForce GTX 1080 and GTX 1070 (but I'm still adamant that they won't be called that).

There will be two GP104-based video cards, with the GTX 1080 featuring 8GB of GDDR5X on a 384-bit memory bus with 320GB of bandwidth. The GTX 1070 will use the same GP104 GPU, but drop down to GDDR5 with a 256-bit memory bus and 256GB/sec bandwidth. We should note that the current GM200-based GTX 980 Ti has 6GB of GDDR5 on a 384-bit memory bus with 334GB/sec - so the new GTX 1080 would have similar memory bandwidth, but with the new Pascal architecture, on the 16nm FinFET process.

Last month during the Game Developers Conference, AMD hosted its huge Capsaicin event, where it officially unveiled the new Radeon Pro Duo.

AMD's new Radeon Pro Duo is a dual-GPU video card with two Fiji GPUs, 8GB of HBM1 (4GB per GPU), four DisplayPort outputs, and 16 TFlops of performance. Well, it hits retail on April 26 with a not-so-bad price tag of $1500.

This isn't a video card for most people, as it'll be the glory card for all things VR - gaming, production and engineering with VR, as well as the insane enthusiasts (like myself) who just want the most powerful video cards they can get.

NVIDIA unveiled its impressive Tesla P100 video card at GTC 2016, powered by the new Pascal architecture. P100 is built from the impressive 16nm FinFET process, as well as the hugely transformative HBM2.

Well, what about the consumer side? Well, the GP104 will reportedly be unveiled later this month or early May, but new reports are pointing towards NVIDIA using GDDR5 on the next-gen GeForce video cards. We should expect 8GB of newer, denser, faster GDDR5 RAM. Both the GP104- and GP106-based solutions should feature GDDR5.

The newer GDDR5 should be clocked at 8Gbps, which will provide a decent upgrade over the 7Gbps on the current GDDR5 used in the GTX 980 Ti and Titan X. We will bring you up to date information on all next-gen video cards, so expect a bigger wave of GPU related news in the coming months.

Multiple sources around the world, as well as our own, are saying that NVIDIA will unveil their next-gen video cards in the coming weeks. With Computex right around the corner, this makes perfect sense.

We've already seen the company unveil its new Tesla P100 video card, powered by the new Pascal architecture, rocking HBM2 and the spiffy new 16nm FinFET process. The P100 features 15 billion transistors on the GPU itself, with another 135 billion transistors for the rest of the card - thanks to the incredible HBM2.

Most sites are still reporting that NVIDIA will call the new cards the GeForce GTX 1080 and GTX 1070, but I don't think we'll see this happen. I think we'll see NVIDIA move towards something like GeForce 10K, which sounds much better.