NVIDIA will soon launch its next-generation lineup of graphics cards based on a new and improved "Ampere" architecture. With the first Tesla server cards that are a part of the Ampere lineup going inside Indiana University Big Red 200 supercomputer, we now have some potential specifications and information about its compute performance. Thanks to the Twitter user dylan552p(@dylan522p), who did some math about the potential compute performance of the Ampere GPUs based on NextPlatform's report, we discovered that Ampere is potentially going to feature up to 18 TeraFLOPs of FP64 compute performance.

With Big Red 200 supercomputer being based on Cray's Shasta supercomputer building block, it is being deployed in two phases. The first phase is the deployment of 672 dual-socket nodes powered by AMD's EPYC 7742 "Rome" processors. These CPUs provide 3.15 PetaFLOPs of combined FP64 performance. With a total of 8 PetaFLOPs planned to be achieved by the Big Red 200, that leaves just a bit under 5 PetaFLOPs to be had using GPU+CPU enabled system. Considering the configuration of a node that contains one next-generation AMD "Milan" 64 core CPU, and four of NVIDIA's "Ampere" GPUs alongside it. If we take for a fact that Milan boosts FP64 performance by 25% compared to Rome, then the math shows that the 256 GPUs that will be delivered in the second phase of Big Red 200 deployment will feature up to 18 TeraFLOPs of FP64 compute performance. Even if "Milan" doubles the FP64 compute power of "Rome", there will be around 17.6 TeraFLOPs of FP64 performance for the GPU.

NVIDIA's next-generation of graphics cards codenamed Ampere is set to arrive sometime this year, presumably around GTC 2020 which takes place on March 22nd. Before the CEO of NVIDIA, Jensen Huang officially reveals the specifications of these new GPUs, we have the latest round of rumors coming our way. According to VideoCardz, which cites multiple sources, the die configurations of the upcoming GeForce RTX 3070 and RTX 3080 have been detailed. Using the latest 7 nm manufacturing process from Samsung, this generation of NVIDIA GPU offers a big improvement from the previous generation.

For starters the two dies which have appeared have codenames like GA103 and GA104, standing for RTX 3080 and RTX 3070 respectively. Perhaps the biggest surprise is the Streaming Multiprocessor (SM) count. The smaller GA104 die has as much as 48 SMs, resulting in 3072 CUDA cores, while the bigger, oddly named, GA103 die has as much as 60 SMs that result in 3840 CUDA cores in total. These improvements in SM count should result in a notable performance increase across the board. Alongside the increase in SM count, there is also a new memory bus width. The smaller GA104 die that should end up in RTX 3070 uses a 256-bit memory bus allowing for 8/16 GB of GDDR6 memory, while its bigger brother, the GA103, has a 320-bit wide bus that allows the card to be configured with either 10 or 20 GB of GDDR6 memory. In the images below you can check out the alleged diagrams for yourself and see if this looks fake or not, however, it is recommended to take this rumor with a grain of salt.

As we approach the release of NVIDIA's Ampere GPUs, which are rumored to launch in the second half of this year, more rumors and information about the upcoming graphics cards are appearing. Today, according to the latest report made by Taipei Times, NVIDIA's next-generation of graphics cards based on "Ampere" architecture is rumored to have as much as 50% performance uplift compared to the previous generations of Turing GPUs, while using having half the power consumption.

Built using Samsung's 7 nm manufacturing node, Ampere is poised to be the new king among all future GPUs. The rumored 50% performance increase is not impossible, due to features and improvements that the new 7 nm manufacturing node brings. If utilizing the density alone, NVIDIA can extract at least 50% extra performance that is due to the use of a smaller node. However, performance should increase even further because Ampere will bring new architecture as well. Combining a new manufacturing node and new microarchitecture, Ampere will reduce power consumption in half, making for a very efficient GPU solution. We still don't know if the performance will increase mostly for ray tracing applications, or will NVIDIA put the focus on general graphics performance.

NVIDIA has reportedly codenamed a future GPU architecture "Hopper," in honor of Grace Hopper, an eminent computer scientist who invented one of the first linkers, and programmed the Harvard Mark I computer that aided the American war efforts in World War II. This came to light as Twitter user "@kopite7kimi," who's had a fairly high hit-rate with NVIDIA info tweeted not just the codename, but also a key NVIDIA product design change. The tweets were later deleted, but not before 3DCenter.org reported on them. To begin with, "Hopper" is reportedly succeeding the upcoming "Ampere" architecture slated for the first half of 2020.

"Hopper" is also rumored to introduce MCM (multi-chip module) GPU packages, which are packages with multiple GPU dies. Currently, GPU MCMs are packages that have one GPU die surrounded by memory dies or stacks. This combination of GPU dies could make up "giant cores," at least in the higher end of the performance spectrum. NVIDIA reserves MCMs for only its most expensive Tesla family of compute accelerators, or Quadro professional graphics cards, and seldom offers client-segment GeForce products.

According to the sources over at Igor's Lab, NVIDIA could launch its next generation of GPUs, codenamed "Ampere", as soon as first half of the 2020 arrives. Having just recently launched GeForce RTX Super lineup, NVIDIA could surprise us again in the coming months with replacement for it's Turing lineup of graphics cards. Expected to directly replace high-end GPU models that are currently present, like GeForce RTX 2080 Ti and RTX 2080 Super, Ampere should bring many performance and technology advancements a new graphics card generation is usually associated with.

For starters, we could expect a notable die shrink to take place in form of 7 nm node, which will replace the aging 12 nm process that Turing is currently being built on. This alone should bring more than 50% increase in transistor density, resulting in much more performance and lower power consumption compared to previous generation. NVIDIA's foundry of choice is still unknown, however current speculations are predicting that Samsung will manufacture Ampere, possibly due to delivery issues that are taking place at TSMC. Architectural improvements should take place as well. Ray tracing is expected to persist and get enhanced with possibly more hardware allocated for it, along with better software to support the ray tracing ecosystem of applications.

There's just no quieting the rumor mill. It's like we're walking through a field that's made entirely of small pieces of stone that we inadvertently kick - and under every stone, another tidbit, another speculation, another pointer - a veritable breadcrumb trail that's getting more and more convoluted. Even as we were getting sort of decided in regards to NVIDIA's next-generation hardware and its nomenclature and model number - 1100 series - we now have two distinct sources and reports popping one right after the other that point to a 2000 series - and that also suggests Ampere might be in the cards for the next-gen product after all.

We've previously touched upon whether or not NVIDIA should launch their 1100 or 2000 series of graphics cards ahead of any new product from AMD. At the time, I wrote that I only saw benefits to that approach: earlier time to market -> satisfaction of upgrade itches and entrenchment as the only latest-gen manufacturer -> raised costs over lack of competition -> ability to respond by lowering prices after achieving a war-chest of profits. However, reports of a costly NVIDIA mistake in overestimating demand for its Pascal GPUs does lend some other shades to the whole equation.

Write-offs in inventory are costly (just ask Microsoft), and apparently, NVIDIA has found itself in a miscalculating demeanor: overestimating gamers' and miners' demand for their graphics cards. When it comes to gamers, NVIDIA's Pascal graphics cards have been available in the market for two years now - it's relatively safe to say that the majority of gamers who needed higher-performance graphics cards have already taken the plunge. As to miners, the cryptocurrency market contraction (and other factors) has led to a taper-out of graphics card demand for this particular workload. The result? NVIDIA's demand overestimation has led, according to Seeking Alpha, to a "top three" Taiwan OEM returning 300,000 GPUs to NVIDIA, and "aggressively" increased GDDR5 buying orders from the company, suggesting an excess stock of GPUs that need to be made into boards.

We the media and you enthusiasts are always getting scare jumps every time a high-profile launch is announced - or even hinted at. And few product launches are as enthusing as those of new, refined graphics cards architectures - the possibilities for extra performance, bang for buck improvements, mid-tier performance that belonged in last generation's halo products - it's all a mix of merriment and expectation - even if it sometimes tastes a little sour.

Adding to the previous breadcrumbs neatly laid-out regarding NVIDIA's Hot Chips presentation on a new "Next Generation mainstream GPU", the source for et another piece of bread that would make Grettel proud comes from Power Logic, a fan supplier for numerous AIB partners (company representative holding an EVGA graphics card below), who have recently said they expected "Q3 orders to be through the roof". Such an increase in demand usually means increased orders as AIB partners stock up on materials to produce a substantial enough stock for new product launches, and does fall in line with the NVIDIA Hot Chips presentation in August. Q3 starts in July, though, and while the supply-chain timings are unknown, it seems somewhat tight for a July product launch that coincides with the increased fan orders.

An official Gigabyte UK Notebook representative, who goes by the name of Atom80, over at the OverclockersUK forums has confirmed that NVIDIA's next-generation mobile GPUs will launch towards the end of this year. When asked about whether Gigabyte will be providing a GTX 1080 option for their Aero 15X V8-CF1 notebook, Atom80 stated that there are no plans to upgrade the Aorus notebook family until the next-generation GPUs are available. Since the mobile variants usually launch a few months after the desktop variants, it's possible that we're looking at a summer launch for the desktop models.

In what could bring cheers to PC gamers, and tears to miners, NVIDIA is reportedly wary of a possible drop in cryptocurrencies through 2018. This directly affects the company, since GPUs are used in mining various cryptocurrencies, which triggered inflation in prices of graphics cards from Q2-2017 to Q1-2018. Over the past couple of weeks, prices of popular high-end GPUs such as the GeForce GTX 1080 Ti have cooled, although not back to their original levels. NVIDIA's manufacturing division, which sub-contracts silicon fabrication to TSMC, is calculating the impact a cryptocurrency slump could have on its supply-chain, and are being conservative with their orders to the foundry. A drop in demand could leave the company with vast amounts of unsold inventories based on an old-generation architecture (Pascal, in the wake of Volta/Ampere), which could result in multi-billion-dollar inventory write-offs. According to a Digitimes report, NVIDIA has placed restrictions on its add-in card (AIC) partners on marketing cryptocurrency mining abilities of their graphics cards, and selling directly to large miners.

In addition to a slump in demand for cryptocurrencies, 2018 could see introduction of purpose-built crypto-mining ASICs that are tailored for popular cryptocurrencies. Purpose-built ASICs tend to be extremely economical for medium-thru-large scale miners, in comparison to GPUs. The third horseman is policy. While several governments around the world have developed an appreciation for blockchain technology for its resilience to tampering, fraud, and data-theft (which could be implemented in safekeeping government- and bank-records); governments are, understandably, anti-cryptocurrency, as it undermines sovereign legal tender issued by central banks, and aids tax-evasion. Several governments through 2017-18 have announced measures to crack down on cryptocurrency mining and use as tender. This has led to a further drop in public interest in cryptocurrencies, as large ICO investors are weary of losing money in a highly volatile market. Close to half the ICOs have failed.

It's being reported by Tom's Hardware, citing industry sources, that NVIDIA isn't looking to expand upon its graphics cards lineup at this years' GDC (Game Developers Conference) or GTC (GPU Technology Conference). Even as reports have been hitting the streets that pointed towards NVIDIA announcing (if not launching) their two new product architectures as early as next month, it now seems that won't be the case after all. As a reminder, the architectures we're writing about here are Turing, reportedly for crypto-mining applications, and Ampere, the expected GeForce architecture leapfrogging the current top of the line - and absent from regular consumer shores - Volta.

There's really not much that can be gleaned as of now from industry sources, though. It's clear no one has received any kind of information from NVIDIA when it comes to either of their expected architectures, which means an impending announcement isn't likely. At the same time, NVIDIA really has no interest in pulling the trigger on new products - demand is fine, and competition from AMD is low. As such, reports of a June or later announcement/release are outstandingly credible, as are reports that NVIDIA would put the brakes on a consumer version of Ampere, use it to replace Volta on the professional and server segment, and instead launch Volta - finally - on the consumer segment. This would allow the company to cache in on their Volta architecture, this time on consumer products, for a full generation longer, while innovating the market - of sorts. All scenarios are open right now; but one thing that seems clear is that there will be no announcements next month.

NVIDIA prepares to make its annual tech expo, the 2018 Graphics Technology Conference (GTC) action-packed. The company already surprised us with its next-generation "Volta" architecture based TITAN V graphics card priced at 3 grand; and is working to cash in on the crypto-currency wave and ease pressure on consumer graphics card inventories by designing highly optimized mining accelerators under the new Turing brand. There's now talk that NVIDIA could pole-vault launch of the "Volta" architecture for the consumer-space; by unveiling a GeForce graphics card based on its succeeding architecture, "Ampere."

The oldest reports of NVIDIA unveiling "Ampere" date back to November 2017. At the time it was expected that NVIDIA will only share some PR blurbs on some of the key features it brings to the table, or at best, unveil a specialized (non-gaming) silicon, such as a Drive or machine-learning chip. An Expreview report points at the possibility of a GeForce product, one that you can buy in your friendly neighborhood PC store and play games with. The "Ampere" based GPU will still be based on the 12 nanometer silicon fabrication process at TSMC, and is unlikely to be a big halo chip with exotic HBM stacks. Why NVIDIA chose to leapfrog is uncertain. GTC gets underway late-March.

NVIDIA has reportedly codenamed the GPU architecture that succeeds its upcoming "Volta" architecture after the 18th century French physicist who is one of the pioneers of electromagnetism, André-Marie Ampère, after whom the popular unit of measuring current is named. The new NVIDIA "Ampere" GPU architecture, which succeeds "Volta," will make its debut at the 2018 Graphics Technology Conference (GTC), hosted by NVIDIA. As with GPU architecture launches by the company in recent times, one can expect an unveiling of the architecture, followed by preliminary technical presentations by NVIDIA engineers, with actual products launching a little later, and consumer-grade GeForce product launching much later.

NVIDIA is yet to launch GeForce products based on its upcoming "Volta" architecture as its current "Pascal" architecture turns 18 months old in the consumer graphics space. Should NVIDIA continue on the four-digit model number scheme of its GeForce 10-series "Pascal" family, one can expect those based on "Volta" to follow the GeForce 20-series, and "Ampere" GeForce 30-series. NVIDIA is yet to disclose the defining features of the "Ampere" architecture. We'll probably have to wait until March 2018 to find out.