Learn more about AMD

The EPYC Rome server-grade CPUs have just been revealed to have 64 cores and 2.35 GHz clocks, but it looks like AMD is already working on the successor that will be based on the Zen 3 architecture. According to the U.S. Department of Energy, the upcoming CPUs will be known as the EPYC Milan, will have more than 64 cores, and will be featured in octa-CPU nodes along with Nvidia's 'Volta-Next" GPUs in the Perlmutter supercomputer.

Specs and pricing information for the upcoming AMD Radeon RX 590 based on the Polaris 30 architecture are now available. The Radeon RX 590 is said to have an MSRP of US$279. According to presentation slides accessed by Videocardz, the RX 590 will offer 12% better performance than the current RX 580. It is expected to be offered with a 'blockbuster launch bundle' with three free titles.

While both Zen 2 and the Zen 2 based Rome processor were officially announced by AMD at its Next Horizon event, little information was given on things like clock speed, IPC improvements, and other performance figures and specifications. At this year's Supercomputing conference, however, we got some details on Rome's clock speeds.

Apple has released new models of the 15-inch MacBook Pro with AMD Radeon Pro Vega GPUs. The new MacBook Pros can be configured with either a Radeon Pro Vega 16 or Radeon Pro Vega 20. The new GPU option is only available for the most expensive 15-inch Macbook Pro and will start at US $3,049.

Updated | Buried deep within the footnotes of AMD's announcement of Rome, it seems AMD is claiming an IPC improvement of about 29% when comparing Zen 2 to Zen 1. AMD did not disclose exactly how much more IPC Zen 2 had over Zen 1, but it seems 20% or more for at least some workloads can be expected.

Previous rumors were saying that Navi would only match the GTX 1080 GPUs from Nvidia, and, according to some new information, this is more like a half truth. It looks like AMD is planning to release the mid-range Navi 12 in the first half of 2019, with the Navi 10 and 20 high-end models coming in late 2019.

AMD's Radeon Instinct series consists of high-performance GPUs for applications such as deep learning, science, financial services and other extremely high-performance areas. As promised, its latest variants (the MI50 and MI60) come with 7-nanometer Vega architecture. This pushes their memory bandwidth up to 1024Gbps and their FP64 performance into the teraflops.

Images and the specifications of the XFX Radeon RX 590 Fatboy graphics card have been posted online. The card is obviously based on the upcoming RX 590 that features the Polaris 30 graphics processing unit based on the 12 nm FinFET process. The Fatboy variant will offer a boost clock of 1600 MHz and have a 256-bit memory bus.

While TSMC's new 7nm is already quite popular, being the basis for several processors from companies AMD, Apple, and Huawei, even more are announcing that they too will be fabricating their products at TSMC using their new process. Nvidia, Qualcomm, and more will be ordering chips from TSMC in 2019.

AMD hinted during its recently concluded Next Horizon Event that the Zen 4 microarchitecture is in the design phase. No further details were revealed but chances are that it will be using an advanced node for this architecture that could debut in 2021 or later.

AMD's Next Horizon event held yesterday was important for not only AMD's datacenter business but also for the company as a whole. AMD finally has revealed key information on Zen 2, Rome, the 7nm, new Vega GPUs, and what it means for AMD's business.

Pictures of the Sapphire Radeon RX 590 Nitro+ Special Edition have surfaced online. The exact specs of the card are still unavailable but we should be knowing more information close to the official announcement on November 15. The RX 590 is based on the new Polaris 30 architecture, which is being made in a 12nm process. Sapphire and other AIBs have retained the existing RX 580 designs for this refresh.

Even though AMD may look like it has lost focus of the GPU market, CEO Lisa Su reassures everyone that the red team is not giving up on the high-end gaming GPU market that easily. AMD is currently in the process of developing high-performing quality products, and the first cards are expected to hit the market in 2020.

The Threadripper 2970X and 2990WX flagship processors for this year have already been released. Now, their more 'mid-range' counterparts are available. They have been shown to compete well with some Intel HEDT and workstation CPUs in inital reviews.

The mid-range Radeon RX 500-series from AMD was preferred by crypto-miners, while gamers stuck to the GTX 1050/1060 solutions from Nvidia. Now that AMD took a serious hit with the crypto market decline, the red team is trying to regain some market shares in the mid-range gaming sector with the Radeon RX 590 solution that looks to be slightly faster than Nvidia's GTX 1060 (6 GB).

AMD is now feeling the full hit of the GPU-based crypto mining decline that started back in Q1. Instead of a projected US$50 million downturn in GPU sales, the red team reported losses of US$150 million, thus triggering a share drop of 24%. On the other hand, CPU sales are stronger than ever.

Inside sources claim that AMD's Navi GPUs are performing better than expected and could be launched in the first half of 2019 rather than the second. Previously rumored to be slower than the GTX 1080 from Nvidia, it looks like the red team managed to tune a few things and boost the overall performance. MSRPs are rumored to be in the US$250 zone.

The 10 nm process gave Intel quite a bit of a headache in the past few years. It was supposed to arrive in 2017 initially, but the fabrication process had to be delayed because of design problems, and it looks like Intel may not be able to launch the 10 nm CPUs come 2H 2019 in large amounts either.

The NVIDIA GeForce GTX 1060 is still competitive at the sub-US$400 price bracket. It was originally released with 6GB of GDDR5 memory. A change to its product page on its maker's website now suggests that there are GDDR5X versions of the card.

Intel may be implementing in-silicon solutions against Spectre and meltdown attacks, but not all threat variants can be identified this way. MIT's DAWG solution can detect all variants plus similar speculative attacks, and it can easily be integrated in operating systems without serious performance hits.

Windows 10 19H1, an update already being tested out by some users, is now expected to contain new adaptations to the Spectre-patch problem. These fixes for the more severe Spectre variant 2 vulnerability can reduce performance by up to 30%. The solution is known as the Retpoline kernel.

The Ryzen 2000 series still delivered decent performance gains over the original Ryzen models, even if the shrinking process was minimal. With the jump from 12 nm to 7 nm, the IPC boost is expected to be almost 5 times greater, at least in "scientific tasks." If multimedia and gaming loads get the same boost, this could be huge for AMD.

CineBench benchmarks show the quad-core Core i7-8706G to be just 9 percent behind the Core i5-8300H in the latest XPS 15 9570. This processor is currently not available on any mass market consumer laptop and can only be found on the Precision 5530 2-in-1 workstation for now.

An entry for the upcoming AMD Polaris 30-based Radeon RX 590 has been spotted on 3DMark by noted leakster TUM_APISAK. The new RX 590 is shown to sport higher clocks and is said to be using the same GDDR5 memory subsystem as the current Polaris 20 cards. Overall, the leaked benchmark shows conservative performance improvements over the RX 580.

A BIOS update is now available for Acer Swift 3 SF315-41 series laptops powered by mobile AMD Ryzen APUs that updates the AMD SMU version to 2.11. This removes the 256 MB VRAM limitation for the integrated Vega GPU and allows it access use 1 GB of system RAM as the video buffer. While the new option is indeed welcome, it hardly makes any difference when it comes to practical usage.

Following severe backlash over what has been described as "incompetent" benchmarking, Principled Technologies retested their AMD CPUs with creator mode. In their new results, the 2700X narrowed the gap between the 9900k down to about 15-20% on average, whereas previously the 9900k was sometimes 60% faster than the 2700X. Intel stated that these new results prove that the 9900k is still the "world's best gaming processor."

The industry is one month into a shortage of 14nm CPUs from Intel, and according to HP CEO Dion Weisler and Lenovo COO Gianfranco Lanci this will continue for another 6 months. They have recommended AMD as an alternative thanks to their "good performance" and "fantastic portfolio."

After scathing criticism coming from GamersNexus's Steve Burke and Hardware Unboxed's Steve Walton, Principled Technologies has made statements and defended its i9 9900k performance study. The company said "we are confident in our test methodology and results" following an interview conducted by GamersNexus.

The Radeon RX 570/580 successors are rumored to feature 12 nm Polaris 30 cores with increased frequencies, but same 8 GB GDDR5 VRAM. Performance-wise, these new models will most likely not be faster than the Vega 56/64 cards and will go head to head with Nvidia's GTX 1060/1070 chips.

When Intel commissioned Principled Technologies to benchmark its as of yet unreleased i9 9900k CPU, many voices in the tech press spoke out against such a practice, such as Steve Walton, saying it "undermines what we do." Now, more reliable figures have come out, and it seems like Principled Technologies did not run an unbiased test.

In our HP Spectre x360 review, we found that GPU performance was very poor in games, where it often throttled down to 400 MHz, less than half of its rated clock speed. Thankfully, there is a solution that should prevent such throttling.

About a month ago, Chiphell user Gtx9 leaked what he alleged was a Cinebench score of AMD's "Rome" server architecture. As Cinebench scores are easily manipulated, the weight behind Chiphell's leak was very weak. However, Gtx9 has returned with more convincing proof.

AMD has announced that its 12- and 24-core Ryzen 2 Threadripper chips will be available on October 29. The new chips are counterparts to the more expensive 16- and 32-core chips in the Threadripper lineup and will appeal to enthusiasts looking for a balance between performance and price.

AMD has a very favorable position on the CPU market, especially now that Intel is going through rougher times caused by shortages. The red team should be able to increase the market share with the release of its 7 nm Ryzen 3000, Threadripper 3000 and EPYC "Rome" CPUs, plus the Instinct GPUs in early 2019.

Sources claim that a few Ryzen 3000-series CPU samples have arrived at Radeon Technologies Group in order to tune the GPU drivers for the new interconnects. Tests have revealed that the upcoming CPUs are faster than the i7-8700K from Intel, but are very unstable and crash very often at the moment.

Google has partnered up with the French gaming behemoth Ubisoft to show off the American company’s new Project Stream service. The California-based technology giant is hoping its rather prosaically named offering can take media streaming to a whole new level by allowing gamers to enjoy live interaction with graphically challenging titles, such as the upcoming RPG Assassin’s Creed Odyssey.

While the 14nm CPU shortage did not initially affect US retailers, it has become clear that it finally has. Intel's flagship 8700k CPU has gone from $350 to $399 in under a month, which is nearly a 15% increase in price.

The global AI market is expected to grow at a high annual rate between now and 2023. It will probably be driven by the need to process more and more data as time goes on. The machine learning market is also expected to grow rapidly for the foreseeable future.

AMD is planning to launch a 7 nm gaming GPU codenamed Arcturus some time in 2020, but before that, the red team needs to come up with a viable alternative to Nvidia's new Turing GPUs first. Rumor has it the Navi GPUs scheduled for next year are still on the GTX 1000-series level, but, since the RTX 2000-series is overpriced for now, AMD might have enough time to fine-tune the Navi cards and even push for more aggressive prices.

Some missteps from Intel are projected to cost the company a big chunk of the CPU market. AMD's steady home runs with its Zen architecture, smooth fabrication advancements, and rising shares have been creeping up on Intel.

NVIDIA's target stock price has been downgraded to US$273, and has had its market value suffer, after Morgan Stanley took the relatively restrained improvements of the new GeForce RTX 2080 over older cards. However, a WCCFTech analyst has hit back, asserting that the financial giant does not grasp what the new Turing architecture brings to gaming.

Reports have claimed that Intel will be short of processors for the upcoming 2018 holiday season. They are based on an interview with a JP Morgan analyst, and on a report from a Chinese source claims that the company is reverting from 14nm to 22nm chip production processes. There may be something to them, as AMD's stock is now reported to be rising steadily.

GeForce RTX owners will soon have an objective 3DMark benchmark made specifically for testing their ray tracing capabilities. On the other end of the spectrum, a second entry-level benchmark will be coming next month for testing low-power DirectX 12 devices.