01:12 | Update on GN Fan Testing
We talk about our new case fan testing equipment briefly at this timestamp. Will leave that one to the video.
06:39 | Windows 11 Potentially Inbound
As we’ve mentioned before, Microsoft is brewing a substantial update to Windows 10 under the code name Sun Valley, and it seems the company is finally ready to take the wraps off of what that update entails. Microsoft has announced an event for June 24, where it will detail “the next generation of Windows.”
We know that, to some extent, Sun Valley is a substantial overhaul of the desktop and user interface. Among a new Start menu and Taskbar, we’ve already seen refreshed icons for Windows File Explorer. There’s reportedly also going to be other UI changes, and new fluid animations, among other things.
Beyond the UI, Microsoft is also reportedly set to completely overhaul the Microsoft Store to allow third party apps and commerce inside the store, as well as adjusting the revenue cut it takes on games and software.
These changes may also be enough to prompt Microsoft to trot out a new numbered version of Windows, presumably Windows 11. Teases from Microsoft continue to allude to the number 11, as can be seen in a tweet from Microsoft (note the light shining through the logo looks like an 11). Some have also pointed to the start time of the event: 11 AM ET. Moreover, Microsoft’s Yusuf Mehdi tweeted that he hasn’t “been this excited for a new version of Windows since Windows 95!” Obviously, the keyword there is “version,” and it’s the first time we believe anyone at Microsoft has mentioned a new version as opposed to an update.
At this point, we won’t have to wait long to see what Microsoft has been working on for Windows.
09:19 | Atari Is Launching Its Atari VCS to the Public
Atari is preparing a public retail launch for its Atari VCS console, which was long-delayed but did finally ship to early backers. The price tag, unfortunately, is something of an ask.
The Atari VCS Onyx Base is set at $300, and an “All-in” bundle -- which bundles the console with a wireless controller and wireless joystick -- will go for $400. Despite our criticism of Atari, we did say in our coverage that the Atari VCS console is an interesting design and potentially offered value to early backers who got in at a lower price -- especially if close to $200. Although the hardware itself isn’t exactly competitive. The console is driven by an embedded Ryzen R1606G APU, which is a dual-core, Zen-based part that uses Radeon Vega 3 graphics, and is based on GlobalFoundry’s 14nm process. Other specs include 8GB of DDR4-2400, 32GB of onboard eMCC storage, and an M.2 slot.
At that price, you have a retro-inspired console competing with the Xbox Series S and the PS5 Digital Edition, and we likely don’t need to remind you what hardware those consoles have. Further, you can play many of the Atari VCS’ games catalog legally and for free on Archive.org. Atari is attempting to justify these prices by billing the Atari VCS as a “PC/console hybrid” that can double as a PC if you install Windows. You can color us skeptical about how many users would actually use the machine in such a way.
At any rate, According to Atari, those who weren’t early backers in the crowdfunding campaign can have the opportunity to buy the VCS on June 15th. The console will be available from Atari directly, or retailers like Best Buy, Micro Center, or GameStop.
12:43 | Dell Hit with False Marketing Class Action Complaint
If you’ve been keeping up with our YouTube channel and Twitter, you’ve seen that we’ve detailed our own frustrations with Dell in recent months. These frustrations range from scummy business tactics, to exceedingly low-quality pre-built gaming PCs. Dell’s pre-built G5 5000 isn’t worth considering even if all you wanted was the GPU.
All that said, we now bring you news of consumer frustrations with Dell, in case you got tired of hearing ours. We received an email on behalf of Class Action Attorneys and Whistleblower Brian Mahany, David Kani, and Steve Hochfelsen detailing a new Class Action filed against Dell regarding False Marketing. The email title makes a strong point to accuse Dell of “cheating gamers.”
Robert Felter, who lives in California, is filing a class action lawsuit against Dell. The lawsuit is based on Dell’s Alienware Area-51M R1 gaming laptop, whereby the lawsuit alleged that Dell "intentionally misled and deceived" buyers over the supposed upgradeable components, like the CPU and GPU.
The crux of the issue is Dell’s marketing around an upgradeable laptop. When the Alienware Area-51M R1 was announced and launched in 2019, its marketing was heavily tinged with upgradability as the central feature. According to the lawsuit, Dell never clearly defined the limitations of that upgradability -- at least not before users bought the R1 model.
It seems Dell didn’t clarify how upgradable these machines were until it launched the second model, the Alienware Area-51M R2. This is where the lawsuit starts to gain traction. With the release of the R2, Dell stated that the R1’s upgrade path for CPUs and GPUs would be limited to models within that same generation -- that is, Intel’s 9th-gen CPUs and Nvidia’s RTX 20-series. The newer R9 was much the same, although its CPUs would be locked to Intel’s 10th-gen line. The problem is that systems shipping with the flagship 9000-series CPUs would clearly have no upgrade path at all, being unable to socket 10-series and being devoid of superior 9-series CPUs. It is, therefore, not upgradeable. It is, however, downgradeable, but that’s not really a selling point.
The GPUs were perhaps the most limited, as users could only upgrade the GPU through GPU upgrade kits that used Dell’s proprietary Dell Graphics Form Factor (DGFF) modules. The lawsuit points-out that Dell’s initial marketing and documentation never mentioned this, and misled users into believing that the CPU and GPU were upgradable beyond the current generation, akin to desktops. In fact, Dell’s marketing called this a “mobile desktop,” which the Complaint says “further cements its alleged material capability that the Area 51M is upgradeable in the same way a desktop is upgradeable.”
This apparently became a problem when owners of the original R1 wanted to upgrade their Intel 9th-gen CPU to one of Intel’s 10th-Gen CPUs.
Anyone buying the highest-end model would not be able to upgrade. The easy way out would have been for Dell to only ship this with a 9700K, but that wasn’t the case.
Furthermore, according to the filing:
"After his purchase, Plaintiff later inquired about upgrading his Area 51M’s Core Components but learned that, contrary to Dell’s repeated promises that the Area 51M is upgradable, none of the Area 51M’s Core Components, were in fact, upgradeable in any way. The Area 51M's CPU was not upgradeable to the new INTEL 10th generation CPU, nor was its GPU upgradeable to the new NVIDIA RTX SUPER 2000 series. In fact, the only way Plaintiff could own a laptop with these newly released upgraded Core Components was to spend several thousand dollars more than what an upgrade would cost to purchase the then-newly released Alienware Area 51M R2 or a similarly equipped laptop from another manufacturer."
In short, owners of the R1 would have to buy the R2 to get any meaningful upgrades. And even then, users were still locked into Nvidia’s RTX 20-series line, with the only “new” upgrades for the R2 being the RTX 2080 Super and RTX 2070 Super. And again, they would be forced to buy Dell’s DGFF-based upgrade kits, which were arguably overpriced.
The lawsuit is seeking financial relief, including damages and legal fees, for not only the plaintiff, but for consumers who have purchased the Alienware Area-51M R1 in the following states: Alaska, Arizona, California, Hawaii, Idaho, Montana, Nevada, Oregon, and Washington.
Source: Email to GN via Elite Lawyer Management
18:12 | Dell Kills Its Alienware External GPU Enclosure
Also in Dell news, as confirmed by The Verge, Dell has scrapped its Alienware Graphics Amplifier external GPU enclosure. This news comes on the heels of Dell trotting out the new Alienware X-series of gaming notebooks, with the X15 and X17 models not featuring the proprietary Alienware Graphics Amplifier port. A little digging also shows that the port is also missing from Alienware’s M15 R5 and M15 R6.
In confirming the discontinuance of the port and enclosure to The Verge, Dell didn’t offer any details on why. Reading between the lines a bit, the most obvious reason is the rise of the Thunderbolt interface, and indeed, Aliwenware’s new models come with a Thunderbolt 4 port. In recent years, most eGPUs have relied on the Thunderbolt interface anyway.
The Alienware Graphics Amplifier port and cable were both proprietary, and used a mixture of PCIe and USB signals, and it could be that in getting with the times, Dell is planning a new eGPU enclosure based on Thunderbolt. Or, it could be that the company is exiting the market entirely, as it’s already a fairly niche segment. The Verge’s report notes that the support page for the product was last updated in April of 2021, though support for the latest AMD and Nvidia graphics cards was not added.
19:29 | Steam Hardware Survey for May 2021: AMD’s CPU Share Rises, GPU Share Falls
Steam’s Hardware & Software Survey for May 2021 is out, and it continues to highlight two key trends we’ve been pointing out: The continued rise of AMD’s Ryzen CPU adoption, offset by its recent decline in GPU adoption among Steam users.
The biggest win for AMD is finally crossing the 30% share mark on the Steam CPU charts, as AMD moved from 29.48% in April 2021 to 30.13% in May 2021. This is a trend that has been building for several months now, and has only been compounded by the popularity of AMD’s Ryzen 5000 CPUs.
However, that win comes with a stinging tradeoff, in that AMD is actually losing share in the GPU charts, having tumbled down a bit to a 16.18% share in May. Also, May 2021 marked another month where not a single RX 6000-series card emerged on the Steam charts. We pointed out that the RTX 3060 recently (and finally) made its debut toward the bottom of the charts. This month, that card has notched a 0.10% gain in share, moving up several spots.
Elsewhere, the data in the survey is much the same month-over-month. For instance, the GTX 1060 (8.95%), despite continually shrinking share, is still the most popular GPU among Steam users. And the margin is still surprisingly wide; the closest card is the GTX 1050 Ti, with a 6.71% share. Quad-core CPUs are still king (40.38%), but 8-core CPUs continue to be the fastest growing among Steam users at 13.25%, marking a 0.38 point increase.
21:11 | Nvidia’s DLSS Is Coming to Linux
On top of the RTX 3080 Ti and 3070 Ti announcements, Nvidia also announced expanded DLSS support -- namely that DLSS is coming to Linux. Nvidia’s DLSS will be delivered through Steam’s Proton, a compatibility layer based on Wine, which Linux users have long relied on to play Windows games on Linux distros.
Previously, Nvidia’s DLSS had two exclusive requirements: An RTX GPU and a Windows OS. It appears as if Nvidia is starting to take Linux gaming a bit more seriously, which also aligns with Valve’s previous declaration to better support Linux. Nvidia claims that DLSS support for Vulcan titles is coming later this month, and that support for DirectX games will come this fall.
22:07 | TSMC Technology Symposium 2021 Recap
TSMC recently held its online-only TSMC Technology Symposium 2021, which took place from June 1-2. TSMC’s Technology Symposium serves as a way to highlight new and emerging offerings from TSMC, as well as offering a look at the company’s future roadmaps and the ongoing status of current technologies and process nodes.
Starting with 7nm, TSMC’s 7nm family is seeing a stall in overall adoption as TSMC’s customers migrate to 5nm. To that end, TSMC is only forecasting a 14% capacity increase for 7nm throughout 2021, and that capacity progression will likely wane in the years ahead as TSMC’s other nodes become more mature. We saw a similar trend when 16nm peaked and customers moved onto 7nm.
TSMC’s N6 process, a derivative of its N7 process, will continue to be a very important service for the company. To highlight this, TSMC notes that while in Q4’2020, N6 only accounted for 15% of total N7/7nm output, but that N6 will account for as much as 50% in Q4’2021. Also, and perhaps more interestingly, TSMC states that its N6 D0 (defect density metric) is already achieving the same yields as N7. For TSMC’s 5nm N5, TSMC expects capacity to quadruple by 2023.
Additionally, contributions from TSMC’s new Arizona plant, which it announced it had officially broken ground on, would be on top of that expanded capacity. TSMC’s new Arizona fab isn’t expected to come online until 2024, and will add an additional 20k wafers per month of capacity.
TSMC notes that N5 has already produced better yields than any TSMC 7nm node, and TSMC highlights the production learning curve and the defect density improvements it made with N6/N7 as a reason for this. Of course, TSMC also points to the increased use of EUV on more layers with N5, and TSMC also shared that it accounts for some 65% of cumulative EUV wafers shipped, as well as representing 50% of the EUV lithography machine install base.
TSMC also attributes much of its success with EUV HVM (High Volume Manufacturing) to its use of in-house designed EUV pellicles. Not long ago, we reported that ASML and Mitsui Chemicals had reached a deal for Mitsui to assemble and distribute ASML-designed EUV pellicles. However, as TSMC has been using EUV for some time now (since 2019), it has had to rely on its own pellicle design, as other designs weren’t anywhere near ready. This is a very important and interesting point, and it may offer some insight into how TSMC has forged so far ahead of competing foundries in terms of leading edge nodes. TSMC has seemingly mastered the once arcane EUV process while other foundries have continued to struggle with it, and its custom EUV pellicles are likely a big reason for that.
Moving on, TSMC’s upcoming N4 is an optical shrink of its N5 node, and TSMC notes that it is an enhanced N5 technology. N4 will offer a 6% smaller die area, as well as the usual iterative power/performance improvements. N4 is already producing the same yields as N5, and risk production is scheduled for Q3 '2021.
There was also a brief N3 update, where TSMC reitiered its sticking to its FinFET design (rather than GAAFETs) with volume production set for the back half of 2022. TSMC also introduced two new process nodes: N6RF and N5A. N6RF is an augmented N6 process with optimizations for 5G and WiFi 6/6E silicon, while TSMC’s N5A is aimed at automotive silicon.
25:25 | AMD Confirms Samsung SoC with RDNA Inbound
As AMD’s Computex 2021 keynote was packed pretty densely with hardware and technology updates, this one seemed to kind of fly under the radar. We’ve been reporting on AMD and Samsung’s graphics partnership for some time now, going way back to 2019 when it was first announced. In the years since, there’s been little in the way of updates, other than Samsung confirming that a Exynos SoC with Radeon Silicon was indeed on the way. That SoC is largely believed to be the successor to the in-market Exynos 2100 SoC.
All that said, at Computex 2021, Dr. Lisa Su delivered a brief update on the topic:
“The next place you’ll find RDNA 2 will be the high-performance mobile phone market. AMD has partnered with industry leader Samsung to accelerate graphics innovation in the mobile market, and we’re happy to announce we will bring custom graphics IP to Samsung’s next flagship SoC, with ray-tracing and variable rate shading capabilities. We’re really looking forward to Samsung providing more details later this year,” said Su.
The highlights here are that Su mentioned that the custom graphics IP will be based on RDNA 2, and will bring AMD’s variable rate shading and ray tracing technology to Samsung’s Exynos silicon. There are technically other mobile GPU solutions that use variable rate shading, but this would be the first mobile GPU implementation of ray tracing. For the record, Imagination Technologies has been sitting on ray tracing capable IP for years, such as its GR6500 GPU and C-Series architecture. However, as far as we’re aware, none of that IP has ever come to market -- at least not in a consumer device.
It’s likely we'll know more about this in the back half of the year, once Samsung gears up for its upcoming Galaxy S22 devices.
Editorial, Research: Eric Hamilton
Host: Steve Burke
Video: Keegan Gallick