Memorial Day weekend is typically one of our favorites throughout the tech year here at PCMag, but not for the expectation of a quiet, lazy-hazy, long-needed reporting break. Far from it. It’s time to skip sleep and chug caffeine, and, in less fraught years, the joy of major jet lag as we fly off to the other side of the world from our NYC headquarters to Taipei.
The last Monday in May is the traditional press-event and major launch day of the Computex trade show, held (almost) every year in sprawling fashion across the Taiwanese capital. The large time difference versus US time, however, really means the action begins in earnest a little earlier in the weekend, and the key reveals often happen in the middle of the night for folks in the States.
This year, like 2020 did on a grand scale, has again put a crimp in Computex’s master plan. Last year’s in-person Computex show was first delayed to September 2020, then cancelled outright. Here in 2021, TAITRA, the Computex organizing body, professed early hope and plans that the in-person show would resume, but the fierce resurgence of COVID-19 in some geographies, and the attendant travel restrictions, prevailed. Taiwan has weathered COVID-19 well, but international travel remains a tricky prospect.
The show has been made all-virtual, and will run not for the traditional single first week in June but for the full month of June, officially from May 31 to June 30 on a virtual platform dubbed #COMPUTEXVirtual. The show will be front-loaded with the big stuff early on, though, with heavy hitters AMD, ARM, Intel, and Nvidia planning their big keynotes right at the start.
What emerges at Computex, typically, is what sets the tech table for years to come. Of course, the traditional show vibe is products and green-shoots of technology that will show up years down the road in retail products, so this show can be tougher than most to predict. But some shapes are starting to emerge from the mist that has cloaked PC tech due to more than a year, now, devoid of in-person demos and the onsite encounters that make trade shows like Computex so vital. We won’t be in Taipei, but here are the highlights of what we expect to see.
The World’s Worst-Kept Secret: New GeForce RTX Cards
First, you’d guess it based on tradition: Of course we should eventually see amped-up “Ti” versions of popular GeForce RTX 30-Series “Ampere” cards a year or so after their debut. That’s the way Nvidia rolls, after all. And sure, GeForce RTX cards have been harder to buy for the last six to nine months than Beanie Babies or Cabbage Patch dolls at their peaks. But upticked versions of Nvidia’s most rarified and sought-after cards still seemed inevitable. (After all, we got a GeForce RTX 3060 Ti at the very end of 2020.)
Then, it became a poorly kept secret: Images of ostensible GeForce RTX 3080 Ti cards from MSI on palettes emerged.
Then on May 26, Nvidia semi-spilled its own beans: a teaser video of what looks like liquid metal and the corner of a traditional GeForce Founders Edition card box (?) from the official GeForce Twitter account…
Note that several Nvidia Computex keynotes are on the docket. Our expectations around any cool consumer stuff zeroed in on the one dubbed “The Transformational Power of Accelerated Computing, from Gaming to the Enterprise Data Center.” (It is the first one, pending for June 1 at 1 a.m. ET in the US, and the afternoon of the same day in Taipei.) We’d bank on at least one card hitting the street: the much-exposed GeForce RTX 3080 Ti, based on all the rumors. We’d also expect it to be unobtanium soon after it launches, whenever that is. But maybe a high enough price will help it last a few minutes longer on sale at the outset than its RTX 30-Series siblings have!
Anything else? Fair game. An RTX 3070 Ti? Versions of existing RTX cards with different memory speeds or amounts? More “lite hash rate” (LHR) hash-limited cards to thwart GPU-snarfing cryptominers? We’ll have to see. We’d also expect plenty else from Nvidia on the subject of AI, automotive, and machine learning, but we think the big news for consumers will be cards.
AMD: Team Red, Dead Set on Redemption
As we and plenty of others have been reporting over the last couple of years, AMD has been on a torrid streak in terms of desktop and, more recently, laptop processors. The company is still an underdog in terms of sheer numbers and financials against rival Intel, but it’s far from one on the state of its compute architecture, and in momentum. The launch of “Zen 3” (third-generation) Ryzen desktop chips in late 2020 has been a stake-in-the-ground success, cementing an already strong position in desktop chips, and early 2021 saw AMD making inroads on mobile CPU performance and laptop design wins. (See our Ryzen “Renoir” vs. Core “Tiger Lake” initial face-off from the second half of 2020 and our first look at Zen 3 laptop performance from January of this year.)
The thing is, we’ve heard plenty of buzz around “Zen 4” possibly hitting this year, but Computex 2021 would seem to be too soon for anything much beyond broad outlines and preliminary chatter. CEO Dr. Lisa Su will take the keynote stage on Monday night US East Coast time (Tuesday morning in Taipei) and presumably trace out the coming months of AMD roadmap.
Will it be an update on Zen 4? We’d bank on at least some Zen 4 talk, but what seems most overdue, at least in terms of launch cycles, is an update to AMD’s desktop APUs (its lower-end chips with good integrated graphics). In these days of scarcity around graphics cards, even the older Ryzen 3000 series G-Series APUs are hard to find at reasonable prices, and a new generation would be manna in today’s GPU desert. That said, Dr. Su recently stated that the company has been prioritizing certain chip models over others given the late silicon shortage, with low-end chips often getting short shrift. So it’s possible APUs or, say, new Athlons may go by the silicon wayside.
Alternately, or in addition, what Dr. Su reveals could also be Radeon graphics-related. The long rumored “Big Navi” (high-end versions of AMD’s new-architecture graphics cards) indeed came to pass, and RDNA 2 has proven a worthy competitor to the latest GeForce cards, and perhaps a bit more accessible to buy. We tested cards like the Radeon RX 6800 XT late last year, and it would seem lower-end RX 6000-series offerings with RDNA 2 would be inevitable. Is a Radeon RX 6600 or 6500 desktop card, getting down into the $200 to $300 zone, in the offing? We wouldn’t bet against it. Whether it happens at Computex, though, is anyone’s guess. The same goes for overdue updates to AMD’s mobile Radeon graphics lineup. RDNA 2 has yet to hit AMD-based gaming laptops.
As for the high-end desktop market for CPUs (often dubbed “HEDT”), we also haven’t seen any action on the Ryzen Threadripper front in some time (or for that matter, anything much on Intel’s Core X-Series and its X299-chipset platform, apart from the occasional new motherboard). Might new Threadripper SKUs, a second generation on the TRX40 chipset, emerge? We’d put that as a longer shot given the relative lack of buzz.
Intel: Lots Already Launched, What Could Be New?
A few weeks back, Intel fired off its full magazine of its high-performance “Tiger Lake-H” power-user CPUs (see our performance first look here), which were first teased at CES 2021. Before that was the launch of its lukewarm 11th Generation Core “Rocket Lake-S” desktop CPU line, its second effort on the LGA1200 socket and what will likely be seen as a transitional or “caretaker” CPU family between now and the more evolved, architecturally forward-looking 12th Generation “Alder Lake” line, expected later this year.
And then there was the “Tiger Lake” 11th Generation mobile CPUs for thin laptops, which debuted in 2020’s second half. What could be left for Intel to spring on us at Computex?
The pump would seem likely dry on consumer silicon, but the chip giant never lacks for projects in the offing. As noted above, HEDT/Core X-Series has been quiet, so a move related to X299 and Core X would not surprise us.
And then there is the company’s discrete-graphics efforts, which have been teased for the last several years. The Xe DG1 card is projected to show up in some OEM systems in a low-key way, but is not expected to be a performance monster. Could the Xe DG2, the vaunted Intel debut PC-gaming card, finally see the light of day?
We’d at least expect a tease or two. Hopefully something beyond just card renders, as enticing as they have been.
Emerging Computing Platforms: ARM on Stage
In the days leading up to Computex, Qualcomm pushed out its vision for lower-end compute silicon (Snapdragon 7c Gen 2) and a development kit for ARM-compatible software. The 2019 Computex was the launch pad for the first Qualcomm Snapdragon 8cx laptop, which was 5G-equipped. ARM CEO Simon Segars, as well as Rene Haas of the company’ IP Products Group, will address the conference on June 2 with a talk on “Accelerating Ubiquitous Intelligence,” with a key theme being the insatiable appetite for compute power to tame workloads in AI and machine learning.
Another theme: the momentum toward emerging tech like ARM platforms taking an outsize role in the recovery from the 2020-21 pandemic, and in solving complex global problems and threats like climate change and cybersecurity. AI and machine learning workloads will be at the solution core of many of these big-scale issues, and the ARM brass will lay out the company’s vision for how its emerging architectures will contribute to the challenges of the 2020s.
Could it just be air? Perhaps, but we wouldn’t bet against it. Given the emergence of Apple’s M1 ARM-based SoC platform and the recent Qualcomm moves, plus the buzz around a possible Nvidia acquisition of ARM, could this be start of a Computex that, in subsequent years, will be divided on different lines: not AMD vs. Intel, or AMD vs. Nvidia, but traditional/legacy platforms vs. emerging ones? Or, on the flip side of that, maybe even the two finding more common ground?
Revved-Up SSDs: PCI Express 4.0, Thunderbolt 4 Gain More Momentum
The other core component apart from GPUs and CPUs that sees lots of action during the typical Computex is storage, specifically flash. With the launch of Rocket Lake-S earlier this year and the maturity of PCI Express 4.0 on AMD platforms for a couple of years now, we’d expect to see a mounting flurry of PCI Express 4.0 SSDs emerging, and this type of SSD moving closer and closer to the mainstream, in terms of volume and pricing, over the next year or two. The Computex 2021 show could be the first rotation of that particular downhill snowball.
In our testing, we haven’t seen a whole lot of real-world advantage to PCIe 4.0 vs. PCIe 3.0 for average users, except in cases of sustained large-file and large-folder transfers. But the emergence of PCIe 4.0 support on Intel Rocket Lake motherboards and in the very latest Intel mobile CPUs means that the “Express 4.0” lane is clear for a speed-up. It’s no longer just primarily an AMD-platform thing.
Likewise, with the emergence of Thunderbolt 4 ports on a slowly rising wave of cutting-edge PCs, we would not be surprised to see at Computex the first external Thunderbolt 4 devices taking advantage of the speedy, newly emerging port start to hit the public consciousness. (See our primer about Thunderbolt 4 and why it will increasingly matter in 2021.)
Stay tuned to PCMag.com and our Computex news feed for more on the show as it runs its course through the next month.
