The last few years have seen heightened consumer focus on wireless networking. The industry has also been quite busy, enabling the operation of Wi-Fi 6 (802.11ax) in the 6GHz band with Wi-Fi 6E. In parallel, the 802.11 Working Group had started work on 802.11be with a focus on extremely high throughput (EHT). Wi-Fi 7 is set to become the consumer-facing moniker for 802.11be.

802.11be aims to achieve high throughput primarily through a combination of three aspects:

• Support for up to 16 spatial streams
• Support for 320MHz-wide channels (with operation in 2.4 GHz, 5 GHz, and 6 GHz bands)
• Support for 4096-QAM (4K-QAM) resulting in better utilization of available spectrum (a faster modulation / coding scheme).

Theoretically, these aspects allow for up to around 46 Gbps of wireless throughput. 802.11be also aims to enable usage of Wi-Fi for real-time applications by including features for low-latency communications such as Multi-link operation (MLO). This allows a client and an access point to simultaneously communicate over multiple channels that might even belong to different bands. Interference and co-existence with non-Wi-Fi users of the same spectrum is handled using automatic frequency coordination (AFC).

A number of additional features targeting these two thrust areas are under consideration for the final standard. However, multiple silicon vendors have already started introducing silicon supporting variations of the aspects included in the first 802.11be draft specifications.

Mediatek was one of the first vendors to demonstrate working 802.11be-compliant silicon earlier this year. Though Mediatek indicated that the products would be marketed under the Filogic lineup, concrete technical details and part numbers were not announced. At MWC 2022, Qualcomm provided details of their 802.11be client silicon targeting mobile devices. The FastConnect 7800 is expected to become available in H2 2022, and integrates Bluetooth 5.3 support with key Wi-Fi 7 features.

Broadcom is announcing a comprehensive portfolio of products targeting various Wi-Fi 7 markets today.

The table below summarize the key features of the products targeting access points. All of these support 4K-QAM and Multi-Link Operation.

The AP products support AFC in order to ensure that 6GHz range is not affected (the APs are mandated to receive regularly scheduled clearance from a central authority to prevent interference with 6 GHz spectrum users). Broadcom has applied to be an AFC operator and will offer AFC service with their chips. This will use Open AFC code. Broadcom also believes a vibrant Wi-Fi 7 ecosystem requires an AFC service that the device makers can use independent of the chip vendor. Organizations such as the Wi-Fi Alliance, and Wireless Broadband Alliance have applied to be AFC operators, with Open AFC already gaining traction in this.

Tying these together in Broadcom’s Wi-Fi 7 router reference design is the BCM4916 network processor. This ARM v8-compatible quad-core SoC includes a dual-issue Runner packet processor (DI-XRDP), an integrated NBASE-T (up to 10G) Ethernet PHY for WAN or LAN, four integrated 1GbE PHYs, three USXGMII interfaces, multiple USB ports, and a 32-bit DDR3/DDR4 DRAM interface.

The quad-core CPU is a custom Broadcom design with 64KB of L1 cache and 1MB of L2 cache, providing up to 24K DMIPS horsepower. Without the knowledge of exact clock speeds, it is difficult to compare against the standard Cortex cores from ARM. Based on the DMIPS number, this appears to slot in-between a Cortex-A53 and an A57, but does not include the ARM v8.2 features of the A55.

Broadcom’s customers can play around with the radio configuration in the above design to create products at different price points. Tri-band support becomes compulsory in Wi-Fi 7. The extremely high throughput enabled by this means that 10G support on the WAN/LAN side and NBASE-T support on the LAN front will become the de-facto standard for routers and APs in the coming years.

Broadcom is also introducing the BCM4398 – an integrated Wi-Fi 7 and Bluetooth 5 combo chip targeting smartphones and other mobile applications. It supports two streams of Wi-Fi 7, channel width of up to 320 MHz, and a 6.05 Gbps PHY rate. On the client side, one of the key user-visible updates is low-latency operation. Broadcom claims that the BCM4398 can achieve sub-millisecond latencies for lightly congested environments in the 6 GHz band. The client multi-link operation feature can keep both uplink and downlink latencies between 5 and 10ms in situations where both 5 and 6 GHz bands are highly congested. This provides determinism for AR/VR knowing that worst case latencies will be down to a few milliseconds even during heavy congestion.

Broadcom is the first vendor to announce a complete portfolio of Wi-Fi 7 products, with sampling for key customers already in progress. The new products should give end-consumers a taste of the values delivered by Wi-Fi 7 for different applications a few quarters from now.

Gallery: Broadcom Launches Wi-Fi 7 Portfolio for Access Points and Client Devices

Source: AnandTech – Broadcom Launches Wi-Fi 7 Portfolio for Access Points and Client Devices

Back in January during their CES 2022 keynote, NVIDIA teased the GeForce RTX 3090 Ti, an even more powerful version of NVIDIA’s flagship card for the high-end gaming and content creation markets. At the time, NVIDIA told us to expect more information later in January, only for January (and February) to come and go without further mention of the card. But now, in the waning days of March, the GeForce RTX 3090 Ti’s day has come, as NVIDIA is launching their new flagship video card today.

Source: AnandTech – NVIDIA Releases GeForce RTX 3090 Ti: Ampere the All-Powerful

To bring digital-camera imaging quality to its new smartphones even in challenging captures like high-contrast, low-light, and motion, rather than look for or develop an alternative to established mobile-device CPUs, Global consumer electronics and mobile communications company OPPO designed the new MariSilicon X imaging NPU (Neural Processing Unit) chip.

MariSilicon X combines neural processing hardware with an ISP (“Image Signal Processor”) and a multi-tier memory subsystem into a dedicated component between the smartphone’s cameras and CPU. This lets MariSilicon X run machine learning and AI algorithms many times faster and with a fraction of the energy of previous approaches. The result: jaw-dropping computational photography improvements including superior night and low-light videos, with crisp detail and better color reproduction. OPPO is premiering MariSilicon X and the imaging benefits it brings in its brand-new Find X5 Series smartphones.

Source: AnandTech – Sponsored Post: OPPO’s MariSilicon X Imaging NPU Amps Up Night Video for New Find X5 Smartphones

With both GDC and GTC going on this week, this is a big time for GPUs of all sorts. And today, AMD wants to get in on the game as well, with the release of the PCIe version of their MI200 accelerator family, the MI210.

First unveiled alongside the MI250 and MI250X back in November, when AMD initially launched the Instinct MI200 family, the MI210 is the third and final member of AMD’s latest generation of GPU-based accelerators. Bringing the CDNA 2 architecture into a PCIe card, the MI210 is being aimed at customers who are after the MI200 family’s HPC and machine learning performance, but need it in a standardized form factor for mainstream servers. Overall, the MI200 is being launched widely today as part of AMD moving the entire MI200 product stack to general availability for OEM customers.

Starting with a look at the top-line specifications, the MI210 is an interesting variant to the existing MI250 accelerators. Whereas those two parts were based on a pair of Aldebaran (CDNA 2) dies in an MCM configuration on a single package, for MI210 AMD is paring everything back to a single die and related hardware. With MI250(X) requiring 560W in the OAM form factor, AMD essentially needed to halve the hardware anyhow to get things down to 300W for a PCIe card. So they’ve done so by ditching the second on-package die.

The net result is that the MI210 is essentially half of an MI250, both in regards to physical hardware and expected performance. The CNDA 2 Graphics Compute Die features the same 104 enabled CUs as on MI250, with the chip running at the same peak clockspeed of 1.7GHz. So workload scalability aside, the performance of the MI210 is for all practical purposes half of a MI250.

That halving goes for memory, as well. As MI250 paired 64GB of HBM2e memory with each GCD – for a total of 128GB of memory – MI210 brings that down to 64GB for the single GCD. AMD is using the same 3.2GHz HBM2e memory here, so the overall memory bandwidth for the chip is 1.6 TB/second.

In regards to performance, the use of a single Aldebaran die does make for some odd comparisons to AMD’s previous-generation PCIe card, the Radeon Instinct MI100. While clocked higher, the slightly reduced number of CUs relative to the MI100 means that for some workloads, the old accelerator is, at least on paper, a bit faster. In practice, MI210 has more memory and more memory bandwidth, so it should still have the performance edge the real world, but it’s going to be close. In workloads that can’t take advantage of CDNA 2’s architectural improvements, MI210 is not going to be a step up from MI100.

All of this underscores the overall similarity between the CDNA (1) and CDNA 2 architectures, and how developers need to make use of CDNA 2’s new features to get the most out of the hardware. Where CDNA 2 shines in comparison to CDNA (1) is with FP64 vector workloads, FP64 matrix workloads, and packed FP32 vector workloads. All three use cases benefit from AMD doubling the width of their ALUs to a full 64-bits wide, allowing FP64 operations to be processed at full speed. Meanwhile, when FP32 operations are packed together to completely fill the wider ALU, then they too can benefit from the new ALUs.

But, as we noted in our initial MI250 discussion, like all packed instruction formats, packed FP32 isn’t free. Developers and libraries need to be coded to take advantage of it; packed operands need to be adjacent and aligned to even registers. For software being written specifically for the architecture (e.g. Frontier), this is easily enough done, but more portable software will need updated to take this into account. And it’s for that reason that AMD wisely still advertises its FP32 vector performance at full rate (22.6 TFLOPS), rather than assuming the use of packed instructions.

The launch of the MI210 also marks the introduction of AMD’s improved matrix cores into a PCIe card. For CDNA 2, they’ve been expanded to allow full-speed FP64 matrix operation, bringing them up to the same 256 FLOPS rate as FP32 matrix operations, a 4x improvement over the old 64 FLOPS/clock/CU rate.

Moving on, the PCIe format MI210 also gets a trio of Infinity Fabric 3.0 links along the top of the card, just like the MI100. This allows an MI210 card to be linked up with one or three other cards, forming a 2 or 4-way cluster of cards. Meanwhile, backhaul to the CPU or any other PCIe devices is provided via a PCIe 4.0 x16 connection, which is being powered by one of the flexible IF links from the GCD.

As previously mentioned, the TDP for the MI210 is set at 300W, the same level as the MI100 and MI50 before it – and essentially the limit for a PCIe server card. Like most server accelerators, this is fully passive dual slot card design, relying on significant airflow from the server chassis to keep things cool.  The GPU itself is powered by a combination of the PCIe slot and an 8 pin, EPS12V connector at the rear of the card.

Otherwise, despite the change in form factors, AMD is going after much the same market with MI210 as they have MI250(X). Which is to say HPC users who specifically need a fast FP64 accelerator. Thanks to its heritage as a chip designed first and foremost for supercomputers (i.e. Frontier), the MI200 family currently stands alone in its FP64 vector and FP64 matrix performance, as rival GPUs have focused instead on improving performance at the lower precisions used in most industry/non-scientific workloads. Though even at lower precisions, the MI200 family is nothing to sneeze at with tis 1024 FLOPS-per-CU rate on FP16 and BF16 matrix operations.

Wrapping things up, MI210 is slated to become available today from AMD’s usual server partners, including ASUS, Dell, Supermicro, HPE, and Lenovo. Those vendors are now also offering servers based on AMD’s MI250(X) accelerators, so AMD’s more mainstream customers will have access to systems based on AMD’s full lineup of MI200 accelerators.

Source: AnandTech – AMD Releases Instinct MI210 Accelerator: CDNA 2 On a PCIe Card

Mushkin’s lineup of PCIe 4.0 SSDs has largely remained a Phison affair. The Delta series was based on the Phison E16 and the Gamma on the Phison E18. Recently, the company launched a new series of PCIe 4.0 SSDs – the Redline VORTEX. The key here seems to be the usage of a new SSD controller – the Innogrit Rainier IG5236. It appears to be taking over the flagship mantle from the Gamma – besting it in both read and write random access IOPS and also sequential read speeds. However, unlike the Delta and Gamma, which came to the market in 1TB, 2TB, and 4TB flavors, the Redline VORTEX series has three capacity points – 512GB, 1TB, and 2TB. Detailed specifications are provided in the table below.

<!–

–>

The SSD adopts a graphene heat dissipating label for its thermal solution – typical for the price point targeted. The performance numbers (aided by dynamic SLC caching) make it sitable for content creation and gaming – workloads that typically benefit from the capabilities provided by PCIe 4.0 SSDs.

Mushkin is not the first to market with the Innogrit Rainier controller. The Patriot Viper VP4300 series and the ADATA XPG GAMMIX S70 Blade were introduced late last year. While the Viper VP4300 is priced quite high, the Mushkin Redline VORTEX manages to undercut the XPG GAMMIX S70 blade by $12 at the 512GB capacity point and $5 at the 1TB point (based on current street pricing). The company is yet to announce availability and pricing for the 2TB SKU. The appearance of more affordable PCIe 4.0 NVMe SSDs in the market is good news for consumers.

Source: AnandTech – Mushkin Redline VORTEX PCIe 4.0 NVMe SSD Launched: Affordable Flagship

Alongside their spring driver update, AMD this morning is also unveiling the first nugget of information about the next generation of their FidelityFX Super Resolution (FSR) technology. Dubbed FSR 2.0, the next generation of AMD’s upscaling technology will be taking the logical leap into adding temporal data, giving FSR more data to work with, and thus improving its ability to generate details. And, while AMD is being coy with details for today’s early teaser, at a high level this technology should put AMD much closer to competing with NVIDIA’s temporal-based DLSS 2.0 upscaling technology, as well as Intel’s forthcoming XeSS upscaling tech.

Source: AnandTech – AMD Teases FSR 2.0: Temporal Upscaling Tech for Games Coming in Q2

AMD this morning is releasing the awaited spring update to their AMD Software Adrenalin Edition suite for their GPUs. First unveiled back in January as part of AMD’s keynote address, the spring update (22.3.1) introduces a few quality-of-life features for AMD’s software stack, and is being headlined by the first release of AMD’s FSR 1.0-based Radeon Super Resolution (RSR) technology for driver-based game upscaling.

Source: AnandTech – AMD Releases Adrenalin Software Spring 2022 Update: Adds RSR Upscaling and More

Earlier this month Intel quietly launched its W680 chipset, the company’s workstation-focused chipset for its 12th Gen Core (Alder Lake) processors. Unlike the current generation of consumer desktop chipsets such as Z690, H670, B660, and H610, the W680 adds the capability to use ECC DRAM, including both DDR5 and DDR4 variants. At present, there haven’t been many W680 motherboard announcements, although a couple of vendors, including ASRock Industrial and Supermicro have a few options listed. So we’re giving you the lowdown on W680, what it has to offer, and what technologies it brings for users looking to build a workstation-class desktop with Intel’s latest Alder Lake architecture.

Source: AnandTech – The Intel W680 Chipset Overview: Alder Lake Workstations Get ECC Memory and Overclocking Support