NZXT has announced a pair of new motherboards designed for Intel’s 12th Gen Core processors: the N7 Z690 and N5 Z690. The N7 Z690 positions itself as the premium model of the pairing and comes with armor covering the vast majority of the PCB, Wi-Fi 6E, and support for memory speeds up to DDR4-4800. The N5 Z690 opts for a less aggressive approach at a more affordable price but still comes with Wi-Fi 6E and supports all of Intel’s Alder Lake Core processors.

Starting with the most premium of NZXT’s new motherboard pairing for Intel’s 12th Gen family of processors, the NZXT N7 Z690 follows a similar design to previous iterations of its N series models, including the N7 Z490 which we previously reviewed. As with previous N series models from NZXT, these are designed around its range of chassis, such as the H series models, for a seamless look and design. Perhaps the most prominent feature of the N7 Z690 is that it has armor plating covering practically all of the PCB for a cleaner and sleeker look.

There’s space underneath the sleek armor for up to three PCIe 4.0 x4 M.2 drives and four SATA ports for conventional HDDs and optical media drives. NZXT includes one full-length PCIe 5.0 x16 slot, with the second full-length slot operating at PCIe 4.0 x4 and the bottom full-length slot operating at PCIe 3.0 x4. Inbetween the full-length slots are two PCIe 3.0 x1 slots, while the N7 Z690 includes support for DDR4 memory at speeds up to DDR4-4800, with four slots supporting up to 128 GB in total. 

The NZXT N7 Z690 is available in both black and white and includes a mid-range feature controller set with a Realtek RTL8125BG 2.5 GbE controller, an Intel AX210 Wi-Fi 6E CNVi, and a Realtek ALC1220 HD audio codec.

The NZXT N5 Z690 is the more affordable of the pair and doesn’t include all of the armor of the N7; instead, opting for a more barebone look that users are accustomed to with entry-level motherboards. It does have the same networking configuration as the N7 (RTL8125BG 2.5 GbE and AX210 Wi-Fi 6E), but it uses a cheaper and lower grade Realtek ALC897 HD audio codec, as well as support for DDR4 memroy up to DDR4-5000 speeds.

It has the same PCIe configuration as the N7 with one full-length PCIe 5.0 x16, one full-length PCIe 4.0 x4, one full-length PCIe 3.0 x4, and two PCIe 3.0 x1 slots. The N5 Z690 even includes four PCIe 4.0 x2 M.2 slots which is one more than the more expensive N7 Z690 model.

The NZXT N7 Z690 is available in black or white and can be purchased now directly from NZXT for $300. Without all the armor and a lower grade HD audio codec, the NZXT N5 Z690 is also available in black or white and can be bought from NZXT directly for $240.

Source: NZXT

Read MoreAnandTech

At Computex 2022, the CEO of AMD, Dr. Lisa Sui, unveiled its Ryzen 7000 series of processors, as well as the associated AM5 platform. But while discussing specific details about its new platform for Zen 4 and beyond, AMD inadvertently ended up creating a conflux of confusion around the AM5 platform by quoting different power figures to different groups. Ultimately, at different points AMD was quoting 170 Watts as both the highest nominal TDP supported by the platform, as well as the Power Package Tracking (PPT) rating, which is the absolute highest amount of power a chip can draw under load. It goes without saying that these two claims shouldn’t both be right, and a correction was needed.

As first reported by the Tom’s Hardware crew, AMD has published a statement addressing the confusion, and proving the correct values. In short, the 170 Watt TDP was correct. Meanwhile the PPT value is actually 230 Watts – which at 1.35x the TDP rating, is typical for AMD’s Ryzen processors.

AMD’s full statement is below:

The overall increase in power specification figures for the AM5 platform was not unexpected – part of the benefit of the move to LGA sockets is additional pins for power delivery – but this finally settles the matter of just how much power AMD’s new socket and platform are designed to deliver. Motherboard vendors will no doubt go (well) past this on their high-end boards, of course, but 170W/230W will be the baseline for any motherboard that wants to officially support high-end AM5 chips.

CPU power consumption has been on the rise for the past several years, as we’re now well into the Dark Silicon era. While an individual CPU core still only draws a modest amount of power – on the order of 20W to 30W for a high-performance core – the total power requirement quickly balloons for high-end processors, which pack upwards of 16 cores. As a result, power delivery limits are typically the constraining factor for heavily multi-threaded workloads, as CPUs have to back down on clockspeeds in order to stay within their power envelopes. Increasing platform power limits, in turn, offers more headroom for keeping more cores clocked higher more often.

Though it should be noted that AMD’s clarifications today are for the AM5 socket, not the initial Ryzen 7000 series chips that will use it. AMD doesn’t necessarily have to tap into the full TDP of the socket right away – though for the aforementioned MT performance reasons, there’s good reason to. So officially, we still don’t know what the TDPs of the high-end Ryzen 7000 processors will be; but unofficially, it wouldn’t be surprising to see the top chips approach 170 Watts.

Read MoreAnandTech

It took the semiconductor industry over a decade to prep everything needed for production of chips using extreme ultraviolet (EUV) lithography. It looks like it is going to take a lot less to reach the next level — EUV with High-NA.

Higher Resolution Needed

Nowadays the most advanced chips are made on 5/4-nm-class process using EUV lithography ASML’s Twinscan NXE:3400C (and similar) systems that feature a 0.33 numerical aperture (NA) optics, which provides a 13 nm resolution. This resolution is good enough for a single-pattern approach at 7 nm/6 nm nodes with 36 nm ~ 38 nm pitches and at 5nm with 30 nm ~ 32 nm pitches. But as pitches get below 30 nm (at beyond 5 nm nodes) the 13 nm resolution might call for dual lithographic exposure that is going to be used for years to come. 

For post-3nm nodes, ASML and its partners are working on a brand-new EUV tool — the Twinscan EXE:5000-series — featuring a 0.55 NA (High-NA) lens capable of an 8nm resolution, which is projected to avoid multipatterning at 3 nm and beyond. The new High-NA scanners are still in development, they are expected to be extremely complex, very large, and expensive — each of them will cost over $400 million. High-NA will require not only new optics, but a new light source too, and even new fab buildings to accomodate the larger machines, which will require major investments.

But in a bid to keep scaling performance, power, area, and costs (PPAc) of semiconductors, leading makers of logic chips and memory devices are willing to adopt new technologies, and High-NA EUV scanners are crucially important for post 3-nm nodes. As a result, demand for High-NA tools is, well, pretty high. 

10 to 20 High-NA Systems to Be Delivered

Several weeks ago, ASML disclosed that it had received multiple orders in Q1 2022 for its High-NA Twinscan EXE:5200 systems (EUV 0.55 NA) from both logic and DRAM customers. Last week it clarified that it had five orders for pilot High-NA scanners due to be delivered in 2024 and ‘over five’ orders for subsequent models featuring higher productivity that will be delivered starting from 2025, reports Reuters.

Interestingly, back in 2020 ~ 2021, ASML said that it had has received High-NA commitments from three customers, for a total of up to 12 systems. Keeping in mind that logic makers are usually the first to adopt leading edge tools, it is safe to bet that Intel, Samsung Foundry, and TSMC committed in 2020 ~ 2021 to get pre-production High-NA scanners. Moreover, ASML has already started building the first High-NA system, which will be completed in 2023 and will be used by Imec and ASML customers for research and development purposes.

“On High-NA EUV, we are making good progress and have currently started the integration of the first High-NA system in our new cleanroom in Veldhoven,” said Peter Wennink, chief executive of ASML. “We received multiple orders for our EXE:5200 system in Q1. We also received additional EXE:5200 orders this month, April. With these bookings, we now have High-NA orders from three Logic and two Memory customers. The EXE:5200 is ASML’s next model High-NA system and will provide the next step for lithography performance and productivity.”

ASML’s Twinscan EXE:5200 is considerably more complex than regular Twinscan NXE:3400C machines, so it takes longer time to build these tools. The company hopes that it will be able to deliver as many as 20 High-NA systems in mid-term future, which probably means that its customers will have to compete for these machines.

“We are also discussing with our supply chain partners to secure a capacity of around 20 EUV 0.55NA systems in the medium term,” said Wennink. 

Intel First to Adopt Pre-Production Tools

So far, the only process technology confirmed to use ASML’s High-NA tools is Intel’s 18A node and that one was once scheduled to enter high-volume production in 2025, around the time when ASML starts to deliver its production High-NA EUV systems. But recently Intel pulled-in the start of18A production to the second half of 2024 and indicated that it could use ASML’s Twinscan NXE:3600D or NXE:3800E for its 18A manufacturing, presumably via multi-patterning.

While Intel’s 18A technology would greatly benefit from High-NA EUV tools, it looks like Intel does not necessarily need Twinscan EXE:5200 machines for this node. Usage of multi-patterning for commercial chips means a longer product cycle, lower productivity, higher risks, and potentially lower yields (though the latter is not cast in stone). Yet, it looks like Intel wants its 18A node to arrive as soon as possible, perhaps because it considers it a major tool that will allow it to recapture process technology leadership from TSMC. Consequently, Intel’s updated plans are now to phase in High-NA tooling during 18A’s lifecycle if the tools are completed on time.

Of course, it remains to be seen whether usage of 0.33 NA EUV scanners for 18A will offer enough productivity for Intel and customers of Intel Foundry Services. But, at least in 2024, Intel is not going to have any choice but to use machines that it has.

Other leading makers of semiconductors — TSMC, Samsung, SK Hynix, and Micron — will also inevitably adopt High-NA EUV for high-volume manufacturing of chips. The only question is when exactly this is when exactly this is set to happen. 

Read MoreAnandTech

Mediatek was one of the first vendors to demonstrate working 802.11be-compliant silicon under the Filogic lineup in January 2022. As part of the announcements at this week’s Computex, the company provided additional technical details along with part numbers. The Filogic 880 platform will service access points, routers, and gateways, while the Filogic 380 will be seen in the client devices.

Wi-Fi 7 / 802.11be Background

The 802.11 Working Group focused on extremely high throughput when starting work on 802.11be. This has been achieved primarily through a combination of three different aspects:

Support for up to 16 spatial streams
Support for channel widths up to 320MHz (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).

It must be noted that wider channels are available only in the 5 GHz and 6 GHz bands. 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.

Source: Mediatek Wi-Fi 7 Whitepaper

802.11ax introduced OFDMA to split the channel into multiple resource units (RUs), allowing multiple clients to receive data packets simultaneously. Wi-Fi 7 brings in a Multiple RU (MRU) feature that allows punctured RUs (at a 20 MHz granularity). Puncturing enables interference mitigation, and also increases the OFDMA efficiency.

Source: Mediatek Wi-Fi 7 Whitepaper

The benefits of MRU are brought out in the above picture – on top, puncturing allows improvement in spectral efficiency, and hence, data bandwidth. Below that, we see obvious latency benefits when multiple clients are simultaneously active.

Interference and co-existence with non-Wi-Fi users of the same spectrum is handled using automatic frequency coordination (AFC). While Broadcom’s pledge of support for the Open AFC initiative, and Qualcomm providing its own turnkey solution for its customers, Mediatek has only indicated that the Filogic 880 platform supports AFC, with no additional details.

Mediatek Filogic 880 Platform

Mediatek’s Filogic 880 platform services the Wi-Fi 7 access point and wireless router market. Similar to its competitor’s offerings, the platform supports all the major Wi-Fi 7 features – 320 MHz bandwidth in the 6GHz band, 4K-QAM, Multi-link operation (MLO), Automatic Frequency Coordination (AFC), and Multiple Resource Units.

The platform supports up to five bands with independent RFICs that communicate with the Filogic 680 Wi-Fi 7 baseband chip. While traditional routers may use only three bands, mesh systems and premium routers can make use of the support for two additional bands. At the heart of the reference design is a 1.8 GHz quad-core Cortex A73-based WiSoC. It also includes a network processing unit (NPU) with hardware-accelerated QoS and tunneling offload engines. The WiSoC doesn’t include an integrated switch, though. Designs are expected to adopt an external switch to translate one of the two 10Gbps USXGMII ports to either a single 10GbE, or a mix of NBASE-T and Gigabit Ethernet ports. Various other high-speed I/Os are also available in the WiSoC.

Unlike Qualcomm (whose announcement of the Wi-Fi 7 Networking Pro product suite was covered earlier this month), Mediatek is adopting an open approach in terms of making technical details of the platform available to the public. Broadcom also adopted a similar approach while introducing their product suite.

The Filogic 880 platform’s WiSoC is fabricated in a 6nm process. According to Mediatek, this delivers significant power efficiency compared to its competitors (without public knowledge of the fabrication process for Qualcomm’s and Broadcom’s offerings and/or public availability of their platforms, this claim is difficult to evaluate).

The MAC/baseband is implemented in the Filogic 680 part. Here, Mediatek opts for an interesting approach The 6GHz band also adopts a 4T5R (quad-transmit, penta-receive) antenna configuration to boost throughput and range, particularly for backhauls in mesh systems. The Filogic 680 contains the MAC and baseband logic for all three bands, and also includes MLO control. Mediatek contends that this approach is better than the scheme of discrete MAC/PHYs for each band between the WiSoC and the basebad/RFIC, and putting the MLO logic in the WiSoC – in terms of latency (claims of up to 100x improvement) as well as board design cost.

Mediatek’s reference design is a four-layer PCB, which should lead to a lower cost for its customers. Mediatek claims that the multi-chip MAC / PHY / baseband approach of its competitors results in higher cost due to more complex PCBs (six or eight layer PCBs). These aspects should be readily evident to end users when routers / APs based on platforms from different vendors come to the market.

Mediatek Filogic 380

On the client side, Mediatek is announcing the Filogic 380 combo solution with support for Wi-Fi 7 and Bluetooth 5.3. This combo single-chip solution is also built on a 6nm process. The 2×2 solution supports speeds up to 6.5Gbps with dual-band dual-concurrent operation. 320 Mhz bandwidth, 4K-QAM, MLO, and MRU are supported.

Mediatek also indicated support for an extra receive antenna to improve performance and range for Bluetooth 5.3 connectivity by enabling MRC (multiple receive combining). Bluetooth LE audio is also supported. The Filogic 380 also includes coexistence technology to ensure that BT and Wi-Fi can both operate in the 2.4GHz band without interference. Host connectivity is either via PCIe 4.0 x1 or USB 3.0.

Mediatek is currently demonstrating both platforms at Computex 2022. Based on Mediatek’s claims, it appears that routers and APs based on the Filogic 880 should hit very attractive price points without sacrificing anything in terms of performance or  features. Products based on the new Filogic platforms should start appearing in the market within a few quarters.

Read MoreAnandTech