Premium Segment SoC Gets a Cortex-X CPU Core

A little more after the 2021/2022 product cycleinteresting 2023 turned out to be a much simpler year for the prolific SoC and cellular modem vendor than Qualcomm probably hoped. After releasing the first part of the Gen 2 family with his flagship class Snapdragon 8 Gen 2 earlier this year, the company is preparing to repeat the next step in its product stack with his Snapdragon 7+ Gen 2. Qualcomm’s traditional $400 to $600 “premium” market segment has focused on flagship-level features with more modest performance and cost for the Snapdragon 7+ Gen 2, giving the platform significant performance Aims to provide a boost.

Positioned as the successor to last year’s Snapdragon 7 Gen 1, this year’s iteration of Snapdragon 7 is broadly focused on improving performance rather than adding features. and new CPU and GPU architectures (especially CPU cores on the Armv9 architecture), but only a handful of new features this year. Instead, however, we have what Qualcomm is touting as his one of the biggest performance boosts for his Snapdragon 7 family. This was made possible in large part by Samsung’s very welcome pivot from its beleaguered 4nm process to TSMC’s 4nm process. This reflects the switch Qualcomm made to his well-received midcycle Snapdragon 8+ Gen 1 part last year.

New this year, Qualcomm has dropped a hint that this isn’t the only Snapdragon 7 Gen 2 part we’ll see this year in response to its decision to launch the first Gen 2 part as the 7+ instead of the 7. . Simply put, launching as a Snapdragon 7+ part leaves Qualcomm room to launch his vanilla Snapdragon 7 part later. Granted, Qualcomm hasn’t explicitly announced such a part right now, but unless they’re planning something below 7+, there’s little reason for him to launch 7+ in the first place. Otherwise, the always one-chip stack could have launched as his 7 part of the Snapdragon 7 Gen 1.

In terms of CPU configurations, the Snapdragon 7+ Gen 2 retains the same 1+3+4 CPU core configuration seen in the last few generations of the Snapdragon 7 family. The big news here is that Qualcomm has switched from using slightly higher clocked mid-cores to using a more performant CPU architecture, resulting in a significant performance boost for their top performing Prime cores.

So for the first time in a Snapdragon 7 part, Qualcomm is utilizing one of Arm’s Cortex-X cores for the Prime core. The Cortex-X2 used here is technically Arm’s previous generation design, so it won’t step on the toes of the Snapdragon 8 Gen 2 and its Cortex-X3 cores. Compared to the A710 core used for the core (and the mid-core for 7+ Gen 2), the Cortex-X2 shows significant improvements in both IPC and clock speeds. As a result, the Prime core’s peak clock speed goes from 2.4GHz to 2.91GHz, further synergizing with his IPC gains on the more complex cores.

Overall, Qualcomm touts the 7+ Gen 2’s CPU performance as “up to” 50% better than the 7 Gen 1. Virtually all of this comes from the new Prime cores.

The trade-off is that there is only one Cortex-X2 core, so such a significant performance boost is really only available for single-threaded workloads. The three mid (performance) cores are Cortex-A710 based again and are 2% higher than before. As such, Gen 1 beyond Gen 7 does not provide significant benefits for multithreaded workloads. The improved power efficiency of TSMC’s 4nm process should have some gains there, but part of those gains is to make that power-hungry Cortex-X2 viable in terms of battery life. is invested in

The 7+ Gen 2, on the other hand, also incorporates a faster Adreno GPU. As Qualcomm’s integrated GPUs have been for several generations, the company hasn’t assigned a product number, limiting the amount of details it can share, let alone disclose important architectural details. Based on that, this doesn’t look like it’s using the new 8th Gen 2 GPU architecture. So it looks like Qualcomm has included larger versions of their existing GPUs and almost certainly boosted their clock speeds.

Either way, GPU performance expectations are high for the new SoC, with Qualcomm boasting a significant 2x performance improvement over the 7th Gen 1. This platform offered just 20% more performance than his predecessor. These are not his SoCs in the flagship class, but Qualcomm wants to position the Snapdragon 7 series as suitable for gaming smartphones, especially in China.

Overall, Qualcomm touts a 13% improvement in power efficiency over the 7th Gen 1, at least on a “long daily use” basis. TSMC’s switch to the 4nm process should yield significant gains, as evidenced by some of last year’s Gen 8 and above Gen 1, but at the same time, Qualcomm will likely take a sizable chunk of those gains. It’s clear that we’re investing in improving overall performance.

Feeding the dragon is a 32-bit (dual 16-bit) LPDDR5 memory controller. Unlike the Snapdragon 8 Gen 2, the 7+ Gen 2 does not receive support for faster LPDDR5X memory. In other words, the Snapdragon 7 family remains as-is. In this case, this means support for memory speeds up to his LPDDR5-6400, equivalent to 25.6GB/s of memory bandwidth. Qualcomm’s cache and memory subsystems will be under more pressure to keep the various processing blocks fed, in contrast to the massive performance gains in CPUs and GPUs.

Speaking of which, it wasn’t just the CPU and GPU blocks that saw a big performance boost. Qualcomm’s Hexagon DSP/AI engine block has also received significant performance tweaks, matching his 2x increase in GPUs. Qualcomm skimped on the technical details here, but in our briefing we talked about features like INT4 and micro-tiling (his two key features of the next-generation Hexagon block in 8th Gen 2). was not mentioned. is a greatly enhanced version of his Hexagon block used in the previous 7 Gen 1.

However, one of the Snapdragon 8 technologies moving to Snapdragon 7 is the triple 18-bit Spectra ISP. The 7+ Gen 2’s 18-bit unit replaces the 14-bit unit found on the previous generation platform and supports HDR video capture with triple exposure calculations and low-light photography, which Qualcomm calls Mega Low. improve. light function. The net result is that with its zero shutter lag feature, the 7+ Gen 2 can capture higher resolutions, and combined with its updated GPU, it can now record 4K video at up to 60 fps, making the 7 Gen Doubles the 4K30 limit of 1.

Finally, rounding out the package is a reissue of Qualcomm’s Snapdragon X62 integrated modem. Similar to last year’s SoC, this is a mmWave + Sub-6 Release 16 design capable of achieving a theoretical maximum download speed of 4.4Gbps. However, this year’s design has a twist. Dual SIM Dual Active (DSDA) support. This is a first for the Snapdragon 7 platform. Both active radios on the 7+ Gen 2 will support his 5G and 4G communications, so dual SIM users can basically use whatever network they like on either radio. This is another premium feature that was previously limited to his Qualcomm’s Snapdragon 8 platform.

As for non-cellular connectivity, the 7+ Gen 2 uses the FastConnect 6900 radio system. This is a relatively modest update to the previous 6700 radio, bumping Bluetooth support to version 5.3 of the protocol and boosting the peak bandwidth of the 2×2 stream Wi-Fi 6E radio to 3.6Gbps thanks to Dual Band Simultaneous (DBS) support. is increasing to .

In summary, the Snapdragon 7+ Gen 2 hits the market very quickly. According to Qualcomm, handsets using the SoC will be available later this month, and among OEMs Redmi and Realme will release phones based on the new chips.