Affiliate links on Android Authority may earn us a commission. Learn more.
Exclusive: Here's what Qualcomm didn't tell you about the Snapdragon X series
After years of struggling to compete in the PC space, Qualcomm finally has a platform that should be able to compete. The Snapdragon X platform is built on its custom Oryon CPU architecture, which it claims is powerful enough to match — and possibly even beat — Apple, Intel, and AMD.
Qualcomm has shared quite a bit about the performance of its upcoming chips and recently detailed more information about the exact models to be available — the X Elite and X Plus. Thanks to a source familiar with the matter, we can share a little more than Qualcomm let slip by previewing the additional specs for these upcoming chips.
The complete Snapdragon X lineup
Qualcomm has split its Snapdragon X lineup into two product lines this generation: X1 Elite for 12-core variants and X1 Plus for 10 and 8-core chips. The lineup Qualcomm is currently offering to manufacturers consists of five parts: three under the X1 Elite name and two under X1 Plus. However, Qualcomm only detailed four of these chips during its recent announcement, missing out the lowest-end X1 Plus (X1P-42-100).
Additionally, all X1 chips except the lowest-end X1 Plus are the same exact silicon die, just locked down in different ways. This can be easily noticed by looking at the specs table, where the chips sharing the same die only differ in CPU and GPU specs, while the X1-P42-100 also sacrifices four PCIe 4.0 lanes and half of the video hardware.
X1 Elite (X1E-84-100) | X1 Elite (X1E-80-100) | X1 Elite (X1E-78-100) | X1 Plus (X1P-64-100) | X1 Plus (X1P-42-100) | |
---|---|---|---|---|---|
Performance cores | X1 Elite (X1E-84-100) 8x Oryon - 3.8GHz | X1 Elite (X1E-80-100) 8x Oryon - 3.4GHz | X1 Elite (X1E-78-100) 8x Oryon - 3.4GHz | X1 Plus (X1P-64-100) 6x Oryon - 3.4GHz | X1 Plus (X1P-42-100) 8x Oryon - ? |
Efficiency cores | X1 Elite (X1E-84-100) 4x Oryon - 3.8GHz | X1 Elite (X1E-80-100) 4x Oryon - 3.4GHz | X1 Elite (X1E-78-100) 4x Oryon - 3.4GHz | X1 Plus (X1P-64-100) 4x Oryon - 3.4GHz | X1 Plus (X1P-42-100) - |
Turbo | X1 Elite (X1E-84-100) 4.2GHz up to two cores | X1 Elite (X1E-80-100) 4.0GHz up to two cores | X1 Elite (X1E-78-100) - | X1 Plus (X1P-64-100) - | X1 Plus (X1P-42-100) - |
GPU frequency | X1 Elite (X1E-84-100) 1.5GHz | X1 Elite (X1E-80-100) 1.25GHz | X1 Elite (X1E-78-100) 1.25GHz | X1 Plus (X1P-64-100) 1.25GHz | X1 Plus (X1P-42-100) ? |
Video hardware | X1 Elite (X1E-84-100)X1 Elite (X1E-80-100)X1 Elite (X1E-78-100)X1 Plus (X1P-64-100) 4K 60 fps encode, 4K 120 fps decode | X1 Plus (X1P-42-100) 4K 30 fps encode, 4K 60 fps decode | |||
PCIe | X1 Elite (X1E-84-100)X1 Elite (X1E-80-100)X1 Elite (X1E-78-100)X1 Plus (X1P-64-100) 8+4 lanes of PCIe 4.0, 2+2 lanes of PCIe 3.0 | X1 Plus (X1P-42-100) 4+4 lanes of PCIe 4.0, 2+2 lanes of PCIe 3.0 |
Promising performance
Qualcomm provides partners with reference performance and power consumption numbers for all new platforms, so they can easily verify their devices are working as intended. These numbers aren’t fully representative of the final performance numbers we will see from retail devices (especially since, a lot of the time, they are done in reference devices with thermal characteristics that are far from ideal), but they are still a good reference to see how they will perform. We summarized the performance numbers below, with competing chips from Apple and Intel for comparison:
X1 Elite (X1E80100) | X1 Elite (X1E78100) | X1 Plus (X1P64100) | Intel Ultra 7 155H | Apple M3 Pro (18-core GPU, 12-core CPU) | |
---|---|---|---|---|---|
CPU TESTS | |||||
Geekbench 6 (single-threaded) | X1 Elite (X1E80100) 2790 | X1 Elite (X1E78100) 2418 | X1 Plus (X1P64100) 2419 | Intel Ultra 7 155H ~2300 | Apple M3 Pro (18-core GPU, 12-core CPU) ~3200 |
Geekbench 6 (multi-threaded) | X1 Elite (X1E80100) 14309 | X1 Elite (X1E78100) 14077 | X1 Plus (X1P64100) 13139 | Intel Ultra 7 155H ~13000 | Apple M3 Pro (18-core GPU, 12-core CPU) ~16000 |
Cinebench 2024 (single-threaded) | X1 Elite (X1E80100) 122.83 | X1 Elite (X1E78100) 107.4 | X1 Plus (X1P64100) 109.1 | Intel Ultra 7 155H ~100 | Apple M3 Pro (18-core GPU, 12-core CPU) ~140 |
Cinebench 2024 (multi-threaded) | X1 Elite (X1E80100) ~1100 (depends on thermals) | X1 Elite (X1E78100) 891.7 | X1 Plus (X1P64100) 841.5 | Intel Ultra 7 155H ~900 | Apple M3 Pro (18-core GPU, 12-core CPU) ~1100 |
GPU TESTS | |||||
3DMark Wildlife Extreme Score | X1 Elite (X1E80100) 6051 | X1 Elite (X1E78100) 6208 | X1 Plus (X1P64100) 6245 | Intel Ultra 7 155H ~5000 | Apple M3 Pro (18-core GPU, 12-core CPU) ~14500 |
GFXBench Aztec Ruins Normal Tier | X1 Elite (X1E80100) 262.373 | X1 Elite (X1E78100) 276.59 | X1 Plus (X1P64100) 276.26 | Intel Ultra 7 155H ~240 | Apple M3 Pro (18-core GPU, 12-core CPU) ~600 |
AI TESTS | |||||
Procyon AI total score | X1 Elite (X1E80100) 1746 | X1 Elite (X1E78100) 1772 | X1 Plus (X1P64100) 1779 | Intel Ultra 7 155H ~200 | Apple M3 Pro (18-core GPU, 12-core CPU) ~900 |
As we can see, the results look really promising. The Oryon cores of the Snapdragon X1 easily match up with Intel’s Meteor Lake chips and, in the case of the X1E-80-100, even score a win in single-thread performance. Apple’s M3 is still ahead. However, we have not yet seen the results for the flagship X1E-84-100 chip, which should offer even higher performance than the aforementioned X1E-80-100. This is also true for the integrated GPU performance: Qualcomm’s Adreno beats Intel’s Arc but loses to Apple’s solution with a considerably larger gap than the CPU deficit.
One area the Snapdragon X1 truly excels at is AI performance (tested in the Procyon AI benchmark): it’s 2x faster than the Apple M3 Pro and 8x faster than the Intel Ultra 7. This is thanks to Qualcomm’s mature AI acceleration architecture, based on its Hexagon NPUs (providing 45 TOPS of performance in this case), as well as the Adreno GPU and Oryon CPU (an additional 30 TOPS). This will be useful as it was recently revealed that Microsoft’s Copilot AI on Windows will run locally and that future versions of Windows will impose a 40 TOPS minimum NPU performance requirement. The Snapdragon X1 is fairly future-proof in this increasingly popular area of computing.
All these values show Qualcomm is finally serious about PCs: the Snapdragon X1 goes head-to-head with chips that are usually included in higher-tier laptops that start at $1,000 and are often even more expensive than that. Thanks to the fact that there are more SKUs in the lineup this year, we should see the chips used in more product classes, and not just high-end devices, as was the case with previous generations of Qualcomm’s laptop chips.
Snapdragon X1 appears competitive with the best from Apple and Intel.
As for power consumption, Qualcomm provides them in a variety of scenarios; for simplicity, we will list them for a CPU-only load. There’s a lot of variance in silicon devices, so Qualcomm lists the values as they should be able to be achieved by 95% and 50% of the manufactured chips, respectively. In practice, the better parts will simply get binned as the higher-end SKUs, though, so the distinction doesn’t have any significant implications in reality. However, it’s still important to keep in mind.
X1 Elite (X1E84100) | X1 Elite (X1E80100) | X1 Plus (X1P64100) | |
---|---|---|---|
Total package power (95% parts) | X1 Elite (X1E84100) 98.50W | X1 Elite (X1E80100) 52.92W | X1 Plus (X1P64100) 42.52W |
Total package power (50% parts) | X1 Elite (X1E84100) 82.33W | X1 Elite (X1E80100) 43.40W | X1 Plus (X1P64100) 35.01W |
For context, Apple’s M3 Pro (built on TSMC’s 3 nm process node) consumes around 42W under CPU loads, and the older M2 Pro (TSMC’s 4 nm process, similar – if not identical – to the one used for the Snapdragon X1 series) uses around 55W. Intel’s Core Ultra 7 155h, meanwhile, uses around 80W.
The huge jump between X1E-84-100 and X1E-80-100 might seem strange, given the difference between the two is just 400MHz on the CPU cores, but it is explainable. Semiconductor chips tend to require much more power to push them past a certain point on their operating points curves. 3.8GHz seems to be relatively high; Qualcomm is basically “overclocking” the chip out of the factory. Additionally, the higher-end chip requires different power delivery circuitry, which lets it consume way more power.
Not just PC chips
While the main point of this article is to talk about the Snapdragon X, we have also been tipped about a potential Oryon-powered server chip in development. The product in question is called the “SD1” and has the following specs:
- 80 Oryon cores at up to 3.8GHz
- 16 channels of DDR5 up to 5600MHz
- 70 PCIe 5.0 lanes
- CXL v1.1 support
- 9470-pin LGA socket (98.0×95.0mm)
- Support for a two-socket configuration
- Built on TSMC’s 5nm process (N5P)
While our source couldn’t confirm the status of this project, we have been told that Qualcomm’s partners were briefed about it in late 2021 and early 2022, which aligns with earlier rumors about such a chip.
This wouldn’t be Qualcomm’s first foray into server space. Qualcomm launched Centriq in 2017, a line of Arm-based server chips that were unfortunately canceled just a year later.
Windows on Arm is finally here
As shown by the benchmarks above, Qualcomm finally has a competitive solution for Arm-based PCs. The fastest models go head-to-head with Apple’s and Intel’s offerings, and the broader lineup of chips should cover most of the typical price points.
We can’t wait to see how these chips fare in practice. Fortunately, the wait shouldn’t be too long; the first Snapdragon X-based laptops are expected to launch in mid-2024.