Snapdragon X Elite vs Apple M3: Qualcomm Does an Apple!

Apr. 25, 2024

The entire lineup ofSnapdragon X series processorshas been announced. Among them, the Snapdragon X Elite processor, particularly the X1E-84-100 variant is the most powerful chipset. It features 12 Oryon CPU cores, clocked up to 4.2GHz. So in this post, we compare the Snapdragon X Elite vs Apple M3 with deep analysis of CPU, GPU, NPU, and more. In addition, we have compared benchmark scores of Snapdragon X Elite and Apple M3. On that note, let’s jump in.

Note:Snapdragon X Elite’s top-end variant (X1E-84-100) features 12 CPU cores. So for better parity, we will be comparing Qualcomm’s PC chipset with the Apple M3 Pro that packs 12 CPU cores.

Specification Comparison

Let’s compare the CPU of the Snapdragon X Elite and Apple M3 first. The Snapdragon X Elite is fabricated on TSMC’s proven 4nm process node. The chipset features 12 high-performance Oryon cores and does not pack any efficiency cores. The Snapdragon X Elite has a maximum frequency of 3.8GHz and two cores can beturbo boosted up to a mighty 4.2GHz.

On paper, the Snapdragon X Elite looks more potent than the Apple M3 Pro as two of its cores can run at a slightly higher frequency. However, to get a clear picture, we need to look at the Geekbench scores of Snapdragon X Elite and Apple M3 Pro.

We have not seen leaked benchmarks of the top-tier variant, Snapdragon X Elite (X1E-84-100) running on a consumer laptop so far. So we are going to useofficial benchmark scoresfrom Qualcomm’s reference device in this comparison.

In the multi-core test, the Snapdragon X Elite did perform better than the M3 Pro, but again by a slight margin. And yes, the top-end SD X Elite variant indeed performsbetter than the Apple M2 Pro and base Apple M3.

Overall, I would conclude that both chipsets have powerful CPU cores and arenearly matching in performance. However, Apple’s CPU cores are able to match performance while running at a lower frequency. It means that in terms of performance-per-watt, Apple still has a slight lead.

While we are yet to test the Snapdragon X Elite, Signal65 has done an early comparison betweenSurface Laptop 7th Edition and MacBook Air M3. The analysis was commissioned by Microsoft. The conclusion of the report is that Snapdragon X Elite is plenty powerful and outranks Apple M3 in multi-core tasks. However, for single-core tasks, M3 is around 15% faster than X Elite.

In terms of GPU, Apple M3’s 10-core GPU is about 30% faster than Snapdragon X Elite’s Adreno GPU. In NPU performance, however, Snapdragon X Elite’s Hexagon NPU performs 2x better than Apple M3’s 16-core Neural Engine.

The battery life is also quite good on the Snapdragon X Elite with active cooling support. The Surface Laptop 7 delivered about 21 hours of video playback on a single charge, outranking the MacBook Air M3 by 16%.

Having said all of that, the thermal performance of X Elite is slightly disappointing as it runs slightly hotter than MacBook Air. It requires active cooling and yet crosses 50 degrees C when the whole system is stressed for a long duration. The fanless MacBook Air M3, on the other hand, remained close to 45.8 degrees C.

In our comparison betweenA17 Pro and Snapdragon 8 Gen 3, we found that 8 Gen 3’s Adreno GPU is both more powerful and efficient than Apple’s new mobile GPU. But, does the Adreno GPU have the prowess to defeat Apple’s next-gen GPU developed on a newer architecture for M-series SoCs? Let’s find out.

The Snapdragon X Elite comes with an integrated Adreno GPU onboard. We don’t know the GPU core count, but the top-end variant (X1E-84-100) canperform 4.6 TFLOPS, which is higher than the 10-core GPU on the base Apple M2 chip (3.6 TFLOPS).

However, Apple has taken its GPU to the next level this year with the M3 chipset. It has introduced a new GPU architecture and has also broughtsupport forray tracingand Mesh Shaders. Not to mention, Dynamic Caching is supposed to make graphics performance even better on macOS with proper memory management.

Apple is planning to make macOS a powerful gaming platform, which is something the Cupertino giant has missed out on for many years. As things stand right now, the GPU on the Apple M3 chipset seems to be powerful, and Qualcomm needs to do more to entice thegaming communityto move from x86 to an ARM-based Windows platform.

Qualcomm recently said thatWindows games will work just fine on Snapdragon X Elitedue to improved emulation, but will have to test it out before giving a final judgment.

Having said that, Qualcomm released some GPU benchmark numbers for the X Elite in comparison to the older Apple M2 chipset. The Snapdragon X Eliteachieved 44.5 FPSin the 3DMark Wildlife Extreme test whereas the Apple M2 achieved 40.8 FPS. In the Aztec Ruins benchmark test, the X Elite scored 350 FPS and the M2 scored 295 FPS.

If we consider the Geekbench GPU score of both chipsets based on the OpenCL API, it seems the Adreno GPU is behind the Apple M3 GPU. When compared to the Apple M3 Pro which packs 14 GPU cores, the difference becomes even more substantial. So yeah, the Adreno GPU on the Snapdragon X Elitedoesn’t beat the Apple M3 GPU, however, it’s slightly better than the M2 GPU.

As we transition to the AI age, the competition in the silicon industry is heating up to deliver the best on-device AI experience. Qualcomm has packed its powerful Hexagon AI Engine on the Snapdragon X Elite and it can perform45 TOPS(trillion operations per second) with the NPU alone.

But what’s even more impressive is that when using all the compute units together including CPU, GPU, and NPU, it candeliver up to 75 TOPS. Not to mention, it can generate 30 tokens per second while running a 7B LLM model.

In comparison, the 16-core Neural Engine on the Apple M3 chipset can onlyperform up to 18 TOPS. It’s quite puzzling that Apple’s A17 Pro, a mobile chipset with 16 Neural Engine cores can perform 35 TOPS, but Apple’s latest M-series processors are limited to just 18 TOPS.

However, with the M3 family, you get an option of 128GB Unified memory to load large language models whereas you can only run modelsup to 64GBon X Elite. Simply put, in terms of on-device AI capability, the Snapdragon X Elite seems more capable than the Apple M3.

The Snapdragon X Eliteuses 16-bit 8-channel LPDDR5Xmemory, and it supports capacity up to 64GB. It has a memory bandwidth of 135GBps. On the other hand, as we mentioned in ourApple M3 vs Apple M2comparison, the memory speed has been downgraded on the M3 Pro specifically.

Apple is using a192-bit memory bus(down from 256-bit), which reduces the LPDDR5 memorybandwidth to 150GBpsfrom 200GBps. Despite all of that, you’ll get better memory speed on the Apple M3 Pro than the Snapdragon X Elite.

The Snapdragon X Elite comes with the latest connectivity options. It includesWi-Fi 7, Bluetooth 5.4, and LEsupport. Laptop makers also have the option to add a 5G modem to bring an always-connected PC experience. Qualcomm has said that its discrete Snapdragon X65 5G modem can be paired with the Snapdragon X Elite.

Not to forget, Snapdragon X Elite supports AV1 encoding and decoding for efficient video streaming and processing. On the Apple M3, you get Wi-Fi 6E and Bluetooth 5.3 support, but there is no support for mobile connectivity on the go. You willget AV1 decodingsupport, but can’t encode videos with AV1 codecs.

As far as Oryon CPU cores are concerned, it appears Qualcomm has truly developed a powerful CPU and itrivals the Apple M3 chipset. The Nuvia team has managed to deliver powerful performance which is amazing. In the GPU department, Qualcomm needs to make big changes to compete with a laptop-grade GPU. Currently, it can only compete with the older M2 10-core GPU, which is not bad per se.

While Qualcomm says that emulation has improved, the company along with Microsoft needs to work with the developer community to bring native games to the ARM platform. Windows is already the choice for gamers, and if Qualcomm wants to succeed and replace x86 as the preferred hardware platform, it needs to work onsoftware compatibility.

Finally, in the NPU and connectivity departments, Qualcomm is already leading the race and setting a benchmark foron-device AI experiences. Overall, there’s just one conclusion that we can make; whether it’s Qualcomm or Apple, it’s time for ARM to shine bright for years to come. What are your thoughts on this? Let us know in the comments section below.

Passionate about Windows, ChromeOS, Android, security and privacy issues. Have a penchant to solve everyday computing problems.