At Computex Taiwan last May, Intel announced the next generation of Core Ultra 200V Architecturecodenamed Lunar Lake. Intel’s new Lunar Lake architecture aims to deliver competitive performance at ultra-low power consumption for thin and compact laptops. After all, Qualcomm’s ARM-based Snapdragon X Elite has entered the Windows PC ecosystem and is already making headlines for its efficiency. So, let’s learn more about Intel’s Lunar Lake architecture and how it has been redesigned for efficiency.
Intel Lunar Lake Architecture
With Meteor Lake last year, Intel moved from its age-old monolithic design to tile-based chip design. And Lunar Lake goes even further. Unlike Meteor Lake where the Compute tile only housed the CPU and cache, the Compute tile on Lunar Lake processors will also house the CPU, cache, GPU, and NPU.
This means that the Compute tile is the the largest tile on the dieand the best part this year is that it is made on TSMC N3B Process NodeSure, TSMC’s N3B has lower yields than the much improved latest N3E node, but Intel is finally moving away from the Intel foundry to TSMC’s advanced 3nm process node, which is great.
As for the platform controller that provides the I/O and connectivity, it’s built on TSMC’s 6nm node (N6), like last year’s Meteor Lake. This is the first time Intel has designed its processor, but TSMC is building it. Finally, Intel is packaging the entire chipset using its own Foveros 3D technology.
Image courtesy of Intel
Not only that, Intel is also move memory to processorThis means that unified memory, similar to Apple’s M-series chips, will be available on Lunar Lake chips. LPDDR5X-8533 RAM Memory is available in 16GB or 32GB capacity.
Overall, the Lunar Lake architecture has undergone significant changes. The CPU, GPU, NPU, and cache are now part of the Compute tile and are manufactured on TSMC’s 3nm process node (N3B), which should lead to much better efficiency. And finally, memory is available directly on the SoC to reduce power consumption and space and improve bandwidth.
At the Computex event, Michelle Holthaus, Intel’s executive vice president and general manager, said“We are going to break the myth that [x86] can’t be as efficient.” Intel says x86-based Lunar Lake processors reduce energy consumption by 40%.
It seems that Intel is taking all the necessary steps to improve the efficiency of Lunar Lake processors. Now let’s take a look at the new Lunar Lake CPU cores.
Intel Lunar Lake Processor
The lunar lake will have 8 processor cores – 4 performance (P) cores named Lion Cove and 4 efficiency (E) cores named Skymont. As I mentioned above, the CPU is part of the Compute tile. Intel claims that the Lion Cove P-core on Lunar Lake brings a 14% IPC gain over the Redwood Cove P-core on Meteor Lake.
Image courtesy of Intel via YouTube
Intel has done something very different this time around. The chipmaker has SMT completely removed (Simultaneous Multi-threading) after more than two decades of using its processor. SMT, commonly known as HyperThreading, allows a core to perform two tasks in parallel. Intel claims that removing SMT helps improve performance per watt by 5%.
To compensate for the lack of HyperThreading, Intel says Lunar Lake processors can execute more instructions per cycle instead of relying on parallel execution. This allows the processor to perform better in single-threaded tasks.
Image courtesy of Intel via YouTube
Now let’s move on to the E-core. I think Skymont is the standout feature of Lunar Lake processors. Intel claims that Skymont delivers a massive 68% improvement in IPC over Meteor’s Crestmont Lake E-core. The 4-core Skymont cluster remains separated into a “low-power island” from the P-core cluster with access to its own L3 cache.
As a result, Skymont consumes one-third the power needed to match Crestmont’s peak performance. In total, Skymont delivers twice the performance of the Crestmont core in single-threaded tasks.
Image courtesy of Intel via YouTube
In addition to this, Intel has brought granularity to clock speed increases with Lunar Lake. Instead of increasing the clock speed by 100 MHz, which consumes more power, the Lunar Lake architecture can increase the clock speed by 16.67 MHz to manage the power budget of any task.
The reduced frequency range will result in lower power consumption. Overall, Intel claims that the Lunar Lake processor can match the single-threaded performance of Meteor Lake for only half the power, which is quite impressive.
Lunar Lake Geekbench Score (Leaked)
While Lunar Lake is set to launch on September 3rd, some Geekbench scores have already leaked. Running the lowest-end SKU (Ultra 5 228V Core), the 8-core processor scored 2,530 in the single-core test and 9,875 in the multi-core test. The SKU reaches 4.5 GHz with a TDP of 17 W (max turbo power of 30 W).
And the highest-end SKU (Core Ultra 9 288V) from Lunar Lake manages to score 2,790 in the single-core test and 11,048 in the multi-core test. other racesit even managed to cross the 2,900 mark in single-threaded tasks. This particular SKU goes up to 5.1 GHz and has a TDP of 30 W.
Intel Lunar Lake: new Xe2 GPU
The integrated GPU on Lunar Lake is built on the Battlemage graphics architecture and it includes 8 2nd Generation Intel Xe CoresIt also features 8 ray tracing units for enhanced gaming performance and real-time ray tracing. Plus, for AI tasks, the new Lunar Lake GPU alone can perform 67 trillion operations per second (TOPS). Pretty impressive, right?
Image courtesy of Intel via YouTube
Compared to the Meteor Lake GPU, the Lunar Lake GPU is 1.5x faster and also features XeSS AI-based upscaling. Its display engine can drive three 4K HDR displays at 60Hz and a single 8K HDR display at 60Hz. Finally, Lunar Lake processors also support AV1 encoding and decoding.
Intel Lunar Lake CPU Processor
Much has been made of Meteor Lake’s weak NPU that could only run 10 TOPS, but with Lunar Lake, Intel will power a line of Copilot+ PCs for local AI workloads. The new Lunar Lake NPU 4 can run up to 48 TOPS on its own, which is more than Microsoft’s 40 TOPS eligibility cap for Copilot+ PCs.
Image courtesy of Intel via YouTube
Taking into account all the computing units, the CPU can perform up to 120 TOPS. The GPU can perform up to 67 TOPS, the CPU up to 5 TOPS and the NPU up to 48 TOPS. totaling 120 TOPS. It’s even hhigher than Qualcomm’s total of 75 TOPS processing capacity on the Snapdragon X Elite. Keep in mind that the TOPS figure is based on the INT8 data type.
Intel Lunar Lake: references leaked
Below you can check out all the leaked SKUs of the Core Ulra processors based on the Lunar Lake architecture. There are nine different SKUs featuring eight CPU cores on each of them. The distinguishing factors are memory, CPU/GPU clock speed, and NPU capacity.
| Lunar Lake References | Cores/Threads | Memory | Maximum processor frequency | Maximum GPU frequency | NPU (TOP) | TDP Beach |
|---|---|---|---|---|---|---|
| Core Ultra 9 288V | 8C/8T | 32 GB | 5.1 GHz | 2.05 GHz | 48 | 30 W – 30 W |
| Core Ultra 7 268V | 8C/8T | 32 GB | 5.0 GHz | 2.00 GHz | 48 | 17 W – 30 W |
| Core Ultra 7 266V | 8C/8T | 16 GB | 5.0 GHz | 2.00 GHz | 48 | 17 W – 30 W |
| Core Ultra 7 258V | 8C/8T | 32 GB | 4.8 GHz | 1.95 GHz | 47 | 17 W – 30 W |
| Core Ultra 7 256V | 8C/8T | 16 GB | 4.8 GHz | 1.95 GHz | 47 | 17 W – 30 W |
| Ultra 5 238 V Core | 8C/8T | 32 GB | 4.7 GHz | 1.85 GHz | 40 | 17 W – 30 W |
| Core Ultra 5 236V | 8C/8T | 16 GB | 4.7 GHz | 1.85 GHz | 40 | 17 W – 30 W |
| Ultra 5 228V Core | 8C/8T | 32 GB | 4.5 GHz | 1.85 GHz | 40 | 17 W – 30 W |
| Ultra 5 226V Core | 8C/8T | 16 GB | 4.5 GHz | 1.85 GHz | 40 | 17 W – 30 W |
Related articles
Apple M4 vs Snapdragon X Elite: Can Qualcomm Keep Up?
Arjun Sha May 8, 2024
Snapdragon X Elite vs Apple M3: Qualcomm makes an Apple!
Arjun Sha April 25, 2024
Intel Lunar Lake: Further Improvements
As mentioned above, RAM is now part of the SoC. This means that the CPU, GPU, or NPU can quickly access the memory. Intel claims that moving the memory to the SoC also helps free up space on the motherboard. Since the memory is physically closer to the compute tile, bandwidth improves with reduced latency and results in about a 40% reduction in power consumption.
Of course, with integrated memory, users won’t be able to upgrade or replace the memory, which might not please some. Aside from that, Intel says Thread Director has been improved to allocate tasks to the appropriate cores. Intel is also using machine learning to tell the system scheduler to better guide tasks.
Finally, the The TDP range of Lunar Lake processors is between 17W and 30W. Overall, I’m very excited about the Lunar Lake processors that are expected to arrive on September 3, 2024. It’s going to be an exciting time for consumers as Intel takes on Qualcomm and AMD in the AI PC race. We might finally see improved battery life on Windows laptops with x86 processors.
