MLID brings a lot of AMD mobile platform coverage in 2024, covering high and low end product lines.

This time, not only the CPU architecture has been upgraded to Zen5, but also the GPU architecture and performance have been improved by leaps and leaps, even reaching the mobile version of the RTX 4070 level!

MLID first points out,Since the inception of the Zen architecture, AMD has made capturing the server market a top priorityBecause server chips are very profitable, AMD chiplet small chip design is also very suitable for the server market, which is more competitive against Intel Xeon.

AMD expects to capture it by the end of 202335-40％Its server market share is high enough, and the rest of the market will be extremely difficult for Intel to take away from it.

AMD now has the financial firepower to invest heavily in several market segments, starting with mobile laptops in 2024.

MLID also reminds you,Processor design cycles are typically three to four years longTherefore, the Rnoir Ryzen 4000 series and Cezanne Ryzen 5000 series that you have seen before were all designed when AMD was still very poor, and the specifications and performance were inevitably conservative.

Now the product,Hawk Point debuted in the first quarter of 2024, targeting the mainstream mobile market with a power consumption range of 15-45W.

It will replace the current 4nm Phoenix with little change in specifications and is expected to be a slightly tweaked upgrade.

In Q2 / Q3 2024, "Strix Point" will arrive, targeting the high-end mobile market with power consumption range of 15-45W. The original target is the first half of 2024, which seems to be a bit delayed.

It will be a 4nm process, single chip design, with a large and small core design for the CPU part, equippedMaximum 4 Zen5, 8 Zen5c, total up to 12 core 24 threads, sharing 24MB level 3 cache.

The GPU part has been upgraded to RDNA3+ architecture (or RDNA 3.5), CU units have been increased from up to 12 to up to 16 (1024 stream processors), although there is still no unlimited cache, and performance is expected to be slightly lower than the 35W RTX 3050. But significantly ahead of Intel Meteor Lake 14 generation Core.

Memory upgrade support 128-bit LPDDR5X, in addition to AI engine, up to 20TOPS.

In the second half of 2024, Sarlak (formerly codenamed Strix Hero) is ready to be used in all phases of the 20-120W power range. The competition is no longer Intel, but Apple's M Series!

Also 4nm process, but using small chip, pure large core design, integrationUp to 16 Zen5 CPU coresThe early test data is about 25% higher than the current 16 core performance. Meanwhile, there are different specifications of 12 core, 8 core and 6 core, among which 8/6 core power consumption is lower than 28W.

The GPU part is also RDNA3+ architecture, up to 40 CU units (2560 stream processors), and 32MB unlimited cache, performance is expected to rival 95W power RTX 4070!

There are also different versions of 32, 24, and 20 CU units, which are expected to perform in the RTX 4060, 4050, and 3050 levels respectively.

Memory support to 256-bit wide LPDDR5X, AI computing power up to 40TOPS.

There will also be the ultimate mobile platform, Fire Range, in the second half of 2024.

It also has 16 Zen5 CPU cores, but with the GPU architecture downgraded to RDNA2, it should also be small in scale, clearly a top platform focused on CPU computing performance.

By 2025, the road map has two new names: Kracken and Eshcer", both from German, are located in the mainstream, but the exact circumstances are unknown.

At the entry level, Mendocino, a combination of 6nm technology and Zen2 and RDNA2 architectures, will have a strong presence in 2025.