AMD Ryzen AI Max 390 vs Intel Xeon Phi 7290 Comparison

FAQ

Q: Which processor has more cores and threads?

A: The Intel Xeon Phi 7290 has 72 cores and 288 threads, while the AMD Ryzen AI Max 390 has 12 cores and 24 threads. The Xeon Phi offers six times the core count and twelve times the thread count on paper.

Q: How do the two chips compare in single-threaded performance?

A: The AMD Ryzen AI Max 390 dominates in single-threaded workloads. In Cinebench R23 single-core, it scores 5091 versus the Xeon Phi's 2140, a 137.9% advantage. In PassMark single-thread, the lead expands dramatically: 4028 versus 485, a 730.5% delta.

Q: Which processor wins in multi-core Cinebench tests?

A: The AMD Ryzen AI Max 390 wins every multi-core Cinebench run. In Cinebench R23 multicore, it scores 36064 versus 15163 for the Xeon Phi, a 137.8% lead. The same pattern holds in R20 (15146 vs 6368) and R15 (3635 vs 1528).

Q: Does the Xeon Phi win any benchmark at all?

A: Yes, the Xeon Phi 7290 wins three PassMark subtests: data compression (564535 vs 487145, a 13.7% lead), extended instructions (41517 vs 38716, a 6.7% lead), and random string sorting (68593 vs 53113, a 22.6% lead).

Q: What are the process nodes and foundries for each?

A: The AMD Ryzen AI Max 390 is built on TSMC's 4 nm process, while the Intel Xeon Phi 7290 uses Intel's 14 nm process. The Xeon Phi also has a transistor count of 8,000 million.

Q: What memory types do they support?

A: The AMD Ryzen AI Max 390 supports LPDDR5X memory with a quad-channel bus and 256.0 GB/s bandwidth. The Intel Xeon Phi 7290 supports DDR4 memory, though its bus width and bandwidth figures are not recorded in the database.

Architecture Differences

The two processors come from opposite ends of the design spectrum. The AMD Ryzen AI Max 390 uses the Zen 5 architecture under the Strix Halo codename, manufactured on a 4 nm process at TSMC. Its die is composed of two 70.6 mm² chiplets. The Intel Xeon Phi 7290 uses the Knights Landing architecture, built on Intel's 14 nm process with a transistor count of 8,000 million.

Cache hierarchies diverge sharply. The Ryzen AI Max 390 has 80 KB of L1 per core, 1 MB of L2 per core, and a large 64 MB shared L3 cache. The Xeon Phi 7290 has 32 KB of L1 per core and 512 KB of L2 per core, with no recorded L3 cache in the database. The Ryzen's 64 MB shared L3 is a substantial resource for data reuse, while the Xeon Phi relies on per-core L2 slices.

Memory support reflects their different market segments. The Ryzen AI Max 390 is a mobile part on AMD Socket FP11, supporting LPDDR5X with a quad-channel bus and 256.0 GB/s bandwidth. The Xeon Phi 7290 is a server or workstation part on Intel Socket 3647, supporting DDR4, though its memory bus details are not recorded. Both support ECC memory.

The Ryzen AI Max 390 includes an integrated Radeon 8050S GPU, while the Xeon Phi 7290 has no integrated graphics. The Ryzen also supports PCIe Gen 4 with 16 CPU-only lanes; the Xeon Phi's PCIe configuration is not listed. The Ryzen is built for low power with a 55 W TDP, while the Xeon Phi draws 245 W.

Clock speeds tell a story of specialization. The Ryzen AI Max 390 has a base clock of 3.20 GHz and a boost clock of 5.00 GHz. The Xeon Phi 7290 has a base clock of 1.50 GHz and a boost of 1.70 GHz. The Ryzen's much higher frequencies help explain its single-thread dominance, while the Xeon Phi's lower clocks are offset by its massive core count, though that advantage does not materialize in most recorded benchmarks.

Production status differs too: the Ryzen AI Max 390 is listed as active, while the Xeon Phi 7290's status is not recorded. The Ryzen was released in January 2025, the Xeon Phi in June 2016. Neither chip has an unlocked multiplier.

The Verdict

The data points to a clear overall winner for general and multi-threaded compute: the AMD Ryzen AI Max 390. It wins 14 of the 17 head-to-head benchmark comparisons, including every Cinebench test and the PassMark multithread test. Its average benchmark score of 56273 sits above the Xeon Phi's 53469, though both land in the 91st percentile of all CPUs.

The Xeon Phi 7290 is not without merit. It wins in data compression, extended instructions, and random string sorting, suggesting that specific workloads involving those operations can favor its architecture. But those wins are narrow in percentage terms (6.7% to 22.6%) compared to the Ryzen's often massive leads.

For anyone running typical productivity, rendering, encryption, physics, or integer math tasks, the Ryzen AI Max 390 is the better choice. Its single-thread performance is in a different league, and its multi-core results are consistently ahead despite having far fewer cores. The Xeon Phi 7290 is only worth considering for niche workloads that match its three winning benchmark categories, and even then, the margins are modest.

Specification Differences

• Cores: 12 (AMD) vs 72 (Intel)
• Threads: 24 (AMD) vs 288 (Intel)
• Base Clock: 3.20 GHz (AMD) vs 1.50 GHz (Intel)
• Boost Clock: 5.00 GHz (AMD) vs 1.70 GHz (Intel)
• TDP: 55 W (AMD) vs 245 W (Intel)
• Socket: AMD Socket FP11 (AMD) vs Intel Socket 3647 (Intel)
• Architecture: Zen 5 (AMD) vs Knights Landing (Intel)
• Codename: Strix Halo (AMD) vs Knights Landing (Intel)
• Process Node: 4 nm (AMD) vs 14 nm (Intel)
• Foundry: TSMC (AMD) vs Intel (Intel)
• Transistors: Not recorded (AMD) vs 8,000 million (Intel)
• Die Size: 2x 70.6 mm² (AMD) vs Not recorded (Intel)
• L1 Cache: 80 KB per core (AMD) vs 32 KB per core (Intel)
• L2 Cache: 1 MB per core (AMD) vs 512 KB per core (Intel)
• L3 Cache: 64 MB shared (AMD) vs None recorded (Intel)
• Memory Support: LPDDR5X (AMD) vs DDR4 (Intel)
• Memory Bus: Quad-channel (AMD) vs Not recorded (Intel)
• Memory Bandwidth: 256.0 GB/s (AMD) vs Not recorded (Intel)
• Integrated Graphics: Radeon 8050S (AMD) vs None (Intel)
• Market Segment: Mobile (AMD) vs Server/Workstation (Intel)
• Release Date: 2025-01-05 (AMD) vs 2016-06-19 (Intel)
• Part Number: 100-000001423 (AMD) vs SR2WY (Intel)

Head-to-Head Benchmarks

The most striking result is in PassMark single-thread, where the Ryzen AI Max 390 scores 4028 versus the Xeon Phi's 485, a 730.5% advantage. This is the largest delta across all recorded tests and underscores the massive per-core performance gap between a modern 5.00 GHz Zen 5 core and a 1.70 GHz Knights Landing core.

In Cinebench R23 multicore, the Ryzen scores 36064 against the Xeon Phi's 15163, a 137.8% lead. The R20 multicore result mirrors this: 15146 versus 6368, a 137.8% delta. R15 multicore shows 3635 versus 1528, a 137.9% lead. These consistent margins suggest a fundamental throughput advantage for the Ryzen despite its 12-core count versus the Xeon Phi's 72 cores.

PassMark multithread shows the Ryzen at 41737 versus 17839, a 134% win. Integer math goes to the Ryzen at 146519 versus 126922, a 15.4% lead. Floating point math is a bigger win: 90594 versus 47417, a 91.1% advantage. Data encryption is another decisive Ryzen win: 25097 versus 12505, a 100.7% delta. Find prime numbers favors the Ryzen heavily: 316 versus 114, a 177.2% lead. Physics also goes to the Ryzen: 2761 versus 2257, a 22.3% win.

The Xeon Phi's three wins are narrower. Random string sorting shows 68593 versus 53113, a 22.6% margin for Intel. Data compression gives 564535 versus 487145, a 13.7% lead. Extended instructions sees 41517 versus 38716, a 6.7% advantage. These wins suggest the Xeon Phi's many cores can still shine in specific memory-heavy or instruction-heavy patterns.

Where Each One Wins

The AMD Ryzen AI Max 390 wins in every scenario that depends on high clock speeds, modern architecture, or balanced multi-core throughput. That covers rendering workloads as measured by Cinebench, general multithreaded processing, integer and floating point math, encryption, prime number searches, and physics simulations. Its 5.00 GHz boost clock and 64 MB shared L3 cache make it the stronger choice for interactive work, compilation, scientific computing, and any task where single-thread latency matters.

The Intel Xeon Phi 7290 wins in three specific PassMark subtests: data compression, extended instructions, and random string sorting. These are workloads where its 288 threads and per-core L2 cache can be marshalled into parallel data shuffling. The 22.6% lead in random string sorting and the 13.7% lead in data compression indicate that certain data-movement-heavy tasks can exploit its massive thread count. Extended instructions, a 6.7% win, hints at specialized vector or SIMD workloads where Knights Landing's design still holds a niche.

Neither processor is universally superior. The Ryzen AI Max 390 is the default choice for almost any modern application, with 14 wins out of 17 tests and an average benchmark score of 56273. The Xeon Phi 7290, with its 91st percentile ranking and average score of 53469, remains relevant only for workloads that align with its three victories. For those specific tasks, the Xeon Phi's 72 cores and 288 threads can still outperform a much newer 12-core chip. For everything else, the data overwhelmingly favors the Ryzen.