AMD EPYC 4465P vs AMD EPYC 4484PX Comparison

The AMD EPYC 4484PX and AMD EPYC 4465P are both 12-core, 24-thread server processors on the AM5 socket, but they deliver distinctly different performance profiles. The 4484PX wins the majority of benchmark comparisons, 13 out of 17, but the 4465P counters with decisive victories in math-heavy workloads and single-threaded tasks. The data shows a clear split: the 4484PX dominates in cache-sensitive and encryption tasks, while the 4465P takes the lead in raw arithmetic throughput and single-core speed.

Head-to-Head Benchmarks

The most lopsided win for the AMD EPYC 4484PX comes in PassMark physics, where it scores 4938 against the 4465P’s 3647, a 35.4% advantage. This is the largest delta in the entire head-to-head set and indicates a substantial edge in simulation or physics-based computation. Similarly, the 4484PX crushes the 4465P in PassMark find prime numbers, posting 425 versus 338, a 25.7% gap. These two results alone suggest the 4484PX is far stronger in workloads that rely on rapid integer iteration or specific algorithm execution.

Encryption is another major victory for the 4484PX. It scores 36499 in PassMark data encryption, beating the 4465P’s 32184 by 13.4%. The 4484PX also wins PassMark data compression with 603371 versus 577210, a 4.5% lead, and PassMark random string sorting with 71642 versus 67597, a 6% advantage. These wins point to the 4484PX’s larger cache and 3D V-Cache slice being highly effective for data manipulation tasks.

In the Cinebench suite, the 4484PX wins every single test, but the margins are razor-thin. In Cinebench R23 multi-core, it scores 42964 versus 42918, a 0.1% difference. Single-core R23 shows the same 0.1% edge, 6065 versus 6059. The story repeats in Cinebench R20 (18044 versus 18025 multi-core, 2547 versus 2544 single-core) and R15 (4330 versus 4326 multi-core, 611 versus 610 single-core). These results indicate that for standard rendering workloads, the two processors are effectively tied, with the 4484PX holding a marginal but consistent lead.

The AMD EPYC 4465P fights back in PassMark floating point math, scoring 105833 against the 4484PX’s 96446, an 8.9% win. It also takes PassMark integer math, 174551 versus 162993, a 6.6% advantage. Most notably, the 4465P wins PassMark single-thread with a score of 4575 versus 4119, a 10% margin. This single-thread dominance is significant, suggesting the Zen 5 architecture in the 4465P delivers higher per-core performance despite lower clock speeds on paper.

The overall average benchmark scores reflect this mixed picture. The 4484PX has an average benchmark score of 67822, placing it in the 94th percentile of all CPUs, while the 4465P averages 66925, in the 93rd percentile. The 4484PX’s nearest rival, the AMD Ryzen Threadripper PRO 5955WX, scores 67868, a near-perfect match at -0.1% delta. The 4465P’s closest competitor is the Intel Core Ultra 5 250K Plus at 66855, just 0.1% behind.

FAQ

Q: Which processor has a higher boost clock?

A: The AMD EPYC 4484PX has a boost clock of 5.60 GHz, while the AMD EPYC 4465P boosts to 5.40 GHz.

Q: Do both processors support ECC memory?

A: Yes, both the AMD EPYC 4484PX and the AMD EPYC 4465P support ECC memory.

Q: What is the difference in L3 cache size?

A: The AMD EPYC 4484PX has 128 MB of shared L3 cache, which includes a 1x 64MB 3D V-Cache slice. The AMD EPYC 4465P has 64 MB of shared L3 cache with no 3D V-Cache.

Q: Which processor has a higher memory bandwidth?

A: The AMD EPYC 4465P has a higher memory bandwidth at 89.6 GB/s, compared to 83.2 GB/s for the AMD EPYC 4484PX.

Q: How many PCIe Gen 5 lanes does each CPU provide?

A: The AMD EPYC 4484PX provides 28 lanes of PCIe Gen 5, while the AMD EPYC 4465P provides 24 lanes.

Q: Which processor is newer?

A: The AMD EPYC 4465P was released on 2025-05-12, while the AMD EPYC 4484PX was released on 2024-05-20.

Architecture Differences

The two processors are built on different architectures and process nodes. The AMD EPYC 4484PX uses the Zen 4 architecture, codenamed Raphael, manufactured on a 5 nm process by TSMC. The AMD EPYC 4465P uses the newer Zen 5 architecture, codenamed Grado, on a 4 nm process, also from TSMC. This node shrink likely contributes to the 4465P’s superior single-thread performance and lower power draw.

The cache configurations are markedly different. The 4484PX features a 64 KB L1 cache per core, 1 MB L2 per core, and 128 MB of shared L3 cache, which includes a dedicated 1x 64MB 3D V-Cache slice. The 4465P has a larger L1 cache at 80 KB per core, the same 1 MB L2 per core, but only 64 MB of shared L3 cache, with no 3D V-Cache. This 2x difference in L3 capacity explains the 4484PX’s wins in data compression and random string sorting.

Transistor counts and die sizes also differ. The 4484PX contains 17,840 million transistors across a 2x 71 mm² die, while the 4465P has 16,630 million transistors on a 2x 70.6 mm² die. Despite the lower transistor count, the 4465P achieves higher clock-for-clock efficiency in some tasks due to the Zen 5 architecture.

Memory support is DDR5 for both, with dual-channel buses, but the 4465P has a slight edge in bandwidth at 89.6 GB/s versus 83.2 GB/s. Both have integrated Radeon Graphics, and neither has an unlocked multiplier. The 4484PX has a base clock of 4.40 GHz and a TDP of 120 W, while the 4465P has a base clock of 3.40 GHz and a TDP of just 65 W.

The Verdict

The data clearly favors the AMD EPYC 4484PX for most multi-threaded and cache-dependent server workloads. Its win count of 13 out of 17 benchmarks, including a 35.4% lead in physics and a 13.4% lead in encryption, makes it the stronger choice for database, virtualization, and data processing tasks where large caches matter. The 4484PX also holds the overall average benchmark lead, 67822 versus 66925, and a higher percentile ranking of 94 versus 93.

However, the AMD EPYC 4465P is the better option for single-threaded performance and math-heavy computation. Its 10% lead in PassMark single-thread and 8.9% lead in floating point math are substantial. For workloads that are not cache-limited and rely on raw arithmetic speed, such as certain scientific simulations or financial modeling, the 4465P’s Zen 5 cores provide a measurable advantage. Additionally, its significantly lower TDP of 65 W versus 120 W makes it the more efficient choice for power-constrained environments.

Pick the 4484PX if your priority is maximum cache size, encryption throughput, or physics-based simulation performance. Pick the 4465P if you value single-thread speed, floating point performance, or lower power consumption over the 4484PX’s cache advantages.

Specification Differences

• Architecture: AMD EPYC 4484PX is Zen 4 (Raphael); AMD EPYC 4465P is Zen 5 (Grado).
• Process Node: AMD EPYC 4484PX is 5 nm; AMD EPYC 4465P is 4 nm.
• Base Clock: AMD EPYC 4484PX is 4.40 GHz; AMD EPYC 4465P is 3.40 GHz.
• Boost Clock: AMD EPYC 4484PX is 5.60 GHz; AMD EPYC 4465P is 5.40 GHz.
• TDP: AMD EPYC 4484PX is 120 W; AMD EPYC 4465P is 65 W.
• L1 Cache: AMD EPYC 4484PX is 64 KB (per core); AMD EPYC 4465P is 80 KB (per core).
• L3 Cache: AMD EPYC 4484PX is 128 MB (shared); AMD EPYC 4465P is 64 MB (shared).
• 3D V-Cache: AMD EPYC 4484PX has 1x 64MB Slice; AMD EPYC 4465P has none.
• Transistors: AMD EPYC 4484PX is 17,840 million; AMD EPYC 4465P is 16,630 million.
• Die Size: AMD EPYC 4484PX is 2x 71 mm²; AMD EPYC 4465P is 2x 70.6 mm².
• Memory Bandwidth: AMD EPYC 4484PX is 83.2 GB/s; AMD EPYC 4465P is 89.6 GB/s.
• PCIe Lanes: AMD EPYC 4484PX is Gen 5, 28 Lanes; AMD EPYC 4465P is Gen 5, 24 Lanes.
• Release Date: AMD EPYC 4484PX is 2024-05-20; AMD EPYC 4465P is 2025-05-12.
• Launch MSRP: AMD EPYC 4484PX is $599; AMD EPYC 4465P is $399.

Where Each One Wins

The AMD EPYC 4484PX wins in all Cinebench tests, PassMark data compression, data encryption, extended instructions, find prime numbers, multithread, physics, and random string sorting. This makes it the go-to choice for rendering, data compression pipelines, encryption-heavy security workloads, and any task that benefits from its 128 MB of L3 cache and 3D V-Cache slice. Its 1.4% lead in PassMark multithread and 6% lead in random string sorting reinforce its suitability for heavily threaded server applications.

The AMD EPYC 4465P wins in PassMark floating point math, integer math, and single-thread (both PassMark single_thread and singlethread tests). Its 10% single-thread advantage is the largest single-core delta between the two, making it the better pick for lightly threaded applications, real-time processing, or any workload where per-core speed is the bottleneck. The 8.9% floating point win and 6.6% integer win also make it the superior choice for scientific computing and numerical analysis that relies on arithmetic throughput rather than cache residency.