AMD EPYC 9384X vs AMD Ryzen Threadripper PRO 5965WX Comparison

FAQ

Q: Which processor has more cores?

A: The AMD EPYC 9384X has 32 cores and 64 threads, while the AMD Ryzen Threadripper PRO 5965WX has 24 cores and 48 threads.

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

A: The Ryzen Threadripper PRO 5965WX wins in PassMark single-thread testing with a score of 3337 versus 3015, a 9.6% advantage. However, the EPYC 9384X wins all Cinebench single-core tests by 5% each.

Q: Which chip has the larger L3 cache?

A: The EPYC 9384X features 768 MB of shared L3 cache, versus 128 MB on the Ryzen Threadripper PRO 5965WX.

Q: What are the memory specifications for each?

A: The EPYC 9384X uses DDR5 with a twelve-channel memory bus and 460.8 GB/s bandwidth. The Ryzen Threadripper PRO 5965WX uses DDR4 with an eight-channel bus and 204.8 GB/s bandwidth.

Q: How many benchmark wins does each processor claim?

A: The EPYC 9384X wins 15 of the 17 head-to-head benchmark comparisons, while the Ryzen Threadripper PRO 5965WX wins 2.

Q: What is the launch MSRP for each?

A: The EPYC 9384X had a launch MSRP of $5529, and the Ryzen Threadripper PRO 5965WX had a launch MSRP of $2399.

Architecture Differences

The EPYC 9384X belongs to the EPYC 9004 series built on the Zen 4 architecture with the Genoa-X codename, produced on TSMC's 5 nm process. It integrates 90,160 million transistors across a die configuration of 8x 72 mm². The Ryzen Threadripper PRO 5965WX is a 5000 series part based on Zen 3 with the Chagall PRO codename, manufactured on TSMC's 7 nm process with 16,600 million transistors across 4x 81 mm² dies.

The cache hierarchy diverges significantly. Both processors share 64 KB of L1 cache per core, but the EPYC 9384X has 1 MB of L2 per core versus 512 KB on the Ryzen Threadripper. The L3 cache difference is the most pronounced: 768 MB shared on the Genoa-X part versus 128 MB on the Chagall PRO part. This massive L3 allocation is characteristic of the 3D V-Cache equipped Genoa-X line and directly influences workloads that depend on large resident datasets.

Memory architecture also differs at the platform level. The EPYC 9384X supports DDR5 memory through a twelve-channel bus with a theoretical bandwidth of 460.8 GB/s. The Ryzen Threadripper PRO 5965WX pairs with DDR4 through an eight-channel bus at 204.8 GB/s. Both support ECC memory and provide 128 PCIe lanes from the CPU, but the EPYC uses Gen 5 while the Threadripper uses Gen 4.

The sockets are incompatible: the EPYC 9384X mounts in AMD Socket SP5, while the Ryzen Threadripper PRO 5965WX uses AMD Socket WRX8. Clock behavior differs as well, with the EPYC listing a 3.10 GHz base and 3.90 GHz boost, while the Threadripper runs a 3.80 GHz base and 4.50 GHz boost. The EPYC carries a 320 W TDP, and the Threadripper is rated at 280 W. Neither processor has an unlocked multiplier.

Where Each One Wins

The EPYC 9384X dominates the head-to-head record with 15 wins. Its advantages are largest in physics simulation, where it posts a 119.5% higher score than the Threadripper, and in random string sorting, where it leads by 20.3%. Data encryption shows a 15.1% lead, prime number finding a 14% lead, and floating-point math a 10.7% lead. These results point to server-oriented workloads that exploit the combination of 32 Zen 4 cores, the 768 MB L3 cache, and the higher memory bandwidth of DDR5 with twelve channels. The EPYC also wins every Cinebench R15, R20, and R23 test, both multi-core and single-core, by 5% each.

The Ryzen Threadripper PRO 5965WX wins only the PassMark single-thread and singlethread tests, scoring 3337 versus 3015, a 9.6% margin. This reflects its higher boost clock of 4.50 GHz versus 3.90 GHz on the EPYC. For workloads that are lightly threaded and sensitive to peak frequency, the Threadripper holds a measurable edge.

The split is clear: the EPYC 9384X suits heavily parallel, cache-hungry, memory-bandwidth-intensive server workloads, while the Ryzen Threadripper PRO 5965WX suits single-threaded tasks where raw clock speed dominates. The EPYC's physics score of 9332 versus 4251 is the single largest relative gap in the comparison, suggesting that the EPYC's architecture provides a substantial benefit in simulation-style floating-point workloads. Neither chip offers integrated graphics, so both require a discrete GPU.

Specification Differences

The two processors differ across nearly every major specification category. Core count: the EPYC 9384X has 32 cores and 64 threads; the Threadripper has 24 cores and 48 threads. Clock speeds: the EPYC runs at 3.10 GHz base and 3.90 GHz boost, while the Threadripper runs at 3.80 GHz base and 4.50 GHz boost. TDP: 320 W versus 280 W.

Cache: L1 is identical at 64 KB per core, but L2 differs at 1 MB per core versus 512 KB per core, and L3 differs massively at 768 MB shared versus 128 MB. Process technology: 5 nm versus 7 nm, both from TSMC. Transistor count: 90,160 million versus 16,600 million. Die configuration: 8x 72 mm² versus 4x 81 mm².

Memory support: DDR5 versus DDR4. Memory bus: twelve-channel versus eight-channel. Memory bandwidth: 460.8 GB/s versus 204.8 GB/s. PCIe generation: Gen 5 versus Gen 4, both with 128 lanes from the CPU. Socket: SP5 versus WRX8. Architecture generation: Zen 4 (Genoa-X) versus Zen 3 (Chagall PRO). Release dates: the EPYC launched 2023-06-12, and the Threadripper launched 2022-03-07. The Threadripper has a listed part number of 100-000000446, while the EPYC has none recorded.

Head-to-Head Benchmarks

The head-to-head results show a consistent pattern: the EPYC 9384X wins every Cinebench test by exactly 5%. In Cinebench R15 multi-core, it scores 5968 to 5685. In R15 single-core, it scores 842 to 802. R20 multi-core shows 24870 to 23688, and R20 single-core shows 3510 to 3344. R23 multi-core shows 59215 to 56400, and R23 single-core shows 8359 to 7962. These margins are uniform across both multi-threaded and single-threaded Cinebench variants, indicating a consistent architectural advantage in this rendering workload despite the Threadripper's higher boost clock.

PassMark results reveal larger and more varied gaps. The EPYC's biggest victory is in physics, where it scores 9332 against 4251, a 119.5% lead. This is the standout result of the entire comparison and suggests the EPYC's 32 cores and 768 MB L3 cache deliver exceptional performance in physics simulation. Random string sorting shows a 20.3% gap, with scores of 119440 versus 99287. Data encryption is 15.1% higher at 72631 versus 63113. Prime number finding is 14% higher at 596 versus 523. Floating-point math is 10.7% higher at 174630 versus 157708. Data compression is 10.9% higher at 1119983 versus 1010067. Extended instructions are 10.5% higher at 74363 versus 67300. Integer math is 5.9% higher at 297833 versus 281247. PassMark multi-thread shows a 5% lead at 69665 versus 66353.

The Ryzen Threadripper PRO 5965WX takes the PassMark single-thread test with 3337 versus 3015, a 9.6% margin, and the duplicate singlethread test with the same scores. This is the only category where the Threadripper's higher boost clock translates into a win. The average benchmark score in the database reinforces the overall picture: the EPYC 9384X averages 120427, while the Threadripper averages 98504. Both sit at the 97th percentile among all CPUs.

The EPYC's nearest rivals in the database include the Intel Xeon w7-3565X at 118307 (1.8% behind the EPYC), the AMD EPYC 7642 at 124006 (2.9% ahead), the AMD Ryzen Threadripper PRO 5975WX at 124171 (3% ahead), and the AMD EPYC 9255 at 116388 (3.5% behind). The Threadripper's nearest rivals include the AMD EPYC 4585PX at 99324 (0.8% ahead), the AMD Ryzen 9 9955HX3D at 97453 (1.1% behind), the AMD Ryzen 9 PRO 9945 at 96083 (2.5% behind), and the AMD Ryzen Threadripper PRO 9955WX at 101041 (2.5% ahead). These placement data show that the EPYC 9384X competes in a higher average performance tier than the Threadripper, even though both rank at the same percentile overall.