AMD EPYC 7351P vs AMD Ryzen Threadripper 2920X Comparison

FAQ

Q: Which processor wins more benchmark comparisons in the database?

A: The AMD EPYC 7351P wins 6 of the 8 head-to-head tests, while the AMD Ryzen Threadripper 2920X wins the remaining 2.

Q: How do the two chips compare in average benchmark score?

A: The Ryzen Threadripper 2920X has an average benchmark score of 5730, while the EPYC 7351P scores 5469. The nearest rival data shows the Threadripper is only 0.4% ahead of the EPYC 7351 in average score.

Q: What is the biggest performance gap between the two in either direction?

A: The largest delta is in Geekbench single-core, where the Ryzen Threadripper 2920X leads by 72.7%. The largest EPYC win is 3.5% in Cinebench R15 multi-core.

Q: Do both processors share the same memory controller configuration?

A: No. The Ryzen Threadripper 2920X uses quad-channel DDR4 with 93.9 GB/s bandwidth, while the EPYC 7351P uses eight-channel DDR4 with 170.6 GB/s bandwidth.

Q: What are the core and thread counts for each part?

A: The Ryzen Threadripper 2920X has 12 cores and 24 threads. The EPYC 7351P has 16 cores and 32 threads.

Q: Which processor supports ECC memory?

A: The EPYC 7351P supports ECC memory. The Ryzen Threadripper 2920X does not list ECC support in the database.

The Verdict

The data paints a clear split: the AMD EPYC 7351P dominates the Cinebench suite across every version and both multi-core and single-core tests, while the AMD Ryzen Threadripper 2920X dominates Geekbench by massive margins. The EPYC 7351P wins 6 tests, the Threadripper 2920X wins 2.

For workloads that resemble Cinebench rendering, the EPYC 7351P is the stronger pick. Its wins range from 3.2% to 3.5% across all six Cinebench tests, which is a consistent, if modest, advantage. The EPYC also brings 16 cores versus 12, eight-channel memory versus quad-channel, and ECC support, all of which matter for sustained server or workstation compute.

For single-threaded and latency-sensitive tasks as measured by Geekbench, the Ryzen Threadripper 2920X is the clear winner. It leads by 62.2% in multi-core Geekbench and by 72.7% in single-core Geekbench. Those are enormous deltas, indicating that the Threadripper’s higher clock speeds (3.50 GHz base, 4.30 GHz boost versus 2.40 GHz base, 2.90 GHz boost) translate into dramatically better performance in that benchmark’s workload profile.

The average benchmark scores tell a nuanced story. The Threadripper 2920X averages 5730, the EPYC 7351P averages 5469, a difference of about 4.8% in favor of the Threadripper. Yet the closest rival data shows the Threadripper is only 0.4% ahead of the EPYC 7351 in that specific comparison, while the EPYC 7351P is 0.5% ahead of the Intel Xeon W-1290P and 0.8% ahead of the AMD Ryzen Threadripper 1920.

The verdict depends on the workload. The EPYC 7351P is the better choice for rendering and multi-threaded server tasks, where its extra cores, larger L3 cache, and higher memory bandwidth show up as consistent Cinebench wins. The Ryzen Threadripper 2920X is the better choice for Geekbench-style workloads, where its clock speed advantage produces overwhelming results. Buyers should match the chip to the benchmark profile that mirrors their actual use case.

Head-to-Head Benchmarks

The Cinebench results are uniformly in favor of the EPYC 7351P. In Cinebench R15 multi-core, the EPYC scores 2231 against the Threadripper’s 2154, a 3.5% lead. In R15 single-core, the EPYC scores 314 versus 304, a 3.2% lead. The pattern repeats in R20 multi-core: 9296 versus 8979, a 3.4% lead. In R20 single-core, the EPYC scores 1312 against 1267, again a 3.4% lead. Cinebench R23 multi-core shows 22135 versus 21380, a 3.4% lead, and R23 single-core shows 3125 versus 3018, also a 3.4% lead.

These six wins are consistent and narrow. The EPYC never wins by more than 3.5%, but it wins every Cinebench iteration. That consistency suggests a structural advantage in rendering workloads, likely tied to the EPYC’s 16 cores versus 12 and its 64 MB shared L3 cache versus 32 MB. The Threadripper’s higher clocks do not overcome those differences in Cinebench.

The Geekbench results flip the picture entirely. In Geekbench multi-core, the Threadripper scores 7473 against the EPYC’s 4607, a 62.2% lead. In Geekbench single-core, the Threadripper scores 1266 against the EPYC’s 733, a 72.7% lead. These are not marginal differences; they are dominant wins. The Threadripper’s boost clock of 4.30 GHz versus 2.90 GHz on the EPYC explains much of this, as Geekbench tends to reward higher frequencies and lower latency.

The overall win count is 6 to 2 in favor of the EPYC, but the magnitude of the Threadripper’s two wins is far larger than the magnitude of any single EPYC win. The EPYC’s largest delta is 3.5%, while the Threadripper’s smallest delta is 62.2%. In terms of average benchmark score, the Threadripper leads with 5730 versus 5469, which reflects the outsized Geekbench scores pulling its average up.

For reference, the Threadripper’s nearest rivals include the AMD EPYC 7351 at 0.4% behind, the AMD EPYC 7F32 at 0.5% behind, the Intel Xeon W-2175 at 0.7% ahead, and the Intel Core i9-7920X at 1% behind. The EPYC 7351P’s nearest rivals include the Intel Xeon W-1290P at 0.5% behind, the AMD Ryzen Threadripper 1920 at 0.8% behind, the Intel Core i9-10900X at 1.5% ahead, and the Intel Celeron G6900 at 1.7% ahead. These rival comparisons show both chips sit in a tight cluster of similar performers, with the Threadripper marginally ahead on average.

Specification Differences

The core counts differ substantially. The Ryzen Threadripper 2920X has 12 cores and 24 threads, while the EPYC 7351P has 16 cores and 32 threads. The Threadripper runs at a 3.50 GHz base clock and 4.30 GHz boost clock, while the EPYC runs at 2.40 GHz base and 2.90 GHz boost. The Threadripper has a 180 W TDP, the EPYC has a 170 W TDP.

The sockets are different. The Threadripper uses AMD Socket SP3r2, while the EPYC uses AMD Socket SP3. The memory configurations also differ: the Threadripper uses quad-channel DDR4 with 93.9 GB/s bandwidth, while the EPYC uses eight-channel DDR4 with 170.6 GB/s bandwidth. ECC memory support is present on the EPYC but not listed for the Threadripper.

PCIe lanes differ as well. The Threadripper lists PCIe Gen 3 with 60 lanes (CPU only), while the EPYC lists PCIe Gen 3 without a lane count. The market segments are different: the Threadripper is classified as Desktop, the EPYC as Server/Workstation. The release dates differ by over a year, with the Threadripper launching on 2018-10-02 and the EPYC on 2017-06-28.

The Threadripper has a launch MSRP of $649. The EPYC has no recorded launch MSRP. The Threadripper’s part number is YD292XA8UC9AF, while the EPYC’s is PS735PBEVGPAF. Both have unlocked multipliers.

Architecture Differences

The two processors come from different architectural generations. The Ryzen Threadripper 2920X is based on Zen+ with the codename Colfax, while the EPYC 7351P is based on the original Zen with the codename Naples. This generation gap shows up in the process node: the Threadripper is built on a 12 nm process, while the EPYC uses a 14 nm process. Both are fabricated by GlobalFoundries.

The transistor counts differ significantly. The Threadripper packs 9,600 million transistors across a dual-die design with each die measuring 213 mm², for a total of 2x 213 mm². The EPYC uses a single 213 mm² die with 4,800 million transistors. The Threadripper’s die size is effectively double the EPYC’s, which reflects its higher transistor budget.

Cache configurations differ in capacity and organization. Both have 96 KB L1 per core and 512 KB L2 per core. The L3 cache differs: the Threadripper has 32 MB total, while the EPYC has 64 MB shared. This gives the EPYC a 2x advantage in L3 capacity, which helps in multi-threaded workloads that share data.

The EPYC’s memory subsystem is substantially wider. Its eight-channel memory controller with 170.6 GB/s bandwidth is nearly double the Threadripper’s quad-channel 93.9 GB/s. This is a major architectural differentiator for memory-bound server workloads. The EPYC also supports ECC memory, which the Threadripper does not, making the EPYC more suitable for error-sensitive compute environments.

The generation gap also means the Threadripper benefits from the Zen+ refresh, which typically improves clocks and memory latency over the original Zen. The benchmark data supports this: the Threadripper’s Geekbench scores are vastly higher, despite the EPYC having more cores. The architecture difference, rather than core count, appears to drive the Geekbench outcome.

Where Each One Wins

The EPYC 7351P wins in rendering and multi-threaded compute as measured by Cinebench. It takes all six Cinebench tests, with leads of 3.2% to 3.5%. The EPYC’s 16 cores, 64 MB shared L3, and eight-channel memory bandwidth give it an edge in workloads that scale with core count and memory throughput. The EPYC also wins on ECC support and server-class memory bandwidth, making it the appropriate choice for data integrity-focused environments.

The Ryzen Threadripper 2920X wins in Geekbench, both multi-core and single-core, by margins of 62.2% and 72.7% respectively. These are the largest deltas in the entire comparison. The Threadripper’s higher base and boost clocks, 3.50 GHz and 4.30 GHz versus 2.40 GHz and 2.90 GHz, are the likely drivers. For workloads that favor high clock speeds and lower per-core latency, the Threadripper is the clear winner.

The average benchmark score favors the Threadripper, 5730 versus 5469, but this is heavily influenced by the Geekbench outliers. The nearest rival data shows both chips sit within a tight competitive cluster. The Threadripper is 0.4% ahead of the EPYC 7351 and 1% ahead of the Intel Core i9-7920X, while the EPYC 7351P is 0.5% behind the Intel Xeon W-1290P and 0.8% ahead of the AMD Ryzen Threadripper 1920.

Use-case split: choose the EPYC 7351P for Cinebench-style rendering, multi-threaded server workloads, memory-heavy tasks, and environments requiring ECC. Choose the Ryzen Threadripper 2920X for Geekbench-style latency-sensitive applications, high-frequency single-threaded tasks, and desktop-oriented compute where clock speed matters more than core count. The EPYC wins the majority of tests, but the Threadripper wins the two tests by overwhelming margins, so the correct choice depends entirely on which benchmark profile matches the intended workload.