AMD Ryzen Threadripper 9970X vs AMD Ryzen Threadripper PRO 9985WX Comparison

The AMD Ryzen Threadripper PRO 9985WX is the definitive multi-core champion, winning 9 out of 11 head-to-head benchmarks against the AMD Ryzen Threadripper 9970X, often by margins exceeding 70%. The 9970X, however, secures a narrow but real victory in single-threaded workloads, making it the better choice for latency-sensitive tasks that rely on only one or two cores. This is a clear split: the PRO 9985WX for massive parallel throughput, the 9970X for raw single-core responsiveness.

Where Each One Wins

The benchmark data paints a stark picture of workload specialization. The PRO 9985WX is the undisputed king of parallel processing. Its 64 cores and 128 threads provide a massive scaling advantage in every multithreaded test. In integer math, it scores 872,710 versus 465,378 for the 9970X, a staggering 87.5% advantage. Floating-point math shows a similar story, with the PRO part leading 553,348 to 309,719, a 78.7% gap. This is the silicon for rendering farms, scientific simulations, and any workload that can saturate 128 threads.

The PRO 9985WX also dominates in memory-intensive and I/O-heavy tasks. Its eight-channel memory bus provides 409.6 GB/s of bandwidth, double the 204.8 GB/s of the 9970X's quad-channel design. This translates directly to a 65.7% lead in data compression (2,912,972 vs 1,757,998) and a 71.9% lead in random string sorting (329,014 vs 191,445). The physics test is the most lopsided of all, with the PRO part scoring 13,783 versus the 9970X's 6,835, a 101.7% advantage that underscores its dominance in complex, multi-body simulations.

The 9970X wins exactly two head-to-head tests, both measuring single-threaded performance. It scores 4,530 in the Passmark single-thread test, edging out the PRO 9985WX's 4,482. This 1.1% advantage, while small, is consistent and repeatable across the two related test fields. This suggests that the 9970X's lower core count allows for slightly higher sustained single-core clocks or lower latency, making it the preferable choice for legacy software, game physics on a single thread, or other code paths that cannot utilize more than a few cores.

The Verdict

Choose the AMD Ryzen Threadripper PRO 9985WX if your primary metric is throughput per second. The data shows it is, on average, 14.6% faster than the 9970X across all benchmark categories (average benchmark score of 320,749 vs 279,778). More importantly, it wins every single multithreaded test by a minimum of 39.7% (in the multithread test) and often by more than 80%. It also holds a 99th percentile ranking among all CPUs, placing it in the top tier of available processors. Its 128 PCIe Gen 5 lanes and eight-channel memory also make it the superior platform for multi-GPU and high-bandwidth storage configurations.

Pick the AMD Ryzen Threadripper 9970X if your workflow is dominated by single-threaded performance and you do not need the extreme core count. Its 1.1% lead in the single-thread benchmark is the only performance category where it wins, but it is a legitimate one. It also offers a lower barrier to entry with a launch MSRP of $2499 compared to the PRO 9985WX's $7999, though you must consider that the PRO part's 80 additional PCIe Gen 5 lanes and double memory bandwidth justify its position as a top-tier workstation part. The 9970X still holds a 99th percentile ranking, so it is by no means slow, but its 32 cores are better suited for a high-end desktop than a multi-socket server rack.

Head-to-Head Benchmarks

The most significant victories for the PRO 9985WX are in compute-heavy tasks. The 87.5% delta in integer math (872,710 vs 465,378) is the largest of any test and highlights the raw power of having double the cores. Following closely is the 85% delta in finding prime numbers (1,138 vs 615), which is a pure test of integer throughput and core scaling. The floating-point math test, with a 78.7% advantage (553,348 vs 309,719), confirms that this scaling advantage extends to scientific and engineering workloads. The physics test result is even more pronounced, with a 101.7% delta (13,783 vs 6,835), meaning the PRO part is more than twice as fast in this specific simulation.

In memory and data movement tasks, the PRO 9985WX also shines. Data encryption shows a 78.4% advantage (154,824 vs 86,765), likely benefiting from both the extra cores and the wider memory bus. Data compression is 65.7% faster (2,912,972 vs 1,757,998), and random string sorting is 71.9% faster (329,014 vs 191,445). The Passmark multithread score, a general indicator of parallel performance, is 39.7% higher for the PRO part (150,071 vs 107,399), which is a substantial lead but smaller than the more focused tests, suggesting some workloads do not scale perfectly with core count.

The 9970X's wins are narrow. In the Passmark single-thread test, it scores 4,530 against the PRO 9985WX's 4,482. This 1.1% delta is the only positive result for the 32-core part. This indicates that the PRO 9985WX's architecture, with its larger 256 MB L3 cache and 8 CCDs, does not incur a significant penalty in lightly-threaded tasks, but the 9970X is still measurably quicker. The 9970X also has a higher base clock of 4.00 GHz versus 3.20 GHz for the PRO part, which likely contributes to this single-core edge despite both chips boosting to the same 5.40 GHz.

FAQ

Q: Which processor is better for multithreaded rendering?

A: The AMD Ryzen Threadripper PRO 9985WX is overwhelmingly better. It wins the multi-core Cinebench tests with scores like 132,859 in R23, and in the head-to-head Passmark multithread test, it leads by 39.7% (150,071 vs 107,399). Its 64 cores and 128 threads provide the necessary scaling for this workload.

Q: Is the AMD Ryzen Threadripper 9970X faster in any way?

A: Yes, in single-threaded performance. It scores 4,530 in the Passmark single-thread test, which is 1.1% higher than the PRO 9985WX's 4,482. This is the only benchmark category where the 9970X wins.

Q: How do these two CPUs compare to other high-end processors?

A: The PRO 9985WX has an average score of 320,749, placing it just 0.3% behind the AMD Ryzen Threadripper 9980X and 1.7% ahead of the Intel Xeon 6781P. The 9970X, with an average score of 279,778, is 0.2% behind the Intel Xeon 6780E and 2.5% ahead of the AMD EPYC 9555P. Both are in the 99th percentile of all CPUs.

Q: What is the difference in memory bandwidth?

A: The PRO 9985WX supports eight-channel memory, providing 409.6 GB/s of bandwidth. The 9970X uses quad-channel memory, which halves that to 204.8 GB/s. This is a major factor in the PRO part's lead in memory-intensive benchmarks.

Q: Do both processors support ECC memory?

A: Yes, both the AMD Ryzen Threadripper PRO 9985WX and the AMD Ryzen Threadripper 9970X support ECC memory. This is a critical feature for workstation and server stability.

Q: Are these processors unlocked for overclocking?

A: Yes, both parts have an unlocked multiplier, allowing for overclocking. They also share the same 350 W TDP and the same AMD Socket sTR5.

Architecture Differences

Both processors are built on the same Zen 5 architecture and codename "Shimada Peak", using TSMC's 4 nm process node. The fundamental difference lies in the physical construction. The PRO 9985WX uses a configuration of 8 CCDs (Core Complex Dies), each with a die size of 70.6 mm², for a total of 66,520 million transistors. The 9970X, in contrast, uses only 4 of these same 70.6 mm² CCDs, totaling 33,260 million transistors. This is why the PRO part has twice the L3 cache (256 MB vs 128 MB) and double the core count. The L1 and L2 caches are identical per core at 64 KB and 1 MB, respectively, meaning the per-core cache latency and bandwidth are the same; the PRO part simply has twice as many cores.

Specification Differences

The two processors differ in almost every major specification category. The PRO 9985WX has 64 cores and 128 threads, while the 9970X has 32 cores and 64 threads. The base clock differs, with the 9970X running at 4.00 GHz compared to the PRO part's 3.20 GHz, though both boost to the same 5.40 GHz. The most significant platform differences are the memory and I/O. The PRO 9985WX features eight-channel memory, delivering 409.6 GB/s of bandwidth, and offers 128 PCIe Gen 5 lanes. The 9970X is limited to quad-channel memory at 204.8 GB/s and 80 PCIe Gen 5 lanes. They also target different market segments, with the PRO part aimed at Server/Workstation and the 9970X at Desktop. Their release dates are a week apart, July 22, 2025, for the PRO and July 29, 2025, for the 9970X. The PRO 9985WX has a launch MSRP of $7999, while the 9970X is $2499.