AMD EPYC 72F3 vs Intel Core i9-10920X Comparison

Where Each One Wins

The benchmark data is unusually one-sided in this pairing. Across all six recorded Cinebench tests, the AMD EPYC 72F3 takes every single win. The Intel Core i9-10920X does not claim a single head-to-head victory in the database, whether the workload is multi-threaded or single-threaded.

For multi-threaded workloads, the EPYC 72F3 leads consistently by roughly 5.6% in every Cinebench version. In Cinebench R15 multicore, the EPYC scores 2334 against 2203 for the Intel part. That margin carries into Cinebench R20 multicore, where the EPYC posts 9728 versus 9181, and into Cinebench R23 multicore, where the EPYC reaches 23164 and the Intel chip finishes at 21861. The pattern holds regardless of the renderer version, which suggests a stable architectural advantage rather than a workload-specific quirk.

Single-threaded performance tells the same story, albeit with slightly different margins. The EPYC 72F3 wins Cinebench R15 singlecore with 329 points against 311, a 5.5% gap. In Cinebench R20 singlecore, the EPYC scores 1373 versus 1296, and in R23 singlecore it posts 3270 against 3086. The Intel chip is not competitive in any recorded metric, but the margins are moderate rather than overwhelming.

The Geekbench results are only recorded for the Intel i9-10920X in the database, so no direct comparison is possible there. The i9-10920X scores 11165 in Geekbench multicore and 1673 in singlecore. These numbers stand alone without a rival score to measure against.

The practical interpretation is straightforward. If the workload is Cinebench rendering of any kind, the EPYC 72F3 is the faster processor in every scenario tested. The Intel part does not have a single recorded win to point to. There is no "wins by workload type" split to draw; the data shows one processor dominating all tested categories.

Architecture Differences

The two processors come from fundamentally different design philosophies and manufacturing ecosystems.

The Intel Core i9-10920X is built on the Cascade Lake architecture, specifically Cascade Lake-X, using Intel's 14 nm process node at Intel's own foundry. It packs 12 cores and 24 threads. The base clock is 3.50 GHz with a boost clock of 4.80 GHz. The thermal design power is 165 watts. It uses the Intel Socket 2066 platform, targets the desktop market segment, and is classified as the Core i9 X-Series 10th Gen product line.

The AMD EPYC 72F3 is a Zen 3 part, codenamed Milan, fabricated on TSMC's 7 nm process. It has 8 cores and 16 threads, fewer cores than the Intel part, but a more advanced process node. Base clock is 3.70 GHz, boost clock is 4.10 GHz, and TDP is 180 watts. It sits on the AMD Socket SP3 platform and targets the server and workstation market segment. The EPYC 72F3 carries 33,200 million transistors spread across a die size of 8x 81 mm², and it was released in March 2021, roughly a year and a half after the Intel part's October 2019 launch.

Cache configuration is where the architectures diverge sharply. The Intel chip provides 64 KB of L1 per core, 1 MB of L2 per core, and 19.25 MB of shared L3 cache. The AMD chip also provides 64 KB of L1 per core, but only 512 KB of L2 per core, and a massive 256 MB of shared L3 cache. That 256 MB L3 is more than thirteen times larger than the Intel part's L3, and it is a defining feature of the Milan server architecture.

Memory support also differs. Both support DDR4, but the Intel chip runs a quad-channel memory bus while the AMD part runs an eight-channel memory bus. The EPYC 72F3 has a recorded memory bandwidth of 204.8 GB/s; no memory bandwidth figure is recorded for the Intel chip. The EPYC 72F3 also supports ECC memory, while the Intel i9-10920X does not.

PCIe capability is another clear split. The Intel chip offers Gen 3 PCIe, while the AMD part offers Gen 4 with 128 lanes on the CPU. The Intel chip has an unlocked multiplier; the AMD part is locked. The EPYC 72F3 has a part number recorded in the database (100-000000327100-100000327WOF), while the Intel part has none listed. Neither processor includes integrated graphics.

The process node difference of 14 nm versus 7 nm explains much of the efficiency and performance gap. The AMD part delivers higher performance across all Cinebench tests despite having only 8 cores against the Intel part's 12 cores. The combination of Zen 3's architectural improvements, the larger L3 cache, and the more advanced process node allows the EPYC to overcome a 4-core deficit.

FAQ

Q: Which processor has more cores?

A: The Intel Core i9-10920X has 12 cores and 24 threads. The AMD EPYC 72F3 has 8 cores and 16 threads. The Intel part has a 4-core advantage on paper.

Q: Why does the EPYC 72F3 win despite having fewer cores?

A: The recorded data shows the EPYC 72F3 winning all six Cinebench tests by roughly 5.5 to 5.6%. Its advantage comes from the Zen 3 architecture, the 7 nm process node, an eight-channel memory bus with 204.8 GB/s of bandwidth, and a 256 MB shared L3 cache. These factors offset the core count deficit.

Q: Does the Intel chip win any benchmark in the database?

A: No. The head-to-head benchmark records show the AMD EPYC 72F3 winning all six Cinebench tests. The Intel i9-10920X has zero recorded wins in the head-to-head comparison.

Q: Which processor supports ECC memory?

A: The AMD EPYC 72F3 supports ECC memory. The Intel Core i9-10920X does not support ECC memory according to the recorded specifications.

Q: What is the L3 cache difference?

A: The Intel Core i9-10920X has 19.25 MB of shared L3 cache. The AMD EPYC 72F3 has 256 MB of shared L3 cache, which is over thirteen times larger.

Q: Which processor has a higher boost clock?

A: The Intel Core i9-10920X has a boost clock of 4.80 GHz. The AMD EPYC 72F3 has a boost clock of 4.10 GHz. The Intel part has the higher boost clock, but the EPYC still wins single-threaded Cinebench tests.

Specification Differences

The two processors differ across nearly every recorded specification category.

Cores and threads: Intel has 12 cores and 24 threads. AMD has 8 cores and 16 threads.

Clocks: Intel runs a 3.50 GHz base and 4.80 GHz boost. AMD runs a 3.70 GHz base and 4.10 GHz boost. Intel has the higher boost, AMD has the higher base.

TDP: Intel is rated at 165 watts. AMD is rated at 180 watts.

Socket: Intel uses Intel Socket 2066. AMD uses AMD Socket SP3.

Process node: Intel uses 14 nm at its own foundry. AMD uses 7 nm at TSMC.

Transistors and die size: Intel has no transistor count or die size recorded. AMD has 33,200 million transistors and a die size of 8x 81 mm².

Cache: Intel has 64 KB L1 per core, 1 MB L2 per core, and 19.25 MB shared L3. AMD has 64 KB L1 per core, 512 KB L2 per core, and 256 MB shared L3.

Memory: Both support DDR4, but Intel runs quad-channel while AMD runs eight-channel. AMD has a recorded memory bandwidth of 204.8 GB/s; Intel has none recorded. AMD supports ECC, Intel does not.

PCIe: Intel offers Gen 3. AMD offers Gen 4 with 128 lanes on the CPU.

Market segment: Intel targets desktop. AMD targets server and workstation.

Multiplier: Intel is unlocked. AMD is locked.

Release date: Intel launched October 2019. AMD launched March 2021.

Generation: Intel is Core i9 X-Series 10th Gen. AMD is EPYC Zen 3 (Milan).

Part number: AMD has a recorded part number. Intel does not.

Average benchmark score: Intel averages 6347 across its recorded benchmarks. AMD averages 6700. Both sit at the 62nd percentile among all CPUs.

Nearest rivals: The Intel part sits near the AMD EPYC 7551 (0.7% behind), the AMD Ryzen 3 3100U (1.6% ahead), the AMD Ryzen Threadripper 1950X (1.9% ahead), and the Intel Core i7-12800HE (1.9% behind). The AMD part sits near the Intel Xeon Gold 5317 (0.1% behind), the Intel Xeon D-2775TE (0.2% behind), the Intel Core i9-9940X (0.6% ahead), and the AMD Ryzen Threadripper 2950X (0.8% ahead).

Head-to-Head Benchmarks

The head-to-head records show a clean sweep for the AMD EPYC 72F3 across all six Cinebench tests.

In Cinebench R15 multicore, the EPYC 72F3 scores 2334 against the Intel i9-10920X's 2203. The delta is 5.6% in favor of the AMD chip. The R15 singlecore test shows the EPYC at 329 and the Intel at 311, a 5.5% margin.

Cinebench R20 multicore follows the same pattern. The EPYC scores 9728, the Intel scores 9181, and the delta is again 5.6%. In R20 singlecore, the EPYC posts 1373 against 1296, a 5.6% margin.

Cinebench R23 multicore has the EPYC at 23164 and the Intel at 21861, a 5.6% difference. R23 singlecore has the EPYC at 3270 and the Intel at 3086, also a 5.6% margin.

The consistency of these deltas is notable. Every multicore test across three Cinebench versions lands at exactly 5.6%, and every singlecore test lands at either 5.5% or 5.6%. This uniformity suggests the performance gap is structural, a product of architectural efficiency rather than a workload-dependent outcome. The Intel chip's higher boost clock of 4.80 GHz does not translate into single-threaded wins; the EPYC's lower 4.10 GHz boost still produces higher single-threaded scores.

The Intel chip's 12-core advantage also fails to materialize in any multicore test. Even with 50% more cores, the Intel part trails by the same 5.6% margin in every multicore workload. The EPYC's higher base clock of 3.70 GHz versus 3.50 GHz, combined with its larger cache and newer architecture, appears to compensate fully for the core deficit.

Geekbench results are only recorded for the Intel part, so no head-to-head delta can be computed there. The Intel chip's Geekbench multicore score of 11165 and singlecore score of 1673 stand without a comparison point.

The Verdict

The data points to one clear conclusion: the AMD EPYC 72F3 is the faster processor in every recorded benchmark. It wins all six Cinebench head-to-head tests with consistent margins of 5.5% to 5.6%. The Intel Core i9-10920X does not win a single test in the database.

For rendering workloads using Cinebench, the choice is unambiguous. The EPYC 72F3 delivers higher scores in both single-threaded and multi-threaded tests across R15, R20, and R23 versions. The Intel chip's 12 cores and higher boost clock do not overcome the EPYC's architectural advantages.

The EPYC 72F3 also brings platform benefits that matter for server and workstation buyers. It supports ECC memory, runs an eight-channel memory bus with 204.8 GB/s of bandwidth, offers PCIe Gen 4 with 128 lanes, and carries a 256 MB L3 cache. These are enterprise-class features. The Intel chip offers a quad-channel memory bus, PCIe Gen 3, no ECC support, and a much smaller 19.25 MB L3 cache.

The Intel i9-10920X is a desktop part with an unlocked multiplier, which may appeal to enthusiasts who want overclocking flexibility. Its 12 cores and 24 threads give it more parallel hardware on paper. But the recorded benchmark data shows that hardware advantage does not translate into performance wins against the EPYC 72F3.

The average benchmark scores reinforce the picture. The EPYC 72F3 averages 6700 across its recorded benchmarks, while the Intel chip averages 6347. Both sit at the 62nd percentile among all CPUs, but the raw average favors the AMD part by roughly 5.6%, which matches the head-to-head deltas precisely.

Buyers should pick the AMD EPYC 72F3 if they need the fastest Cinebench performance, ECC memory support, high memory bandwidth, or PCIe Gen 4 connectivity. It is the stronger choice for server and workstation workloads where those features matter. Buyers should consider the Intel Core i9-10920X only if they specifically need a desktop-platform part with an unlocked multiplier and are willing to accept lower benchmark scores across the board. The recorded data does not provide a single metric where the Intel chip comes out ahead.