AMD Ryzen Threadripper 2920X vs Intel Core i9-10900X Comparison

The AMD Ryzen Threadripper 2920X and Intel Core i9-10900X occupy the same desktop high-end bracket, and the database shows them landing in an identical percentile (61) against all recorded CPUs, with average benchmark scores of 5730 for the Threadripper and 5555 for the Core i9. That broad similarity conceals a sharp split in behavior: the Threadripper sweeps every Cinebench test, while the Core i9-10900X takes both Geekbench tests by a wide margin. The recorded data makes this a workload-dependent matchup rather than a one-sided contest.

Head-to-Head Benchmarks

The Cinebench results are strikingly consistent. In Cinebench R15 multi-core the Threadripper 2920X scores 2154 against 1910, a 12.8% win. Cinebench R15 single-core goes the same way, 304 to 269, a 13% advantage. Cinebench R20 multi-core reads 8979 versus 7961, and R20 single-core reads 1267 versus 1123, both 12.8% in AMD's favor. Cinebench R23 repeats the pattern: 21380 to 18957 in multi-core and 3018 to 2676 in single-core, again 12.8% apart. Across all six Cinebench tests the gap sits at roughly 13%, which indicates the Threadripper's two extra cores and four extra threads translate directly into rendering throughput, and that its per-core output is also ahead in these tests despite the Intel part's higher boost clock.

Geekbench flips the script. In Geekbench multi-core the Core i9-10900X scores 10104 against the Threadripper's 7473, a 26% lead for Intel. Geekbench single-core also goes to Intel, 1438 to 1266, a 12% margin. That -26% multi-core figure is the single largest delta in the entire head-to-head set, and it is unusual: the chip with fewer cores wins a multi-core test decisively. The data suggests Geekbench's workload characteristics favor the Cascade Lake architecture's per-thread performance strongly enough to overcome the Threadripper's 12-versus-10 core advantage.

The tally is six wins for AMD and two for Intel, but those two Intel wins are the biggest percentage margins on the board.

Architecture Differences

The two chips come from opposite design philosophies. The Threadripper 2920X is a Zen+ (Colfax) design on GlobalFoundries' 12 nm node, with 9600 million transistors spread across two dies of 213 mm² each. The Core i9-10900X is a Cascade Lake-X design built by Intel on its 14 nm node; the database records no transistor count or die size for it. The Intel part is a monolithic-style approach with a higher base clock of 3.70 GHz versus 3.50 GHz and a higher boost clock of 4.70 GHz versus 4.30 GHz, yet it posts lower single-core scores in every Cinebench generation. That is the most interesting architectural signal in the data: clock speed alone does not determine per-core output here.

Core and cache topology diverge sharply. AMD offers 12 cores and 24 threads with 96 KB of L1 and 512 KB of L2 per core plus 32 MB of L3. Intel offers 10 cores and 20 threads with 64 KB of L1 per core, a larger 1 MB of L2 per core, and 19.25 MB of shared L3. AMD doubles down on L3 capacity; Intel compensates with more L2 per core.

Platform features favor AMD on expansion. The Threadripper provides PCIe Gen 3 with 60 lanes from the CPU, while the Core i9-10900X lists PCIe Gen 3 without a lane count in the database. Both support quad-channel DDR4, both lack integrated graphics, and neither supports ECC memory. AMD records a memory bandwidth figure of 93.9 GB/s; no comparable figure is recorded for the Intel part. The Threadripper fits Socket SP3r2, the Intel chip fits Socket 2066, so the two are not platform-compatible. TDP differs modestly: 180 W for AMD, 165 W for Intel. Both have unlocked multipliers. AMD's launch MSRP was $649; no launch MSRP is recorded for the Core i9-10900X.

Where Each One Wins

The Threadripper 2920X wins every rendering scenario the database covers. All six Cinebench results, spanning R15, R20, and R23 in both single-core and multi-core modes, favor it by roughly 13%. Anyone whose primary workload is 3D rendering, or anything that behaves like Cinebench's sustained multi-threaded compute, gets measurably more throughput from the AMD chip. Its 24 threads, 32 MB of L3, and 60 PCIe lanes also make it the natural fit for expansion-heavy workstations with many add-in cards.

The Core i9-10900X wins in the Geekbench profile. Its 26% multi-core and 12% single-core Geekbench leads indicate that general-purpose mixed workloads, of the kind Geekbench samples, run substantially faster on Cascade Lake. Its lower TDP of 165 W is another recorded advantage, drawing less power at the package level than the Threadripper's 180 W while still delivering those Geekbench wins.

Context from the rivals list reinforces how close these two sit overall. The Threadripper's nearest rivals include the Intel Xeon W-2175 (avgScore 5770, 0.7% ahead of it), the AMD EPYC 7351 (5710, 0.4% behind), the AMD EPYC 7F32 (5703, 0.5% behind), and the Intel Core i9-7920X (5673, 1% behind). The Core i9-10900X's rivals cluster even tighter, from the Intel Celeron G6900 (5561, within 0.1%) to the Intel Core i7-12700T (5606, 0.9% ahead). Both chips sit in the same crowded performance band despite the very different head-to-head results on individual tests.

FAQ

Q: Which CPU is faster in Cinebench R23 multi-core?

A: The Threadripper 2920X, scoring 21380 versus 18957, a 12.8% lead over the Core i9-10900X.

Q: Which CPU wins in Geekbench?

A: The Core i9-10900X wins both Geekbench tests: 10104 to 7473 in multi-core (26%) and 1438 to 1266 in single-core (12%).

Q: How many cores and threads does each chip have?

A: The Threadripper 2920X has 12 cores and 24 threads; the Core i9-10900X has 10 cores and 20 threads.

Q: Do these CPUs use the same socket?

A: No. The Threadripper uses AMD Socket SP3r2 and the Core i9-10900X uses Intel Socket 2066.

Q: Which chip has more L3 cache?

A: The Threadripper 2920X, with 32 MB versus 19.25 MB shared on the Core i9-10900X. Intel does hold more L2 per core, 1 MB versus 512 KB.

Q: Do they support ECC memory?

A: No, the database records ECC memory as unsupported on both.

Specification Differences

• Cores/threads: 12/24 (AMD) versus 10/20 (Intel)
• Base clock: 3.50 GHz (AMD) versus 3.70 GHz (Intel)
• Boost clock: 4.30 GHz (AMD) versus 4.70 GHz (Intel)
• TDP: 180 W (AMD) versus 165 W (Intel)
• Socket: AMD Socket SP3r2 versus Intel Socket 2066
• Architecture: Zen+ (Colfax) versus Cascade Lake (Cascade Lake-X)
• Process node: 12 nm at GlobalFoundries versus 14 nm at Intel
• L1 cache: 96 KB per core (AMD) versus 64 KB per core (Intel)
• L2 cache: 512 KB per core (AMD) versus 1 MB per core (Intel)
• L3 cache: 32 MB (AMD) versus 19.25 MB shared (Intel)
• Memory bandwidth: 93.9 GB/s recorded for AMD; not recorded for Intel
• PCIe: Gen 3 with 60 CPU lanes (AMD) versus Gen 3 (Intel)
• Release date: 2018-10-02 (AMD) versus 2019-10-18 (Intel)
• Transistors/die size: 9600 million across 2x 213 mm² dies (AMD); not recorded (Intel)
• Part number: YD292XA8UC9AF (AMD); not recorded (Intel)

Both support DDR4 on a quad-channel bus, both lack integrated graphics, both have unlocked multipliers, and both are listed as Active desktop parts in the same 61st percentile.

The Verdict

The recorded data points to a workload split. If the job looks like Cinebench, meaning sustained multi-threaded rendering, the Threadripper 2920X is the clear pick: it wins all six Cinebench tests by about 13%, and it brings 24 threads, twice the L3 capacity, 60 PCIe lanes, and a recorded 93.9 GB/s of memory bandwidth for workstation duty. If the job looks like Geekbench, meaning mixed general-purpose compute, the Core i9-10900X is the stronger choice, winning multi-core by 26% and single-core by 12% while drawing a lower 165 W TDP. Six of the eight head-to-head tests go to AMD, but the two Intel wins are the largest margins recorded. Buyers should match the chip to their dominant workload, because on aggregate the two are nearly even, separated by just 175 points in average benchmark score.