AMD EPYC 7352 vs Intel Core Ultra 9 275HX Comparison

AMD
AMD

AMD EPYC 7352

CORE STATE Rome
CORE SPECS 24 Cores / 48 Threads
CLOCK SPEED 2.3 Base / 3.2 GHz Turbo
CACHE 32 MB (per die)
MAX TDP 155W
ARCHITECTURE Zen 2
nm
PROCESS 7 nm
LAUNCH DATE 2019
VS
Intel
INTEL

Core Ultra 9 275HX

CORE STATE Arrow Lake-HX
CORE SPECS 24 Cores / 24 Threads
CLOCK SPEED 2.7 Base / 5.4 GHz Turbo
CACHE 36 MB (shared)
MAX TDP 55W
ARCHITECTURE Arrow Lake
nm
PROCESS 3 nm
LAUNCH DATE 2025

PERFORMANCE BENCHMARKS

cinebench_cinebench_r15_multicore
3,458
5,619.5
cinebench_cinebench_r15_singlecore
488
334
cinebench_cinebench_r20_multicore
14,411
19,899
cinebench_cinebench_r20_singlecore
2,034
2,809
cinebench_cinebench_r23_multicore
34,314
35,589
cinebench_cinebench_r23_singlecore
4,844
2,204
passmark_data_compression
660,712
608,381
passmark_data_encryption
44,426
47,112
passmark_extended_instructions
40,203
47,016
passmark_find_prime_numbers
301
448
passmark_floating_point_math
87,969
191,186
passmark_integer_math
148,605
155,218
passmark_multithread
40,370
55,759
passmark_physics
2,688
3,338
passmark_random_string_sorting
69,231
74,320
passmark_single_thread
1,979
4,713
passmark_singlethread
1,979
4,713
geekbench_multicore
N/A
20,795
geekbench_singlecore
N/A
2,458

Analysis: AMD EPYC 7352 vs Intel Core Ultra 9 275HX

The AMD EPYC 7352 and Intel Core Ultra 9 275HX represent opposite ends of the computing spectrum, yet both achieve the 94th percentile in overall CPU rankings. The EPYC 7352 is a 24-core, 48-thread Zen 2 server processor designed for Socket SP3, while the Core Ultra 9 275HX is a 24-core, 24-thread Arrow Lake mobile chip on BGA 2114. Their average benchmark scores are nearly identical — 68118 for the AMD part versus 67469 for the Intel part — but the distribution of wins is decisively lopsided: the Intel chip wins 14 of 17 head-to-head tests, while the AMD chip takes only 3. This split reflects fundamentally different architectural priorities rather than a simple performance gap.

Where Each One Wins

The AMD EPYC 7352 claims victory in exactly three benchmark categories: Cinebench R15 single-core, Cinebench R23 single-core, and PassMark data compression. The single-core wins are striking because they are not marginal — the EPYC leads by 46.1% in R15 single-core (488 vs 334) and by a massive 119.8% in R23 single-core (4844 vs 2204). This suggests that for legacy single-threaded workloads or applications that rely on older instruction paths, the Zen 2 architecture retains an edge that the newer Arrow Lake design does not match. The data compression win (660712 vs 608381, an 8.6% delta) indicates the EPYC’s 128 MB total L3 cache and 48 threads provide a tangible benefit for compression algorithms that scale with cache capacity and thread count.

The Intel Core Ultra 9 275HX dominates everything else, including all modern multi-core rendering tests, math workloads, and single-threaded PassMark tests. Its wins range from a modest 3.6% margin in Cinebench R23 multi-core (35589 vs 34314) to a crushing 54% lead in PassMark floating-point math (191186 vs 87969) and a 58% advantage in PassMark single-thread (4713 vs 1979). The Intel part also wins Cinebench R15 multi-core by 38.5% (5619.5 vs 3458), Cinebench R20 multi-core by 27.6% (19899 vs 14411), and PassMark multi-thread by 27.6% (55759 vs 40370). The pattern is clear: any workload that leverages high clock speeds, modern instruction sets, or floating-point throughput favors the Intel part decisively.

Architecture Differences

The two processors are built on fundamentally different foundations. The EPYC 7352 uses TSMC’s 7 nm process with 15,200 million transistors spread across four 74 mm² dies, while the Core Ultra 9 275HX uses TSMC’s 3 nm process with 17,800 million transistors on a single 243 mm² die. The AMD part is a server-class design with 24 cores and 48 threads via simultaneous multithreading, whereas the Intel part has 24 cores and 24 threads — no SMT — which explains why the EPYC can match or beat the Intel chip in thread-scaled compression despite lower clock speeds.

Clock behavior diverges sharply. The EPYC 7352 runs at a 2.30 GHz base and 3.20 GHz boost, while the Core Ultra 9 275HX starts at 2.70 GHz and boosts to 5.40 GHz. That 2.2 GHz boost advantage is the primary driver of the Intel part’s single-thread PassMark dominance. Cache hierarchies are also dissimilar: the EPYC provides 96 KB L1 and 512 KB L2 per core, plus 32 MB L3 per die for a 128 MB total, whereas the Intel part offers 192 KB L1 and 3 MB L2 per core, with 36 MB shared L3. The Intel L2 cache is six times larger per core, which helps latency-sensitive workloads.

Memory and I/O configurations reinforce their market roles. The EPYC 7352 supports DDR4 over an eight-channel bus with 204.8 GB/s bandwidth and ECC memory, plus 128 PCIe Gen 4 lanes. The Core Ultra 9 275HX uses dual-channel DDR5 with 102.4 GB/s bandwidth, no ECC, and 20 PCIe Gen 5 lanes. The EPYC also integrates no graphics, while the Intel part includes Arc Xe-LPG Graphics 64EU. Power envelopes are equally divergent: the EPYC is rated at 155 W TDP, the Intel part at 55 W, reflecting the server versus mobile design targets.

Head-to-Head Benchmarks

The biggest win for the AMD EPYC 7352 is Cinebench R23 single-core, where it scores 4844 against the Intel’s 2204, a 119.8% advantage. This is an outlier in the data — no other test shows such a dramatic AMD lead. The R15 single-core result (488 vs 334, 46.1% delta) confirms this pattern but at a smaller magnitude. The data compression test (660712 vs 608381, 8.6% delta) is the only other AMD victory, and it is modest.

The Intel Core Ultra 9 275HX’s largest win is PassMark floating-point math at 191186 versus 87969, a 54% delta. PassMark single-thread (4713 vs 1979) shows a 58% delta, and Cinebench R15 multi-core (5619.5 vs 3458) shows a 38.5% delta. The Intel part also leads Cinebench R20 single-core by 27.6% (2809 vs 2034) and PassMark multi-thread by 27.6% (55759 vs 40370). In PassMark find prime numbers, the Intel chip scores 448 versus 301, a 32.8% delta. PassMark extended instructions (47016 vs 40203, 14.5% delta) and PassMark physics (3338 vs 2688, 19.5% delta) further extend the Intel lead. The narrowest Intel wins are Cinebench R23 multi-core (35589 vs 34314, 3.6% delta) and PassMark integer math (155218 vs 148605, 4.3% delta).

The overall win count — 14 for Intel, 3 for AMD — understates the qualitative split. The AMD wins are concentrated in single-core Cinebench tests and compression, while the Intel wins span rendering, math, encryption, sorting, and physics. The Cinebench R23 single-core result is particularly anomalous: the AMD part beats the Intel part by over 2x, yet loses Cinebench R20 single-core by 27.6%. This inconsistency suggests the two chips respond very differently to specific benchmark revisions.

The Verdict

For server and workstation workloads that depend on massive thread counts and large cache pools, the AMD EPYC 7352 is the defensible choice. Its 48 threads and 128 MB L3 cache deliver a win in data compression, and its single-core Cinebench R23 score of 4844 is unmatched by the Intel part in this comparison. The eight-channel DDR4 memory subsystem with 204.8 GB/s bandwidth and 128 PCIe Gen 4 lanes also suits high-throughput enterprise environments where ECC and expandability are non-negotiable.

For everything else, the Intel Core Ultra 9 275HX is the superior processor. Its 5.40 GHz boost clock drives a 58% lead in PassMark single-thread and a 54% lead in floating-point math. It wins all three Cinebench multi-core tests, including a 27.6% margin in R20 and a 3.6% margin in R23, despite having half the thread count of the AMD part. The 55 W TDP makes it dramatically more power-efficient, and the integrated Arc Xe-LPG Graphics 64EU adds capability the EPYC lacks entirely. The 3 nm process node and 17,800 million transistors provide a modern foundation that the 7 nm Zen 2 design cannot match.

The data does not support a single universal winner. The EPYC 7352 is a specialist for legacy single-threaded Cinebench and compression workloads; the Core Ultra 9 275HX is the general-purpose champion across modern benchmarks. Given that the Intel part wins 14 of 17 tests and its losses are concentrated in two Cinebench single-core variants plus one compression test, the Core Ultra 9 275HX is the stronger overall processor for most users. The EPYC 7352’s 94th percentile ranking and near-identical average score (68118 vs 67469) keep it competitive, but the distribution of wins favors Intel decisively.

FAQ

Q: Which processor has more threads?

A: The AMD EPYC 7352 has 48 threads across 24 cores, while the Intel Core Ultra 9 275HX has 24 threads across 24 cores.

Q: What is the largest benchmark margin between the two?

A: The AMD EPYC 7352 wins Cinebench R23 single-core by 119.8% (4844 vs 2204), which is the largest delta in either direction.

Q: Does the Intel chip win any multi-core rendering test?

A: Yes, the Intel Core Ultra 9 275HX wins Cinebench R15 multi-core by 38.5%, Cinebench R20 multi-core by 27.6%, and Cinebench R23 multi-core by 3.6%.

Q: What memory bandwidth does each processor support?

A: The AMD EPYC 7352 supports 204.8 GB/s over eight-channel DDR4, while the Intel Core Ultra 9 275HX supports 102.4 GB/s over dual-channel DDR5.

Q: Which chip has a higher boost clock?

A: The Intel Core Ultra 9 275HX has a 5.40 GHz boost clock, compared to the AMD EPYC 7352’s 3.20 GHz.

Q: How do their average benchmark scores compare?

A: The AMD EPYC 7352 averages 68118, and the Intel Core Ultra 9 275HX averages 67469, a difference of less than 1%.

Q: Does the Intel chip include integrated graphics?

A: Yes, the Intel Core Ultra 9 275HX includes Arc Xe-LPG Graphics 64EU, while the AMD EPYC 7352 has no integrated graphics.

DETAILED SPECIFICATIONS

SPECIFICATION
EPYC 7352
Ultra 9 275HX
Core Specs
Cores
24
24 0.0%
Threads
48
24 -50.0%
Base Clock (GHz)
2.3
2.7 +17.4%
Boost Clock (GHz)
3.2
5.4 +68.8%
Frequency (GHz)
2.3
2.7 +17.4%
Turbo Clock (GHz)
3.2
5.4 +68.8%
Multiplier
23
27 +17.4%
SMP CPUs
2
1 -50.0%
Cache
L1 Cache
96 KB (per core)
192 KB (per core)
L2 Cache
512 KB (per core)
3 MB (per core)
L3 Cache
32 MB (per die)
36 MB (shared)
Total L3
128 MB
—
Power
TDP (W)
155
55 -64.5%
PL1
—
55 W
PL2
—
160 W
Configurable TDP
180 W
—
Architecture
Architecture
Zen 2
Arrow Lake
Codename
Rome
Arrow Lake-HX
Generation
EPYC (Zen 2 (Rome))
Ultra 9 (Arrow Lake-HX)
Process Size
7 nm
3 nm
Transistors
15,200 million
17,800 million
Die Size
4x 74 mm²
243 mm²
Foundry
TSMC
TSMC
Memory
Memory Support
DDR4
DDR5
Memory Bus
Eight-channel
Dual-channel
Memory Bandwidth
204.8 GB/s
102.4 GB/s
ECC Memory
Yes
No
Platform
Socket
AMD Socket SP3
Intel BGA 2114
Chipsets
—
WM880, HM870
PCIe
Gen 4, 128 Lanes(CPU only)
Gen 5, 20 Lanes(CPU only)
Intel Hybrid
Hybrid Cores
—
P-Cores: 8 E-Cores: 16
E-Core Frequency
—
2.1 GHz up to 4.6 GHz
AMD Multi-Die
CCDs
4
—
Cores per CCD
6
—
IO Process Size
14 nm
—
AI/NPU
NPU
—
Yes / 13 TOPS
Graphics
Integrated Graphics
—
Arc Xe-LPG Graphics 64EU
Other
Market
Server/Workstation
Mobile
Production Status
Active
Active
Launch Price
$1350
—
Part Number
100-000000077
SRVFK
Package
FCLGA-4094
FC-BGA
Tj Max
—
105°C
View EPYC 7352 Details View Core Ultra 9 275HX Details