Intel Xeon E5-2643 v3 vs Intel Xeon E5-4640 v3 Comparison

The Intel Xeon E5-4640 v3 and the Intel Xeon E5-2643 v3 are both Haswell-EP server processors on the LGA 2011-3 platform, yet they are engineered for opposite ends of the workload spectrum. The data shows a clear, consistent victory for the E5-4640 v3 across every benchmark in the head-to-head comparison, but the margin is remarkably narrow given the significant architectural differences. The E5-4640 v3 wins all six Cinebench tests, with the largest deltas appearing in single-core tests at 1.6% ahead, while multi-core tests show a 1.5% advantage. This outcome is counterintuitive at first glance, as the E5-4640 v3 packs twice the core count of the E5-2643 v3. The explanation lies in clock speed: the E5-2643 v3 operates at a 3.40 GHz base clock and 3.70 GHz boost, while the E5-4640 v3 is constrained to 1.90 GHz base and 2.60 GHz boost. The E5-4640 v3's 12 cores and 24 threads ultimately edge out the 6-core, 12-thread E5-2643 v3 in aggregate throughput, but the higher-frequency part keeps the gap to a whisper.

Head-to-Head Benchmarks

The benchmark results paint a picture of near-total dominance for the Intel Xeon E5-4640 v3, albeit with razor-thin margins. In Cinebench R15 multi-core, the E5-4640 v3 scores 888 against the E5-2643 v3's 875, a 1.5% advantage. The single-core R15 test shows a similar story: 125 for the E5-4640 v3 versus 123 for the E5-2643 v3, a 1.6% lead. Moving to Cinebench R20, the pattern holds exactly — the E5-4640 v3 posts 3702 in multi-core versus 3646 for the E5-2643 v3 (1.5% delta), and 522 versus 514 in single-core (1.6% delta). The newest test generation, Cinebench R23, reinforces the trend with the E5-4640 v3 achieving 8816 in multi-core against 8682 (1.5% delta) and 1244 in single-core against 1225 (1.6% delta).

The consistency of these deltas across all six tests is notable. Every multi-core test shows exactly 1.5% separation, and every single-core test shows exactly 1.6% separation. This uniformity suggests that the advantage is not workload-specific but rather a fundamental throughput characteristic. The E5-4640 v3's 12-core configuration provides enough parallel execution capacity to overcome its substantial clock deficit relative to the 6-core E5-2643 v3. However, the magnitude of the win is modest — a 1.5-1.6% edge is within the range of run-to-run variance for many workloads. The data indicates the E5-4640 v3 is the faster part, but the E5-2643 v3 is far from being left behind. In fact, the E5-2643 v3's per-core performance is superior, evidenced by its ability to nearly match a part with double the core count. The average benchmark scores confirm this closeness: the E5-4640 v3 averages 2550 compared to 2511 for the E5-2643 v3, a difference of roughly 1.5% overall.

Architecture Differences

Both processors share the same fundamental Haswell-EP architecture, manufactured on Intel's 22 nm process with 2,600 million transistors on a 356 mm² die. The socket is identical — Intel Socket 2011-3 — and both support DDR4 memory across a quad-channel bus with ECC memory enabled. The PCIe configuration is also the same: Gen 3 with 40 lanes from the CPU. The critical divergence lies in core configuration and cache allocation. The E5-4640 v3 integrates 12 cores with 24 threads, while the E5-2643 v3 halves that to 6 cores and 12 threads. This core count difference directly impacts the shared L3 cache: the E5-4640 v3 offers 30 MB shared, whereas the E5-2643 v3 provides 20 MB shared. Per-core L1 cache remains 64 KB and L2 cache 256 KB for both parts, but the total cache resources scale with the core count.

Clock speeds present the most striking architectural contrast. The E5-2643 v3's base clock of 3.40 GHz and boost of 3.70 GHz far exceed the E5-4640 v3's 1.90 GHz base and 2.60 GHz boost. This 1.50 GHz base clock difference and 1.10 GHz boost difference explain why the 6-core part can compete with the 12-core part. The thermal design power reflects this trade-off inversely: the E5-4640 v3 draws 105 W despite more cores, while the E5-2643 v3 consumes 135 W. Memory bandwidth also favors the E5-2643 v3 at 68.3 GB/s versus 59.7 GB/s for the E5-4640 v3, an advantage that likely stems from the higher clock domain. The release dates differ by roughly eight months, with the E5-2643 v3 launching on September 7, 2014, and the E5-4640 v3 on May 31, 2015. Both parts are end-of-life and locked with no unlocked multiplier. The part numbers differ as SR22L for the E5-4640 v3 and SR204SR1XQ for the E5-2643 v3.

FAQ

Q: Which processor has more cores?

A: The Intel Xeon E5-4640 v3 has 12 cores and 24 threads, while the Intel Xeon E5-2643 v3 has 6 cores and 12 threads.

Q: Why does the E5-2643 v3 nearly match the E5-4640 v3 despite half the cores?

A: The E5-2643 v3 operates at a much higher clock speed — 3.40 GHz base and 3.70 GHz boost versus 1.90 GHz base and 2.60 GHz boost for the E5-4640 v3. This clock advantage compensates for the core deficit in most workloads.

Q: What is the cache difference between these two CPUs?

A: Both have 64 KB L1 and 256 KB L2 per core, but the shared L3 cache differs: the E5-4640 v3 has 30 MB shared, while the E5-2643 v3 has 20 MB shared.

Q: Do both processors support ECC memory?

A: Yes, both support ECC memory and use DDR4 across a quad-channel memory bus. The E5-2643 v3 has higher memory bandwidth at 68.3 GB/s versus 59.7 GB/s for the E5-4640 v3.

Q: Which CPU has a higher thermal design power?

A: The Intel Xeon E5-2643 v3 has a TDP of 135 W, while the Intel Xeon E5-4640 v3 has a lower TDP of 105 W despite having twice the cores.

Q: Are these processors from the same architecture generation?

A: Yes, both are based on the Haswell architecture with the Haswell-EP codename, built on a 22 nm process with 2,600 million transistors and a 356 mm² die size.

Specification Differences

| Specification | Intel Xeon E5-4640 v3 | Intel Xeon E5-2643 v3 |

|---|---|---|

| Cores | 12 | 6 |

| Threads | 24 | 12 |

| Base Clock | 1.90 GHz | 3.40 GHz |

| Boost Clock | 2.60 GHz | 3.70 GHz |

| TDP | 105 W | 135 W |

| L3 Cache | 30 MB (shared) | 20 MB (shared) |

| Memory Bandwidth | 59.7 GB/s | 68.3 GB/s |

| Release Date | 2015-05-31 | 2014-09-07 |

| Launch MSRP | $2859 | $1552 |

| Part Number | SR22L | SR204SR1XQ |

The shared specifications include the Intel Socket 2011-3, Haswell-EP architecture, 22 nm process node, 2,600 million transistors, 356 mm² die size, DDR4 memory support with quad-channel bus, ECC memory support, PCIe Gen 3 with 40 lanes, no integrated graphics, server/workstation market segment, end-of-life production status, and locked multipliers.

The Verdict

The benchmark data is unambiguous: the Intel Xeon E5-4640 v3 is the faster processor in every measured test. The head-to-head results show a clean sweep — 6 wins for the E5-4640 v3 and 0 for the E5-2643 v3. However, the margins are so small that the practical difference is negligible for most workloads. The E5-4640 v3's 12-core advantage delivers only a 1.5% multi-core lead in Cinebench R15, R20, and R23, and a 1.6% single-core lead across the same tests. This means the E5-2643 v3's higher clock speeds nearly neutralize the core count disparity.

For buyers prioritizing raw multi-threaded throughput, the E5-4640 v3 is the data-backed choice. Its 12 cores and 24 threads provide a structural advantage that, while modest in percentage terms, scales with sustained parallel workloads. The lower TDP of 105 W compared to 135 W is an additional operational benefit. For those who need maximum per-core responsiveness or memory bandwidth, the E5-2643 v3 offers a compelling alternative with 68.3 GB/s versus 59.7 GB/s and substantially higher clock speeds. The E5-2643 v3 also carries a lower launch MSRP of $1552 compared to $2859 for the E5-4640 v3, though both are end-of-life products.

The percentile rankings place both CPUs at the 49th percentile among all processors, indicating they sit in the middle of the performance distribution. The average benchmark scores of 2550 for the E5-4640 v3 and 2511 for the E5-2643 v3 reflect their near-identical standing. The nearest rival data reinforces this: the E5-4640 v3 sits within 0.9% of the Intel Xeon E-2244G, while the E5-2643 v3 is 0.6% behind the Intel Core i5-1155G7. Both parts are competitive with modern mid-range processors despite their age.

Where Each One Wins

Intel Xeon E5-4640 v3 — wins in multi-threaded and cache-sensitive scenarios. The 12-core, 24-thread configuration provides superior parallel execution capacity, which manifests in the 1.5% multi-core Cinebench advantages across R15, R20, and R23. The 30 MB shared L3 cache is 50% larger than the E5-2643 v3's 20 MB, offering better data locality for workloads with large working sets. The lower 105 W TDP makes it a more thermally efficient choice for dense server deployments. This processor is the pick for heavily threaded workloads such as virtualization, database processing, and content creation that scales across cores.

Intel Xeon E5-2643 v3 — wins in single-threaded responsiveness and memory throughput. Despite losing the single-core benchmarks by 1.6%, the E5-2643 v3's 3.40 GHz base and 3.70 GHz boost clocks provide significantly higher per-core frequency, which translates to lower latency for lightly threaded tasks. The 68.3 GB/s memory bandwidth versus 59.7 GB/s gives it a measurable advantage in memory-bound applications. The 135 W TDP reflects the higher clock domain. This processor suits workloads where clock speed matters more than core count, such as real-time transaction processing, lightly threaded scientific simulations, or as a high-frequency compute node in a cluster. The 6-core, 12-thread configuration still provides respectable parallelism, evidenced by its ability to nearly match the 12-core E5-4640 v3 in multi-core tests.