Intel Xeon E5-1620 v4 vs Intel Xeon E5-2637 v3 Comparison

Intel Xeon E5-1620 v4 and Intel Xeon E5-2637 v3 are both 4-core, 8-thread workstation processors sharing the same socket and memory type, but they represent different architectural generations. The recorded data shows a consistent, albeit narrow, performance advantage for the newer Broadwell-based E5-1620 v4 across every Cinebench workload, while the older Haswell-based E5-2637 v3 counters with a larger L3 cache and a higher launch MSRP. Neither processor is a clear knockout winner; instead, the choice hinges on specific workload priorities and platform context.

Where Each One Wins

The benchmark results are unequivocal regarding raw compute performance: the Intel Xeon E5-1620 v4 wins all six recorded head-to-head Cinebench tests. The margins are slim, ranging from 1.1% to 1.7%, but they are consistent across both single-core and multi-core workloads. In Cinebench R15 multicore, the E5-1620 v4 scores 627 against 617 for the E5-2637 v3, a 1.6% lead. The single-core test in R15 shows 88 versus 87, a 1.1% advantage. This pattern repeats in R20, where the E5-1620 v4 posts 2616 multicore and 369 single-core, against 2574 and 363 respectively, and in R23, where it reaches 6230 multicore and 879 single-core, versus 6130 and 865.

For users prioritizing peak rendering throughput or single-threaded responsiveness, the E5-1620 v4 is the clear pick. Its wins in every category mean there is no workload in this benchmark suite where the E5-2637 v3 pulls ahead. However, the E5-2637 v3 does hold a structural advantage in cache capacity, with 15 MB of shared L3 versus 10 MB on the E5-1620 v4. This does not translate into a performance win in the recorded Cinebench tests, but it could matter for other, cache-sensitive workloads not represented here. The E5-2637 v3 also carries a different market segment designation, being classified as Server/Workstation, while the E5-1620 v4 is classified as Desktop, which may influence platform expectations.

The overall database percentile ranking places the E5-1620 v4 at the 42nd percentile of all CPUs, with an average benchmark score of 1802, while the E5-2637 v3 sits at the 41st percentile with an average score of 1773. This confirms that the performance gap is real but modest, placing both chips in a similar performance tier among all processors. For anyone choosing between these two specific parts, the E5-1620 v4 wins on measured performance, but the E5-2637 v3 is not far behind.

Architecture Differences

The two processors come from different Intel microarchitectures and manufacturing processes. The E5-1620 v4 is built on the Broadwell architecture, specifically Broadwell-EP, using a 14 nm process node. The E5-2637 v3 is based on Haswell, specifically Haswell-EP, on a 22 nm node. This process shrink is significant, as the 14 nm node allows for a smaller die size of 246 mm² for the E5-1620 v4, compared to 356 mm² for the E5-2637 v3. Interestingly, the transistor count tells a different story: the E5-1620 v4 packs 3,400 million transistors into that smaller die, while the E5-2637 v3 has 2,600 million on its larger die. This means the Broadwell chip achieves higher density, which typically enables better power efficiency and potentially higher clock speeds for a given power envelope.

The core and thread counts are identical: both have 4 cores and 8 threads. Base clocks match at 3.50 GHz, but the boost clock differs, with the E5-1620 v4 reaching 3.80 GHz versus 3.70 GHz on the E5-2637 v3. This 0.10 GHz boost advantage likely contributes to the single-core benchmark wins. The L1 and L2 caches are identical per core (64 KB and 256 KB respectively), but the shared L3 cache diverges sharply: 10 MB on the E5-1620 v4 versus 15 MB on the E5-2637 v3. The larger cache on the Haswell part is a notable feature, potentially benefiting workloads with large working sets that fit within the L3.

Memory support is similar, with both supporting DDR4 and a quad-channel memory bus. However, the recorded memory bandwidth differs: the E5-1620 v4 is rated at 76.8 GB/s, while the E5-2637 v3 is rated at 68.3 GB/s. This 8.5 GB/s difference favors the newer part and could influence memory-bound tasks. Both processors support ECC memory, both have the same PCIe configuration (Gen 3, 40 lanes from the CPU), and neither has integrated graphics. The sockets are identical (Intel Socket 2011-3), making them platform-compatible, but the generations differ: the E5-1620 v4 was released in June 2016, while the E5-2637 v3 was released in September 2014.

The TDP figures are close, with the E5-1620 v4 rated at 140 W and the E5-2637 v3 at 135 W. The newer process node does not result in a lower TDP, as the E5-1620 v4 actually draws slightly more power, likely due to the higher boost clock and additional transistors. Both parts are marked as end-of-life, and neither has an unlocked multiplier.

FAQ

Q: Which processor has a higher boost clock?

A: The Intel Xeon E5-1620 v4 has a boost clock of 3.80 GHz, while the Intel Xeon E5-2637 v3 has a boost clock of 3.70 GHz.

Q: How much larger is the L3 cache on the E5-2637 v3?

A: The E5-2637 v3 has 15 MB of shared L3 cache, which is 5 MB more than the 10 MB shared L3 cache on the E5-1620 v4.

Q: Do these processors support ECC memory?

A: Yes, both the Intel Xeon E5-1620 v4 and the Intel Xeon E5-2637 v3 support ECC memory.

Q: What is the difference in average benchmark scores between the two?

A: The E5-1620 v4 has an average benchmark score of 1802, which is 29 points higher than the 1773 average score of the E5-2637 v3.

Q: Are the memory bandwidth ratings different?

A: Yes, the E5-1620 v4 is rated for 76.8 GB/s of memory bandwidth, whereas the E5-2637 v3 is rated for 68.3 GB/s.

Q: Which processor has a larger die size?

A: The E5-2637 v3 has a die size of 356 mm², which is larger than the 246 mm² die of the E5-1620 v4, despite the older 22 nm process node.

Specification Differences

The following specifications differ between the Intel Xeon E5-1620 v4 and the Intel Xeon E5-2637 v3, based on the recorded data:

• Architecture: Broadwell for the E5-1620 v4, Haswell for the E5-2637 v3.
• Process Node: 14 nm for the E5-1620 v4, 22 nm for the E5-2637 v3.
• Transistors: 3,400 million for the E5-1620 v4, 2,600 million for the E5-2637 v3.
• Die Size: 246 mm² for the E5-1620 v4, 356 mm² for the E5-2637 v3.
• Boost Clock: 3.80 GHz for the E5-1620 v4, 3.70 GHz for the E5-2637 v3.
• L3 Cache: 10 MB (shared) for the E5-1620 v4, 15 MB (shared) for the E5-2637 v3.
• Memory Bandwidth: 76.8 GB/s for the E5-1620 v4, 68.3 GB/s for the E5-2637 v3.
• TDP: 140 W for the E5-1620 v4, 135 W for the E5-2637 v3.
• Market Segment: Desktop for the E5-1620 v4, Server/Workstation for the E5-2637 v3.
• Release Date: June 2016 for the E5-1620 v4, September 2014 for the E5-2637 v3.
• Launch MSRP: $294 for the E5-1620 v4, $996 for the E5-2637 v3.
• Part Number: SR2P6 for the E5-1620 v4, SR202 for the E5-2637 v3.
• Average Benchmark Score: 1802 for the E5-1620 v4, 1773 for the E5-2637 v3.
• Percentile Rank: 42nd for the E5-1620 v4, 41st for the E5-2637 v3.

Identical specifications include cores (4), threads (8), base clock (3.50 GHz), socket (Intel Socket 2011-3), L1 cache (64 KB per core), L2 cache (256 KB per core), memory type (DDR4), memory bus (Quad-channel), ECC support (true), PCIe (Gen 3, 40 lanes), integrated graphics (none), production status (end-of-life), and multiplier lock (locked).

Head-to-Head Benchmarks

The head-to-head Cinebench results show a uniform pattern: the E5-1620 v4 wins every test, but by a small margin. The largest relative win is in Cinebench R20 single-core, where the E5-1620 v4 scores 369 against 363 for the E5-2637 v3, a delta of 1.7%. This single-core advantage is consistent with the 0.10 GHz boost clock difference. The smallest win is in Cinebench R15 single-core, with scores of 88 and 87, a 1.1% delta.

In multi-core workloads, the wins are also consistent. Cinebench R15 multicore shows 627 versus 617, a 1.6% delta. Cinebench R20 multicore shows 2616 versus 2574, again a 1.6% delta. Cinebench R23 multicore shows 6230 versus 6130, also a 1.6% delta. The uniformity of these multi-core deltas suggests that the performance advantage is driven by the higher boost clock and possibly the more efficient 14 nm process, rather than any core count or thread count differences, which are identical.

The average benchmark score difference is 29 points (1802 versus 1773), which is a 1.6% overall advantage for the E5-1620 v4. This aligns closely with the per-test deltas. The nearest rivals for the E5-1620 v4 include the Intel Core i3-9100 (avg score 1806, delta -0.2%), the Intel Core i3-9320 (avg score 1808, delta -0.4%), and the AMD Ryzen 5 PRO 4650U (avg score 1810, delta -0.4%). For the E5-2637 v3, the nearest rivals include the Intel Xeon E5-4610 v3 (avg score 1777, delta -0.2%), the Intel Core i7-4870HQ (avg score 1769, delta 0.2%), and the AMD Ryzen 3 PRO 1300 (avg score 1778, delta -0.3%). The data shows that both Xeon parts sit within a tight performance cluster, with the E5-1620 v4 slightly ahead of its nearest rivals and the E5-2637 v3 slightly behind its own nearest rivals in average score.

In summary, the E5-1620 v4 is the faster processor in every measured Cinebench test, with a consistent 1.6% lead in most cases and a slightly larger 1.7% lead in R20 single-core. The E5-2637 v3 offers a larger L3 cache but does not convert that into a benchmark win. The choice is straightforward for performance, but the E5-2637 v3's cache capacity and lower TDP are counterpoints that may be relevant in specific server or workstation scenarios.