Intel Xeon E5-1660 v4 vs Intel Xeon E5-1680 v3 Comparison

The Intel Xeon E5-1680 v3 and the Intel Xeon E5-1660 v4 are near-identical performers in the benchmark database, with the v3 edging out the v4 in every single recorded test, though by margins so small they round to zero. The average benchmark score for the E5-1680 v3 is 3265, while the E5-1660 v4 sits at 3264, a delta of 0%. Both processors hold the 53rd percentile among all CPUs, and the data positions them as direct rivals, with the AMD Ryzen Embedded V2516 and Intel Core i3-12100T trailing by 0.4% and the Intel Xeon E-2374G also at 0.4% behind. This is a matchup defined by statistical equivalence, but the architecture differences tell a more interesting story.

Head-to-Head Benchmarks

The benchmark results are remarkably consistent across all six Cinebench tests. In Cinebench R15 multicore, both CPUs score exactly 1137, with the E5-1680 v3 listed as the winner due to a 0% delta. The single-core R15 test shows the same pattern: both score 160. Moving to Cinebench R20, the E5-1680 v3 posts a multicore score of 4741 against the E5-1660 v4's 4739 — a two-point lead that represents a 0% delta. The R20 single-core test is a dead tie at 669 for both. In Cinebench R23, the v3 again leads in multicore with 11289 versus 11285, a four-point gap, while single-core scores are identical at 1593.

The data awards the E5-1680 v3 six wins and the E5-1660 v4 zero, but the interpretation should be cautious. A two-point difference in a metric that spans thousands is within any reasonable margin of error. The head-to-head table lists the v3 as the winner in all six tests, but the deltas are all zero. For practical purposes, these two processors deliver identical Cinebench performance. The v3's edge in R20 and R23 multicore tests is consistent — it leads by two and four points respectively — suggesting a tiny but repeatable advantage, possibly attributable to architectural tuning rather than clock differences, since both run at 3.20 GHz base and 3.80 GHz boost. The E5-1660 v4 never leads in any test, but it also never falls behind by more than a rounding error.

Architecture Differences

The fundamental distinction is the manufacturing process and microarchitecture. The E5-1680 v3 uses the Haswell architecture on a 22 nm process node, while the E5-1660 v4 uses Broadwell on a 14 nm node. This process shrink from 22 nm to 14 nm is significant for efficiency and transistor density, but it does not translate into a performance lead in the benchmark data. The transistor counts reflect this: the v3 packs 2,600 million transistors on a 356 mm² die, while the v4 crams 3,400 million transistors onto a smaller 246 mm² die. The higher transistor count on a smaller die for the v4 confirms the density improvement from the 14 nm node.

Both processors share identical core and thread configurations: 8 cores and 16 threads. Cache hierarchies are also the same, with 64 KB of L1 per core, 256 KB of L2 per core, and a shared 20 MB L3 cache. Memory support is DDR4 with a quad-channel bus for both, but the memory bandwidth differs: the E5-1680 v3 achieves 68.3 GB/s, while the E5-1660 v4 reaches 76.8 GB/s, a 12% advantage for the v4. This bandwidth difference is notable but does not appear in the Cinebench scores, which are more compute-bound than memory-bound. PCIe connectivity is identical: Gen 3 with 40 lanes (CPU only). Neither has integrated graphics. Both support ECC memory.

The v3 is classified as a Server/Workstation part, while the v4 is marked as a Desktop part, despite sharing the same Intel Socket 2011-3. The v3 has an unlocked multiplier, whereas the v4 is locked. The v4's launch MSRP is $1113; the v3 has no launch MSRP listed. The release dates differ by roughly two years: the v3 launched in September 2014, the v4 in June 2016. Both are end-of-life products.

The Verdict

From the data, the Intel Xeon E5-1680 v3 is the nominal winner, taking all six head-to-head benchmark comparisons. However, the magnitude of the victory is negligible — every delta is 0%, and the largest absolute gap is four points in Cinebench R23 multicore. The E5-1660 v4 counters with a 12.4% higher memory bandwidth (76.8 GB/s vs 68.3 GB/s), a more advanced 14 nm process, and a higher transistor count, but none of these translate into a single benchmark win. The v3 also offers an unlocked multiplier, which the v4 lacks, making it the more flexible part for overclocking scenarios. The v4's higher memory bandwidth could benefit memory-intensive workloads, but the Cinebench suite does not capture that advantage. For raw compute performance as measured by Cinebench, the v3 is the pick. For systems where memory throughput matters more, the v4's specification sheet is stronger. The v4's launch MSRP of $1113 provides a reference point, but no price comparison is made here.

Specification Differences

The two processors differ in several specification fields. The architecture changes from Haswell (v3) to Broadwell (v4). The process node shrinks from 22 nm to 14 nm. Transistors increase from 2,600 million to 3,400 million, while die size decreases from 356 mm² to 246 mm². Memory bandwidth rises from 68.3 GB/s to 76.8 GB/s. The market segment shifts from Server/Workstation (v3) to Desktop (v4). The multiplier is unlocked on the v3 but locked on the v4. The release date moves from September 2014 to June 2016. The v4 has a launch MSRP of $1113; the v3 has none. The part numbers differ (QFSSSR20H for v3, SR2PK for v4). All other fields — cores (8), threads (16), base clock (3.20 GHz), boost clock (3.80 GHz), TDP (140 W), socket (Intel Socket 2011-3), cache sizes (64 KB L1, 256 KB L2, 20 MB L3), memory type (DDR4), memory bus (quad-channel), ECC support (true), PCIe (Gen 3, 40 lanes), integrated graphics (none), and production status (end-of-life) — are identical.

FAQ

Q: Which processor has a higher average benchmark score?

A: The Intel Xeon E5-1680 v3 has an average benchmark score of 3265, while the E5-1660 v4 scores 3264, a 0% delta.

Q: Are the single-core performance scores different between the two CPUs?

A: No, both CPUs score 160 in Cinebench R15 single-core, 669 in R20 single-core, and 1593 in R23 single-core.

Q: What is the most significant architectural difference?

A: The E5-1680 v3 uses a 22 nm Haswell process, while the E5-1660 v4 uses a 14 nm Broadwell process, with the v4 packing 3,400 million transistors versus the v3's 2,600 million.

Q: Do both CPUs support the same memory configuration?

A: Yes, both support DDR4 with a quad-channel bus, but the v4 has higher memory bandwidth at 76.8 GB/s versus the v3's 68.3 GB/s.

Q: Is one CPU unlocked for overclocking?

A: The E5-1680 v3 has an unlocked multiplier, while the E5-1660 v4's multiplier is locked.

Q: How many benchmark wins does each CPU have?

A: The E5-1680 v3 has 6 wins, and the E5-1660 v4 has 0 wins, though all deltas are 0%.

Where Each One Wins

The Intel Xeon E5-1680 v3 wins in every Cinebench benchmark category: multicore and single-core across R15, R20, and R23. Its leads are minuscule — two points in R20 multicore and four points in R23 multicore — but they are consistent. The v3 also wins on flexibility with its unlocked multiplier, which allows manual clock adjustment, a feature absent on the v4. For workloads that rely purely on CPU compute throughput as measured by Cinebench, the v3 is the superior choice, even if the margin is effectively a tie.

The Intel Xeon E5-1660 v4 wins on memory bandwidth, delivering 76.8 GB/s compared to the v3's 68.3 GB/s. This is a 12.4% advantage that could matter in memory-bound applications such as large dataset processing or virtualized environments, though the Cinebench suite does not reflect this. The v4 also benefits from a more advanced 14 nm process, which typically yields better power efficiency per transistor, and its higher transistor count (3,400 million vs 2,600 million) on a smaller die (246 mm² vs 356 mm²) suggests denser integration. The v4's Desktop market segment classification is its only other distinct attribute. For workloads that stress memory throughput over raw compute, the v4's specification sheet offers a tangible edge, even if the benchmark data shows no performance gap.