Intel Xeon E5-1660 v3 vs Intel Xeon E5-2650 v4 Comparison

Where Each One Wins

The benchmark split is unambiguous: the Intel Xeon E5-2650 v4 wins every recorded Cinebench test, both multi-core and single-core, across all three versions of the benchmark. That is six wins out of six head-to-head comparisons. The Intel Xeon E5-1660 v3 does not claim a single victory in the measured data.

For multi-core workloads, the E5-2650 v4’s advantage comes from its 12 cores and 24 threads, which outnumber the E5-1660 v3’s 8 cores and 16 threads. The margin is consistent, hovering just above 2% in every multi-core test. The largest multi-core gap appears in Cinebench R15, where the E5-2650 v4 scores 1098 against 1074, a 2.2% lead. In Cinebench R20 and R23, the lead is 2.3% in both cases.

Single-core performance is also a win for the E5-2650 v4, though the margin is slightly larger in R15 (2.6%) than in R20 (2.4%) or R23 (2.3%). This is notable because the E5-1660 v3 has a higher base clock (3.00 GHz versus 2.20 GHz) and a higher boost clock (3.50 GHz versus 2.90 GHz). Despite the clock disadvantage, the newer Broadwell architecture of the E5-2650 v4 delivers higher single-thread scores, suggesting architectural efficiency outweighs raw clock speed in these tests.

The database places the E5-2650 v4 at the 53rd percentile among all CPUs, while the E5-1660 v3 sits at the 52nd percentile. The average benchmark score for the E5-2650 v4 is 3154, against 3082 for the E5-1660 v3. That 72-point gap in average score reflects a modest but consistent overall performance advantage.

In short, the E5-2650 v4 is the pick for any workload that benefits from more cores or from slightly faster single-thread execution. The E5-1660 v3 offers no measured advantage in the benchmark suite recorded here.

FAQ

Q: Which processor has more cores?

A: The Intel Xeon E5-2650 v4 has 12 cores and 24 threads. The Intel Xeon E5-1660 v3 has 8 cores and 16 threads.

Q: Does the E5-1660 v3 ever beat the E5-2650 v4 in any benchmark?

A: No. In the recorded head-to-head results, the E5-2650 v4 wins all six tests, with delta percentages ranging from 2.2% to 2.6%.

Q: How does single-core performance compare?

A: The E5-2650 v4 leads in single-core across all three Cinebench versions. In R15, it scores 155 against 151 (2.6% higher). In R20, it scores 646 against 631 (2.4% higher). In R23, it scores 1539 against 1504 (2.3% higher).

Q: What is the average benchmark score difference?

A: The E5-2650 v4 has an average benchmark score of 3154, while the E5-1660 v3 averages 3082. The E5-2650 v4 sits at the 53rd percentile, one point above the E5-1660 v3 at the 52nd percentile.

Q: Do both processors support ECC memory?

A: Yes, both support ECC memory, and both use DDR4 with quad-channel memory buses and 68.3 GB/s of memory bandwidth.

Q: Are both processors end-of-life?

A: Yes, both are marked as end-of-life. The E5-2650 v4 was released on 2016-03-15, while the E5-1660 v3 was released on 2014-09-07.

Head-to-Head Benchmarks

The recorded data tells a consistent story. Across six Cinebench tests, the E5-2650 v4 wins every time, with margins that never dip below 2.2% and never exceed 2.6%. This uniformity suggests a stable performance advantage rather than a workload-specific quirk.

Starting with multi-core results, the smallest gap appears in Cinebench R15, where the E5-2650 v4 scores 1098 and the E5-1660 v3 scores 1074. That is a 2.2% lead. In Cinebench R20, the E5-2650 v4 scores 4579 against 4475, a 2.3% gap. The same 2.3% margin appears in Cinebench R23, with scores of 10904 and 10656 respectively. The multi-core lead is therefore remarkably consistent, hovering around 2.2% to 2.3% regardless of benchmark version.

Single-core results show a slightly wider spread. In Cinebench R15, the E5-2650 v4 scores 155 against 151, a 2.6% advantage. In R20, the scores are 646 and 631, a 2.4% gap. In R23, they are 1539 and 1504, again a 2.3% margin. The single-core edge is marginally larger in the older R15 test but narrows slightly in newer versions.

The biggest single win for the E5-2650 v4 comes in Cinebench R15 single-core, where the 2.6% delta is the largest of any test. The smallest win is the 2.2% margin in R15 multi-core. In absolute terms, the largest score difference is 248 points in Cinebench R23 multi-core (10904 versus 10656). The smallest absolute difference is 4 points in R15 single-core (155 versus 151).

These numbers indicate that the E5-2650 v4 does not simply win on core count. It also edges out the E5-1660 v3 in single-threaded workloads, despite the E5-1660 v3’s higher clocks. The architectural advantage of Broadwell over Haswell appears to offset the clock disadvantage in every measured test.

Specification Differences

The two processors differ in several key specifications. The E5-2650 v4 has 12 cores and 24 threads, while the E5-1660 v3 has 8 cores and 16 threads. Base clocks differ substantially: the E5-2650 v4 runs at 2.20 GHz, the E5-1660 v3 at 3.00 GHz. Boost clocks differ as well: 2.90 GHz for the E5-2650 v4, 3.50 GHz for the E5-1660 v3.

Thermal design power is not equal. The E5-2650 v4 is rated at 105 W, while the E5-1660 v3 draws 140 W. This means the E5-2650 v4 delivers more cores and faster benchmark scores while consuming less power according to the TDP figures.

Cache sizes also differ. The E5-2650 v4 has 30 MB of shared L3 cache, while the E5-1660 v3 has 20 MB. Both have 64 KB of L1 cache per core and 256 KB of L2 cache per core.

Memory support is identical: both use DDR4, both have quad-channel memory buses, and both have 68.3 GB/s of memory bandwidth. Both support ECC memory. Both use PCIe Gen 3 with 40 lanes (CPU only).

The E5-2650 v4 has a launch MSRP of $1166. The E5-1660 v3 does not have a recorded launch MSRP in the database. The E5-2650 v4 has a locked multiplier, while the E5-1660 v3 has an unlocked multiplier. The part numbers are SR2N3 for the E5-2650 v4 and SR20N for the E5-1660 v3.

Architecture Differences

The E5-2650 v4 is built on Broadwell-EP architecture, while the E5-1660 v3 uses Haswell-EP. This is a generational difference: the E5-2650 v4 uses a 14 nm process node, the E5-1660 v3 uses a 22 nm node. The smaller node allows the E5-2650 v4 to pack 3,400 million transistors into a 246 mm² die. The E5-1660 v3 has 2,600 million transistors on a larger 356 mm² die.

The die size difference is striking. The E5-1660 v3 is physically larger despite having fewer cores and less L3 cache. This reflects the older 22 nm manufacturing process. The E5-2650 v4 achieves more cores, more cache, and lower TDP on a smaller die, which is a direct benefit of the newer 14 nm node.

Both processors share the same socket (Intel Socket 2011-3), the same PCIe generation and lane count, and the same memory configuration. The architectural gap is therefore limited to the CPU core design, cache hierarchy, and process node.

The unlocked multiplier on the E5-1660 v3 is a feature difference that could matter for users who plan to overclock. However, the recorded benchmark data does not include any overclocked results, so the performance implications are not measured here.

The E5-2650 v4’s release date is 2016-03-15, about 18 months after the E5-1660 v3’s release on 2014-09-07. That extra time allowed Intel to move from 22 nm Haswell to 14 nm Broadwell, which explains the efficiency and performance improvements.

The Verdict

The data points to a clear choice. The Intel Xeon E5-2650 v4 wins every recorded benchmark, in both multi-core and single-core tests, with margins between 2.2% and 2.6%. It does so with 12 cores against 8, 24 threads against 16, and a TDP of 105 W against 140 W. The average benchmark score favors the E5-2650 v4 by 72 points (3154 versus 3082), and the percentile ranking is one point higher.

Users who need maximum multi-threaded throughput should choose the E5-2650 v4. The 12-core configuration provides a measurable advantage in every Cinebench multi-core test, and the lower TDP makes it the more efficient option for sustained workloads.

Users who value single-thread speed might have expected the E5-1660 v3 to win, given its higher base and boost clocks. The data shows otherwise. The E5-2650 v4 wins single-core tests by 2.3% to 2.6%, meaning the Broadwell architecture’s efficiency overcomes the clock deficit.

The only feature that favors the E5-1660 v3 is the unlocked multiplier. That could appeal to users who intend to overclock, but the database contains no overclocked benchmark results. Without those measurements, the performance benefit cannot be quantified.

There is no scenario in the recorded data where the E5-1660 v3 outperforms the E5-2650 v4. The verdict is straightforward: the E5-2650 v4 is the better processor for both multi-core and single-core workloads, with lower TDP and a smaller die. The E5-1660 v3 remains a capable option for those who need an unlocked multiplier, but its measured performance trails in every test.