CPU Comparison

The Intel Xeon E5645 and Intel Xeon D-1520 are two server processors from different eras of Intel’s roadmap, yet benchmark results place them in a remarkably tight performance band. The E5645 is a Westmere-EP part from the 2010 generation, while the D-1520 is a Broadwell-DE system-on-chip from 2015. Despite a five-year gap in release timing and significant architectural differences, their average benchmark scores are nearly identical: the E5645 posts an average of 1211, and the D-1520 posts 1209. Both sit at the 34th percentile of all CPUs tracked, indicating they occupy the same performance tier in the broader market. The head-to-head data shows the E5645 winning all six benchmark comparisons, but the margins are razor-thin, with the largest delta being just 0.4%.

Head-to-Head Benchmarks

The benchmark suite reveals a pattern of near-parity punctuated by marginal victories for the Intel Xeon E5645. In Cinebench R15 multicore, both processors score exactly 421, producing a delta of 0%. The single-core test in R15 is likewise identical, with both parts scoring 59. This absolute tie in the older test suite suggests that the per-thread performance of the two architectures is effectively indistinguishable in that workload.

Moving to Cinebench R20, the E5645 edges ahead. In the multicore test, the E5645 scores 1758 against the D-1520’s 1755, a delta of 0.2%. The single-core test shows a slightly larger gap: 248 for the E5645 versus 247 for the D-1520, a 0.4% difference. The final test, Cinebench R23, mirrors the R20 results. The E5645 scores 4187 in multicore, while the D-1520 scores 4179, again a 0.2% delta. In single-core R23, the E5645 scores 591 against 590, another 0.2% margin.

The data indicates that the E5645 wins all six head-to-head comparisons, but the victories are purely nominal. A 0.4% difference in a single benchmark is within run-to-run variance for most systems. The practical conclusion from these numbers is that neither chip offers a meaningful performance advantage over the other in Cinebench workloads. The E5645’s six wins are more a function of rounding than actual superiority, as the largest absolute gap is a single point in the R20 single-core test.

Architecture Differences

The underlying designs of these two CPUs could hardly be more different, even though their benchmark outputs align. The E5645 is built on the Westmere-EP architecture using a 32 nm process node, with 1,170 million transistors packed into a 239 mm² die. It features 6 cores and 12 threads, with a base clock of 2.40 GHz and a boost clock of 2.67 GHz. Its cache hierarchy includes 64 KB of L1 per core, 256 KB of L2 per core, and a shared 12 MB L3 cache. The memory controller supports DDR3 in a triple-channel configuration, and the platform uses Intel Socket 1366. PCIe connectivity is limited to Gen 2.

The D-1520, in contrast, is a Broadwell-DE part fabricated on a 14 nm process. It integrates 3,200 million transistors on a slightly larger 246 mm² die. The core count is lower at 4 cores and 8 threads, with a base clock of 2.20 GHz and a boost clock of 2.60 GHz. The cache layout is different: it retains 64 KB L1 and 256 KB L2 per core, but the L3 is allocated as 1.5 MB per core rather than a single shared pool. Memory support is more flexible, accepting both DDR3 and DDR4 in a dual-channel configuration. The socket is Intel BGA 1667, and PCIe is upgraded to Gen 3 with 24 lanes available from the CPU.

The process node difference is stark: 32 nm versus 14 nm. This explains the D-1520’s dramatically lower thermal design power of 45 watts against the E5645’s 80 watts. The D-1520 also integrates more than twice the transistor count, though the die size is only marginally larger. The E5645 compensates for its older process with a higher core count and a larger shared L3 cache, which likely explains its ability to match the newer chip in multicore workloads despite the architectural generation gap.

Where Each One Wins

The benchmark data shows the E5645 winning all six comparisons, but the margins are so small that the practical answer is that neither chip wins in any meaningful sense. The E5645’s wins are confined to Cinebench R15, R20, and R23 across both single-core and multicore tests. In R15, the scores are identical, so the E5645’s "win" is a tie broken by the data structure. In R20 and R23, the deltas range from 0.2% to 0.4%, which is negligible for real-world applications.

For use cases, the E5645’s advantage lies in its core count. With 6 cores and 12 threads, it offers more parallel processing capacity than the D-1520’s 4 cores and 8 threads. Workloads that scale with thread count, such as heavily threaded rendering or virtualization, might see a slight edge from the E5645, though the benchmark data does not show a significant difference. The D-1520, on the other hand, compensates with a 45 W TDP, making it far more power-efficient. For dense server deployments where thermal envelope is a constraint, the D-1520 is the clear choice despite its lower core count. The D-1520 also supports DDR4 memory and PCIe Gen 3, offering modern platform features that the E5645 cannot match with its DDR3 and PCIe Gen 2.

Specification Differences

The two processors diverge on nearly every major specification except for L1 and L2 cache sizes, which are identical at 64 KB and 256 KB per core respectively. The core and thread counts differ: the E5645 has 6 cores and 12 threads, while the D-1520 has 4 cores and 8 threads. Base clocks are close but not equal, with the E5645 at 2.40 GHz and the D-1520 at 2.20 GHz. Boost clocks are 2.67 GHz for the E5645 and 2.60 GHz for the D-1520.

Thermal design power is a major differentiator: 80 watts for the E5645 versus 45 watts for the D-1520. The sockets are incompatible, with the E5645 using Intel Socket 1366 and the D-1520 using Intel BGA 1667. The process node favors the D-1520 at 14 nm versus 32 nm. Transistor counts are 1,170 million for the E5645 and 3,200 million for the D-1520. Die sizes are 239 mm² and 246 mm² respectively.

Cache hierarchy differs in L3 allocation: the E5645 has 12 MB shared, while the D-1520 has 1.5 MB per core. Memory support is DDR3 triple-channel for the E5645 and DDR3/DDR4 dual-channel for the D-1520. PCIe generation is Gen 2 for the E5645 and Gen 3 with 24 lanes for the D-1520. The production status also differs: the E5645 is end-of-life, while the D-1520 remains active. The release dates are March 2010 for the E5645 and March 2015 for the D-1520. The D-1520 has a launch MSRP of $199.

FAQ

Q: Which processor has a higher core count?

A: The Intel Xeon E5645 has 6 cores and 12 threads, while the Intel Xeon D-1520 has 4 cores and 8 threads.

Q: How do the two CPUs compare in Cinebench R23 multicore performance?

A: The E5645 scores 4187 in Cinebench R23 multicore, while the D-1520 scores 4179, a delta of 0.2% in favor of the E5645.

Q: What is the thermal design power difference between the two?

A: The E5645 has a TDP of 80 watts, whereas the D-1520 has a TDP of 45 watts, making the D-1520 significantly more power-efficient.

Q: Which processor supports DDR4 memory?

A: The Intel Xeon D-1520 supports both DDR3 and DDR4 memory, while the Intel Xeon E5645 supports only DDR3.

Q: Are there any benchmark tests where the two processors score identically?

A: Yes, in Cinebench R15 multicore, both score 421, and in Cinebench R15 single-core, both score 59.

Q: What is the process node for each chip?

A: The E5645 is fabricated on a 32 nm process, while the D-1520 uses a 14 nm process.