Intel Xeon E5-2628 v3

The Intel Xeon E5-2628 v3 is an 8-core, 16-thread server/workstation processor built on Intel’s 22 nm Haswell-EP architecture, designed for the Intel Socket 2011-3 platform. With a base clock of 2.50 GHz and a boost clock of 3.00 GHz, this chip sits in the mid-range of the Xeon E5 v3 family, targeting single-socket and dual-socket configurations where balanced throughput and power efficiency are priorities. Released in September 2014 and now end-of-life, the E5-2628 v3 remains relevant in benchmark databases as a reference point for older enterprise silicon, with its performance metrics placing it near the 48th percentile of all CPUs tracked.

Platform and Compatibility

The E5-2628 v3 uses the Intel Socket 2011-3, a platform shared with Haswell-EP and Broadwell-EP Xeon processors. This socket supports quad-channel DDR4 memory, and the processor’s memory bus is rated at 68.3 GB/s of peak bandwidth. ECC memory is supported, which is a critical feature for server and workstation reliability, ensuring data integrity in compute-heavy or long-running workloads. The memory controller handles DDR4 modules, though the FACT PACK does not specify maximum capacity or speed tiers, so those details remain undefined here.

For expansion, the CPU provides 40 PCIe Gen 3 lanes (CPU only), which is a generous allocation for a mid-range Xeon. This allows multiple GPUs, NVMe storage adapters, or high-speed network cards to be attached directly to the processor, reducing latency and avoiding chipset bottlenecks. The absence of integrated graphics means a discrete GPU is mandatory for any display output, which is typical for server/workstation parts. Upgrade path considerations are tied to the Socket 2011-3 ecosystem: the same board can accommodate other Haswell-EP or Broadwell-EP Xeons, but the FACT PACK does not list specific compatible models, so the analysis stops at the socket and memory support levels. The architecture is Haswell, codenamed Haswell-EP, with a die size of 356 mm² containing 2,600 million transistors, indicating a dense, mature design for its era.

Power and Thermals

The E5-2628 v3 has a TDP of 85 watts, classifying it as a low-to-mid power draw part within the Xeon E5 lineup. This TDP figure implies that a capable air cooler is sufficient for most installations, as the heat output is modest compared to higher-core-count Xeons that often exceed 100 watts. For dual-socket servers, two such processors would total 170 watts of thermal load, which still fits within standard 1U or 2U chassis cooling designs. The 22 nm process node helps keep power efficiency reasonable, and the 2.50 GHz base clock with a 3.00 GHz boost suggests the chip can sustain moderate all-core loads without exotic cooling. Since the multiplier is locked (multiplierUnlocked: false), overclocking is not an option, so thermals are governed entirely by the stock frequency behavior and workload demands. The data does not provide specific cooling requirements beyond the TDP, so any recommendation must remain qualitative: an 85 W TDP is well within the range of tower air coolers or low-profile server heatsinks, and no liquid cooling is necessary.

Benchmark Performance

Benchmark results for the E5-2628 v3 show a processor that is competent for older multi-threaded tasks but lags in single-thread performance. In Cinebench R15, the multi-core score is 723, while the single-core score is 102. Moving to Cinebench R20, the multi-core score rises to 3015 and single-core to 425. In the more recent Cinebench R23, the multi-core score is 7179 and single-core is 1013. These numbers reveal a pattern: the multi-core scaling is strong, with the R23 multi-core score being about 7.1 times the single-core score, which is close to the theoretical maximum for 8 cores with 16 threads, indicating efficient thread utilization. However, the absolute single-core scores are low by modern standards, reflecting the modest 3.00 GHz boost clock and the Haswell architecture’s older IPC (instructions per clock).

The average benchmark score across all recorded tests is 2076, placing the chip at the 48th percentile of all CPUs in the database. This means it outperforms roughly half of all tracked processors, but it is firmly in the mid-range tier. Compared to its nearest rivals, the performance differences are marginal. The Intel Core i5-1130G7 has an average score of 2084, which is 0.4% higher than the E5-2628 v3, a negligible difference that puts them effectively on par. The AMD Ryzen Embedded R2544 scores 2086, 0.5% higher, again within noise. The Intel Xeon E3-1585 v5 scores 2067, which is 0.5% lower than the E5-2628 v3, and the Intel Core i5-8500T scores 2090, 0.7% higher. None of these deltas exceed 1%, so the E5-2628 v3 sits in a tightly clustered performance band where no rival holds a meaningful advantage in aggregate benchmarks. This suggests that for general-purpose workloads, the choice between these processors would come down to platform features (memory channels, PCIe lanes, ECC support) rather than raw compute speed.

FAQ

Q: What is the socket type for the Intel Xeon E5-2628 v3?

A: The processor uses Intel Socket 2011-3, which is shared with other Haswell-EP and Broadwell-EP Xeon processors.

Q: Does the E5-2628 v3 support ECC memory?

A: Yes, ECC memory support is enabled, which is essential for error-correcting workloads in servers and workstations.

Q: How many PCIe lanes does the CPU provide?

A: The processor provides 40 PCIe Gen 3 lanes, CPU only, suitable for multiple add-in cards.

Q: What is the memory bandwidth of the E5-2628 v3?

A: The quad-channel DDR4 memory bus delivers a peak bandwidth of 68.3 GB/s.

Q: Is the multiplier unlocked for overclocking?

A: No, the multiplier is locked, so the processor runs at its specified 2.50 GHz base and 3.00 GHz boost clocks.

Q: How does the E5-2628 v3 compare to the Intel Core i5-8500T in average benchmark score?

A: The Core i5-8500T has an average score of 2090, which is 0.7% higher than the E5-2628 v3’s 2076, a negligible difference.

How It Compares

Against the Intel Core i5-1130G7, the E5-2628 v3 is essentially tied, with the i5-1130G7 scoring 2084 versus the Xeon’s 2076, a delta of -0.4%. The i5-1130G7 is a low-power mobile part, so this comparison highlights that the older Xeon’s extra cores and memory bandwidth do not translate into a lead in aggregate benchmarks, likely because the i5’s newer architecture compensates in single-thread and lightly threaded tasks.

The Intel Xeon E3-1585 v5 is the only rival that the E5-2628 v3 beats, with the E3-1585 v5 scoring 2067, which is 0.5% lower. Both are Xeon parts, but the E3-1585 v5 is a single-socket, lower-core-count design; the slight advantage for the E5-2628 v3 comes from its 8 cores and 16 threads, which aid in multi-threaded Cinebench runs.

The AMD Ryzen Embedded R2544 scores 2086, putting it 0.5% ahead of the E5-2628 v3. This embedded part likely benefits from a more modern architecture and higher clocks, but the margin is so slim that the Xeon’s quad-channel memory and 40 PCIe lanes offer a tangible platform-level advantage for server workloads that the Ryzen cannot match.

The Intel Core i5-8500T scores 2090, 0.7% higher than the E5-2628 v3. The i5-8500T is a desktop-derived low-power chip with 6 cores, so its higher aggregate score suggests better single-thread performance, but the Xeon counters with more threads and ECC support, making it the better fit for reliability-focused environments.

Who Should Consider It

The E5-2628 v3 is best suited for workloads that scale with core count and thread count, given its 8-core/16-thread configuration and strong multi-core Cinebench scores. In Cinebench R23, the multi-core score of 7179 is roughly 7.1 times the single-core score of 1013, indicating excellent parallel scaling. This makes it a reasonable choice for older rendering tasks, batch processing, or virtualization hosts where multiple virtual machines can utilize the threads. However, its single-core performance is weak, with an R23 single-core score of 1013, so it is not recommended for gaming or any application that relies heavily on single-threaded performance, such as older game engines or lightly threaded productivity apps.

For office and general productivity, the E5-2628 v3 is overkill in core count but lacks the per-core speed of modern chips. The average benchmark score of 2076 places it at the 48th percentile, meaning it will handle everyday tasks without issue, but there is no advantage over cheaper alternatives. For content creation, the multi-threaded Cinebench scores suggest competence in video encoding or 3D rendering, but the 85 W TDP and end-of-life status mean newer parts with similar scores offer better power efficiency. In a server context, the quad-channel DDR4 memory (68.3 GB/s bandwidth), ECC support, and 40 PCIe Gen 3 lanes make it a solid foundation for a dual-socket build where aggregate throughput matters more than single-core speed. The data does not support recommending it for high-frequency trading or real-time analytics, where single-thread latency is critical. Ultimately, this processor fits a narrow niche: legacy enterprise systems that require ECC, many PCIe lanes, and reliable multi-threading, where absolute performance per core is secondary to platform robustness.