Performance problems with Xeon E5 2620 V2 processors
Performance problems with Xeon E5 2620 V2 processors
Hey everyone, I’m dealing with a puzzling problem with my old E5 2620 v2. It’s acting strange under load. I’ve made some headway so far, but I’m stuck again. I built a "server" using components from an older IBM machine—specs are 32GB DDR3 RAM, 1600MHz (4x8GB) processor, and a peculiar Chinese motherboard (possibly the cause). Here’s the link: https://www.amazon.com.au/dp/B0CBV5YG29?...ct_details.
Initially, the CPU clock speed stayed flat at 2.1GHz, which improved after tweaking BIOS settings. Now it’s stuck capping around 2.4GHz even when pushing higher loads. I’ve tried installing Windows to help diagnose, but performance remains limited. I’m pretty sure my BIOS is correctly configured for full boost, and the motherboard seems fine.
The issue might lie in how the power delivery system handles the CPU under stress. The Chinese model feels different, so I don’t have much reference material. Any insights or suggestions would be greatly appreciated—thanks a lot!
CPU temperatures stay around the low 30s while power consumption remains near 55-60W during operation
This was the top offer I discovered for a matx or smaller LGA 2011 board. It came with some risk, but the board functions well and remains reliable (excluding the clock speed concern).
The engine in this vehicle is functioning properly. It can provide the power I need, though it sometimes struggles to meet my expectations—it works when possible but isn’t always optimal. ...If that level was sufficient for you, perhaps you wouldn’t be here discussing it... https://ark.intel.com/content/www/us/en/...0-ghz.html This is mainly about the socket and chipset; unlocking all the CPU capabilities might require a different motherboard. You might want to explore a Supermicro mainboard, as it’s more typical for server environments and offers broader support.
Consider testing ThrottleStop version 9.6 and capturing screenshots of the TRL and C States interface during idle. A 2.6 GHz frequency is achievable only with one core running. Ensure the BIOS enables the core C states; some Chinese boards may not do this properly. Execute the TS Bench, configure it for a single-threaded run, and record a screenshot of ThrottleStop while the CPU is loading. Close Task Manager and other monitoring tools during testing. These processors aren't built to operate at 2.6 GHz when all cores are engaged.
In my experience, people often ask the same thing about these Chinese salvage boards—specifically, their lack of turbo boost performance compared to other issues. It seems you’re likely facing a similar concern.
Brother, I'm not aiming to criticize you. I'm happy you have the ability to create bios for these boards. However, it seems our main target probably doesn't have a chipset or bios setup that supports turbo boost. I'm trying to explain why this problem is happening.
He confirmed the turbo boost functions properly now. The main issue seems to be his expectations. Locked CPUs cannot handle all core overclocking with any motherboard or chipset. These processors don’t reach 2.60 GHz when more than one core is running. Even BIOS updates can’t alter this restriction.
Basically… if it holds up, it's okay. You don't have to overcomplicate things to set up a NAS. SMB works well on one thread, and with gigabit connections it should handle fine. For a basic home setup, the processor will suffice, and a few hundred MHz won't be an issue. Just keep it stable if you can.