Encountered a hardware issue and seeking guidance on what to do next.
Encountered a hardware issue and seeking guidance on what to do next.
I own an older setup that I have overclocked to its limits while keeping it stable, but today I faced a problem that forced me to slow down the RAM to roughly half its previous speed. This was necessary to prevent crashes caused by memory issues or leaks. Since I haven’t kept up with hardware trends since building this system, I’m seeking advice on how to proceed given the current situation. The system runs a Rampage IV Black with a [email protected] GHz and originally had 48 GB of RAM (now rated at 1600 GB). The GPUs are dual 980 Ti Lighting cards at 1500 MHz, which have never caused problems for me, and I’ve never had issues after tuning them. There’s also a sound card between the two GPUs, so that should be considered when suggesting a new board. Everything has been functioning smoothly since installation.
I’m leaning toward replacing both the CPU and RAM entirely rather than sticking with mismatched components. I’m not very familiar with the latest hardware but am thinking about upgrading to something like the newest 2011-3 or similar models that support quad-channel memory and 40 PCIe lanes. I’m open to AMD options or any configuration that delivers comparable or better performance. My plan is to wait until DDR4 is fully phased out before making any changes, so I’ll likely replace the CPU and RAM now. However, I’m considering a used 4960 for now because replacing it outright would cost around $1000, and there’s always a risk it won’t maintain its performance. Could you help me understand my choices?
You're not seeing the reason for replacing your CPU. Based on what you've shared, it looks like most probably just one or a few RAM modules are faulty. You can try running MEMTEST86+ on each module individually until you find the problem.
The reason for replacing your CPU isn't immediately clear from your description. It probably points to a damaged RAM module. You can test one at a time using MEMTEST86+ until you find the problem.
It's unlikely your CPU has failed completely, especially if temperatures and voltages have been managed well. CPUs are quite durable. Even after clearing all memory tests, a motherboard issue is more probable than a CPU failure, given how long it's been running.
A used replacement wouldn't be worth considering because new CPUs cost around $275—just enough for a Ryzen 3600 or X570 board. It has 16 PCIe lanes and dual-channel memory, but only 16 PCIe 4.0 lanes and speeds from 3200Mhz to 3600Mhz. Few tasks really benefit from quad channel memory, which is usually limited. I’d expect faster DDR4 RAM to make up for any shortcomings. The CPU is quicker and more energy-efficient. Unless you’re fully utilizing all those PCIe lanes, they’re probably unnecessary. A dual GPU works well in a PCIe 3.0 x8/x8 setup, let alone a PCIe 4.0 one.
Regarding GPUs, swapping out the two GTX 980 Ti for a single RTX 2070 would be a much better upgrade, especially since it would outperform them in most games—except perhaps for extremely rare SLI-optimized titles. Overall, it would save you money and power, and new components usually come with updated warranties.
My suggestion is to locate and replace the faulty RAM module. If it’s worse, replacing your hardware entirely makes more sense from a cost and efficiency standpoint. As you didn’t mention using it for workstation tasks, it seems you’re not running resource-intensive applications. You’d be better off with multiple high-speed NVMe SSDs, fiber optics, and other bandwidth-heavy devices. Spending on 40+ PCIe lanes and quad channel memory isn’t practical.
It crashes on every one of the 8 sticks, with higher clock speeds causing it to fail more quickly and fewer sticks leading to faster crashes. I can extract an hour by running all of them at 1333. If this is true, I plan to build a system with a 9900k or 3950x based setup using a 2080ti, not upgrading just yet, but you're confirming it's not worth the parts cost. I'll re-test with memcheck since I forgot it existed.
I have identified the problem and determined the next steps, opting for the suggested new system. Thanks velocity!