PCI Express (PCIe) is the high-speed interface connecting your GPU, NVMe SSDs, and expansion cards to the CPU and motherboard. Understanding bandwidth and lanes helps you make smarter build decisions.
PCIe Generations and Speed
Each PCIe generation doubles the bandwidth of the previous: PCIe 3.0 provides ~1GB/s per lane, PCIe 4.0 ~2GB/s per lane, PCIe 5.0 ~4GB/s per lane. A GPU in an x16 slot gets 16 lanes: PCIe 4.0 x16 = 32GB/s bidirectional bandwidth. This is far beyond what any current consumer GPU's internal bandwidth needs — even the RTX 5090 doesn't saturate a PCIe 4.0 x16 connection.
x16 vs x8 for GPUs
Running a GPU at x8 PCIe 4.0 (common when multiple M.2 slots are populated on some motherboards) loses at most 1–2% performance in real-world gaming benchmarks. PCIe bandwidth is not the GPU bottleneck for consumer gaming workloads. Only in highly synthetic bandwidth-saturating benchmarks does x8 vs x16 show meaningful difference.
NVMe SSDs and PCIe Generation
PCIe 5.0 NVMe SSDs achieve 10,000+ MB/s sequential reads. PCIe 4.0 achieves 7,000 MB/s. PCIe 3.0 achieves 3,500 MB/s. For game loading and general OS use, the difference between 3.0 and 4.0 is subtle. For large file transfers (video production, large dataset work), PCIe 4.0 and 5.0 make a real difference in time-to-complete.
Checking Your System's PCIe Allocation
In Windows: Device Manager → View → Resources by connection shows PCIe device assignments. GPU-Z shows what PCIe version and width your GPU is currently running at. Check this if you suspect a slot or routing issue — some boards automatically drop to x8 when both PCIe slots are populated.