benchmark-comparison.md

Benchmark Comparison

Compared results from:

• zedis-benchmark-matrix.txt
• valkey-benchmark-matrix.txt

Valkey version used for comparison: Valkey 9.0.3 (6e63ad9c/1) 64 bit

Target in both cases: 127.0.0.1:6379

Hardware used for these runs:

• CPU: AMD Ryzen 9 7900X 12-Core Processor
• Cores / Threads: 12 / 24
• Max clock: 5.74 GHz
• Cache: L1d 384 KiB, L1i 384 KiB, L2 12 MiB, L3 64 MiB
• RAM: 30 GiB

Throughput

Case	Zedis SET	Valkey SET	Winner	Zedis GET	Valkey GET	Winner
Single client	73,099.41	68,775.79	Zedis +6.3%	73,855.24	66,844.91	Zedis +10.5%
50 clients	132,450.33	146,627.56	Valkey +10.7%	140,252.45	146,842.88	Valkey +4.7%
100 clients	145,985.41	145,560.41	Zedis +0.3%	144,508.67	147,058.83	Valkey +1.8%
128B payload	134,228.19	147,929.00	Valkey +10.2%	144,508.67	146,198.83	Valkey +1.2%
1KB payload	132,978.73	140,845.08	Valkey +5.9%	142,653.36	146,412.88	Valkey +2.6%

p50 Latency

Case	Zedis SET	Valkey SET	Zedis GET	Valkey GET
Single client	0.015 ms	0.015 ms	0.015 ms	0.015 ms
50 clients	0.199 ms	0.175 ms	0.183 ms	0.175 ms
100 clients	0.351 ms	0.351 ms	0.351 ms	0.351 ms
128B payload	0.191 ms	0.175 ms	0.183 ms	0.175 ms
1KB payload	0.199 ms	0.183 ms	0.183 ms	0.175 ms

Summary

• Zedis wins the single-client benchmark for both SET and GET.
• Valkey wins most multi-client and larger-payload cases.
• At 100 clients, SET throughput is nearly tied.

Notes

• These numbers come directly from the two benchmark matrix output files in this repository.
• Results are sensitive to machine load, server build mode, and whether both servers were configured identically beyond the default port.