AMD’s next-gen Ryzen 9000 processors are scheduled to launch on the AM5 platform next month. Featuring the Zen 5 core architecture, these CPUs boast a revamped front-end, wider execution, and a faster memory sub-system. The below sample, allegedly a Ryzen 5 9600X ES, exhibits a monumental increase in the L1 and L2 cache bandwidth over the 7600X. The cache read/write speeds are nearly 2x as fast as the Zen 4 chips, indicating substantial changes to the memory subsystem and the Infinity Fabric clocks.
The Ryzen 5 9600X (Zen 5) L1 and L2 read, write, and copy speeds are summarized below, compared to the average R5 7600X (Zen 4) sample.
AMD Ryzen 9000 (Zen 5) vs Ryzen 7000 (Zen 4) Cache Speeds
| SKU | Read Speed (GB/s) | Write Speed (GB/s) | Copy Speed (GB/s) | Latency (ns) |
|---|---|---|---|---|
| Ryzen 9600X (L1) | 3756.4 | 1884.4 | 3755.9 | 0.8 |
| Ryzen 7600X (L1) | 2029.6 | 1026.9 | 2048.1 | 0.7 |
| Ryzen 9600X (L2) | 1874.6 | 1795.1 | 1859.7 | 2.8 |
| Ryzen 7600X (L2) | 1028.5 | 1017 | 1017.6 | 2.6 |
The Ryzen 5 9600X delivers 80-90% higher L1 and L2 cache bandwidths, with slightly higher latencies. The fact that the L3 bandwidth is unchanged indicates that inter-core bandwidth/latency has likely not improved. Furthermore, the Infinity Fabric clock is set to 3 GHz, the same as most Ryzen 7000 EXPO profiles.
Minimal Improvement in CPU-Z IPC
The below CPU-Z benchmark was conducted on a Ryzen 5 9600X sample overclocked to 5.7 GHz. Even with a hefty +300 MHz overclock, it only manages a 13-14% increase in single and multi-core performance over the 7600X.
| Ryzen 5 7600X | Ryzen 5 9600X ES | Ryzen 5 9600X OC | |
|---|---|---|---|
| 1T | 764 | 776 | 871 |
| nT | 6190 | 6201 | 7096 |
This isn’t surprising as CPU-Z doesn’t feature in AMD’s Zen 5 IPC uplift chart with a paltry 1% uplift per the Zen 4 IPC chart. Multiple reports show CPU-Z is a deeply flawed benchmark, so take the above numbers with a grain of salt.
Source: HXL.
