Figures

AMD RDNA 3 (AMD Radeon RX 7900 XTX): recorded figures
Metric Value Evidence Basis Method Source
Clock frequency gpu 2300 MHz Confirmed Vendor-rated Nominal clock Stated by the vendor The lower declared clock. The separately named upper clock is a different quantity and uses peak-clock.
Clock frequency gpu 2500 MHz Confirmed Vendor-rated Peak clock Stated by the vendor AMD calls this the boost clock. Its separate game clock is a nominal-clock measurement.
Peak FP32 rate gpu 61.42 TFLOP/s Confirmed Theoretical Theoretical peak Stated by the vendor 12,288 stream processors × 2 FP32 operations per cycle × 2500 MHz boost clock. AMD’s 61.42 TFLOPS uses 12,288 stream processors × two FP32 operations per cycle × the 2500 MHz boost clock, never the 2300 MHz game clock.
Geekbench Compute score, opencl gpu 208 566 Confirmed Measured Benchmark chart average Published independently geekbench-compute 7 , opencl A dated capture of the benchmark publisher’s rolling average for this graphics processor, read from the 25 July 2026 OpenCL chart. Version 7 calibrates a compute result against a baseline of 100,000, so it may be compared only with another Geekbench 7 OpenCL figure from this curation round, never with a Metal figure, an earlier version or a processor score. The chart also lists this processor under “Radeon RX 7900 XTX”, at 212 610; the row naming the product as its maker does was taken, and the disagreement is recorded in the conflict ledger.
Rated power consumption gpu 355 W Confirmed Vendor-rated Rated power Stated by the vendor The vendor’s board-power rating, not a measured system draw.

Conflicting sources

Geekbench Compute score

  1. 208 566 accepted

    Device and Score table, row "AMD Radeon RX 7900 XTX": 208566.

  2. 212 610

    Device and Score table, row "Radeon RX 7900 XTX": 212610.

Editorial decision 2026-08-18

The chart keys its rows on the string a driver reports, so one product reaches it under several names with several averages. The rule this catalog applies, here and in every later round: take the row that names the product as its maker names it, and treat a string carrying tokens the maker does not use — a “GPU” suffix, a die designation, another company’s prefix — as the driver’s name rather than the product’s. The same prefixed-versus-bare pair as the GTX 1080, decided the same way. The unprefixed row scores two percent higher, and taking it for that reason would make the choice of name a performance claim.

Fitted to

Editorial notes

The general-shader FP32 peak is separate from tensor, AI and ray-tracing rates.

Sources

Back to top