Compare two Radeon GPUs

Useful pairs

Pre-rendered, shareable URLs. Every other catalog pair is also a static page — use the picker or the lists below.

• 32 GB vs 48 GB

  R9700 vs PRO W7900

  R9700 is gfx1201; PRO W7900 is gfx1100 — do not mix those ROCm pins. R9700 is 32 GB; PRO W7900 is 48 GB. House 35B Q4_K_M fits both at 4K on the calculator. R9700 has measured 35B llama.cpp cards on the local desktop. PRO W7900 speed stays not measured. llama.cpp first.

• 32 GB vs 32 GB

  R9700 vs PRO W7800

  R9700 is gfx1201; PRO W7800 is gfx1100 — do not mix those ROCm pins. R9700 is 32 GB; PRO W7800 is 32 GB. House 35B Q4_K_M fits both at 4K on the calculator. R9700 has measured 35B llama.cpp cards on the local desktop. PRO W7800 speed stays not measured. llama.cpp first.

• 48 GB vs 24 GB

  PRO W7900 vs RX 7900 XTX

  Same gfx1100 family package. PRO W7900 is 48 GB; RX 7900 XTX is 24 GB. House 35B Q4_K_M fits both at 4K on the calculator. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 48 GB vs 32 GB

  PRO W7900 vs PRO W7800

  Same gfx1100 family package. PRO W7900 is 48 GB; PRO W7800 is 32 GB. House 35B Q4_K_M fits both at 4K on the calculator. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 32 GB vs 32 GB

  PRO W7800 vs PRO W6800

  PRO W7800 is gfx1100; PRO W6800 is gfx1030 — do not mix those ROCm pins. PRO W7800 is 32 GB; PRO W6800 is 32 GB. House 35B Q4_K_M fits both at 4K on the calculator. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 12 GB

  PRO W7700 vs RX 7700 XT

  Same gfx1101 family package. PRO W7700 is 16 GB; RX 7700 XT is 12 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 32 GB vs 16 GB

  PRO W6800 vs RX 6800 XT

  Same gfx1030 family package. PRO W6800 is 32 GB; RX 6800 XT is 16 GB. House 35B Q4–Q6 files fit PRO W6800 at 4K on the calculator and spill on RX 6800 XT. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 32 GB vs 16 GB

  R9700 vs RX 9070 XT

  Same gfx1201 family package. R9700 is 32 GB; RX 9070 XT is 16 GB. House 35B Q4–Q6 files fit R9700 at 4K on the calculator and spill on RX 9070 XT. R9700 has measured 35B llama.cpp cards on the local desktop. RX 9070 XT speed stays not measured. llama.cpp first.

• 32 GB vs 24 GB

  R9700 vs RX 7900 XTX

  R9700 is gfx1201; RX 7900 XTX is gfx1100 — do not mix those ROCm pins. R9700 is 32 GB; RX 7900 XTX is 24 GB. House 35B Q4_K_M fits both at 4K on the calculator. R9700 has measured 35B llama.cpp cards on the local desktop. RX 7900 XTX speed stays not measured. llama.cpp first.

• 24 GB vs 20 GB

  RX 7900 XTX vs RX 7900 XT

  Same gfx1100 family package. RX 7900 XTX is 24 GB; RX 7900 XT is 20 GB. House 35B Q4–Q6 files fit RX 7900 XTX at 4K on the calculator and spill on RX 7900 XT. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 24 GB vs 16 GB

  RX 7900 XTX vs RX 9070 XT

  RX 7900 XTX is gfx1100; RX 9070 XT is gfx1201 — do not mix those ROCm pins. RX 7900 XTX is 24 GB; RX 9070 XT is 16 GB. House 35B Q4–Q6 files fit RX 7900 XTX at 4K on the calculator and spill on RX 9070 XT. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 16 GB

  RX 9070 XT vs RX 9060 XT

  RX 9070 XT is gfx1201; RX 9060 XT is gfx1200 — do not mix those ROCm pins. RX 9070 XT is 16 GB; RX 9060 XT is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 32 GB vs 16 GB

  R9700 vs RX 9070

  Same gfx1201 family package. R9700 is 32 GB; RX 9070 is 16 GB. House 35B Q4–Q6 files fit R9700 at 4K on the calculator and spill on RX 9070. R9700 has measured 35B llama.cpp cards on the local desktop. RX 9070 speed stays not measured. llama.cpp first.

• 16 GB vs 16 GB

  RX 9070 XT vs RX 9070

  Same gfx1201 family package. RX 9070 XT is 16 GB; RX 9070 is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 12 GB

  RX 9070 vs RX 9070 GRE

  Same gfx1201 family package. RX 9070 is 16 GB; RX 9070 GRE is 12 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 12 GB vs 16 GB

  RX 9070 GRE vs RX 9060 XT

  RX 9070 GRE is gfx1201; RX 9060 XT is gfx1200 — do not mix those ROCm pins. RX 9070 GRE is 12 GB; RX 9060 XT is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 32 GB vs 16 GB

  PRO W6800 vs RX 6950 XT

  Same gfx1030 family package. PRO W6800 is 32 GB; RX 6950 XT is 16 GB. House 35B Q4–Q6 files fit PRO W6800 at 4K on the calculator and spill on RX 6950 XT. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 16 GB

  RX 6800 XT vs RX 6950 XT

  Same gfx1030 family package. RX 6800 XT is 16 GB; RX 6950 XT is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 32 GB vs 64 GB shared kit

  R9700 vs Strix Halo

  R9700 is gfx1201; Strix Halo is gfx1151 — do not mix those ROCm pins. R9700 is 32 GB; Strix Halo is 64 GB shared kit. House 35B Q4_K_M fits both at 4K on the calculator. R9700 has measured 35B llama.cpp cards on the local desktop. Strix Halo speed stays not measured. llama.cpp first.

• 16 GB vs 16 GB

  RX 9070 XT vs RX 7900 GRE

  RX 9070 XT is gfx1201; RX 7900 GRE is gfx1100 — do not mix those ROCm pins. RX 9070 XT is 16 GB; RX 7900 GRE is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 20 GB vs 16 GB

  RX 7900 XT vs RX 7900 GRE

  Same gfx1100 family package. RX 7900 XT is 20 GB; RX 7900 GRE is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 12 GB vs 16 GB

  RX 7700 XT vs RX 6800 XT

  RX 7700 XT is gfx1101; RX 6800 XT is gfx1030 — do not mix those ROCm pins. RX 7700 XT is 12 GB; RX 6800 XT is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 12 GB

  RX 7800 XT vs RX 7700 XT

  Same gfx1101 family package. RX 7800 XT is 16 GB; RX 7700 XT is 12 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 16 GB

  RX 9070 XT vs RX 7800 XT

  RX 9070 XT is gfx1201; RX 7800 XT is gfx1101 — do not mix those ROCm pins. RX 9070 XT is 16 GB; RX 7800 XT is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 8 GB

  RX 7600 XT vs RX 7600

  Same gfx1102 family package. RX 7600 XT is 16 GB; RX 7600 is 8 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 12 GB vs 12 GB

  RX 7700 XT vs RX 6700 XT

  RX 7700 XT is gfx1101; RX 6700 XT is gfx1031 — do not mix those ROCm pins. RX 7700 XT is 12 GB; RX 6700 XT is 12 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 32 GB vs 16 GB

  R9700 vs RX 9060 XT

  R9700 is gfx1201; RX 9060 XT is gfx1200 — do not mix those ROCm pins. R9700 is 32 GB; RX 9060 XT is 16 GB. House 35B Q4–Q6 files fit R9700 at 4K on the calculator and spill on RX 9060 XT. R9700 has measured 35B llama.cpp cards on the local desktop. RX 9060 XT speed stays not measured. llama.cpp first.

• 16 GB vs 16 GB

  RX 7600 XT vs RX 6800 XT

  RX 7600 XT is gfx1102; RX 6800 XT is gfx1030 — do not mix those ROCm pins. RX 7600 XT is 16 GB; RX 6800 XT is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 64 GB shared kit

  RX 9070 XT vs Strix Halo

  RX 9070 XT is gfx1201; Strix Halo is gfx1151 — do not mix those ROCm pins. RX 9070 XT is 16 GB; Strix Halo is 64 GB shared kit. House 35B Q4–Q6 files fit Strix Halo at 4K on the calculator and spill on RX 9070 XT. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 12 GB

  RX 6800 XT vs RX 6700 XT

  RX 6800 XT is gfx1030; RX 6700 XT is gfx1031 — do not mix those ROCm pins. RX 6800 XT is 16 GB; RX 6700 XT is 12 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

• 16 GB vs 16 GB

  RX 9060 XT vs RX 7600 XT

  RX 9060 XT is gfx1200; RX 7600 XT is gfx1102 — do not mix those ROCm pins. RX 9060 XT is 16 GB; RX 7600 XT is 16 GB. House 35B Q4–Q6 files do not sit in either VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Neither card was measured on the local desktop. Fit numbers are calculator estimates. llama.cpp first.

Every sheet vs every other

Works with JavaScript disabled. Each heading is one GPU; the links are static pair pages.