Manufacturers ship RX 9070 XT cards with GDDR6 memory from either Hynix or Samsung. Clocks are the same, yet performance and thermals can differ slightly—and the same goes for overclocking. Samsung chips tend to reach higher memory clocks, which suggests better performance, but it isn’t that simple. We will compare the practical differences between both variants using the same card: Gigabyte’s Radeon RX 9070 XT Gaming OC.
First, to illustrate, let’s see how overclocking improved performance across a suite of tests on the Hynix-equipped card. I didn’t aim for the absolute limit—the goal was a fast run and an optimistic scenario where the benefit of a higher memory clock is clearly visible, so we can compare how memory clock affects performance.
If you skipped the previous chapter, here’s a reminder: with tighter timings, the memory won’t overclock as high as it does at stock timings. And higher clock speeds can easily outweigh the gains from tighter timings.
| Gigabyte RX 9070 XT GOC Ice (Hynix), default | Fast Timing, ~2640 MHz | Default timing, ~2740 MHz | Fast Timing, ~2640 MHz, rozdíl | Default timing, ~2740 MHz, rozdíl | |
|---|---|---|---|---|---|
| Cyberpunk 2077, Ultra, scaling off, 1920 × 1080 | 195,8 | 197,4 | 201,4 | 0.8 % | 2.9 % |
| Cyberpunk 2077, Ultra, FSR3 nat. AA, 1920 × 1080 | 141,1 | 141,3 | 142,9 | 0.1 % | 1.3 % |
| Cyberpunk 2077, Ultra, FSR3 bal., 1920 × 1080 | 239,5 | 239,7 | 243,7 | 0.1 % | 1.8 % |
| Cyberpunk 2077, Ultra, FSR3 bal., fg 2×, 1920 × 1080 | 459 | 455,8 | 466 | -0.7 % | 1.5 % |
| Cyberpunk 2077, RT Ultra, scaling off, 1920 × 1080 | 83,5 | 83,5 | 84,1 | 0.0 % | 0.7 % |
| Cyberpunk 2077, RT Ultra, FSR3 nat. AA, 1920 × 1080 | 81,4 | 81,2 | 81,8 | -0.3 % | 0.5 % |
| Cyberpunk 2077, RT Ultra, FSR3 bal., 1920 × 1080 | 146,6 | 146,1 | 148 | -0.3 % | 1.0 % |
| Cyberpunk 2077, RT Ultra, FSR3 bal., fg 2×, 1920 × 1080 | 285,7 | 284,4 | 288,2 | -0.5 % | 0.9 % |
| Cyberpunk 2077, RT Overdrive, scaling off, 1920 × 1080 | 42 | 42 | 42,4 | 0.0 % | 1.0 % |
| Cyberpunk 2077, RT Overdrive, FSR3 nat. AA, 1920 × 1080 | 41,3 | 41,4 | 41,7 | 0.2 % | 1.0 % |
| Cyberpunk 2077, RT Overdrive, FSR3 perf., 1920 × 1080 | 104,2 | 104,2 | 105,4 | 0.0 % | 1.2 % |
| Cyberpunk 2077, RT Overdrive, FSR3 perf., fg 2×, 1920 × 1080 | 204,5 | 204,6 | 207,3 | 0.1 % | 1.4 % |
| Cyberpunk 2077, Ultra, scaling off, 2560 × 1440 | 132,8 | 133,2 | 134,9 | 0.3 % | 1.6 % |
| Cyberpunk 2077, Ultra, FSR3 nat. AA, 2560 × 1440 | 86,6 | 86,5 | 87,3 | -0.1 % | 0.8 % |
| Cyberpunk 2077, Ultra, FSR3 bal., 2560 × 1440 | 178,3 | 178,4 | 181,3 | 0.1 % | 1.7 % |
| Cyberpunk 2077, Ultra, FSR3 bal., fg 2×, 2560 × 1440 | 335,4 | 334,4 | 338,8 | -0.3 % | 1.0 % |
| Cyberpunk 2077, RT Ultra, scaling off, 2560 × 1440 | 53,5 | 53,5 | 53,9 | 0.0 % | 0.8 % |
| Cyberpunk 2077, RT Ultra, FSR3 nat. AA, 2560 × 1440 | 51,8 | 51,9 | 52,2 | 0.2 % | 0.8 % |
| Cyberpunk 2077, RT Ultra, FSR3 bal., 2560 × 1440 | 106,8 | 106,4 | 107,6 | -0.4 % | 0.8 % |
| Cyberpunk 2077, RT Ultra, FSR3 bal., fg 2×, 2560 × 1440 | 205,2 | 204,1 | 206,6 | -0.5 % | 0.7 % |
| Cyberpunk 2077, RT Overdrive, scaling off, 2560 × 1440 | 25,6 | 25,7 | 25,8 | 0.4 % | 0.8 % |
| Cyberpunk 2077, RT Overdrive, FSR3 nat. AA, 2560 × 1440 | 25,2 | 25,2 | 25,4 | 0.0 % | 0.8 % |
| Cyberpunk 2077, RT Overdrive, FSR3 perf., 2560 × 1440 | 72,9 | 73,3 | 73,8 | 0.6 % | 1.2 % |
| Cyberpunk 2077, RT Overdrive, FSR3 perf., fg 2×, 2560 × 1440 | 141,9 | 142,6 | 143,1 | 0.5 % | 0.9 % |
| Cyberpunk 2077, Ultra, scaling off, 3840 × 2160 | 63 | 63,5 | 63,9 | 0.8 % | 1.4 % |
| Cyberpunk 2077, Ultra, FSR3 nat. AA, 3840 × 2160 | 38,2 | 38,4 | 38,6 | 0.5 % | 1.1 % |
| Cyberpunk 2077, Ultra, FSR3 bal., 3840 × 2160 | 94,7 | 95,3 | 95,7 | 0.6 % | 1.1 % |
| Cyberpunk 2077, Ultra, FSR3 bal., fg 2×, 3840 × 2160 | 168,3 | 169,5 | 170,3 | 0.7 % | 1.2 % |
| Cyberpunk 2077, RT Ultra, scaling off, 3840 × 2160 | 26,1 | 26,2 | 26,3 | 0.4 % | 0.8 % |
| Cyberpunk 2077, RT Ultra, FSR3 nat. AA, 3840 × 2160 | 25,2 | 25,3 | 25,4 | 0.4 % | 0.8 % |
| Cyberpunk 2077, RT Ultra, FSR3 bal., 3840 × 2160 | 58,5 | 58,8 | 59,2 | 0.5 % | 1.2 % |
| Cyberpunk 2077, RT Ultra, FSR3 bal., fg 2×, 3840 × 2160 | 109 | 109,7 | 110,4 | 0.6 % | 1.3 % |
| Cyberpunk 2077, RT Overdrive, scaling off, 3840 × 2160 | 12 | 12 | 12,1 | 0.0 % | 0.8 % |
| Cyberpunk 2077, RT Overdrive, FSR3 nat. AA, 3840 × 2160 | 11,8 | 11,8 | 11,9 | 0.0 % | 0.9 % |
| Cyberpunk 2077, RT Overdrive, FSR3 perf., 3840 × 2160 | 39,1 | 39,3 | 39,5 | 0.5 % | 1.0 % |
| Cyberpunk 2077, RT Overdrive, FSR3 perf., fg 2×, 3840 × 2160 | 74,6 | 75 | 75,6 | 0.5 % | 1.3 % |
| F1 24, high, TAA, 1920 × 1080 | 310,7 | 309,3 | 313,7 | -0.5 % | 1.0 % |
| F1 24, high, TAA, 2560 × 1440 | 251,3 | 250 | 254,3 | -0.5 % | 1.2 % |
| F1 24, high, TAA, 3480 × 2160 | 159,7 | 162 | 161,7 | 1.4 % | 1.3 % |
| F1 24, ultra high, TAA, 1920 × 1080 | 132,3 | 133,3 | 134 | 0.8 % | 1.3 % |
| F1 24, ultra high, TAA, 2560 × 1440 | 97,3 | 98,3 | 98,7 | 1.0 % | 1.4 % |
| F1 24, ultra high, TAA, 3480 × 2160 | 55 | 55,3 | 56 | 0.6 % | 1.8 % |
| F1 24, ultra high, FSR bal., 1920 × 1080 | 176,7 | 177 | 178 | 0.2 % | 0.7 % |
| F1 24, ultra high, FSR bal., 2560 × 1440 | 147,3 | 148,3 | 148,7 | 0.7 % | 1.0 % |
| F1 24, ultra high, FSR bal., 3480 × 2160 | 98 | 98,3 | 99,3 | 0.3 % | 1.3 % |
| F1 24, ultra high, FSR bal. + frame gen. 2×, 1920 × 1080 | 339 | 339,7 | 341,3 | 0.2 % | 0.7 % |
| F1 24, ultra high, FSR bal. + frame gen. 2×, 2560 × 1440 | 288 | 289,3 | 291 | 0.5 % | 1.0 % |
| F1 24, ultra high, FSR bal. + frame gen. 2×, 3480 × 2160 | 188,3 | 189,7 | 191,7 | 0.7 % | 1.8 % |
| Fire Strike Extreme Score | 32926,5 | 33151,5 | 33333,5 | 0.7 % | 1.2 % |
| Fire Strike Extreme Graphics score | 35135 | 35426 | 35731 | 0.8 % | 1.7 % |
| Fire Strike Extreme Graphics test 1 [fps] | 187 | 188 | 190 | 0.5 % | 1.6 % |
| Fire Strike Extreme Graphics test 2 [fps] | 129,2 | 130,6 | 131,6 | 1.1 % | 1.9 % |
| Time Spy Score | 26642 | 26905 | 26965 | 1.0 % | 1.2 % |
| Time Spy Graphics score | 29818 | 30136 | 30310 | 1.1 % | 1.7 % |
| Time Spy Graphics test 1 [fps] | 197,3 | 199,8 | 200,6 | 1.3 % | 1.7 % |
| Time Spy Graphics test 2 [fps] | 168,8 | 170,3 | 171,5 | 0.9 % | 1.6 % |
| Time Spy Extreme Score | 12682 | 12801 | 12877 | 0.9 % | 1.5 % |
| Time Spy Extreme Graphics score | 14512 | 14698 | 14811 | 1.3 % | 2.1 % |
| Time Spy Extreme Graphics test 1 [fps] | 92,5 | 93,7 | 94,4 | 1.3 % | 2.1 % |
| Time Spy Extreme Graphics test 2 [fps] | 84,9 | 86 | 86,6 | 1.3 % | 2.0 % |
| Speed Way Score | 6434,5 | 6595 | 6714 | 2.5 % | 4.3 % |
| Speed Way Graphics test [fps] | 64 | 66 | 67 | 3.1 % | 4.7 % |
| Port Royal Score | 18247 | 18384 | 18512,5 | 0.8 % | 1.5 % |
| Port Royal Graphics test [fps] | 84,5 | 85,1 | 85,7 | 0.7 % | 1.4 % |
| Steel Nomad Score (DX12) | 7184 | 7229 | 7285 | 0.6 % | 1.4 % |
| Steel Nomad Graphics test (DX12) [fps] | 71,9 | 72,3 | 72,9 | 0.6 % | 1.4 % |
| Steel Nomad Light Score (DX12) | 26528 | 26614 | 26792 | 0.3 % | 1.0 % |
| Steel Nomad Light Graphics test (DX12)[fps] | 196,5 | 197,1 | 198,5 | 0.3 % | 1.0 % |
| Steel Nomad Light Unlimited Score (DX12) | 26553 | 26691 | 26827 | 0.5 % | 1.0 % |
| Steel Nomad Light Unlimited Graphics test (DX12) [fps] | 196,7 | 197,7 | 198,7 | 0.5 % | 1.0 % |
| Steel Nomad Score (Vulkan) | 7415 | 7463 | 7497 | 0.7 % | 1.1 % |
| Steel Nomad Graphics test (Vulkan) [fps] | 74,2 | 74,6 | 75 | 0.5 % | 1.1 % |
| Steel Nomad Light Score (Vulkan) | 27367 | 27236 | 27391 | -0.5 % | 0.1 % |
| Steel Nomad Light Graphics test (Vulkan) [fps] | 202,7 | 201,8 | 202,9 | -0.4 % | 0.1 % |
| Steel Nomad Light Unlimited Score (Vulkan) | 27559 | 27361 | 27427 | -0.7 % | -0.5 % |
| Steel Nomad Light Unlimited Score Graphics test (Vulkan) [fps] | 204,1 | 202,7 | 203,2 | -0.7 % | -0.4 % |
| Solar Bay Score | 127327 | 128150 | 128689 | 0.7 % | 1.1 % |
| Solar Bay Graphics test [fps] | 484,1 | 487,3 | 489,3 | 0.7 % | 1.1 % |
| Solar Bay Unlimited Score | 128737 | 129292 | 130019 | 0.4 % | 1.0 % |
| Solar Bay Unlimited Graphics test [fps] | 489,5 | 491,6 | 494,4 | 0.4 % | 1.0 % |
| Wild Life Score | 144774 | 145478 | 145935 | 0.5 % | 0.8 % |
| Wild Life Graphics test [fps] | 866,9 | 871,1 | 873,9 | 0.5 % | 0.8 % |
| Wild Life Extreme Score | 56107 | 56520 | 57156 | 0.7 % | 1.9 % |
| Wild Life Extreme Graphics test [fps] | 336 | 338,5 | 342,3 | 0.7 % | 1.9 % |
| Wild Life Unlimited Score | 147307 | 149351 | 149737 | 1.4 % | 1.7 % |
| Wild Life Unlimited Graphics test [fps] | 882,1 | 894,3 | 896,6 | 1.4 % | 1.6 % |
| Wild Life Extreme Unlimited Score [fps] | 57128 | 57530 | 58245,5 | 0.7 % | 2.0 % |
| Wild Life Extreme Unlimited Graphics test [fps] | 342 | 344 | 349 | 0.6 % | 2.1 % |
| DirectX Raytracing feature test Score [fps] | 63 | 63 | 63 | 0.0 % | 0.0 % |
In some tests, the performance differences with the higher memory clock were negative. Without further testing, it’s hard to say to what extent this could be normal measurement error and to what extent it might be due to performance degradation from error correction. Given that the problematic measurements are mainly those where the performance could be limited by the CPU (such as Steel Nomad Light or F1 24 at Ultra and Full HD resolution), I’m inclined to attribute it more to measurement error. Overall, performance increased, even if not as significantly as with standard timings and a 100 MHz higher memory clock.
⠀
