I wanted to get some feedback about GPU Memory Currents reported by HWiNFO for two AMD Navi 10 variants running the same application software.
In various workloads, GPU Memory Current (VDDIO) of a 5700 is about 1.5 times greater than VDDIO of a 5700 XT.
VDDIO for the 8 GB GDDR6 of the 5700 running above ~20A (or ~27 W) is puzzling to me. Does that seem too high to be possible to anyone?
GPUs
AMD Radeon RX 5700 (Navi 10 Rev 4) with 8GB Micron GDDR6 @ 1746 MHz
AMD Radeon RX 5700 XT (Navi 10 Rev 1) with 8GB Samsung GDDR6 @ 1742 MHz
HWiNFO64 v6.43-4385
Windows 10 OS Build 19042.804
Radeon Software Version 21.2.3
Notes:
GPU PPT Limit is set to 104 W using Wattman Power Tuning. This is an operational requirement.
HWiNFO64 v6.43-4385 is used because it reports a maximum GPU Memory Voltage (VDDIO) correctly (I think) as 1.356V rather than HWiNFO64 v6.42 which reports GPU VDDIO as 0.850V.
The currrent (A) values below are averages.
GPU, APP, VDDCR_GFX_A, VDDIO_A, VDDCR_SOC_A, VDDCI_MEM_A, MEM_CTL_UTIL_PCT, APP_OUTPUT
5700_GDDR6_Micron, Furmark1.25_1920x1080, 58, 19, 14, 21, 33, 61
5700XT_GDDR6_Samsung, Furmark1.25_1920x1080, 67, 12, 8, 15, 31, 72
5700_GDDR6_Micron, OCCT3_VRAM_5minute, 38, 17, 13, 20, 33, pass
5700XT_GDDR6_Samsung, OCCT3_VRAM_5minute, 35, 11, 6, 15, 23, pass
5700_GDDR6_Micron, OCCT3_3D_5minute, 79, 11, 9, 20, 5, pass
5700XT_GDDR6_Samsung, OCCT3_3D_5minute, 91, 7, 4, 15, 4, pass
5700_GDDR6_Micron, Luxmark31_luxball, 49, 24, 14, 20, 63, 25164
5700XT_GDDR6_Samsung, Luxmark31_luxball, 60, 15, 8, 15, 50, 26234
5700_GDDR6_Micron, lolminer1.21_benchmark_epoch_385, 33, 32, 16, 20, 75, 30
5700XT_GDDR6_Samsung, lolminer1.21_ benchmark_epoch_385, 50, 21, 15, 16, 90, 47
In various workloads, GPU Memory Current (VDDIO) of a 5700 is about 1.5 times greater than VDDIO of a 5700 XT.
VDDIO for the 8 GB GDDR6 of the 5700 running above ~20A (or ~27 W) is puzzling to me. Does that seem too high to be possible to anyone?
GPUs
AMD Radeon RX 5700 (Navi 10 Rev 4) with 8GB Micron GDDR6 @ 1746 MHz
AMD Radeon RX 5700 XT (Navi 10 Rev 1) with 8GB Samsung GDDR6 @ 1742 MHz
HWiNFO64 v6.43-4385
Windows 10 OS Build 19042.804
Radeon Software Version 21.2.3
Notes:
GPU PPT Limit is set to 104 W using Wattman Power Tuning. This is an operational requirement.
HWiNFO64 v6.43-4385 is used because it reports a maximum GPU Memory Voltage (VDDIO) correctly (I think) as 1.356V rather than HWiNFO64 v6.42 which reports GPU VDDIO as 0.850V.
The currrent (A) values below are averages.
GPU, APP, VDDCR_GFX_A, VDDIO_A, VDDCR_SOC_A, VDDCI_MEM_A, MEM_CTL_UTIL_PCT, APP_OUTPUT
5700_GDDR6_Micron, Furmark1.25_1920x1080, 58, 19, 14, 21, 33, 61
5700XT_GDDR6_Samsung, Furmark1.25_1920x1080, 67, 12, 8, 15, 31, 72
5700_GDDR6_Micron, OCCT3_VRAM_5minute, 38, 17, 13, 20, 33, pass
5700XT_GDDR6_Samsung, OCCT3_VRAM_5minute, 35, 11, 6, 15, 23, pass
5700_GDDR6_Micron, OCCT3_3D_5minute, 79, 11, 9, 20, 5, pass
5700XT_GDDR6_Samsung, OCCT3_3D_5minute, 91, 7, 4, 15, 4, pass
5700_GDDR6_Micron, Luxmark31_luxball, 49, 24, 14, 20, 63, 25164
5700XT_GDDR6_Samsung, Luxmark31_luxball, 60, 15, 8, 15, 50, 26234
5700_GDDR6_Micron, lolminer1.21_benchmark_epoch_385, 33, 32, 16, 20, 75, 30
5700XT_GDDR6_Samsung, lolminer1.21_ benchmark_epoch_385, 50, 21, 15, 16, 90, 47
Last edited: