Mashy81
Jr. Member
 Offline
Activity: 225
Merit: 1
|
 |
June 30, 2019, 02:01:53 AM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas |
|
|
|
|
|
BogdanCo
|
|
June 30, 2019, 05:45:54 AM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
For a 6 card Vega64 rig you should burn ~1100-1200 W. This is consumption at wall, so includes everything. 1800W is way to much. |
|
|
|
|
Jorkoto
Newbie
Offline
Activity: 3
Merit: 0
|
|
June 30, 2019, 09:16:39 AM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas I'm mining cnr with one vega64 (Samsung memory) windows10, 19.6.1 driver, core 1587/855mv mem 1100/855mv at 2.386 kh/s, power consumption 170W. Using only AMD Memory Tweak tool for all adjustments. |
|
|
|
|
netmebtc
Newbie
Offline
Activity: 72
Merit: 0
|
|
June 30, 2019, 09:45:39 AM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas I do some undervolt and ppt moding . set cclock to 1100/860mv mclock 1100/860mv , I got the 1450w power consumption at wall and 11.7Kh mining cnR running trm 0.5.2 with cn_config 15*14:AAA [2019-06-30 17:39:10] Mining cnr with 6 GPU workers [2019-06-30 17:39:10] GPU 0 (06:00.0) 64 CUs Temp 63C Fan 49.00% (2408 RPM) Core 1062 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 1 (19:00.0) 64 CUs Temp 63C Fan 49.00% (2446 RPM) Core 1058 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 2 (13:00.0) 64 CUs Temp 51C Fan 58.00% (2874 RPM) Core 1055 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 3 (16:00.0) 64 CUs Temp 63C Fan 52.00% (2575 RPM) Core 1053 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 4 (03:00.0) 64 CUs Temp 62C Fan 56.00% (2788 RPM) Core 1069 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 5 (09:00.0) 64 CUs Temp 62C Fan 53.00% (2599 RPM) Core 1040 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] [2019-06-30 17:39:10] Current CN configs: [2019-06-30 17:39:10] GPU 0 15*14:AAA [2019-06-30 17:39:10] GPU 1 15*14:AAA [2019-06-30 17:39:10] GPU 2 15*14:AAA [2019-06-30 17:39:10] GPU 3 15*14:AAA [2019-06-30 17:39:10] GPU 4 15*14:AAA [2019-06-30 17:39:10] GPU 5 15*14:AAA [2019-06-30 17:39:10] [2019-06-30 17:39:10] Stats Uptime: 0 days, 00:16:53 [2019-06-30 17:39:10] GPU 0 [63C, fan 49%] cnr: 1.951kh/s, avg 1.951kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] GPU 1 [63C, fan 49%] cnr: 1.949kh/s, avg 1.950kh/s, pool 1.393kh/s a:14 r:0 hw:0 [2019-06-30 17:39:10] GPU 2 [51C, fan 58%] cnr: 1.937kh/s, avg 1.945kh/s, pool 2.290kh/s a:23 r:0 hw:0 [2019-06-30 17:39:10] GPU 3 [63C, fan 52%] cnr: 1.939kh/s, avg 1.942kh/s, pool 1.892kh/s a:19 r:0 hw:0 [2019-06-30 17:39:10] GPU 4 [62C, fan 56%] cnr: 1.957kh/s, avg 1.962kh/s, pool 2.191kh/s a:22 r:0 hw:0 [2019-06-30 17:39:10] GPU 5 [62C, fan 53%] cnr: 1.928kh/s, avg 1.927kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] Total cnr: 11.66kh/s, avg 11.68kh/s, pool 12.54kh/s a:126 r:0 hw:0 I modified the PPT with the registry. GPU_P0=852;800 GPU_P1=991;810 GPU_P2=1084;820 GPU_P3=1138;830 GPU_P4=1150;840 GPU_P5=1202;850 GPU_P6=1212;850 GPU_P7=1100;860 Mem_P0=167;800 Mem_P1=500;800 Mem_P2=800;810 Mem_P3=1100;860 Fan_Min=1588 Fan_Max=4900 Fan_Target=63 Fan_Acoustic=2400 Power_Temp=75 Power_Target=0 The above parameters I have been through a number of tests, found a relatively stable data, if the voltage is set below 860mv, some cards will randomly appear crash How can I further optimize and reduce the power consumption to 1200w? |
|
|
|
|
kibatronic
Newbie
Offline
Activity: 22
Merit: 1
|
|
June 30, 2019, 03:36:55 PM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas I do some undervolt and ppt moding . set cclock to 1100/860mv mclock 1100/860mv , I got the 1450w power consumption at wall and 11.7Kh mining cnR running trm 0.5.2 with cn_config 15*14:AAA [2019-06-30 17:39:10] Mining cnr with 6 GPU workers [2019-06-30 17:39:10] GPU 0 (06:00.0) 64 CUs Temp 63C Fan 49.00% (2408 RPM) Core 1062 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 1 (19:00.0) 64 CUs Temp 63C Fan 49.00% (2446 RPM) Core 1058 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 2 (13:00.0) 64 CUs Temp 51C Fan 58.00% (2874 RPM) Core 1055 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 3 (16:00.0) 64 CUs Temp 63C Fan 52.00% (2575 RPM) Core 1053 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 4 (03:00.0) 64 CUs Temp 62C Fan 56.00% (2788 RPM) Core 1069 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 5 (09:00.0) 64 CUs Temp 62C Fan 53.00% (2599 RPM) Core 1040 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] [2019-06-30 17:39:10] Current CN configs: [2019-06-30 17:39:10] GPU 0 15*14:AAA [2019-06-30 17:39:10] GPU 1 15*14:AAA [2019-06-30 17:39:10] GPU 2 15*14:AAA [2019-06-30 17:39:10] GPU 3 15*14:AAA [2019-06-30 17:39:10] GPU 4 15*14:AAA [2019-06-30 17:39:10] GPU 5 15*14:AAA [2019-06-30 17:39:10] [2019-06-30 17:39:10] Stats Uptime: 0 days, 00:16:53 [2019-06-30 17:39:10] GPU 0 [63C, fan 49%] cnr: 1.951kh/s, avg 1.951kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] GPU 1 [63C, fan 49%] cnr: 1.949kh/s, avg 1.950kh/s, pool 1.393kh/s a:14 r:0 hw:0 [2019-06-30 17:39:10] GPU 2 [51C, fan 58%] cnr: 1.937kh/s, avg 1.945kh/s, pool 2.290kh/s a:23 r:0 hw:0 [2019-06-30 17:39:10] GPU 3 [63C, fan 52%] cnr: 1.939kh/s, avg 1.942kh/s, pool 1.892kh/s a:19 r:0 hw:0 [2019-06-30 17:39:10] GPU 4 [62C, fan 56%] cnr: 1.957kh/s, avg 1.962kh/s, pool 2.191kh/s a:22 r:0 hw:0 [2019-06-30 17:39:10] GPU 5 [62C, fan 53%] cnr: 1.928kh/s, avg 1.927kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] Total cnr: 11.66kh/s, avg 11.68kh/s, pool 12.54kh/s a:126 r:0 hw:0 I modified the PPT with the registry. GPU_P0=852;800 GPU_P1=991;810 GPU_P2=1084;820 GPU_P3=1138;830 GPU_P4=1150;840 GPU_P5=1202;850 GPU_P6=1212;850 GPU_P7=1100;860 Mem_P0=167;800 Mem_P1=500;800 Mem_P2=800;810 Mem_P3=1100;860 Fan_Min=1588 Fan_Max=4900 Fan_Target=63 Fan_Acoustic=2400 Power_Temp=75 Power_Target=0 The above parameters I have been through a number of tests, found a relatively stable data, if the voltage is set below 860mv, some cards will randomly appear crash How can I further optimize and reduce the power consumption to 1200w? If you are running the core at such so low frequency you should be able to lower the vcore to 810-820, but only if you also lower your HBM to 1028MHz or less. The reason is that the SOC will need more voltage, ie 860mV at 1107MHz than at 1028MHz. And the SOC frequency always needs to be higher or the same as the HBM. These are my settings for 4 Vega 56 pulling 650w from the wall: http://m3server.se/img/QuadVega56XMR.png |
|
|
|
|
jdhot1
Newbie
Offline
Activity: 2
Merit: 0
|
|
June 30, 2019, 09:24:25 PM |
|
Hi mate, thanks for the info below! Would you be able to provide us with the ppt for your config? I'm using a Win 10 rig with 18.5.1 drivers. Anytime I set the voltage on any of my Vega 56s lower than 850mv the rig restarts when logging in to the pc remotely using Chrome Desktop (core is set to 1143mhz with HBM @ 920mhz). The Vegas all have stock bios however I'm running mem tweak. I was never able to get the voltage down prior to mem tweak either. Your assistance would be greatly appreciated. Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas I do some undervolt and ppt moding . set cclock to 1100/860mv mclock 1100/860mv , I got the 1450w power consumption at wall and 11.7Kh mining cnR running trm 0.5.2 with cn_config 15*14:AAA [2019-06-30 17:39:10] Mining cnr with 6 GPU workers [2019-06-30 17:39:10] GPU 0 (06:00.0) 64 CUs Temp 63C Fan 49.00% (2408 RPM) Core 1062 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 1 (19:00.0) 64 CUs Temp 63C Fan 49.00% (2446 RPM) Core 1058 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 2 (13:00.0) 64 CUs Temp 51C Fan 58.00% (2874 RPM) Core 1055 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 3 (16:00.0) 64 CUs Temp 63C Fan 52.00% (2575 RPM) Core 1053 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 4 (03:00.0) 64 CUs Temp 62C Fan 56.00% (2788 RPM) Core 1069 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 5 (09:00.0) 64 CUs Temp 62C Fan 53.00% (2599 RPM) Core 1040 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] [2019-06-30 17:39:10] Current CN configs: [2019-06-30 17:39:10] GPU 0 15*14:AAA [2019-06-30 17:39:10] GPU 1 15*14:AAA [2019-06-30 17:39:10] GPU 2 15*14:AAA [2019-06-30 17:39:10] GPU 3 15*14:AAA [2019-06-30 17:39:10] GPU 4 15*14:AAA [2019-06-30 17:39:10] GPU 5 15*14:AAA [2019-06-30 17:39:10] [2019-06-30 17:39:10] Stats Uptime: 0 days, 00:16:53 [2019-06-30 17:39:10] GPU 0 [63C, fan 49%] cnr: 1.951kh/s, avg 1.951kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] GPU 1 [63C, fan 49%] cnr: 1.949kh/s, avg 1.950kh/s, pool 1.393kh/s a:14 r:0 hw:0 [2019-06-30 17:39:10] GPU 2 [51C, fan 58%] cnr: 1.937kh/s, avg 1.945kh/s, pool 2.290kh/s a:23 r:0 hw:0 [2019-06-30 17:39:10] GPU 3 [63C, fan 52%] cnr: 1.939kh/s, avg 1.942kh/s, pool 1.892kh/s a:19 r:0 hw:0 [2019-06-30 17:39:10] GPU 4 [62C, fan 56%] cnr: 1.957kh/s, avg 1.962kh/s, pool 2.191kh/s a:22 r:0 hw:0 [2019-06-30 17:39:10] GPU 5 [62C, fan 53%] cnr: 1.928kh/s, avg 1.927kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] Total cnr: 11.66kh/s, avg 11.68kh/s, pool 12.54kh/s a:126 r:0 hw:0 I modified the PPT with the registry. GPU_P0=852;800 GPU_P1=991;810 GPU_P2=1084;820 GPU_P3=1138;830 GPU_P4=1150;840 GPU_P5=1202;850 GPU_P6=1212;850 GPU_P7=1100;860 Mem_P0=167;800 Mem_P1=500;800 Mem_P2=800;810 Mem_P3=1100;860 Fan_Min=1588 Fan_Max=4900 Fan_Target=63 Fan_Acoustic=2400 Power_Temp=75 Power_Target=0 The above parameters I have been through a number of tests, found a relatively stable data, if the voltage is set below 860mv, some cards will randomly appear crash How can I further optimize and reduce the power consumption to 1200w? If you are running the core at such so low frequency you should be able to lower the vcore to 810-820, but only if you also lower your HBM to 1028MHz or less. The reason is that the SOC will need more voltage, ie 860mV at 1107MHz than at 1028MHz. And the SOC frequency always needs to be higher or the same as the HBM. These are my settings for 4 Vega 56 pulling 650w from the wall: http://m3server.se/img/QuadVega56XMR.png |
|
|
|
|
pbfarmer
Member
Offline
Activity: 340
Merit: 29
|
|
June 30, 2019, 10:02:19 PM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas I do some undervolt and ppt moding . set cclock to 1100/860mv mclock 1100/860mv , I got the 1450w power consumption at wall and 11.7Kh mining cnR ... [2019-06-30 17:39:10] Mining cnr with 6 GPU workers [2019-06-30 17:39:10] GPU 0 (06:00.0) 64 CUs Temp 63C Fan 49.00% (2408 RPM) Core 1062 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 1 (19:00.0) 64 CUs Temp 63C Fan 49.00% (2446 RPM) Core 1058 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 2 (13:00.0) 64 CUs Temp 51C Fan 58.00% (2874 RPM) Core 1055 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 3 (16:00.0) 64 CUs Temp 63C Fan 52.00% (2575 RPM) Core 1053 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 4 (03:00.0) 64 CUs Temp 62C Fan 56.00% (2788 RPM) Core 1069 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 5 (09:00.0) 64 CUs Temp 62C Fan 53.00% (2599 RPM) Core 1040 MHz Mem 1100 MHz, VDDC 862.0 mV Normally when someone complains of high power use, I would suggest checking that your settings are actually being used, but your TRM output above confirms they are indeed as expected. Instead, your ATW power measurements sound suspect - make sure your measurements are over a sufficient time period. Unless you're on a 240v circuit, 1800w sustained load should most likely be tripping your breaker (not to mention, pretty much maxing out a gold rated 1600w PSU, though a decent PSU should be able to handle that.) Also, 200w for the base platform is really high unless you're also CPU mining - even my OC'd 8700k only runs ~100w idle. I would re-check your power measurements - my 64s with ~1375 cclock (actual) / 1107 mclock / <= 850mv produce ~2450h/s @ 178w avg. At 900mv like your initial config, I would expect 200-215w each. Other than that, make sure your SOC is running at 1107 and not 1200mhz, though that seems unlikely to be the problem, as you are able to run @ 862mv (your extremely low core may possibly be allowing that tho.) |
|
|
|
|
kibatronic
Newbie
Offline
Activity: 22
Merit: 1
|
|
June 30, 2019, 11:09:06 PM |
|
Hi mate, thanks for the info below!
Would you be able to provide us with the ppt for your config?
I'm using a Win 10 rig with 18.5.1 drivers. Anytime I set the voltage on any of my Vega 56s lower than 850mv the rig restarts when logging in to the pc remotely using Chrome Desktop (core is set to 1143mhz with HBM @ 920mhz).
The Vegas all have stock bios however I'm running mem tweak.
I was never able to get the voltage down prior to mem tweak either.
Your assistance would be greatly appreciated.
.....
Sure, here are the settings I am using: http://m3server.se/img/vegasettings.pngI did end up loosening the RC to 44 since there were some HW errors. |
|
|
|
|
netmebtc
Newbie
Offline
Activity: 72
Merit: 0
|
|
July 01, 2019, 05:00:55 AM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas I do some undervolt and ppt moding . set cclock to 1100/860mv mclock 1100/860mv , I got the 1450w power consumption at wall and 11.7Kh mining cnR running trm 0.5.2 with cn_config 15*14:AAA [2019-06-30 17:39:10] Mining cnr with 6 GPU workers [2019-06-30 17:39:10] GPU 0 (06:00.0) 64 CUs Temp 63C Fan 49.00% (2408 RPM) Core 1062 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 1 (19:00.0) 64 CUs Temp 63C Fan 49.00% (2446 RPM) Core 1058 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 2 (13:00.0) 64 CUs Temp 51C Fan 58.00% (2874 RPM) Core 1055 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 3 (16:00.0) 64 CUs Temp 63C Fan 52.00% (2575 RPM) Core 1053 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 4 (03:00.0) 64 CUs Temp 62C Fan 56.00% (2788 RPM) Core 1069 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 5 (09:00.0) 64 CUs Temp 62C Fan 53.00% (2599 RPM) Core 1040 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] [2019-06-30 17:39:10] Current CN configs: [2019-06-30 17:39:10] GPU 0 15*14:AAA [2019-06-30 17:39:10] GPU 1 15*14:AAA [2019-06-30 17:39:10] GPU 2 15*14:AAA [2019-06-30 17:39:10] GPU 3 15*14:AAA [2019-06-30 17:39:10] GPU 4 15*14:AAA [2019-06-30 17:39:10] GPU 5 15*14:AAA [2019-06-30 17:39:10] [2019-06-30 17:39:10] Stats Uptime: 0 days, 00:16:53 [2019-06-30 17:39:10] GPU 0 [63C, fan 49%] cnr: 1.951kh/s, avg 1.951kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] GPU 1 [63C, fan 49%] cnr: 1.949kh/s, avg 1.950kh/s, pool 1.393kh/s a:14 r:0 hw:0 [2019-06-30 17:39:10] GPU 2 [51C, fan 58%] cnr: 1.937kh/s, avg 1.945kh/s, pool 2.290kh/s a:23 r:0 hw:0 [2019-06-30 17:39:10] GPU 3 [63C, fan 52%] cnr: 1.939kh/s, avg 1.942kh/s, pool 1.892kh/s a:19 r:0 hw:0 [2019-06-30 17:39:10] GPU 4 [62C, fan 56%] cnr: 1.957kh/s, avg 1.962kh/s, pool 2.191kh/s a:22 r:0 hw:0 [2019-06-30 17:39:10] GPU 5 [62C, fan 53%] cnr: 1.928kh/s, avg 1.927kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] Total cnr: 11.66kh/s, avg 11.68kh/s, pool 12.54kh/s a:126 r:0 hw:0 I modified the PPT with the registry. GPU_P0=852;800 GPU_P1=991;810 GPU_P2=1084;820 GPU_P3=1138;830 GPU_P4=1150;840 GPU_P5=1202;850 GPU_P6=1212;850 GPU_P7=1100;860 Mem_P0=167;800 Mem_P1=500;800 Mem_P2=800;810 Mem_P3=1100;860 Fan_Min=1588 Fan_Max=4900 Fan_Target=63 Fan_Acoustic=2400 Power_Temp=75 Power_Target=0 The above parameters I have been through a number of tests, found a relatively stable data, if the voltage is set below 860mv, some cards will randomly appear crash How can I further optimize and reduce the power consumption to 1200w? If you are running the core at such so low frequency you should be able to lower the vcore to 810-820, but only if you also lower your HBM to 1028MHz or less. The reason is that the SOC will need more voltage, ie 860mV at 1107MHz than at 1028MHz. And the SOC frequency always needs to be higher or the same as the HBM. These are my settings for 4 Vega 56 pulling 650w from the wall: http://m3server.se/img/QuadVega56XMR.pngHello mate, very appreciate for your guiding. Unfortunately, I set the core 1300mhz and mem 975mhz and voltage below 820mv, then vega does not work, it is easy to crash. |
|
|
|
|
seefatlow
Jr. Member
Offline
Activity: 80
Merit: 1
|
|
July 01, 2019, 06:23:07 AM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas I do some undervolt and ppt moding . set cclock to 1100/860mv mclock 1100/860mv , I got the 1450w power consumption at wall and 11.7Kh mining cnR running trm 0.5.2 with cn_config 15*14:AAA [2019-06-30 17:39:10] Mining cnr with 6 GPU workers [2019-06-30 17:39:10] GPU 0 (06:00.0) 64 CUs Temp 63C Fan 49.00% (2408 RPM) Core 1062 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 1 (19:00.0) 64 CUs Temp 63C Fan 49.00% (2446 RPM) Core 1058 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 2 (13:00.0) 64 CUs Temp 51C Fan 58.00% (2874 RPM) Core 1055 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 3 (16:00.0) 64 CUs Temp 63C Fan 52.00% (2575 RPM) Core 1053 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 4 (03:00.0) 64 CUs Temp 62C Fan 56.00% (2788 RPM) Core 1069 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 5 (09:00.0) 64 CUs Temp 62C Fan 53.00% (2599 RPM) Core 1040 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] [2019-06-30 17:39:10] Current CN configs: [2019-06-30 17:39:10] GPU 0 15*14:AAA [2019-06-30 17:39:10] GPU 1 15*14:AAA [2019-06-30 17:39:10] GPU 2 15*14:AAA [2019-06-30 17:39:10] GPU 3 15*14:AAA [2019-06-30 17:39:10] GPU 4 15*14:AAA [2019-06-30 17:39:10] GPU 5 15*14:AAA [2019-06-30 17:39:10] [2019-06-30 17:39:10] Stats Uptime: 0 days, 00:16:53 [2019-06-30 17:39:10] GPU 0 [63C, fan 49%] cnr: 1.951kh/s, avg 1.951kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] GPU 1 [63C, fan 49%] cnr: 1.949kh/s, avg 1.950kh/s, pool 1.393kh/s a:14 r:0 hw:0 [2019-06-30 17:39:10] GPU 2 [51C, fan 58%] cnr: 1.937kh/s, avg 1.945kh/s, pool 2.290kh/s a:23 r:0 hw:0 [2019-06-30 17:39:10] GPU 3 [63C, fan 52%] cnr: 1.939kh/s, avg 1.942kh/s, pool 1.892kh/s a:19 r:0 hw:0 [2019-06-30 17:39:10] GPU 4 [62C, fan 56%] cnr: 1.957kh/s, avg 1.962kh/s, pool 2.191kh/s a:22 r:0 hw:0 [2019-06-30 17:39:10] GPU 5 [62C, fan 53%] cnr: 1.928kh/s, avg 1.927kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] Total cnr: 11.66kh/s, avg 11.68kh/s, pool 12.54kh/s a:126 r:0 hw:0 I modified the PPT with the registry. GPU_P0=852;800 GPU_P1=991;810 GPU_P2=1084;820 GPU_P3=1138;830 GPU_P4=1150;840 GPU_P5=1202;850 GPU_P6=1212;850 GPU_P7=1100;860 Mem_P0=167;800 Mem_P1=500;800 Mem_P2=800;810 Mem_P3=1100;860 Fan_Min=1588 Fan_Max=4900 Fan_Target=63 Fan_Acoustic=2400 Power_Temp=75 Power_Target=0 The above parameters I have been through a number of tests, found a relatively stable data, if the voltage is set below 860mv, some cards will randomly appear crash How can I further optimize and reduce the power consumption to 1200w? If you are running the core at such so low frequency you should be able to lower the vcore to 810-820, but only if you also lower your HBM to 1028MHz or less. The reason is that the SOC will need more voltage, ie 860mV at 1107MHz than at 1028MHz. And the SOC frequency always needs to be higher or the same as the HBM. These are my settings for 4 Vega 56 pulling 650w from the wall:  Hello mate, very appreciate for your guiding. Unfortunately, I set the core 1300mhz and mem 975mhz and voltage below 820mv, then vega does not work, it is easy to crash. @netmebtc I can't do <850mV voltage either for my Vega56s even though my settings and ppt tables are the same. Think this has something to do with the temperature are cards are running at. Like yours, my cards run at temp >60C. Just think this might be the reason why we can't run at lower voltages. This is also the reason why the mem tweak settings don't work on my cards either. Setting my REF higher than the default value will crash my cards. |
|
|
|
|
kibatronic
Newbie
Offline
Activity: 22
Merit: 1
|
|
July 01, 2019, 11:32:56 AM
Last edit: July 01, 2019, 12:11:42 PM by kibatronic |
|
.....
Hello mate, very appreciate for your guiding. Unfortunately, I set the core 1300mhz and mem 975mhz and voltage below 820mv, then vega does not work, it is easy to crash. @netmebtc I can't do <850mV voltage either for my Vega56s even though my settings and ppt tables are the same. Think this has something to do with the temperature are cards are running at. Like yours, my cards run at temp >60C. Just think this might be the reason why we can't run at lower voltages. This is also the reason why the mem tweak settings don't work on my cards either. Setting my REF higher than the default value will crash my cards. [/quote] Did you make sure to disable all other powertstates in Overdriventool except P5? If you flash a Vega 64 BIOS you will also get a higher voltage to the HBM adding stability. |
|
|
|
|
mariano_quilmes
Newbie
Offline
Activity: 10
Merit: 0
|
|
July 01, 2019, 12:38:28 PM |
|
Hello, I would like to know if anyone can help me. I have a problem with the autotune Quick and Full option in TRM. I use three Vegas 56 but when the tunning starts, only two are making all the process at the same time, the third one is working at half the hash rate of the other two. For example, if two of them are being tested at 1900 hs, the third one is hashing at 1300 at it stays there. I have replaced risers, PSU cables also reinstalled drivers, and always one of them is slower.
I am using an EVGA G3 1000 and I have checked the wall consumption with a wattmeter which is 500w.
Thanks in advance.
Mariano.
|
|
|
|
|
netmebtc
Newbie
Offline
Activity: 72
Merit: 0
|
|
July 01, 2019, 01:09:06 PM |
|
.....
Hello mate, very appreciate for your guiding. Unfortunately, I set the core 1300mhz and mem 975mhz and voltage below 820mv, then vega does not work, it is easy to crash. @netmebtc I can't do <850mV voltage either for my Vega56s even though my settings and ppt tables are the same. Think this has something to do with the temperature are cards are running at. Like yours, my cards run at temp >60C. Just think this might be the reason why we can't run at lower voltages. This is also the reason why the mem tweak settings don't work on my cards either. Setting my REF higher than the default value will crash my cards. Did you make sure to disable all other powertstates in Overdriventool except P5? If you flash a Vega 64 BIOS you will also get a higher voltage to the HBM adding stability. [/quote] I done it. I just change the SOC clock to 1028Mhz, and set voltage to 810mv, then all vegas working normal. GPU_P0=852;800;0 GPU_P1=991;800;0 GPU_P2=1084;800;0 GPU_P3=1138;800;0 GPU_P4=1150;800;0 GPU_P5=1202;800;0 GPU_P6=1212;810;0 GPU_P7=1337;810 Mem_P0=167;800;0 Mem_P1=500;800;0 Mem_P2=800;810;0 Mem_P3=975;810 Mem_TimingLevel=0 Fan_P0=30;11 Fan_P1=50;12 Fan_P2=55;25 Fan_P3=63;48 Fan_P4=75;78 Fan_ZeroRPM=0 Fan_Acoustic=2400 Power_Target=0 But with these setting, the power consumption on wall is still too high at 1400w, It seems that only lower the core clock could lower the power consumption. By the way ,I have not used AMD Memory Tweak tool |
|
|
|
|
seefatlow
Jr. Member
Offline
Activity: 80
Merit: 1
|
|
July 01, 2019, 03:44:44 PM |
|
.....
Hello mate, very appreciate for your guiding. Unfortunately, I set the core 1300mhz and mem 975mhz and voltage below 820mv, then vega does not work, it is easy to crash. @netmebtc I can't do <850mV voltage either for my Vega56s even though my settings and ppt tables are the same. Think this has something to do with the temperature are cards are running at. Like yours, my cards run at temp >60C. Just think this might be the reason why we can't run at lower voltages. This is also the reason why the mem tweak settings don't work on my cards either. Setting my REF higher than the default value will crash my cards. Did you make sure to disable all other powertstates in Overdriventool except P5? If you flash a Vega 64 BIOS you will also get a higher voltage to the HBM adding stability. [/quote] What’s special about the P5 state ? Also are all your Vega56 flashed to 64? |
|
|
|
|
seefatlow
Jr. Member
Offline
Activity: 80
Merit: 1
|
|
July 01, 2019, 03:46:50 PM |
|
Hello, I would like to know if anyone can help me. I have a problem with the autotune Quick and Full option in TRM. I use three Vegas 56 but when the tunning starts, only two are making all the process at the same time, the third one is working at half the hash rate of the other two. For example, if two of them are being tested at 1900 hs, the third one is hashing at 1300 at it stays there. I have replaced risers, PSU cables also reinstalled drivers, and always one of them is slower.
I am using an EVGA G3 1000 and I have checked the wall consumption with a wattmeter which is 500w.
Thanks in advance.
Mariano.
What happens if you don’t run tuning and set all you GPUs to following config 16+14:AAA. Is the 3rd card still hashing with 1 thread? That’s basically what is happening when the card hashes at just 1300 |
|
|
|
|
seefatlow
Jr. Member
Offline
Activity: 80
Merit: 1
|
|
July 01, 2019, 03:58:41 PM |
|
Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?
Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?
That power use is very high. Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas I do some undervolt and ppt moding . set cclock to 1100/860mv mclock 1100/860mv , I got the 1450w power consumption at wall and 11.7Kh mining cnR running trm 0.5.2 with cn_config 15*14:AAA [2019-06-30 17:39:10] Mining cnr with 6 GPU workers [2019-06-30 17:39:10] GPU 0 (06:00.0) 64 CUs Temp 63C Fan 49.00% (2408 RPM) Core 1062 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 1 (19:00.0) 64 CUs Temp 63C Fan 49.00% (2446 RPM) Core 1058 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 2 (13:00.0) 64 CUs Temp 51C Fan 58.00% (2874 RPM) Core 1055 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 3 (16:00.0) 64 CUs Temp 63C Fan 52.00% (2575 RPM) Core 1053 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 4 (03:00.0) 64 CUs Temp 62C Fan 56.00% (2788 RPM) Core 1069 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] GPU 5 (09:00.0) 64 CUs Temp 62C Fan 53.00% (2599 RPM) Core 1040 MHz Mem 1100 MHz, VDDC 862.0 mV [2019-06-30 17:39:10] [2019-06-30 17:39:10] Current CN configs: [2019-06-30 17:39:10] GPU 0 15*14:AAA [2019-06-30 17:39:10] GPU 1 15*14:AAA [2019-06-30 17:39:10] GPU 2 15*14:AAA [2019-06-30 17:39:10] GPU 3 15*14:AAA [2019-06-30 17:39:10] GPU 4 15*14:AAA [2019-06-30 17:39:10] GPU 5 15*14:AAA [2019-06-30 17:39:10] [2019-06-30 17:39:10] Stats Uptime: 0 days, 00:16:53 [2019-06-30 17:39:10] GPU 0 [63C, fan 49%] cnr: 1.951kh/s, avg 1.951kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] GPU 1 [63C, fan 49%] cnr: 1.949kh/s, avg 1.950kh/s, pool 1.393kh/s a:14 r:0 hw:0 [2019-06-30 17:39:10] GPU 2 [51C, fan 58%] cnr: 1.937kh/s, avg 1.945kh/s, pool 2.290kh/s a:23 r:0 hw:0 [2019-06-30 17:39:10] GPU 3 [63C, fan 52%] cnr: 1.939kh/s, avg 1.942kh/s, pool 1.892kh/s a:19 r:0 hw:0 [2019-06-30 17:39:10] GPU 4 [62C, fan 56%] cnr: 1.957kh/s, avg 1.962kh/s, pool 2.191kh/s a:22 r:0 hw:0 [2019-06-30 17:39:10] GPU 5 [62C, fan 53%] cnr: 1.928kh/s, avg 1.927kh/s, pool 2.388kh/s a:24 r:0 hw:0 [2019-06-30 17:39:10] Total cnr: 11.66kh/s, avg 11.68kh/s, pool 12.54kh/s a:126 r:0 hw:0 I modified the PPT with the registry. GPU_P0=852;800 GPU_P1=991;810 GPU_P2=1084;820 GPU_P3=1138;830 GPU_P4=1150;840 GPU_P5=1202;850 GPU_P6=1212;850 GPU_P7=1100;860 Mem_P0=167;800 Mem_P1=500;800 Mem_P2=800;810 Mem_P3=1100;860 Fan_Min=1588 Fan_Max=4900 Fan_Target=63 Fan_Acoustic=2400 Power_Temp=75 Power_Target=0 The above parameters I have been through a number of tests, found a relatively stable data, if the voltage is set below 860mv, some cards will randomly appear crash How can I further optimize and reduce the power consumption to 1200w? Your driver seems dated. Have you tried using the latest 19.xx drivers? |
|
|
|
|
kibatronic
Newbie
Offline
Activity: 22
Merit: 1
|
|
July 01, 2019, 10:46:09 PM |
|
.....
What’s special about the P5 state ? Also are all your Vega56 flashed to 64?
I found that by limiting the Vegas to only use the P5 state I could use a lower voltage without crashing. My Vega 56's are all flashed with a Vega 64 BIOS. |
|
|
|
|
|
|
seefatlow
Jr. Member
Offline
Activity: 80
Merit: 1
|
|
July 03, 2019, 03:47:52 PM |
|
.....
What’s special about the P5 state ? Also are all your Vega56 flashed to 64?
I found that by limiting the Vegas to only use the P5 state I could use a lower voltage without crashing. My Vega 56's are all flashed with a Vega 64 BIOS. You are absolutely right. It works. Thanks for the tip. Did you also flash non reference Vega56(Hynix) to 64? Can this be done? |
|
|
|
|
todxx (OP)
Member
Offline
Activity: 176
Merit: 76
|
|
July 03, 2019, 06:43:52 PM |
|
Team Red Miner v0.5.3 releasedhttps://github.com/todxx/teamredminer/releasesChanges in v0.5.3 - Added MTP algo for Zcoin/XZC (please read MTP_MINING.txt before mining).
- Further small stabilization fixes for CN variants, primarily 4MB algos.
Now supporting the MTP algorithm for Zcoin  !!!!! MTP IS EXTREMELY POWER HUNGRY AND HOT !!!! Please make sure to lower your clocks and have good cooling for your GPUs when testing MTP. MTP can consume double the power of other algos for un-tuned cards.
|
|
|
|
|
|
