I would love to use the avalon miners but I cringe when looking at their efficiency.
Ja... Out of 6 I purchased only 3 are online at the work farm, 1 at home is offline replaced by a M10 and I have 2 more still in their un-opened boxes. 
Topic: AvalonMiner 921 released - 20 TH/s @ 1800W - page 2. (Read 19456 times)
I would love to use the avalon miners but I cringe when looking at their efficiency.
My A841's are gone = really good easy clocking options. that really make better watts per th
My A851 also gone = bad clocking that did not save watts per th
My a921 at the Clifton array.
Since I am driving to Clifton early this Sat morning to repair bad m10 (fingers crossed)
I could pull the a921 bring it home and test the new firmware.
maybe it will improve watts per th.
My A851 also gone = bad clocking that did not save watts per th
My a921 at the Clifton array.
Since I am driving to Clifton early this Sat morning to repair bad m10 (fingers crossed)
I could pull the a921 bring it home and test the new firmware.
maybe it will improve watts per th.
@minefarmbuy
My latest demo and review is for the A851
My latest demo and review is for the A851
I would have thought you or philip got demo's, maybe that the 851 or I need more coffee.
It would be interesting to see if the power efficiency is better with the new firmware.
After running for 24 hours now, I'm thinking it is a power setting. Voltage Level Offset seems to behave as it should now.
So after ~34 hours of mining with the new firmware (and adjusting voltage offsets) I can say w/ certainty that the new MM reduces the hashrate. Of the two machines I installed it on, both lost 1.5-2TH/s. I've rolled back to the old firmware and the machines are back at their normal 20TH/s. If I had the machines locally I would test their power consumption with the new firmware running. I'm wondering if the loss of ~1.5TH/s came with a decent reduction in current; can anyone with new firmware running test this and report their findings?
So after ~34 hours of mining with the new firmware (and adjusting voltage offsets) I can say w/ certainty that the new MM reduces the hashrate. Of the two machines I installed it on, both lost 1.5-2TH/s. I've rolled back to the old firmware and the machines are back at their normal 20TH/s. If I had the machines locally I would test their power consumption with the new firmware running. I'm wondering if the loss of ~1.5TH/s came with a decent reduction in current; can anyone with new firmware running test this and report their findings?
P set..
maybe a power or performance related parameter?
maybe a power or performance related parameter?
New firmware was just released for 921 (and the 920). I wish they were more descriptive in the changelog. All that's noted is:
Code:
* Add support for AVA9_P_SET_SS_PARA_EN
A920 is part of the flash sale, confirmed by Lily.
A921 is out of stock. A851 is the only other available option from Canaan.
A921 is out of stock. A851 is the only other available option from Canaan.
So 20th at 2050w is potential max draw?
the a921 appears to be better it is available at a much better price then the a920
which means they are having trouble selling the a921's out at the discount price of 349
the a920 is 2x the price of the blowout on the a921
Things are a struggle for avalon as the eff is poor on the a920 or the a921
I won't mention the name of different company has a 31th machine using 2200 watts for only 799.
Avalon needs to get a decent item out soon.
I have always supported them and they did not attack other algos with new asics.
So I did get the A921 and managed to set it to do 19.5 th at 1700 watts with a downclock
at 349 + shipping I paid 385 for the a921 that keeps it in line against higher price gear if you have cheap power
get them.
So 20th at 2050w is potential max draw?
Yeah I'm not sure if there is a typo or few in the A920 technical specs.
Reading the specs, apart from the hashrate and power draw everything else is identical. They do make a mistake in the efficiency though.
921 = 2000/18000 .09 j/GH
920 = 1700/18000 = .094 j/GH
This reminds me of the 741 and 740 models. I think that's what it was. The model ending in 0 were under performing units they sold in the Asian markets. All I really want to see is the 941.
921 = 2000/18000 .09 j/GH
920 = 1700/18000 = .094 j/GH
This reminds me of the 741 and 740 models. I think that's what it was. The model ending in 0 were under performing units they sold in the Asian markets. All I really want to see is the 941.
Sexy. We'll get a listing up shortly.
And yet another Avalon 9 series model appears in the web shop! 
AvalonMiner 920
Link: https://canaan.io/product/avalonminer-920/
AvalonMiner 920
Link: https://canaan.io/product/avalonminer-920/
Quote
| Model Name | AvalonMiner 920 | |
| Hash Rate | 18TH/s, -5% ~ +10% | |
| Power | 1700W, +0% ~ +20% @ 20TH/s, 25°C, 94% PSU Efficiency, 220V AC, Wall-Plug | |
| Consumption | (Minimum, Variant power consumption at different temperature and speed) | |
| Power Efficiency | 0.089J / GigaHash (Minimum) | |
| Rated Input Voltage | 12 ~ 12.6V | |
| Chips | 104 x A3206 7nm ASIC | |
| Frequency | Smart Speed | |
| Network Connection | AUC3 + AUC3 I2C to AvalonMiner Controller | |
| Controller | AvalonMiner Controller | |
| AUC3 | Up to 5 devices / AUC3, Daisy-Chain Connection | |
| Operating Temperature | -5℃ ~ 30℃ | |
| Air-intake Temperature | -5℃ ~ 30℃ | |
| Cooling | 1 x 14038 FAN | |
| Noise | 70db (Typical) | |
| Gross Dimensions | 420mm*235mm*225mm | |
| Net Dimensions | 378mm*170mm*155mm | |
| Gross Weight | 6 kg | |
| Net Weight | 5.5 kg |
Thanks for info will check on it.
Gave you a merit point.
It does work
these links show it works now
https://i.imgur.com/iT8ZCSw.png
https://i.imgur.com/YBlpFOk.png
Gave you a merit point.
It does work
these links show it works now
https://i.imgur.com/iT8ZCSw.png
https://i.imgur.com/YBlpFOk.png
It appears that Canaan fixed the previously non-functioning Voltage Level Offset drop-down menu on October 6, 2018.
The Voltage Level Offset drop-down menu in version 20181011 and later should therefore work properly now. It should no longer be necessary to add the --avalon9-voltage-level-offset command-line option in the More Options field to change the voltage level offset setting.
To check whether the Voltage Level Offset drop-down menu is working, navigate to the Processes page (under Status) and scroll to the bottom. The --avalon9-voltage-level-offset option in the row that is displaying the cgminer process should reflect the selected value from the Voltage Level Offset drop-down menu.
The Voltage Level Offset drop-down menu in version 20181011 and later should therefore work properly now. It should no longer be necessary to add the --avalon9-voltage-level-offset command-line option in the More Options field to change the voltage level offset setting.
To check whether the Voltage Level Offset drop-down menu is working, navigate to the Processes page (under Status) and scroll to the bottom. The --avalon9-voltage-level-offset option in the row that is displaying the cgminer process should reflect the selected value from the Voltage Level Offset drop-down menu.
multiple partitions don't mean much in this case.
but the one board running at much lower numbers is an indication of some thing wrong
my numbers were always much closer to equal then your MW1 board is doing.
I suggest getting a different power supply to see if your power supple is doing that.
Are you near New Jersey?
if so I could lend this one to you
https://www.youtube.com/watch?v=aXZQmKtv2-E&t
it runs on 240 volts and will put out 2400 watts.
I can tell you right now to change your power settings as your results are bad terrible in fact.
and 12.6volts is high but 12.8volts is really high.
I would do this setting on the bottom line
see if you get my results
https://bitcointalksearch.org/topic/m.48823226
I used this psu
https://www.youtube.com/watch?v=7s6jmcEptbo
but the one board running at much lower numbers is an indication of some thing wrong
my numbers were always much closer to equal then your MW1 board is doing.
I suggest getting a different power supply to see if your power supple is doing that.
Are you near New Jersey?
if so I could lend this one to you
https://www.youtube.com/watch?v=aXZQmKtv2-E&t
it runs on 240 volts and will put out 2400 watts.
I can tell you right now to change your power settings as your results are bad terrible in fact.
and 12.6volts is high but 12.8volts is really high.
I would do this setting on the bottom line
Quote
More Options --avalon9-voltage-level -2 --avalon9-voltage-level-offset -2
see if you get my results
https://bitcointalksearch.org/topic/m.48823226
I used this psu
https://www.youtube.com/watch?v=7s6jmcEptbo
Thanks Frodo, that was very helpful. I had read this before but did not think to apply it to the 921 series.
I think I see an issue with one of the hash boards. MW1 is showing much lower numbers of work than the other 3 hash boards. Several single digit numbers, 5 zeros and just low overall when compared to the other hash boards. I pasted this section of my CGMiner API Log below. Hopefully playing with the power supply will lead to a problem diagnosis of either power or hash board.
I am running the latest controller image from October and the latest MM firmware. In case it matters, the Avalon 9xx controller SD Card had a faulty or incompatible image new out of the box. I could not get into the Advanced mode. I flashed the latest image to this SD-Card, but I did not format it first. It came with two partitions and the Win32DiskImager program completed very quickly, in just 10 or 15 seconds. I am assuming the 2nd partition remains intact as it was from the factory. Should I remove both partitions first or does that really matter?
I think I see an issue with one of the hash boards. MW1 is showing much lower numbers of work than the other 3 hash boards. Several single digit numbers, 5 zeros and just low overall when compared to the other hash boards. I pasted this section of my CGMiner API Log below. Hopefully playing with the power supply will lead to a problem diagnosis of either power or hash board.
Code:
MW0[1117 1149 1118 1138 1148 1175 1123 1176 1223 1113 1106 1098 1151 1107 1162 1075 1038 1096 1137 1095 1128 1197 1180 1102 1101 1093]
MW1[320 1133 825 0 316 61 1150 1071 0 0 1115 1 2 1 1107 527 0 548 911 1140 1173 811 979 1004 1192 835]
MW2[1224 845 1180 1099 1174 1168 1206 1156 1167 1167 1005 1044 1060 1082 1027 1108 1097 738 577 1095 1145 1080 1150 1144 1152 1176]
MW3[1213 1129 1249 1168 1165 1143 1089 1093 1082 1192 1174 1134 1158 1144 1123 1092 1104 1182 1195 1087 1172 1136 1194 1229 1246 1071]
MW1[320 1133 825 0 316 61 1150 1071 0 0 1115 1 2 1 1107 527 0 548 911 1140 1173 811 979 1004 1192 835]
MW2[1224 845 1180 1099 1174 1168 1206 1156 1167 1167 1005 1044 1060 1082 1027 1108 1097 738 577 1095 1145 1080 1150 1144 1152 1176]
MW3[1213 1129 1249 1168 1165 1143 1089 1093 1082 1192 1174 1134 1158 1144 1123 1092 1104 1182 1195 1087 1172 1136 1194 1229 1246 1071]
I am running the latest controller image from October and the latest MM firmware. In case it matters, the Avalon 9xx controller SD Card had a faulty or incompatible image new out of the box. I could not get into the Advanced mode. I flashed the latest image to this SD-Card, but I did not format it first. It came with two partitions and the Win32DiskImager program completed very quickly, in just 10 or 15 seconds. I am assuming the 2nd partition remains intact as it was from the factory. Should I remove both partitions first or does that really matter?
Jump to: