Hash rates wrongly associated on HiveOS dashboard
UserDC-LeGrand opened this issue · 12 comments
This issue is strange.
On Gminer-N ETHASH, somehow the hash rates for the 3070 GPU#2, wind up in GPU#0.
The first two 2060 GPU's (0 an1) should hover around the 30-32 Mh/s range, and the last 3070 GPU around the 60-63 MH/s range.
The issue is in the way you build your scripts ... Look at the order of the devices you are passing on the command line - -d 6 7 4 2 0 3 5 1 . This should be removed or properly sequence such as this**: 0 1 2 3 4 5 6 7**
root@rig4:/hive/miners/custom/SWARM.3.6.7.linux/bin/gminer-a-1# /hive/miners/custom/SWARM.3.6.7.linux/bin/gminer-a-1/miner -d 6 7 4 2 0 3 5 1 --api 33001 --server etchash.auto.nicehash.com --port 9200 --algo etchash --proto stratum --user xxxxxxxxxxxxxxxxxxxxxxx.rig4 --logfile '/hive/miners/custom/SWARM.3.6.7.linux/logs/gminer-a-1.log' --pass x --templimit 61
Go to .\debug\busdata.txt. It is a .txt file, but should be just normal json.
Please confirm that it is sorting the gpus correctly by its bus_id. It should appear in order based on its bus_id.
If that is correct, check agent-screen
and see if bus_numbers
are correct in minerstats [ ]
If that is correct, check order of hashrate hs:[ ]
presented in agent-screen
. If it's showing the 64.61 mh/s as first, then this is issue with Gminer, not SWARM.
Follow up- Did you ever check? I was searching code, and I don't see how Swarm atm could display hashrates in wrong slot currently (not to say bug isn't there).
Most miners don't show what hashrate goes to which gpu, just a list of hashrates in api. It is assumed it is displayed in order of gpu bus.
Found the issue, and perhaps it is in the way you build your scripts ... Look at the order of the devices you are passing on the command line - -d 6 7 4 2 0 3 5 1 . This should be removed or properly sequence such as this**: 0 1 2 3 4 5 6 7**
root@rig4:/hive/miners/custom/SWARM.3.6.7.linux/bin/gminer-a-1# /hive/miners/custom/SWARM.3.6.7.linux/bin/gminer-a-1/miner -d 6 7 4 2 0 3 5 1 --api 33001 --server etchash.auto.nicehash.com --port 9200 --algo etchash --proto stratum --user xxxxxxxxxxxxxxxxxxxxxxx.rig4 --logfile '/hive/miners/custom/SWARM.3.6.7.linux/logs/gminer-a-1.log' --pass x --templimit 61
Can you post your debug\devicelist.txt file here? I just want to know how SWARM is recording them.
.\debug\busdata.txt too. If you can post both files, I can figure out issue.
As requested. The busid is correctly sequenced. The device list is not ... Should it not be 0, 1 ,2 ?
root@rig5:/hive/miners/custom/SWARM.3.6.7.linux/debug# cat busdata.txt
[
{
"busid": "03:00.0",
"name": "NVIDIA TU106 [GeForce RTX 2060 Rev. A]",
"brand": "nvidia",
"subvendor": "ASUS",
"mem": "6144 MiB",
"vbios": "90.06.46.00.89",
"plim_min": "105.00 W",
"plim_def": "190.00 W",
"plim_max": "238.00 W"
},
{
"busid": "06:00.0",
"name": "NVIDIA TU106 [GeForce RTX 2060 Rev. A]",
"brand": "nvidia",
"subvendor": "ASUS",
"mem": "6144 MiB",
"vbios": "90.06.46.00.89",
"plim_min": "105.00 W",
"plim_def": "190.00 W",
"plim_max": "238.00 W"
},
{
"busid": "08:00.0",
"name": "NVIDIA GA104 [GeForce RTX 3070 LHR]",
"brand": "nvidia",
"subvendor": "EVGA",
"mem": "8192 MiB",
"vbios": "94.04.46.40.60",
"plim_min": "100.00 W",
"plim_def": "240.00 W",
"plim_max": "270.00 W"
}
]
root@rig5:/hive/miners/custom/SWARM.3.6.7.linux/debug# cat devicelist.txt
{
"AMD": {},
"CPU": {
"11": 11,
"9": 9,
"7": 7,
"6": 6,
"2": 2,
"13": 13,
"12": 12,
"3": 3,
"4": 4,
"8": 8,
"10": 10,
"0": 0,
"5": 5,
"1": 1
},
"NVIDIA": {
"2": "2",
"0": "0",
"1": "1"
}
}
root@rig5:/hive/miners/custom/SWARM.3.6.7.linux/debug#
As requested. The busid is correctly sequenced. The device list is not ... Should it not be 0, 1 ,2 ?
root@rig5:/hive/miners/custom/SWARM.3.6.7.linux/debug# cat busdata.txt [ { "busid": "03:00.0", "name": "NVIDIA TU106 [GeForce RTX 2060 Rev. A]", "brand": "nvidia", "subvendor": "ASUS", "mem": "6144 MiB", "vbios": "90.06.46.00.89", "plim_min": "105.00 W", "plim_def": "190.00 W", "plim_max": "238.00 W" }, { "busid": "06:00.0", "name": "NVIDIA TU106 [GeForce RTX 2060 Rev. A]", "brand": "nvidia", "subvendor": "ASUS", "mem": "6144 MiB", "vbios": "90.06.46.00.89", "plim_min": "105.00 W", "plim_def": "190.00 W", "plim_max": "238.00 W" }, { "busid": "08:00.0", "name": "NVIDIA GA104 [GeForce RTX 3070 LHR]", "brand": "nvidia", "subvendor": "EVGA", "mem": "8192 MiB", "vbios": "94.04.46.40.60", "plim_min": "100.00 W", "plim_def": "240.00 W", "plim_max": "270.00 W" } ] root@rig5:/hive/miners/custom/SWARM.3.6.7.linux/debug# cat devicelist.txt { "AMD": {}, "CPU": { "11": 11, "9": 9, "7": 7, "6": 6, "2": 2, "13": 13, "12": 12, "3": 3, "4": 4, "8": 8, "10": 10, "0": 0, "5": 5, "1": 1 }, "NVIDIA": { "2": "2", "0": "0", "1": "1" } } root@rig5:/hive/miners/custom/SWARM.3.6.7.linux/debug#
This is not same rig.
Rig above has 9 gpus. This devicelist.txt shows only 3.
I think I found issue. It's being Powershell, and adding to the table out of order.
This should fix it, just I don't know the full extent/implications it will have particularly with cross-platform (AMD & NVIDIA) on rig. But it will order them by their device slot, and by extension the -d argument should be put correctly.
If you want to forward test and confirm while I'm working on other stuff, would appreciate it. It's a simple edit to the file notated in the commit.