We will try to reproduce the problem on Windows 7 rig. Is this also happening with older drivers?
To answer this since I mean the same issue by saying that:
* Win7, GTX10 Series has lower hash rates
It's the same issue exactly like this:
Has anyone figured out the low hashrate problem on Windows 7 when using NVIDIA GTX10xx series cards, RTX cards work fine on Win7 and hash at full speed, I tried diff versions of PM and still only get about half the speed, I got 14 total rigs, 6 rigs on Win7 that work great with no problems but had to switch to GMINER for the time being, and 8 rigs on Win10 using PM with no problems.
We will try to reproduce the problem on Windows 7 rig. Is this also happening with older drivers?
Yes, it is. It happens with any Nvidia drivers under Win7.
More details:
The issue suddenly happens a month+ ago after some new DAG was reached,
on any machines that are run Win7 and have card from 10 GPU Series.
There was no any changes to the machine/same drivers as before,
but hash rate reduced by 1/4 or 1/5 for all 10 Series cards.
Older drivers was used 38x.xx [that hash rate was always higher before that]
Tired multiple drivers, more newest, latest nothings helps.
For example:
1060 6GB was 24.xx, and suddenly unable to go more than 19.xx with regardless of any settings or anything.
The hash rate for those cards become crippled when using PhoenixMiner under Win7
Under Win10 is fine, or under Win7 is also fine if another miner is used like Gminer,
who addressed this issue and they fix it quickly.
* Win10, 1060 6GB cards, when GPUs clocks are applied via miner with command line parameters, clocks/voltages/memory,
miner start reporting wrong power consumption.. [unreal low one],
you can check that by comparing the power consumption reported by miner with other tools like GPU-Z and similar.
When all those options are removed, (and miner no longer control the GPU parameters like clocks/voltages)
the miner start reporting the correct power consumption.
We will need logs to reproduce these problems, as we haven't seen anything similar on our test rigs.
Don't have access to those rigs anymore where I saw that issue, but will try to duplicate them on a different one tomorrow.
[This was 100% reproducible issue like other issues.]
* Win10, for 3080 GPU when using command line option "-tmaxmem xx",
it cause brutal share staling for all GPUs (with regardless of GPU's models used in the same rig)
lag of 2888ms, 3000ms, 300ms in that range randomly vary between 300ms - 3000ms.
When this option is removed, all found shares return to normal latency of 38ms for all GPUs
We will need logs to reproduce these problems, as we haven't seen anything similar on our test rigs.
Okay, I've recorded it and sent you the logs for this issue, please check PM.
+
There is another issue that I forgot to tell:
With this mixed setup of 3xxx & 1xxx Series GPUs ,
only power consumptions for 3xxx Series GPUs are reported,
for 1xxx Series power consumptions its missing. [its included in the same logs too]