|
Patrike, ya might want to pick up Bminer 7.0.0 in your next update.
|
|
|
|
Patrike,
I replied to your messages about AB remote.
It's not pulling up the card name which makes trying to do application of OC profiles by card, by algo impossible, unless you specifically know the GPU number. Not sure what you can do about it, but there is also a new 4.5.0 AB that was just released?
I tested on my system. With Afterburner 4.3 and got the GPU names over the MSI Afterburner interface. After upgrading to Afterburner 4.5, I get the same problem as you noticed, the output from the MSI Afterburner interface doesn't include the GPU names anymore. Update: I found a solution that works on my system. I didn't have the latest DLL file for the MSI Afterburner Remote Server software that I link to from the Awesome Miner help pages. Please see the bottom of this page: http://www.awesomeminer.com/help/gpuclocking.aspxAfter replacing the DLL file for MSI Afterburner Remote Server, I do get the GPU names. Yeah - fixed with that dll update!!! Thanks! |
|
|
|
New rasp pi has arrived It works the sd cards were good. the oem pi was a dud. I did a video showing the dead on and the good one. I have this running on -2 offset using a delta 2400 watt plat and I am getting very high power use 1523 watts at 13229 gh No image just a link to a screen shot https://i.imgur.com/TRgGOeH.pngdid you see my comment, you can't just set the -2 in the web UI, that doesn't change the undervolt, you have to include the command line option in the more options line on the web UI. |
|
|
|
The upgrade did not work the first time. It looks like the uninstall part worked but not installing the new version. I had to reinstall 4.7.4 and upgrade a second time for it to work.
In my case the automatic upgrade process worked perfectly and nothing else had to be done .... Same and included the remote upgrade just fine too |
|
|
|
|
Patrike,
I replied to your messages about AB remote.
It's not pulling up the card name which makes trying to do application of OC profiles by card, by algo impossible, unless you specifically know the GPU number. Not sure what you can do about it, but there is also a new 4.5.0 AB that was just released?
|
|
|
|
All you have to do is look under the status->processes tab in the web UI and scroll down till you find the cgminer process and you'll see how it was started and you'll see that if you tell it to be -1 in the UI, it will still be 0 in the processes. Doesn't matter which one you select, it will always be 0. Adding it in the more options will add the string to the end of the command launch string and it will override the 0 hardcoded value to whatever you specify, and it doesn't need to match what is the UI.
Just me, or does that sound kind of like a bug? Clearly the Web UI could make that change for us... it's been that way since the 821's. when I found the issue, I sent Canaan an email, that is when I found out they meant to *NOT* leave the undervolt stuff in the 8 series UI, and always keep it at 0. I completely get why, there are some 821/841's that *won't* run at -2... I have 2 841's that wont for example, but -1, 0, and +1 they run perfectly. That same email is where they told me the reason for the 100% fan, and if you look in the ccminer code, while the 741's had a fan moderation temperature of 90C, the 8's that was changed to 70C. The high end shutdown limit is 105 I believe in both, but they really don't want them going to that limit. |
|
|
|
How did you do that? Because if you just selected -1 in the web UI, that didn't make the change, you have to add the "--avalon8-voltage-level-offset -1" string in the more options line in the configuration the 0 - zero option is hard coded no matter if you selected the -1, -2, etc or not.  ... little known detail. BTW, from 0 to -1 on mine will show an approx .150-.2 Th drop per miner in the rate, but the outlet temp will change by a couple of C too Wow. I didn't know that. Rather presumed using the Avalon Web Interface and selecting the pull down menu for voltage offset would do the trick. That is rather surprising news. All you have to do is look under the status->processes tab in the web UI and scroll down till you find the cgminer process and you'll see how it was started and you'll see that if you tell it to be -1 in the UI, it will still be 0 in the processes. Doesn't matter which one you select, it will always be 0. Adding it in the more options will add the string to the end of the command launch string and it will override the 0 hardcoded value to whatever you specify, and it doesn't need to match what is the UI. |
|
|
|
I guess while I'm on this topic. The undervolt stuff wasn't supposed to be in the firmware, and not all units will undervolt to -2. 2 out of 5 of mine, will not run (or run pretty wounded at -2), but change them to -1 or 0 and they run just fine with on significant delta from the other 3
If one desire maximum hash rate and power is not an issue, should they be running at +1? FWIW - I switched (3) strings of Avalon 8s from +1 to -1 to help cool them down just a touch. Before and after stats: Before: 1a - 11464.6 - single 821 3a - 66555.7 - string of (5) 841s 3b- 64851.9 - string of (5) 841s ---------- 142872.2 After: 1a - 11538.1 3a - 66540.3 3b - 64928.6 --------- 143007.0 So virtually identical numbers. Both test were over 16+ hours. I should note the "after" group spent most of those hours in the evening and early morning with cooler intake air. How did you do that? Because if you just selected -1 in the web UI, that didn't make the change, you have to add the "--avalon8-voltage-level-offset -1" string in the more options line in the configuration the 0 - zero option is hard coded no matter if you selected the -1, -2, etc or not. ... little known detail. BTW, from 0 to -1 on mine will show an approx .150-.2 Th drop per miner in the rate, but the outlet temp will change by a couple of C too |
|
|
|
I noticed this too and made a post here and spoke to Canaan about it. I spoke to Canaan and they said reversing the (sorcerer) fan should be ok if temps stay under 40c. We plan to reverse them too as you did to see as we are run cold/hot aisle setups and curious to see if there is a temp improvement. I'd like a better reply from Canaan other than 'should be' ok. Would be nice for them or their vendor to model & verify just what happens to the various hot spots in a PSU and make yay/nay based on that. Changing from blow to suck will drastically change the flow patterns inside the PSU... Perfect confirmation would be Canaan acting on this and changing the flow direction on future Sorcerer's they sell. But - that said, when I get back in-country next week think I'll give it a whirl. Not very scientific, but from a fingers test. The air coming out is warm - not hot, the 841 output is hotter, and there seems to be no significant temperature difference to the outside of the sorcerer. But as noted, not a very scientific test. |
|
|
|
We can run one with updated firmware as well, going to leave these as is so there is no auto shutoff.
awesome, run all 3 side by each and let us know, that would really be interesting to know. |
|
|
|
Avalon might use more power over 30C, specs pertain to 30C usage.
Still best machine on the market.
These two have been running fine at this temp.
[...]
WOW... and I baby mine watching for over 30C inlet too  ... I wonder if those would run that way with the latest firmware... you guys have one you can put the latest MM code on and run and let us know... that would be a really interesting datapoint where it's suppose to have the over temp shutoff fixed, etc.
Moderator's note: This post was edited by frodocooper to trim the quote from Blokforge. |
|
|
|
So the improvement was around 1C... maybe 2, but it's pretty hard to really find the 2C, but at least all the hot air is going in the same direction as the *HOT* air ... |
|
|
|
This afternoon, I'm planning on seeing if I can flip a fan around in a sorcerer and cause it to suck in the same air that the 841 is and exhaust it forwards like the 841s and see if the inlet temps don't drop by 2-3C as a result.
Curious if anyone else noticed this or resolved it?
I have noticed this with the sorcerer, and it has become a pet peev. I've thought the same thing about going 180° with the fan, but haven't been able to make time to experiment. I've also considered, adding length to the cables. Please be sure to post your findings. Ok, was a pretty easy thing to rotate the fans... Except for the gray goooo... (glue), but it will release if wiggled enough. I'll let you know results. Gonna a be a little while till I can repeat conditions like I had this morning. |
|
|
|
|
All this discussion about power, temps, etc. has caused me to chase a rate hole, but the result was enlightening.
If you use a sorcerer PSU, you will notice that the fan sucks instead of blows. This means that in order to have enough attachment length to the power supply connections, you probably are placing the PSU above the 841/821 with the 2 fans opposite (841 vs PSU), and most likely you are exhausting the heat from the PSU out it's back and right back into the 841.
I ran mine on a different supply that I have that's fan blow the opposite way and the difference was 2-3C in inlet temperature difference.
This afternoon, I'm planning on seeing if I can flip a fan around in a sorcerer and cause it to suck in the same air that the 841 is and exhaust it forwards like the 841s and see if the inlet temps don't drop by 2-3C as a result.
Curious if anyone else noticed this or resolved it?
|
|
|
|
Not sure what that is. Here is all the 841 firmware and note the original had an error in that it won't show you when you are in a bad heat range and potentially damaging the ASICs, but the new one will and will shut you down ... https://github.com/Canaan-Creative/avalon8-docs/wiki/AvalonMiner-841-firmware-downloadsPretty much everything shipped with the 8411802-1e73b80 version of the MM code, and the 8411803-14bf10 came out about a week after the 841s were released to the general public. I updated all of mine and have had no issues... but I also haven't ran them wide open with no regard for the temperatures either. I guess while I'm on this topic. The undervolt stuff wasn't supposed to be in the firmware, and not all units will undervolt to -2. 2 out of 5 of mine, will not run (or run pretty wounded at -2), but change them to -1 or 0 and they run just fine with no significant delta from the other 3 |
|
|
|
Thanks for sharing all details. Let's investigate if MSI Afterburner is returning the GPU names over the control interface that Awesome Miner uses. 1) Open a web browser on the computer where you run MSI Afterburner, where there are no GPU names. 2) Navigate to: http://localhost:82/macm3) Enter username/password like this: MSIAfterburner/17cc95b4017d496f82 4) Please send me the output via mail or PM. Awesome Miner is reading all MSI Afterburner information, including GPU names, for this output. Curious, it doesn't show on the remote system.. Here is the important section  But it does show on the machine with the 950  Notice the device="" isn't there on the first one and is on the second one Both are windows 10, both same version of AM, same version of AB, same version of AB remote server. |
|
|
|
Room temp is 80°f, and everything is stock on the settings. Like I've posted before, I try not to run them this hot. They are currently located in my garage, and I can't "air" it out until I get home from work. Soon, they will be moved to my basement for the summer. I also want to add, I have 2 M3's running with them as well, and I believe that they are the real heat source. They're little torches, compared to the 841's.
Is that with the latest MM firmware change?
Moderator's note: This post was edited by frodocooper to remove a nested quote. |
|
|
|
Patrike,
The GPU settings dialog doesn't list the card name, only the GPU number and a colon. The values come over from the remote servers, but no card type/name, etc
Am I missing something to enable this?
UPDATE: This seems to only be on remote miners, on a local one the GPU names are shown.
Is this the "Map to system monitoring" and GPU tab of Awesome Miner? Are you running Remote Agent or the Windows service (Remote Service) on the remote computers? The Remote Agent application should be able to get the GPU names in the "Map to system monitoring" list. Please share any additional details and I will try to figure out if there is a specific scenario where it isn't working. Thanks! In general, Awesome Miner uses a third-party library that uses the AMD and nVidia API's to get information about the GPU's, including their names. In case it's about MSI Afterburner and the GPU clocking - do you see the GPU names in MSI Afterburner user interface? Patrike,
(see my other issue with no names on remote AB remoteserver)
More on the GPU setting. If you use AM to change the overclocking, with the checkboxes to set x at start and y at stop. Stop does not happen when you manually start mining, the x configuration happens, but if you then manually stop mining, it does not send the y version. (note, I did clone the x, when I created y, but changed the oc profile to the right version).
Could you please verify the log files (or send them to me view mail) for this scenario? Awesome Miner is writing information about clocking profiles being applied and the exact clocking commands sent to MSI Afterburner. Every time a specific clocking command is sent to MSI Afterburner, you can find "Executing GPU clocking command". When stopping a miner, it should be possible to see "Set clocking stop profile". Patrike, From the start Running MSI Afterburner remote server on the remote miners, (they are also running MSI afterburner, which I think it required?) Map to system monitor works correctly, the names appear there whether running remote server or not.  Where the names does show is when you use the GPU settings -> settings via Awesome Miner and Afterburner...  The same is true if you use the options->gpu clocking->configure and select a remote miner I have a GTX 950 on the machine where AM is running and it's able to get the name in both areas, but not the remote miners. Yes, the names are shown in the MSI interface  As to the log and the profile usage, I'll look at that separately. Thanks |
|
|
|
|
Patrike,
(see my other issue with no names on remote AB remoteserver)
More on the GPU setting. If you use AM to change the overclocking, with the checkboxes to set x at start and y at stop. Stop does not happen when you manually start mining, the x configuration happens, but if you then manually stop mining, it does not send the y version. (note, I did clone the x, when I created y, but changed the oc profile to the right version).
|
|
|
|
Patrike,
The GPU settings dialog doesn't list the card name, only the GPU number and a colon. The values come over from the remote servers, but no card type/name, etc
Am I missing something to enable this?
Patrick, when will you fix this mistake? How much can we talk about this error? The GPU tab displays the names of cards from other settings when these cards are not physically located on the selected installation. And on the monitoring tab, their order also does not match. This error should be corrected and not say that it is very difficult. if you would use the web based help, It would have answered most all of your questions. You can already fix the order and have been able to do this for a long time. Use the "map to system monitoring" button and drag and drop them in the order you want them. Beyond that I doubt there is much that can be done because the OS determined the order they are initialized via the driver and there is no way that software can tell the OS how to reorder them. http://www.awesomeminer.com/help/gpumapping.aspxAlso there are some features that require you to buy a legitimate license, if you haven't done that, then I suspect demanding change isn't going to get you very far. |
|
|
|
|
Show Posts
