That should be enough... hmm can you have a look to max memory alloc size?
|
|
|
I just can confirm that there is some part in my stratum code of 0.6 alpha causing the current AION problems. Working hard on new interface to get it fixed asap. Windows 10 get error for all version of Lol miner blockhaine driver 23 august
RX 570 RX 470 4-8 gb 144_5 algo
Warning: Your device with id 1 is currently not supported by lolMiner or by the specific algorithm you selected.
Help please )
There are exactly two potential problems causing this error message: 1) There is not enough memory detected on the device 2) There is a compile error. Can not say which one is valid here, but compile should usually work - at least here I can compile for all RX500 cards. Can you check - with clinfo.exe how many memory is available? (Global memory size)
|
|
|
Did it work with the alpha 1 hive release? Are there any error messages you see? By the way lolMiner moved into Hive 6 core, alpha 3 will come there next days. 0.6 alpha 3 in AION looses speed (20-30%) after a half hour or hour of mining... and then, maybe after 2 hours, looses connection to pools... I tried it on 4 different AMD RX 570 and R9 270/280 rigs and all acted in same way Same happens here. Running AMD 7990 on Aion Which of the problems? Lost connection or performance penalty? Or even both?
|
|
|
0.6 alpha 3 in AION looses speed (20-30%) after a half hour or hour of mining... and then, maybe after 2 hours, looses connection to pools... I tried it on 4 different AMD RX 570 and R9 270/280 rigs and all acted in same way Hmm ... thats odd - loosing connection may happen atm, as mentioned I am not happy with the stratum, but no idea why looses speed. Can you monitor clocks and heat and report if something strange happens in that moment? Here at mine the code was stable so far Excuse me for my inaccuracy, but increasing of parallel threads leads to increasing memory usage, so more threads, more speed. I'm not sure of my reasoning, please correct me. Will be better if 8 Gigs cards to use full memory with increasing of speed
That may indeed be true, two threads instead of one in parallel. But not sure that will bring much - the cores are well loaded and no long idle times in 192/7. Also I do not have a 8G AMD, so can not test ^^ Maybe you can check if its indeed better to run the miner 2x on same cards, but I doubt it will bring much more then 1-2% maybe.
|
|
|
So the speeds are the same for nvidia or can we expect some decent improvements?
The Aion code is approx 16% quicker then in 0.5, the others remained on their level. Maybe got one more idea for 144/5 but need to check that ^^
|
|
|
Hi there, I just published the v0.6 alpha 3 release. You can find it here: https://github.com/Lolliedieb/lolMiner-preview/releases/tag/alpha3Last days were a bit difficult. I did a lot of work in the background of the miner that took a lot of time but has no obvious changes till today. But finally it will end in making all stratum interfaces more stable, so well, it will be worth. Also testing a lot with the 96/5 miner for AMD. Unfortunately it was very quick, but was not able to get it stable. So was at the not that happy decision to skip it or postpone this release. In the end now the alpha 3 has the same AMD code as 0.5 giving me some more time to make the new one stable. But the good thing: alpha 3 now supports all coins that 0.5 did, so the family is complete again and I learned some things that may also help other algos to perform better. So there will be an alpha 4 of 0.6 series before the release. Planned for that one: - New 96/5 code for AMD - Improves 144/5 for AMD - New stratums for all - Api adjustment so Awesome Miner works Hope to serve you well -------------- Is API functionality so that lolminer can be used with Awesome Miner going to be added? Every miner worth mentioning has this functionality. Please let me know if you will be implementing it and when. Than you.
Planned for alpha 4 in ~ 2 weeks +/- It possible to increase Equihash192/7 GPU memory usage over 4G? Actual maximum GPU memory usage on this algorithm is 3.1G, not closer to 4 as per described in OP Environment variables are set correct.
The description is still for 0.5, 0.6 uses less memory. But why you want to use more when you do not need to? Where is the next alpha with this market dying I need this..
Here it is. But yes the market atm hurts a lot Well can not change the market, but believe me it its me same way then you guys...
|
|
|
I get many times "Error: lolMiner detected that device 1 is stucked Shutting down lolMiner" This does not tell me anything in the logs, why issue happened, no overheat, cards running ok. Why this happen on both Nvidea and AMD?
Which version and on which algorithm? When the cards are not OC its often an implementation bug on my side - especially the ZERO / GENX code of 0.5 had this problem because I was not careful enough in some routines so sometimes I write to memory addresses that did not exist - and that caused the problem. I had hope to get rid of this in 0.6, if not - please report with a bit more detail and I will revise the code again
|
|
|
Sorry for the delay here, this time debugging hurts me a lot. Well at least I can resolve the AION game here. Following the table with guesses, best guess in bold and in red the actual value: 1080 FE (~176 in 0.5): User | Guess | mrshaw070 | 190 | squeekie | 200 | catfish78 | 202 | Benchmark 0.6 | 206.5 | invizion | 228 | baraday | 232 |
Rx 580 4G (~106.5 in 0.5): User | Guess | mrshaw070 | 116 | Benchmark 0.6 | 118.3 | squeekie | 120 | catfish78 | 123 | invizion | 135 | baraday | 143 |
May the two winners send me an AION Mainnet Addr as private message so I can send their price? I have to say the optimization was very odd this time. I started with Nvidia and the changing to 0.6 code style brought most of the improvements. So far so good. Then I tried more things I had in mind but ran into a kind of wall - it just did not want to improve behind this point. Or rather - one idea worked locally but then added a penalty later when reconstructing the solutions such that it was almost +-0. AMD was totally different. Here the change to 0.6 code base was very buggy in the beginning and once done it was exactly same speed as 0.5. Something is saturated there and so did not find the bottle neck. But then the +-0 idea of the NV code brought at least approx 4 sol/s. That was a start. Then started playing around and introduced some code part that I did not try anywhere before. But while NV was totally not impressed by this, here it brought about 7% speedup. Strange but well. So, right now fixing more stuff - expect 0.6 alpha 3 in next 4-5 days. ---- Having stability issues running 8 RX cards mining AION. What settings are you guys using? Getting frustrated The mining code is not stable for so many cards. The only way atm is to run 2x4 cards. That works fine. Still working on this issue and thats also why alpha 3 will not be last 0.6 alpha. Hello today I have tried mining BTCZ and my RX 580 Gaming 8GB has only 25-30 Sol / s where is the error? Everybody says they have 580 to 100 Sol / s at RX
Imao perfectly fine. 100 sol/s is impossible for this cards - even a 2080ti without power limit may do 85 to 90 or so but no way more. Who ever told you 100 had no clue. Lolly, is it worth waiting for the hashrate increase in alpha 3 for old card(hawaii, tahiti)?
Hmm ... GCN1 cards is a bit difficult, some of them lacking some of the newer features. But the Hawaii series cards already do not run so bad - at least not slower then the Polaris cards that are much newer. I doubt that there is a specific optimization for them to make them fly higher. If any every trick for Polaris will help them, too. So, plan for alpha 4 is to test some of the things recently learned in the 144/5 code. Lets see if anything there helps the AMDs. Any news on the next Alpha with AMD support for MNX?
I just can confirm that 0.6 alpha 3 will run all algorithms of 0.5, so MNX on AMD will run. But no idea about speed yet so? what is the problem here, anyone
The config file looking ok, but the miner may have problems with the fact that EXAMPLE_AION exists twice. But else it should work with 0.5. 0.6 does not yet support AION but will be changed in next alpha.
|
|
|
I need to know what algo/coin do I need to use for Pirate Coin? I have run through most of the algo/coin manes i n the manual and AION seems to work but no confirmed hashrate from the pool, others hash but don't report. I am using AMD RX580 cards.
Atm pirate coin is not supported since that uses Equihash 200/9. If there would be support in future (no idea yet of that will happen), it would be --coin ARRR By the way can not resolve the game yet - the code is still wip. Have noticed that some parts that worked well for 144/5 did only partially work for 210/9 - in terms of speed. But can already say both codes got faster then their 0.5 cousins but not yet how much ^^
|
|
|
lolMiner.... I think you could not figure out the worse name.
i like the name As if anybody here give a shit about names xD So many strange names in crypto, but what is important is what is behind. Hands down - who did not raise an eyebrow hearing about the eth enlargement pill the first time My guess: 1080 --> 200 580 --> 120 Ok, noted My guess: 1080 ---> 202 580 ---> 123 Did you two talked about that before? I guess 1080 232 sol/s 580 143 sol/s. At least one of us is optimistic Hey man how do I enable logging on Alpha so I can paste you my results?
The only thing you need is an empty folder in your miner directory with name "logs". Then you can add the command line option --logs 1 to enable writing a log file. Or use "LOGS" : 1, in config file. just cause the crash contained a line with "boost" The boost lib is not the problem here. Thats just because if all cards fail the compatibility check there are a few values not written to an internal configuration table ... and this missing value is causing the bug. That will be fixed in alpha 3 So, from clinfo its indeed that your machine does not find the amdgpu-pro driver and uses the Clover platform only ... and this one does not like my kernel code (yet). I will have to set up a test machine on my own to see why this happens (or you manage to set up the pro-drivers ... lets see who succeeds earlier ^^) My 390's rig crashes immediately (BTG 144/5). What is the reason? Is it because of the drivers? I am with crimson 17.7.1 (the last that are supported in windows 8.1).
Good question. I am also running that crimson drivers on my Windows test machine and its running fine there on my 580. Also on an earlier page we had reports about a 390 running well. So, atm no idea why, does this happen at older versions as well? There are two things that may happen: 1) you run with too extreme OC / UV settings. I have once tried to run lolMiner with the ideal settings for mining eth... that try did not last long... equihash has more compute intense components that require a bit more clock and voltage. 2) One user this morning (in telegram) had the problem that his defender refused to unpack some parts of the .zip. As a result he was mixing and old .exe with new Kernel code. That results in a 0 sol/s immediately. Hope one of the two answers help you - else this is going to be an interesting debug...
|
|
|
Ok, das ist seltsam. Der Miner zeigt halt llvm treiber an, die man sonst nur bei amdgpu (nicht-pro) plus Mesa vor der Nase hat. Ich bin übrigends auf amsgpu-pro 18.30 und ubuntu 18.04, das klappt ganz gut. Kannst du mal clinfo laufen lassen ob da zufällig beide Platformen da sind? Den Fall hatte ich bisher nicht, daher weiß ich nicht ob der miner dann den richtigen Pfad nimmt. Du könntest sonst auch mit --DEVICES AMD starten und sehen ob das was ändert.
|
|
|
cant wait for 210,9 (aion) improvement Me too, no idea where it would end up xD I just started doing the concepts... Lets make a game out of this... I have two testing cards, one 1080 and one 580. The 1080 does 176 sol/s on lolMiner 0.5, the 580 does 107 sol/s. Price question: how much will they do in 0.6 alpha 3? I donate 2x10 Aion to the closest guess per card that is commented here before 12 am UTC this Friday. For each user only the first guess is counted Edit: will evaluate this with a half hour benchmark run. cant wait for 210,9 (aion) improvement Same here looking every day that thread to find Aion support at 0.6. Will take at least till end of this week. But working on this. Nvidea stability is not to good, keeps turning off on mnx, some more tuning is needed. What are best settings to try on the 1070ti cards?
Hmm. Can you be more specific on the error or give details in the github bug tracker (that what its meant for). Especially error message and probably your settings may help a lot making it more stable. Only 70sol/s with 1080ti for 144_5 lago, not 80-85. Power Limit 75% on Ryzen1700x cpu
... Average speed (30s): 64.4 sol/s | 66.8 sol/s | 69.3 sol/s | 68.8 sol/s | 73.7 sol/s Total: 343.1 sol/s
I do not own a 1080ti, only a normal Ti. The 85+ benchmark was done by a user with 100% PL. I know from my experiments with the 1080 that lolMiner is very sensitive on clocks, so just lowering pt without tuning the clock back up may kill the performance. I can recommend this guide for a user of ZEL community made how to obtain top performance: https://medium.com/@TaneSavoya/step-to-step-guide-how-to-mine-with-nvidia-gpus-in-windows-10-more-quiet-cooler-and-energy-bf3b63f40919ERROR: Kernel compile error for device 0(Radeon 550 Series (POLARIS12 / DRM 3.26.0 / 4.15.0-36-generic, LLVM 6.0.0) lolMiner will try next kernel in list (if there is any) Warning: Your device with id 0 is currently not supported by lolMiner or by the specific algorithm you selected. Warning: Your device with id 1 is currently not supported by lolMiner or by the specific algorithm you selected.
any support for RX 550 4gb planed soon? would be awesome .. tried 2 previous versions (and the one with additional kernel) couldn't get any to work .. Here offline I can compile for RX 550 with the AMD kernel analyzer, but I think the problem here is the LLVM driver. I only have a machine with the amdgpu-pro here, there the compile for 550 works. Will need to set up a 2nd Ubuntu for the other drivers ... or you switch drivers ... hmmm what is that? @mnx with the alpha 2 ==>> incorrect size of solution
Thats odd ... did not see that reply yet. Can you tell me on what pool you are that return this message? Because it indicates that the encoding of solution lolMiner sends does not fit the pools expectation. Did this happen to only one share or for every one?
|
|
|
Just look the screenshot, its http://localhost:8080/summary when you started with --apiport 8080
|
|
|
Hey Folks, I just made the lolMiner 0.6 alpha 2 release public. You can find the new version here: Readme: https://github.com/Lolliedieb/lolMiner-preview/blob/master/README.mdRelease: https://github.com/Lolliedieb/lolMiner-preview/releasesChanges so far: - A new mining code for 192/7 (Zero, GenX) - This one is based on the better 144/5 code. Compared to 0.5 the Nvidia code is ~25% faster, AMD approx 15%. Also the code now should again work on headless 3G cards (~2860 MB required) - A new mining code for MNX on Nvidia - The AMD code will follow next week - I had some problems getting is full stable. The Nvidia one is ~20% faster then 0.5 code. This will likely not be the final code for full 0.6, because well - it beats ewbf 0.5, but not 0.6, so have enough motivation to improve it further xD - Massive rework in the stratum interface - I fixed a lot of bugs there. Most important are better routines to handle miner stuck when first connection attempt does not work and a fix to the "received new job but stopped mining bug". Also as new feature, lolMiner now is detecting which GPU got accepted shares and it measures the latency of the pool response. GPU 0: Share accepted (14 ms)
- Improved API / Statistics Module - All the stats are now collected at one position in the miner (that was a code cleanup haha ) The API now provides info about the pool latency, submitted and accepted shares for each GPU, the GPU names, the GPUs PCIE address (for connecting miner with external tools) and so on. Hope the new versions serves you well Next then: MNX / AMD and Aion Looking forward to do this ... learned a lot working on this version again. ------------------------------------- Room for your questions: Next alpha build today?
Yep Im having problems with the latest NVIDIA Drivers and my 1070ti card. Trying mine MNX coin - keeps saying my card is stuck and miner restarts. Not sure what I'm doing wrong, the clocks are ok..
Good question - can not say it as well - but maybe check the new version ... maybe thats more stable. I cant connect to suprnova pool for mining btg with lolminer. As I understand it have something to do with personalisation string BgoldPoW but I cant find anywhere how to do that.
This is not required - for suprnova just use --coin BTG and you will be fine. For example I mine BTG with the line lolMiner.exe --coin BTG --server btg.suprnova.cc --port 8866 --user Lolliedieb.gold --pass x
aion1@aion1-TB85:~/Aion_miner$ clinfo -l aion1@aion1-TB85:~/Aion_miner$ clinfo -l amdgpu ignoring unknown command-line parameter amdgpu
What is odd here: clinfo is returning exactly nothing ... Did you use the amdgpu-install with the right parameter for OpenCL, e.g. ./amdgpu-install --opencl=legacy
Because if --opencl is not set it will only install the graphic part and not the compute part of the driver.
|
|
|
this is ok? to put auto192
"DEFAULTS" : { "DEVICES" : "AUTO", "APIPORT" : 0 },
"EXAMPLE_GENX" : { "COIN" : "AUTO192.7", "POOLS" : [ {"POOL" : "genx.suprnova.cc", "PORT" : "9983", "USER" : "jjzsolt.joecoinS", "PASS" : "x"} ]
As far as I know suprnova supports the auto method, so this should work. But you have a typo in: "EXAMPLE_GENX" : { "COIN" : "AUTO192_7", "POOLS" : [ {"POOL" : "genx.suprnova.cc", "PORT" : "9983", "USER" : "jjzsolt.joecoinS", "PASS" : "x"} ] }
This is how it should look like (if the EXAMPLE_GENX entry is the last one in config file ... else you have to put an additional "," after the last "}" I quoted). Note that its AUTO192_7 instead of AUTO192.7
|
|
|
Hello I am having this issue: ... Kernel driver in use: amdgpu
Opencl is installed.
Can you check if clinfo also finds your GPUs? As far as I know the amdgpu driver comes without OpenCL support, this is only included in the amdgpu-pro driver (or via CLOVER / Mesa drivers that are based on the OpenSource amdgpu). That may be the issue. Whats the latest version DL link?
Download link is broken. 404 error
I just updated the download links - now 0.5 and the 0.6 alpha are both on gitHub. Hello, Lolliedieb!I use your miner for GTX 1080. Everything is great! I tried to use for the GTX 960 and got this picture: Reading pool data from command line lolMiner is configured to mine Bitcoin Gold(BTG), Equihash 144/5 Setup Miner... Using device with id 0 (GeForce GTX 960) Does this mean that the miner cannot be used with old video cards or is it something else? (driver, win 10 or other ....) EWBF miner sometimes gives up to 20 sol. Trifle - but nice ... What miner version you use, is this 0.6 or 0.5? I have to admit I do not own a Maxwell series GPU, so I am not sure where this problem comes from. One possibility would be the chip running out of resources at some point, most likely register space or shared mem, but can not say for sure due to lack of a testing system (Someone here with a 960 in a linux system who could rent me a ssh access for a day so I can debug?)
|
|
|
Is there a way we can we just add the necessary instructions to the existing config, to mine 2 sets of devices? I can get help with the above instructions in a day or so, but if this is a common problem maybe adding a run.bat or config file that can be adapted to 2 sets of GPUs on one rig. I'm just thinking out loud here in my ignorance. Forgive me if I am out of line.
I think what you want to get is rather easy to do: Just create two identical profiles in user_config.json, one with "DEVICES" : [0,1,2,3], and second with the other cards. Name one "PROFILE1 and the other "PROFILE2". Then the only thing you need is to do a copy of the run_miner.bat and edit them, so one points to PROFILE1 and the other to PROFILE2. Then double click both and you are ready to go I noticed that setting AMD drivers from Graphics to Compute makes no difference to hashrate? I thought hashrate would increase on Compute mode. Also when the next Alpha for MNX coin mining, thanks..
Next alphs planned for next 5 to 7 days. Will likely include MNX for Nvidia and Zero/GenX for AMD. May sound odd that I mix algos here but makes sense from my perspective since some techniques from both sides will flow into Aion code ^^ Will lolMiner be supported on awesome miner? Thanks,
0.6 API should allow this - I will check that
|
|
|
Hi there, just want to inform you we had a bug with AION last hours that rigs did not want to start up - error was on my side and is resolved, no update required.
Will work on a backup strategy for 0.6 so this type of errors does not happen again then.
|
|
|
But is the 0.5 official hive upgraded to the bugfix version, cause i cant get it working?, it just stops after listing gpus, like the 0.5 pre bugfix did?
I hope they did that - I forwarded the bugfix one day before publishing it here to the hive team - I hope they did that, but if not its not in my control. I hope they did that (else: if it still is the pre-bugfix it will hang on some 144/5 coins - one fix would be using "COIN" : "Auto144_5" and “OVERWRITEPERSONAL” : "TheCoinsPersonalizationString" in the configuration - with that you can mine any 144/5 even in the buggy version. Well for ethereum that may work fine, but I guess for Equihash - that has more core demand - the settings may not fit so well, especially the low voltage. Try to go for 1200 / 1850 at a slightly higher core voltage (-100mv vs stock may be good point to start) . Ethereum is not sensitive for memory and regist bits flipping - that may cause the one or other hash to be invalid, but its a rare occurrence. In Equihash one defect bit can cause the miner to stuck, because its more using data dependent loops.
|
|
|
Ah, I guess you are on the wrong file. The official file for 0.5 hive integration (all algorithms) is: http://download.hiveos.farm/custom/lolminer-0.5.tar.gz(With Hive thread https://forum.hiveos.farm/t/custom-miner-lolminer/8135 ) For 0.6 (Only 144/5) you can load an experimental integration here: https://github.com/Lolliedieb/lolMiner-preview/releasesOK, got it sorted out. There was a problem with registering worker, I used %EWAL% instead of %WAL%. Got 400 Sols from 5xRX570, then 350, then it exited saying a device got stuck. Are there any parameters to tune the miner, like intensity, to avoid this ?
Hmm, ok good to hear that it started but not that it got stuck. Thats one of first stuck bug reports I get for 0.5 and AION. Just fire it on again an see if it happens again (if yes: what are your OC settings?) But I have hope this does not happen again. But will note this down to search in AION mining code for potential 0 sol/s traps.
|
|
|
|