with --xintensity -1 even more stales, prefer to not get any stales like i had with trex, or at least near 0.
Stale shares aren't a problem if you mine on the right pool. etheremine.org, miningpoolhub and many other pools pay you the same for stale and normal shares. Some pools give a penalty on stales and even rejects, so we might case on the pool name and insert better defaults. --xintensity -1 on flexpool or nicehash give a poor result. You need --xintensity 32 or lower. Note that some pools will disconnect you if you have more than 5% stale. But then the hashrate of the 3070 will be pool dependent. Mine on etheremine get 65MHASH, mine on flexpool get 62MHASH 76m [2021-09-14 21:32:28.141] ID BOARD TYPE HASHRATE/W HASHRATE AVERAGE SHARES
76m [2021-09-14 21:32:28.141] GPU0 3070 Cuda 527.65 kH/W 64.90 MH/s 64.47 MH/s 16/3/3 (84.21)
76m [2021-09-14 21:32:28.141] GPU1 3060Ti Cuda 475.88 kH/W 64.24 MH/s 64.25 MH/s 21/0/4 (100.00)
76m [2021-09-14 21:32:28.141] GPU2 3060 Cuda 475.49 kH/W 50.88 MH/s 50.86 MH/s 14/0/3 (100.00)
76m [2021-09-14 21:32:28.141] 1479.01 kH/s 180.02 MH/s 179.58 MH/s 51/3/10 (94.44)
typo? GPU0 Cpu verification failed. Share not sendt to pool. Rejected!
Typo fixed in the next upcoming version v1.05
The Team Black miner V1.04 with improved opencl speed, have some reports of higher profit on the newer AMD cards. Try it out! 6600xt 34MHASH 5700xt 58MHASH But the power usage is up slightly. You can reduce it with a lower --xintensity.
Stats for AMD added on windows with temp control on the gpu like TBMiner on nvidia. If the gpu gets to hot, let is sleep for x seconds. More work needed on the linux version. Note that v1.04 have some issues when running mixed AMD and NVIDIA in the rig. So better to use the --amd-only or --cuda-all-devices. v1.05 Coming soon... [moderator's note: consecutive posts merged] |
|
|
|
with --xintensity -1 even more stales, and 3070 started to get bad shares, have to lower memory oc. Will do more tests, preffer to not get any stales like i had with trex, or at
typo? GPU0 Cpu verification failed. Share not sendt to pool. Rejected!
With the Team black kernel you can increase the performance by reducing the clocks and increase the --xintensity. Run Nvidiaprofile inspector and disable force P2 state, and then retune your rig. For RTX cards: Low coreclock, increase the TDP (powerlimit) high memclock. increasing from 65% tdp to 67% tdp can gain a few MHASH compared to the other miners.
Set ETH_WALLET=
What is the first line for ?
Insert your wallet adress here [moderator's note: consecutive posts merged] |
|
|
|
|
Some pools pay the same for stale shares as normal shares. This is because the ethereum blockchain reward stale work as well. (uncle blocks) On miningpoolhub use --xintensity -1 for a boost in hashrate and profit. The stale shares in the mining window is not reported by the pool, but calculated by the miner.
|
|
|
|
|
Why don't you ask in the Hiveos thread. Right now ubuntu 18.4 linux and windows 10 is up to +10% more profitable than hiveos.
You just need to run
The Team Black Miner v1.04
rtx 3090: ~135MHASH
rtx 3080: ~102MHASH
rtx 3070: ~64MHASH
rtx 3060ti: ~63MHASH
rtx 3060: ~52MHASH
rtx: 2060: ~34MHASH
Download, select a poolscript, change the wallet adress, click and mine. Voila
Overclock in MSI afterburner.
|
|
|
|
Miningpoolhub, Crazypool and ethermine and many others pay the full reward for a stale share. So if the Team Black Miner is configured to produce some stales (--xintensity High or -1(autotune)), you get payed for the extra MHASH'es it produces. Flexpool expects under 1% stale shares, so if you feed them 5% stale , you get 4% less return. Other pools pay a reduced reward for stales. F.ex ezil.me has 70% reward for a stale shares. The Team Black Miner has the lowest fee of 0.5%, and created by trusted opensource devs with a long track record. Move your rigs over to us, and we will give you more profit. |
|
|
|
v1.02 still
Thanks for testing. you should use the new version v1.04 because it has more checks like cpu verification of the dagfile. This is useful when overclocking. The average function take the average of x jobs sent to the gpu. low intensities have a larger x, but if a job is small the average will fluctate more. Perhaps better to display the 10min average here in the future versions. |
|
|
|
miningpoolhub is better than other pools now?
no, but some pools need low --xintensity. nicehash.com has 10% rejected shares with --xintensity -1. 1.8% rejected with --xintensity 32 . flexpool.io give a penalty for stale shares so you need to mine with --xintensity 8 or 16. The ethereum network is paying for stale work (uncle blocks) so when a pool rejects them, they either steal your money or divide the mined money wrong. Rtx 3090 --xintensity -1 can give up to 135MHASH Rtx 3090 --xintensity 8 can give up to 115MHASH (?) 20MHASH wasted on a pool that doesn't compensate the work you do. etheremine, nanopool, f2pool, sparkpool etc all good. |
|
|
|
of course, 1 or 2 shares every 5 min,
Looks like there is a bug, but your shares are credited. Will be fixed in the next release. This bug could affect other stratum 1.0 pools as well.
v1.04 1. Fixed bug when submitting shares in stratum 1.0 sometimes the workername was lost 2. Fixed stats printout if the poolname is too long 3. Added more pools 4. Cpu rejected opencl shares now added to the stats 5. Faster opencl kernel 6. Fixed ethereum classic opencl mining on miningpoolhub https://github.com/sp-hash/TeamBlackMiner/releases/TeamBlackMiner_1_04_cuda_11_2.7z sha-256 fbe5c3e03c4a6ca31d9864f050594779b89a32fa8dea43ef3d80ca888bce9915 TeamBlackMiner_1_04_cuda_11_4.7z sha-256 1af8eaa96aa98d92af5d203f70dcfa48f0899d947abcb7faa24c8cfb6a8c40f7
1.03 cuda 11.2 - still missing from miningpoolhub.
fixed in v1.04. The profit is amazing @ miningpoolhub [moderator's note: consecutive posts merged] |
|
|
|
|
And you get accepted shares in the console? The diff is high on this pool, so it could take some time with 1 card or small rig.
|
|
|
|
i tried two variant before reporting and this is what i had and it did not work:
--wallet user --worker_name name
--wallet user.name
looks like there is something in the code, i do not use %% anywhere.
workaround use both: --wallet user. name -worker_name name |
|
|
|
1.03 cuda 11.2 - still missing from miningpoolhub. With trex it is on the list in ~5 min. So, it is still broken. That text is there for few days. I have 9 other rigs (trex) there and all statistics are updating every few minutes.
I've tried this morning with one of my rigs. (the one in the picture at my post 7:20). Worker and stats are showing fine.  But now I have added another rig, and it doesn't show. So there is an issue here. Will investegate. edit: Looks like the example .bat script is wrong. You need to correct to: --wallet %USERNAME%.%computername% --worker_name %computername% |
|
|
|
Any chance to make your miner work with Miningrigrentals? TeamBlack miner says it is not supported.
Yes, it's in the todo list |
|
|
|
and therefore less stale.
Some pools pay the same for stale shares as normal shares. Ethereum has a 15 second block time, why does a pool spam you with jobs every 200ms. The Team Black miner is running 8% slower on low intensity settings. Selecting the right pool is important here.
teamred miner: 58.8
s-eth.2miners.com:2020 (ethash) DIFFICULTY: 2.03 EPOCH: 440
(At least 2 shares/10 mins at each intensity)
'Stock' settings - 6800xt (2400 clock / 1990 mem)
Xintensity = -1: 62.32
Xintensity = 1: 58.51 - 213W
Xintensity = 12: 61.10
Xintensity = 24: 61.94
Xintensity = 36: 62.03
Xintensity = 48: 62.11
Xintensity = 78: 62.15
Xintensity = 128: 62.33
Xintensity = 240: 62.60
Xintensity = 360: 62.64
Xintensity = 512: 62.70 - 230W
No shares reject
will update this post tomorrow with better memory times then stock (just got the card so haven't set it up fully for mining)
Thanks for the test. Looking good. in v1.02 the stale share estimate count was not added yet. We will release v1.03 today with improvements
 https://github.com/sp-hash/TeamBlackMiner/releases/TeamBlackMiner_1_03_cuda_11_2.7z SHA-256 3154d4c34542a197e5e93b5db174b968353dd80f4751c1065341d0a2d31353b4 TeamBlackMiner_1_03_cuda_11_4.7z SHA-256 4b2ff528171f5fb6fdf2d47376ea26be64434b52cd08ab1d17f1721232a07d35 TeamBlackMiner_1_03_Ubuntu_18_04_Cuda_11_4.zip SHA-256 2b495b63431724dbc3122cabac8ec615107ae58779bf0f97e23508caf97baf60 v1.03 1. CPU verification of the Dag buffer (opencl/cuda) 2. Improved the stale share counter with a better estimation 3. Stale counter added to the opencl solver 4. Bumped the default xintensity to 64 (cuda) 5. Improved cuda 11.4 kernel tuning. abit faster 6. Improved DAG generation to be nicer to high oc. 7. Improved 11.4 kernel to accept higher oc. 8. Nicer on the dag generator to accept higher oc. (cuda) [moderator's note: consecutive posts merged] |
|
|
|
There is a temporarily problem with the pool. Read on the top 
The best stable setting I found with the cuda 11.2 build v1.02 (hashrate/watts) rtx 3060 (lhr 1) 50.50MHASH Disable p2 state (nvidia profile inspector) 370.05 beta driver (x16 riser, hdmi plug) Add --xintensity 64 to the commandline msi afterburner: 65% powerlimit 90 degrees templimit -200 core +1150 mem [moderator's note: consecutive posts merged] |
|
|
|
We have added cpu verification of the dag buffer. If the verification fail, the miner will exit/restart.  |
|
|
|
eth/zil , 3080 non lhr , mem +700 , on intensity -1 , from 250 shares 70 was stale , on default intensity value 1400 shares ~60 stale
Thanks for testing. -1 is for pools like crazypool no penalty for stale shares. And other pools which doesn't spam you with jobs every 300ms. The local stale checker is not 100% accurate, I see accepted shares on the pool that we report as stale. Will look into it. added ubuntu 18.4 build (hiveos) https://github.com/sp-hash/TeamBlackMiner/releases/ |
|
|
|
3060ti, mem 8250 core 1020 pwr 70, I only get invalid shares. I have to bring the mem overclock right down to give me anything useful, resulting in a lower hash rate than t-rex. "Cpu verification failed. Share not sent to pool. Rejected!"
Thanks for testing. We need to add verification code to the dagbuffer, and recreate it with a lower intensity on errors.This is in the todo list. It might help to add som more power. The best clocks on the other miners might not be the best for this one. 3090, mem 10500 core 1050 pwr 95 intensity 64 on flexpool, gives me around 124 MH. Up from 118 on t-rex. 124MHASH is a good result. We have 0.5% fee vs 1% in t-rex so you can add another +0.59MHASH to the profit |
|
|
|
I am testing now with RTX 3080 (Nicehash & --xintensity -1).
On nicehash you shouldn't use -1 since you will get alot of job not found and stale shares. This is because nicehash need very small buffersizes sent to the gpu. With no parameter set, the default is --xintensity 32 Where can one see the dynamic xintensity calculated value (so I can report it  )? you can't because it's calculated on every round based on your hashrate. So it's not a fixed value. But we might added to the stats panel later. Updated every 60 seconds.
v1.02 1. Added stale share check on the submitted shares and display the count. 2. Added --xintensity parameter. To finetune the buffer size sent to the gpu. (NVIDIA/AMD) 3. Fixed memory problem on startup 4. Added support for more pools 5. Fixed crash in the opencl mining. Cards stopped working. 6. Print pool difficulty changes to console 7. Fixed bug in list opencl devices 8. Added gpu memory temp to stats https://github.com/sp-hash/TeamBlackMiner/releases/windows: TeamBlackMiner_1_02_cuda_11_2.7z (for the 470.05 driver) SHA-256 bb517a8e37faf1c1f7c3b4941b941a53fdad3bb09eefc16909d09286642b5659 TeamBlackMiner_1_02_cuda_11_4.7z SHA-256 984587059306b5f9dfcc5f9c918e9f30a6b9d3725fc712ca0bf2faefcbb56ed0 linux: TeamBlackMiner_1_02_Ubuntu_18_04_Cuda_11_4.zip SHA-256 b201ab9d96cbf9c11cca38f9941475d6a3b5aca20a1ceb2ba30924ca09e97a93
my hashrash stays between 57-59. Pool: 2miners, us server, Card: AMD 6800XT on adrenalin 21.30.15.01
We have added the --xintensity parameter to the opencl solver. Can you try again? What hashrates are you getting and what is the --xintensity ?
Ethermine, win 10, core +65, mem +150, driver 460
Edit2: I tried to get this working on my 3060 v1 card with 470.05 driver using the 1.01 Cuda 11.2 build. It starts mining fine. I am even have a increase of about 2mh/s with no changes in OC, but all shares are coming up with error " mining.submit: Pool reponded with error." Then showing invalid POW. I am using Flexpool East. I am not having any issues with shares being submitted with this miner on the 1.02 Cuda 11.4 build on my 2080TI.
The cuda_11_2 v1.02 build seems to be working fine. [moderator's note: consecutive posts merged] |
|
|
|
I've also tried the --xintensity parameter
not supported on opencl yet.
-U [GPU0] and it works now.
select cuda devices with -U [0,2,3,4]
Edit2: I tried to get this working on my 3060 v1 card with 470.05 driver using the 1.01 Cuda 11.2 build. It starts mining fine. I am even have a increase of about 2mh/s with no changes in OC, but all shares are coming up with error " mining.submit: Pool reponded with error." Then showing invalid POW. I am using Flexpool East. I am not having any issues with shares being submitted with this miner on the 1.02 Cuda 11.4 build on my 2080TI.
I guess your problem will be fixed when 1.02 is released for cuda 11.2 [moderator's note: consecutive posts merged] |
|
|
|
I have tried the latest version with gtx 1070 x 8, got 60 shares rejected (100%) and back to t-rex.
Wich pool, platform version and settings?
v1.02 beta 1. Added stale share check on the submitted shares and display the count. 2. Added --xintensity parameter.(-E) To feed the pool(s) what they need. (default is --xintensity 32, valid values --xintensity 1 and up to 10000? --xintensity -1 For the program to calculate the best intensity.) 3. Try to fix memory problem on startup 4. Added support for more pools 5. Fixed crash in the opencl mining. Cards stopped working. windows exe only https://github.com/sp-hash/TeamBlackMiner/releases/If you have time, post the best --xintensity. We need the pool and cards. Nicehash and the other 100ms job pools probobly need a low value to avoid rejected/stale shares. Etheremine, Solominingpools, crazypool, miningpoolhub and a few others works best with --xintensity -1 or perhaps high fixed values. Low xintensity will give poor hashrate, so try to find the sweetspots. [moderator's note: consecutive posts merged] |
|
|
|
|
Show Posts
