this is what i get from t-rex
Slow.. Why isn't it 100++?. We do the unlock different than t-rex and will probobly end up over 100 when the code is done. My comment from 10 days ago.. Getting closer to a 90-100MHASH unlock on 3080ti LHR2
Hashrates fluctate, but stable without triggering the LHR for ~5 minutes at 100% tdp.
|
|
|
|
need more improvment.... 88-90 mhs on t-rex
Not on the latest driver. I only get 72.5 when running t-rex on windows with autotune in (497.09) v1.32 will have LHR detection and reset device when the gpu goes into LHR mode. Then it's easier for the miner to find the optimal tweak / lhr value. Next up is autotune. you have to work on stability and compatibility, in particular with ezil.me, there are a lot of rejections.
ezil.me worked good in earlier versions, but after doing small modifications to fix the hiveon pool it seems to break ezil. Will take another look But you might consider changing the pool to get the most out of your cards with TBM... here from a happy user: pool: 2miners.com --xintensity 6144 (without poolside stale shares and 100% accepted) (10 hours of running) ~90watt  |
|
|
|
|
If this happens a new ethereum classic will spin of the hardfork. GPU miners will probobly sell all ethereum, because they tend to back the coins they make money on. Moving from POW to POS has always failed. There are a number of coins that have collapsed aftwer moving to POS...
|
|
|
|
ezil.me standalone support was added in 1.31. Make sure you are on the latest version. For config look in the example bat file
I use 1.31 ((( use 7 kern on 3080 and 3070 , what xint I can use ?(( hive os 470.86 drivers This only on ezil pool, on hive it's ok xintensity is pool dependent. on f2pool,2miners, miningpoolhub and nanopool you can use up to 4096 without rejects. (Nvidia) For amd divide by 4. (1024) |
|
|
|
|
ezil.me standalone support was added in 1.31. Make sure you are on the latest version. For config look in the example bat file
|
|
|
|
All other miners show the acutally hashrate and not the average hashrate. Example: the rig runs for 5 days with 120mh -> so average hashrate is 120mh, but for any reason, some cards begin to struggle and hash with the half hashrate, than it needs hours to see the hashrate drop. I hope you understand what i mean.
AwesomeMiner show realtime hashrate for all miners, and average only as an extra if available.
Yes, but our average hashtrate is not the average since the miner started, but the last 2000 jobs sendt to the gpu. We increased it because of the LHR mode which is fluctating alot. But I see your point. Can reduce the samplerate in the next version, |
|
|
|
Is there a failure in the API? For my nvidia rig the API show me the realtime hashrate, but on my AMD test rig it shows the average hashrate. So AwesomeMiner show me the wrong hashrate -> we need the realtime hashrate.
Maybe you can add the average hashrate also to the API, to see both. Maybe it is better to ask patrike from AwesomeMiner.
We report the hashrate in out API, currently the average hashrate. The realtime hashrate can be added later, but why do you need it? |
|
|
|
Not a huge issue or anything, but in version 1.31 I'm getting the message "GPU# XIntensity set to ####" twice for each AMD card.
Fixed in v1.32. I just recieved a 3070LHR today so will finetune abit before releasing 1.32. The LHR detection and reset device is working pretty good. The 3080ti almost never goto LHR mode, so our LHR code is stable even on the latest drivers. |
|
|
|
|
LHR detector implemented. So perhaps lhr+autotune in the next build.
|
|
|
|
in my not so stable rig so far.
If your rig is unstable don't mine eth+zil because the dag creation is difficult for the graphic card. Or add --dagintensity 1 . Tested it for 20 minutes, and everything looks normal. 1 reported stale in TBM, 2 on the pool  |
|
|
|
|
Thanks for testing, I will try a rig on Ezil. Do you have a logfile I can look at for more information?
|
|
|
|
I get 34,4 mh on my 6600xt. After 24h no reject share on rustpool and ethermine.
But ETC+Zil will not work for me. Cannot subscribe to the pool.
I have created a ticked for this. We will look into it. https://github.com/sp-hash/TeamBlackMiner/issues/179Upgrade to v1.30. I got less than 0.5% rejects in my test.
I upgraded yesterday, in about 12h only 11 rejected shares (0,22%). Wonderful job, you're very close to perfection in Hiveon pool, huge improvement. I noticed that all the rejected are basically in the first minutes after miner is started, then almost no rejected ones until is restarted, it seems like pool and miner needs some minutes to adjust
Did a small change and now I get 0 rejects on the hiveon pool. Will let it run for a while. edit: 0% rejects and 0% stale in the LHR card testrig on the hiveon pool.. Small changes to the stratum code can break compatibility with other pools, so we need more testing. |
|
|
|
|
But the new version have increased cpu usage for cuda.
building a new version
edit: windows complete
|
|
|
|
@sp_ can you recommand more stable setting to avoid rejected shares? With 1.29 are reduced but still not perfect (about 1%)
Found a possible fix. Testing now.. edit: 30mins without a rejected share now. 85/85 |
|
|
|
Have you tried any other xintensity values for MPH that worked best for your setup?
yes. --xintensity 4096 move the rig up to 115 (+5) mhash and poolside 113 ish. on 1.30 and nvidia 497.09 windows 10 let's try --xintensity [5120] --tweak [7] --kernel [6] on 2miners running at P0 = ON and the latest final MSI AB (i find memclock improve from +1200 (MSI beta version) to +1500 (MSI final version)).  Thanks, you seem to know how to use this program. But @91 watt ?? Sounds too low |
|
|
|
That being said, I'm sure you can make a better argument than complaining that TBM's dev copied lolminer, then highlight two completely different sentences.
I have never tried lolminer, but ppl ask us to add features they need in order to switch their farm over to the more profitable Team Black Miner. https://github.com/sp-hash/TeamBlackMiner/issues
Have you tried any other xintensity values for MPH that worked best for your setup?
yes. --xintensity 4096 move the rig up to 115 (+5) mhash and poolside 113 ish.
It's almost certain that the "improved" reported hashrate is not real and it's actually worse in terms of actual valid share submission rate compared to established miners
This is a pool thing. The reported hashrate is an estimate. For the best pools look in your faq. https://github.com/sp-hash/TeamBlackMiner/blob/main/FAQ.md
baby steps. keep it up. i noticed that for trex, lhr unlock performance is not as good on the latest driver. it stays at 73 stable while it could get up to 76 on previous drivers. the number here is the LHR unlock %
I get 72.5 after a while with the latest t-rex on the latest driver with autotune. This might not be the optimal settings t-rex, but I do use the same settings in TBM and get a hashincrease. 3080ti;
core: 0 default
mem: (+1000) 10500mhz
pl: 70% power
lhr-unlock [1] = 81 MHASH stable for 1 hour.
lhr-unlock [1] --tweak [1] = 85 MHASH stable for 8-15 minutes.
Been running rock stable all night with -lhr-unlock [1]. We will do a linux build and release today. In the next version we probobly will add a LHR detector, and reset the device automaticly. Then we migth be able to do 85 (90 with 11000 mem) stable. on the latest driver.
If you don't know how the LHR MR Sp__ works don't denigrate minors in favor of your rigged software,
The LHR unlocker has cost me a few hundred hours of work already. Alot of Trial and Error. The first LHR unlock version was released too early, but v1.30 and moving forward should give a good boost. Bugs are being fixed, the miner improves and the fee is picking up. This is how Software development of the Team Black Miner is done: live@github:  Alot of hard work.. This is a Marathon not a sprint. [moderator's note: consecutive posts merged] |
|
|
|
|
Show Posts
