excavator by NiceHash - multi-algorithm NVIDIA GPU miner [1.5.6a]

[ViperGuy]

I refuse to use Excavator for anything other than Nicehash. I get better results from the latest cudaminer. Plus, I have a theory that all those "benchmarks" that nicehash requires when a mosquito lands on your GPU are giving them free hashes.

[Tidsdilatation]

I refuse to use Excavator for anything other than Nicehash. I get better results from the latest cudaminer. Plus, I have a theory that all those "benchmarks" that nicehash requires when a mosquito lands on your GPU are giving them free hashes.

The reason i stopped use the nicehash miner 2 is becourse of all those god damn benchmarks. Every time i restart the god damn program i have 2 benchmark eight fucking GPUs?! It takes forever, like 5 hours. Thats alot in mining time.

[Kaan3000]

Just started testing 1.4.3a with my 5 x nvidia 1080ti rig. It seems way more stable than 1.1.0a. I knew excavator had potential against EWBF, this is why I'm giving another try.

What I love is that the GPUs are rather balanced in terms of output (680h/s) and total hashrate doesn't fluctuate much as in EWBF: Last 300 shares are all submitted  around 3.38kh/s.
 
What I don't like though, is that the sum is below 3500 which is the average at EWBF (3400min / 3600 max)

It stopped printing the results after 20 minutes but my readings from the pool look ok.

I wonder how I can set thread and block sizes via the .json file? I'm not sure if excavator is using all the available juice :/

EDIT: I'm mining Zcash "equihash"

[proteus7]

Regarding those neoscrypt speed improvements...I'm seeing almost 2x. Is that real? says Titan XP is now doing 2.14MH?!

[MarkAz]

Any update on when you'll be making the next Linux release (ideally with the fan speed fix)?  You're killing me with all of these Windows releases without Linux builds...

[dropky]

As promised I uploaded a simple benchmark script for excavator - https://github.com/nicehash/excavator/blob/master/excavator-benchmark.py

Python 3 is required!

Examples:

1. benchmark algorithm 1 on device 0
python excavator-benchmark.py --excavator path_to_excavator -a algo1 -d 0 -b

2. benchmark all algorithms on all devices
python excavator-benchmark.py --excavator path_to_excavator -b

3. benchmark on nicehash pool (default is eu server)
python excavator-benchmark.py --excavator path_to_excavator -u btc_address

Default time for benchmark is 300 seconds per algorithm, you can modify it with -t parameter.

Any way to edit the script to use multiple workers?  It would be nice if this could be dynamic based on the algo.

Workers are added at line 210. You can copy the append command multiple times or add a for loop to run multiple workers per device.

Example #1: To add two workers per device change

Code:

for dev in devices:
      workers.append(self.add_worker(alg_id, dev[0], algo[1:]))

to

Code:

for dev in devices:
      workers.append(self.add_worker(alg_id, dev[0], algo[1:]))
      workers.append(self.add_worker(alg_id, dev[0], algo[1:]))

To make it algorithm specific you could always check which algorithm is processed at the moment.

Example #2: Add two workers only when running equihash:

Code:

for dev in devices:
      workers.append(self.add_worker(alg_id, dev[0], algo[1:]))
      if algo[0] == "equihash":
            workers.append(self.add_worker(alg_id, dev[0], algo[1:]))

[dropky]

lol new version 2.0.1.3 removed daggerhashimoto for 3gb cards!!! LOL they dont know till April is profitable?

We decided to drop 3GB cards for daggerhashimoto because the DAG size was getting dangerously close to the 3GB limit and daggerhashimoto didn't even run on these cards especially when they are used as primary cards. This may not be a problem for experienced miners, but the majority of NHM users are casual ones. There are so many things to consider, you know...

I have been trying to figure out why I had no daggerhashimoto on my GTX 1060 3gb rig with NHM. now I know.
But this site below says the dag file is only 2.31gb yet, nowhere near the 3 gb limit ?

https://investoon.com/tools/dag_size

As mentioned before we dropped the daggerhashimoto for 3gb cards because there were a lot of issues/crashes with it among NHM users (especially on primary cards). We could remove the 3gb condition in excavator though, and you can run it without NHM then..

[dropky]

Which is better this or hsr?

I'm assuming you're asking about the Neoscrypt. Well.. It depends on the card

[dropky]

I refuse to use Excavator for anything other than Nicehash. I get better results from the latest cudaminer. Plus, I have a theory that all those "benchmarks" that nicehash requires when a mosquito lands on your GPU are giving them free hashes.

During benchmark excavator doesn't connect to the pool, it performs hashing on a test job which is hardcoded in the miner (always the same one). So, no free hashes for us If you don't believe me, you could always try to monitor the network..

[dropky]

I refuse to use Excavator for anything other than Nicehash. I get better results from the latest cudaminer. Plus, I have a theory that all those "benchmarks" that nicehash requires when a mosquito lands on your GPU are giving them free hashes.

The reason i stopped use the nicehash miner 2 is becourse of all those god damn benchmarks. Every time i restart the god damn program i have 2 benchmark eight fucking GPUs?! It takes forever, like 5 hours. Thats alot in mining time.

Yeah... we're working on a fix.

https://www.reddit.com/r/NiceHash/comments/7we5xq/benchmarking_crashes_and_repetitive_benchmarking/

[dropky]

Just started testing 1.4.3a with my 5 x nvidia 1080ti rig. It seems way more stable than 1.1.0a. I knew excavator had potential against EWBF, this is why I'm giving another try.

What I love is that the GPUs are rather balanced in terms of output (680h/s) and total hashrate doesn't fluctuate much as in EWBF: Last 300 shares are all submitted  around 3.38kh/s.
 
What I don't like though, is that the sum is below 3500 which is the average at EWBF (3400min / 3600 max)

It stopped printing the results after 20 minutes but my readings from the pool look ok.

I wonder how I can set thread and block sizes via the .json file? I'm not sure if excavator is using all the available juice :/

EDIT: I'm mining Zcash "equihash"

Unfortunately, you can't set the number of threads and/or blocks for the equihash (because of its implementation). You could try to run different modes (0,1,2) or try to run multiple workers per device. Other than that there is nothing you can do at the moment, we're working on improvements for equihash as we speak.

Note: To change mode of the algorithm (equihash) you have to add M parameter to worker.add:
{"id":1,"method":"worker.add","params":["0","0","M=0"]}

[dropky]

Regarding those neoscrypt speed improvements...I'm seeing almost 2x. Is that real? says Titan XP is now doing 2.14MH?!

Yes, it's real

[dropky]

Any update on when you'll be making the next Linux release (ideally with the fan speed fix)?  You're killing me with all of these Windows releases without Linux builds...

Sorry about that The latest version includes the Linux build - https://github.com/nicehash/excavator/releases/tag/v1.4.3a (with a fan speed fix).

[Kaan3000]

Just started testing 1.4.3a with my 5 x nvidia 1080ti rig. It seems way more stable than 1.1.0a. I knew excavator had potential against EWBF, this is why I'm giving another try.

What I love is that the GPUs are rather balanced in terms of output (680h/s) and total hashrate doesn't fluctuate much as in EWBF: Last 300 shares are all submitted  around 3.38kh/s.
 
What I don't like though, is that the sum is below 3500 which is the average at EWBF (3400min / 3600 max)

It stopped printing the results after 20 minutes but my readings from the pool look ok.

I wonder how I can set thread and block sizes via the .json file? I'm not sure if excavator is using all the available juice :/

EDIT: I'm mining Zcash "equihash"

Unfortunately, you can't set the number of threads and/or blocks for the equihash (because of its implementation). You could try to run different modes (0,1,2) or try to run multiple workers per device. Other than that there is nothing you can do at the moment, we're working on improvements for equihash as we speak.

Note: To change mode of the algorithm (equihash) you have to add M parameter to worker.add:
{"id":1,"method":"worker.add","params":["0","0","M=0"]}

Cool, I will give it a try.

@dropky By the way, what is the maximum number of workers can be added per card? I have 28SM. What would make sense? 4, 7, 28?

[dropky]

@dropky By the way, what is the maximum number of workers can be added per card? I have 28SM. What would make sense? 4, 7, 28?

You should experiment with it, start with 2, then increase the number if you get better results that way (and if it's stable - at some point excavator will probably crash).

[Dragstar]

question to nvidia users should i mine only neoscrypt algo with excavator  1060-1070 cards or mine all algos that pool supports?

[azioprismmm]

where do i put my wallet address with this miner?

[dropky]

question to nvidia users should i mine only neoscrypt algo with excavator  1060-1070 cards or mine all algos that pool supports?

Well.. that's up to you. Try running NHM with all algos enabled for one day and then standalone excavator with neoscrypt for another day and compare the profitability.

[dropky]

where do i put my wallet address with this miner?

In algorithm.add method:
{"id":1,"method":"algorithm.add","params":["equihash","equihash.eu.nicehash.com:3357","btcAddress"]}

See more here: https://github.com/nicehash/excavator/tree/master/api#algorithm-add

[MarkAz]

During benchmark excavator doesn't connect to the pool, it performs hashing on a test job which is hardcoded in the miner (always the same one). So, no free hashes for us If you don't believe me, you could always try to monitor the network..

Is there any particular reason that you couldn't just mine each particular algo (on the users account obviously), instead of just performing the benchmark?  I would think typically this isn't done because it would take a bunch of pools, but obviously Nicehash does them all, so that isn't an issue... Even if some of them aren't as profitable, it would be better than just burning the power for nothing - and be a real world benchmark.  Just wondering if there is some non-obvious reason...