PhoenixMiner 6.2c: fastest Ethereum/Ethash miner with lowest devfee (Win/Linux) - page 223.

melpheos

jr. member

Activity: 557

Merit: 5

Quote from: sk0mi on October 29, 2020, 04:07:18 PM

Quote from: melpheos on October 27, 2020, 12:27:57 PM

What's going on again ?
I switched from Ethermine to Niceshash (with daggerhashimoto) with my 8gb 1070Ti rig (motherboard has 8Gb of ram) and I see the same issue I had when my motherboard had only 2gb or ram !
The only difference I see between Nicehash and Ethermine is that nicehash request a 4Gb DAG while ethermine starts with 3.8Gb
The speed is halfed and the power usage is reduced by 25% (going from 100w to 75w)
Any ideas ?

Try T-rex miner?
Maybe it's the GPU architecture

Damn, that would be a shame if nvidia 8Gb card cannot mine above 4Gb !!!
Just did a test in -bench with 390 and indeed 4.* Gb seems to kill miners on nvidia 1070ti
Anyone else can test ? (i'm using drivers v 452.06)
I have the same issue with Claymore BTW

Ursul0

sr. member

Activity: 857

Merit: 262

Quote from: sk0mi on October 29, 2020, 05:21:02 PM

Quote from: UnclWish on October 25, 2020, 12:59:26 PM

Quote from: dudee on October 25, 2020, 12:43:04 PM

8x580
phoenix 238mhs, clay 243mhs
why such difference? long time ago mined eth and phoenix was faster.

With Phoenix it's need to use AMD memory tweak tool to set REF parameter. Claymore can do this by -rxboost option. Phoenix not.

With Claymore, since you have Polaris cards, not only -rxboost 30, but -strap 1 or -strap 2 also. (-strap will boost way much more than -rxboost) ...think these are Windows ONLY...maybe I'm wrong
I still use Claymore for injecting straps into my RX470s while using Phoenix for actual mining.
Just start Claymore with -rxboost 30 and -strap 2, disable all cards, minimize it and then run Phoenix. Wink

*found that while I was doing some other work with Claymore in background minimized and then I saw other people do it also with Phoenix
Sadly for me, Claymore won't work with latest drivers, so now I use it on just one rig with 18.12.1.1 driver.
With 20.8.1 Claymore just crashes for me. Sad

It only works for Polaris cards, I think. For other cards AmdMemTweak is doing same job, but one have to know to use it...
I still haven't found right parameters for my RX470s with Hynix memory, don't have enough knowledge and time...some day, maybe, some other cards... Undecided

I'm very happy using overdriventool with individual profiles for each gpu - polaris is @850mv 1120-1180MHz:1900-2200MHz (depending on the memory), same with timings with amdmemtweak, those you can also apply after the dag generation. (all from scheduler with highest privilege) and then you are free from any miner specific configs and can also finetune for different algos
edit: all polarises are bios modded with public (polarisbioseditor) straps
edit2: same for vegas, just instead of bios flash use powerplaytables (and overdriven when needed)

sk0mi

newbie

Activity: 24

Merit: 0

Quote from: UnclWish on October 25, 2020, 12:59:26 PM

Quote from: dudee on October 25, 2020, 12:43:04 PM

8x580
phoenix 238mhs, clay 243mhs
why such difference? long time ago mined eth and phoenix was faster.

With Phoenix it's need to use AMD memory tweak tool to set REF parameter. Claymore can do this by -rxboost option. Phoenix not.

With Claymore, since you have Polaris cards, not only -rxboost 30, but -strap 1 or -strap 2 also. (-strap will boost way much more than -rxboost) ...think these are Windows ONLY...maybe I'm wrong
I still use Claymore for injecting straps into my RX470s while using Phoenix for actual mining.
Just start Claymore with -rxboost 30 and -strap 2, disable all cards, minimize it and then run Phoenix. Wink

*found that while I was doing some other work with Claymore in background minimized and then I saw other people do it also with Phoenix
Sadly for me, Claymore won't work with latest drivers, so now I use it on just one rig with 18.12.1.1 driver.
With 20.8.1 Claymore just crashes for me. Sad

It only works for Polaris cards, I think. For other cards AmdMemTweak is doing same job, but one have to know to use it...
I still haven't found right parameters for my RX470s with Hynix memory, don't have enough knowledge and time...some day, maybe, some other cards... Undecided

mihailo_bojarsky1

newbie

Activity: 2

Merit: 0

When can we expect a new version?

sk0mi

newbie

Activity: 24

Merit: 0

Quote from: Crash95 on October 27, 2020, 07:54:02 AM

Quote from: zeef on October 26, 2020, 01:48:04 PM

Quote from: PhoenixMiner on December 26, 2017, 02:14:26 AM

IMPORTANT! All owners of AMD cards with 4 GB RAM must upgrade to PhoenixMiner 5.0e or later version to continue mining after DAG epoch 350 (both ETH and ETC has already passed DAG epoch 350). Additionally, here are some important tips for longest possible usage of 4 GB AMD cards with PhoenixMiner 5.0e and later:

Just moved to lolminer because today on this epoch (371) my 4gb cards stops hashing, cant realoce VRAM
Any future release to solve this problem?
I got all good configurated and switch from claysmore 2 months ago to PM and worked flawless until today.
I got too one 8gb card on the system and one vega. (This two i use PM)

Thanks

Same here !

For 4GB cards -rvram -1 should solve the problem, they wrote about it in text you just quoted...
Try update driver to latest from given list (20.7.1 or 20.8.1)...or higher, maybe.
One of my two 6 x RX470 4B rigs is working great just with 18.12.1.1, the other with the exact same cards wouldn't hash properly until I tried all and finally updated driver all to 20.7.1 (20.8.1).
I avoid Nicehash with 4GB cards because they mix ETH and ETC...making mess with that switching and giving me way less profit.
Lolminer have that "zombie mode" which still have to be developed, gonna try it later because PM works just fine for me.
Update driver to 20.7.1 or 20.8.1 or higher, it should solve your problem with RX480 4GB...DAG 373 is last my RX470s can handle under Win10.

sk0mi

newbie

Activity: 24

Merit: 0

Quote from: Jonet on October 29, 2020, 07:33:13 AM

I have just moved to linux to be able to mine a bit longer with 4GB

PhoenixMiner 5.1c says : "GPU1: Allocating DAG (3.91) GB; good for epoch up to #373"

Should it not be possible to mine until epoc #383 on linux?

amdgpu-pro-20.20-1098277 on ubuntu-20.04 server

PM have default setting of 2 epochs ahead...
(check readme file:
-eres
   Allocate DAG buffers big enough for n epochs ahead (default: 2) to
   avoid allocating new buffers on each DAG epoch switch, which should
   improve DAG switch stability. You may specify this option per-GPU.)
You can try change that...put -eres 10 and see what it says..

sk0mi

newbie

Activity: 24

Merit: 0

Quote from: melpheos on October 27, 2020, 12:27:57 PM

What's going on again ?
I switched from Ethermine to Niceshash (with daggerhashimoto) with my 8gb 1070Ti rig (motherboard has 8Gb of ram) and I see the same issue I had when my motherboard had only 2gb or ram !
The only difference I see between Nicehash and Ethermine is that nicehash request a 4Gb DAG while ethermine starts with 3.8Gb
The speed is halfed and the power usage is reduced by 25% (going from 100w to 75w)
Any ideas ?

Nicehash is mixing ETC and ETH on same address, because they are not actual pool but hashpower brokers.
That’s why you see higher DAG value, sometimes they start with ETC and switch to ETH, back to ETC and so on...
There is no rule, sometimes they start with ETH and switch to ETC...

Try T-rex miner?
Maybe it's the GPU architecture

alucard20724

sr. member

Activity: 703

Merit: 272

Results for 3070 FE

at stock clocks with power set to 50% in after burner, i'm getting 51.2 mhash/s @ 105 watts on the wattmeter for the GPU.. i haven't oc memory yet. I have the 3070 FE

edit: fresh install win10. 457.09 drivers. amd 3600 cpu stock. 16gb ddr4 2666 mem @47 watt system draw msi af 4.6.2 .. 155 watt total system draw under load. measurements on Kill A WATT meter.
Phoenix miner 5.1c
stock clocks 51.2 mhash/s @ 50% msi. 108 watts for gpu.
mem +500 55.28 mhash/s @ 50% msi. 108 watts for gpu.
mem +1000 58.4 mhash/s @ 50% msi. 108 watts for gpu.
mem +1250 59.08 mhash/s @ 50% msi. 108 watts for gpu.

Jonet

jr. member

Activity: 82

Merit: 3

I have just moved to linux to be able to mine a bit longer with 4GB

PhoenixMiner 5.1c says : "GPU1: Allocating DAG (3.91) GB; good for epoch up to #373"

Should it not be possible to mine until epoc #383 on linux?

amdgpu-pro-20.20-1098277 on ubuntu-20.04 server

VasilyS

member

Activity: 620

Merit: 21

Quote from: ?? on ??

New project for miners, absolute free, this project development "ETHlargement" - EthereumPill release
Support Algo: Ethash/KawPow/ProgPow

Download: ...

...

P.S.: the project is being drowned by competitors, don't trust anyone, check the information yourself

This project is sofisticated scam. It is a trojan horse steeling data from computers and sending it to US, Spanish, UAE and other servers. It is info based on execution of the sample on the analysis machine on joesandbox service.

dwarf0

newbie

Activity: 49

Merit: 0

https://i.postimg.cc/Hnh0V3VR/Screenshot-58.png

why would it go there (5.1c)?

Ursul0

sr. member

Activity: 857

Merit: 262

Quote from: BogdanCo on October 28, 2020, 02:46:25 PM

Quote from: UnclWish on October 28, 2020, 02:33:38 PM

Quote from: Ursul0 on October 28, 2020, 02:26:44 PM

some weirdness for you:

In Windows device manager list GPU's didn't go in order of PCI bus number. They're in alphabetical order there.

Still.... he has 4 Vega and 4 Rx 5700 according to PM

)

still the same fuckery. the dev manager is the correct, as it is also apparent from the hash.
also I run 12gpu rigs on asrock 14nm btc since 2017, I know all about enumeration of busses devices and win registry.

...not even trying to fix it now. it has other issues, more critical.

BogdanCo

hero member

Activity: 729

Merit: 513

Quote from: UnclWish on October 28, 2020, 02:33:38 PM

Quote from: Ursul0 on October 28, 2020, 02:26:44 PM

some weirdness for you:

In Windows device manager list GPU's didn't go in order of PCI bus number. They're in alphabetical order there.

Still.... he has 4 Vega and 4 Rx 5700 according to PM

)

UnclWish

sr. member

Activity: 1484

Merit: 253

Quote from: Ursul0 on October 28, 2020, 02:26:44 PM

some weirdness for you:

In Windows device manager list GPU's didn't go in order of PCI bus number. They're in alphabetical order there.

Ursul0

sr. member

Activity: 857

Merit: 262

some weirdness for you:

fapar

sr. member

Activity: 1414

Merit: 270

Undeads.com - P2E Runner Game

Quote from: zeef on October 27, 2020, 10:26:07 AM

Should the -gt value from an card from an algo be used to mine another algo, or we should should use auto-tune when switch to other algo and after the best value achieved we should fix it in .config file? Cool

For example my -gt values from ETH can be used to mine UBIQ?

Thanks

In your case, it is better to use auto-tuning -gt 0. Because the previously found best value for ETH may be worse for CLO or UBIQ.

NCarter84

jr. member

Activity: 195

Merit: 4

Quote from: jgonzi on October 28, 2020, 03:47:15 AM

Quote from: NCarter84 on October 26, 2020, 09:51:06 AM

Quote from: adaseb on October 25, 2020, 11:01:55 PM

Quote from: Junkey on October 25, 2020, 06:21:59 PM

Anyone have an idea why my rx480 rigs always have lower current hash values to reported on ethermine pool while my 1070 rigs always report more current than reported? They both have been running for a week or more consistently.

What percentage difference are we talking about here?

You are probably experiencing variance. Meaning that the speed you see per GPU says 30MH/s is just an estimate of what hashrate you should get in the long run. However it depends on how many shares you actually found. This number can be higher or lower than your actual hashrate. Most likely always lower since you got dev free mining, stales and might get internet connection problems.

Eventually as time -> infinity you will find its more and more close to your true hashing rate.

Your pool side average should be within a few % at any given time, you are correct. But, when your poolside average is always 3-4% lower, day after day after day then something is wrong with the miner.

Sneak Peak on a project I'm working on.

Expected Hashrate 57.26
Average Hash 55.19
Days At or Above 57.26 | 5
Days Below 57.26 | 42
% off from Expected -3.62%
Average Daily Shares 1192.28

Nice stats… you should try with another miner and give ups the information that Phoenix has the Mhs up that the reallity

I have those other stats for the miners too... Will be releasing all info next month once the test is done.

jgonzi

member

Activity: 639

Merit: 19

Quote from: NCarter84 on October 26, 2020, 09:51:06 AM

Quote from: adaseb on October 25, 2020, 11:01:55 PM

Quote from: Junkey on October 25, 2020, 06:21:59 PM

Anyone have an idea why my rx480 rigs always have lower current hash values to reported on ethermine pool while my 1070 rigs always report more current than reported? They both have been running for a week or more consistently.

What percentage difference are we talking about here?

You are probably experiencing variance. Meaning that the speed you see per GPU says 30MH/s is just an estimate of what hashrate you should get in the long run. However it depends on how many shares you actually found. This number can be higher or lower than your actual hashrate. Most likely always lower since you got dev free mining, stales and might get internet connection problems.

Eventually as time -> infinity you will find its more and more close to your true hashing rate.

Your pool side average should be within a few % at any given time, you are correct. But, when your poolside average is always 3-4% lower, day after day after day then something is wrong with the miner.

Sneak Peak on a project I'm working on.

Expected Hashrate 57.26
Average Hash 55.19
Days At or Above 57.26 | 5
Days Below 57.26 | 42
% off from Expected -3.62%
Average Daily Shares 1192.28

Nice stats… you should try with another miner and give ups the information that Phoenix has the Mhs up that the reallity

melpheos

jr. member

Activity: 557

Merit: 5

What's going on again ?
I switched from Ethermine to Niceshash (with daggerhashimoto) with my 8gb 1070Ti rig (motherboard has 8Gb of ram) and I see the same issue I had when my motherboard had only 2gb or ram !
The only difference I see between Nicehash and Ethermine is that nicehash request a 4Gb DAG while ethermine starts with 3.8Gb
The speed is halfed and the power usage is reduced by 25% (going from 100w to 75w)
Any ideas ?

zeef

newbie

Activity: 301

Merit: 0

Should the -gt value from an card from an algo be used to mine another algo, or we should should use auto-tune when switch to other algo and after the best value achieved we should fix it in .config file? Cool

For example my -gt values from ETH can be used to mine UBIQ?

Thanks

Topic: PhoenixMiner 6.2c: fastest Ethereum/Ethash miner with lowest devfee (Win/Linux) - page 223. (Read 784965 times)