SRBMiner Cryptonight AMD GPU Miner V1.9.3 - native algo switching

[Lio78]

Hi Doktor.

Good work with last 1.6.3 and Vega64, but now I have my 580 with less hashrate.

Vega64
1.6.2 at 2090 Hs
1.6.3 at 2120 Hs

Rx 580
1.6.2 at 930 Hs
1.6.3 at 910 Hs

Thank you

[UnclWish]

With 1.6.3 the same situation as with 1.6.2... On cn-v7 RX 580 8Gb max speed is 1000 h/s on intensity 108, but not every launch. Miner needs to several starts to get this speed. Waiting up to 5 minutes not helps. Mostly speed is 985 h/s. On 1.6.0 speed is up to 1035-1040 h/s with intensity 116-117.

On v7 all of my cards , 470, 480, 570, 580 4gb or 8gb they all get the same speed on every launch, this is how it was in earlier versions too, only heavy had this 'have to run multiple times' to get max speed thing.
I changed almost everything in this kernel like it was in 1.6.0 for Rx cards, also nothing that has impact on intensity setting has been changed since, i don't know last 10 versions, so if 116 was a good value in 1.6.0 it should be a good value now too, i don't understand how can't you set it to 116 as before.
I also used the fragment value you wrote earlier here as a default for 580 8gb, you said you get the best results with it.

For me it's interesting that on a few xfx 580 8gb i have, even not on the same rigs, i can use the same settings on v.1.6.2 as before , and get the same hashrate as before. But i havent changed or tried out new drivers or anything on these rigs, cause they are mining rigs, the most important is that they run always stable, so i stick to a driver that is proven to be good/stable.

Win 7 or Win10, doesn't matter, blockchain drivers installed from day 1.

So i basically just upload a new version of miner when testing to the rig, copy the old config files and run the miner.
This way using absolutely everything same i can know if something is better/worse.

But kernel files didn't work from 1.6.0 on 1.6.3 - this means that you deleted them... Maybe something wrong with kernels builded by 1.6.2... I didn't deleted them...
For me the fact is - 1.6.0 running with 117 intensity gives 1040 h/s, 1.6.3 with the same config gives less. Now will try to delete srb files and compile new ones...

EDIT: Problem solved by rebuilding kernel files on 1.6.3. Now speed is the same as on 1.6.0.

[doktor83]

With 1.6.3 the same situation as with 1.6.2... On cn-v7 RX 580 8Gb max speed is 1000 h/s on intensity 108, but not every launch. Miner needs to several starts to get this speed. Waiting up to 5 minutes not helps. Mostly speed is 985 h/s. On 1.6.0 speed is up to 1035-1040 h/s with intensity 116-117.

On v7 all of my cards , 470, 480, 570, 580 4gb or 8gb they all get the same speed on every launch, this is how it was in earlier versions too, only heavy had this 'have to run multiple times' to get max speed thing.
I changed almost everything in this kernel like it was in 1.6.0 for Rx cards, also nothing that has impact on intensity setting has been changed since, i don't know last 10 versions, so if 116 was a good value in 1.6.0 it should be a good value now too, i don't understand how can't you set it to 116 as before.
I also used the fragment value you wrote earlier here as a default for 580 8gb, you said you get the best results with it.

For me it's interesting that on a few xfx 580 8gb i have, even not on the same rigs, i can use the same settings on v.1.6.2 as before , and get the same hashrate as before. But i havent changed or tried out new drivers or anything on these rigs, cause they are mining rigs, the most important is that they run always stable, so i stick to a driver that is proven to be good/stable.

Win 7 or Win10, doesn't matter, blockchain drivers installed from day 1.

So i basically just upload a new version of miner when testing to the rig, copy the old config files and run the miner.
This way using absolutely everything same i can know if something is better/worse.

But kernel files didn't work from 1.6.0 on 1.6.3 - this means that you deleted them... Maybe something wrong with kernels builded by 1.6.2... I didn't deleted them...
For me the fact is - 1.6.0 running with 117 intensity gives 1040 h/s, 1.6.3 with the same config gives less. Now will try to delete srb files and compile new ones...

EDIT: Problem solved by rebuilding kernel files on 1.6.3. Now speed is the same as on 1.6.0.

halleluiah!
I should write somewhere a notice I wrote a notice that 'don't copy new version of miner over old one, or if you do make sure you delete .srb files before running it'.
Great, now someone with RX550 could test.

[zeef]

Hi doktor,

Nice job,

Testing version 1.6.3 with heavy and cryptov7 i got better hashes in vega (30hs more)

Using the bittube v2 algo i got a less hash , 900 hs.. instead of 15xx..

I tell about this problem some posts ago even in prevous version after a time the hash drop a lot , in bitube algo.

Can you check your kernel for this algo.

Im on 1709 windows version and using 18.6.1 amd drivers  

Thanks

I can't replicate this, as when using the 'pre-heating' trick on bittube algo i get a hashrate of ~1580hs and it stays like that for hours +- a few %.
Also no one else reported this here on the forum except you, but i have a few people i know they mine bittube and don't have this problem.

Its strange but happened..

What is the pre heating trick?

I tested in 2 pcs with one vega56 and its ok   but on vega64 have the problem i mencioned in this version.
In version 1.6.2 a mine with vega64 in 158x hashes 

Could be the 18.6.1 drivers?

Thanks


my config:

{
/* Type can be : normal, normalv7, lite, litev7, heavy, ipbc, artocash, alloy, marketcash, b2n, stellitev4, fast, haven */
"cryptonight_type" : "bittubev2",

/* Intensity 0-> auto intensity, or value from 1-300  */
"intensity" : 0,

/* To use 2 threads per card set double_threads to true  */
"double_threads" : true,

/* THERE ARE A LOT OF OPTIONS/PARAMETERS YOU CAN USE HERE */
/* YOU CAN FIND THEM IN THE README FILE */

"min_rig_speed" : 1500,

/* Advanced settings for each GPU manually */
/* Put in devices that you want to use, if you for ex. don't want to use gpu 2, just don't insert it,like in this example */
/* Id starts from 0 , not from 1 !! */
/* To get a list of available devices with their id's, use --listdevices parameter */
/* This is just an example, edit it and remove comment lines !! */

//Máximo que dá vega 64,  é 63 de intensity no heavy!! a partir de 62 perde performance!!
// Maximo vega 56 é 57


"gpu_conf" :
[
   { "id" : 0, "intensity" : 62 , "worksize" : 8, "threads" : 2}, //  vega 64:::   61 - 1555  é 62- 1585 63 - 878 de intensity no bittubev2!! a partir de 62 perde performance!!
//   { "id" : 1, "intensity" : 40, "worksize" : 8, "threads" : 2},
//   { "id" : 3, "intensity" : 30, "worksize" : 8, "threads" : 2},
//   { "id" : 4, "intensity" : 90, "worksize" : 8, "threads" : 1}
]


}

[treanski]

i get around 200hs more on 4 vegas on cryptonight-heavy. nice work

[UAGet]

1.6.3 version for my RX 560 2Gb produces 15-20 less hashes.

[Krisztian5]

Guys, please change the readme.txt algo section:


Not BitTube V2.... its BitTubeV2 (without the space). With space, the miner wont start.

[doktor83]

Guys, please change the readme.txt algo section:


Not BitTube V2.... its BitTubeV2 (without the space). With space, the miner wont start.

Going to fix readme,thanks.

Btw, when you enter a non-valid algo it returns you valid types:

cryptonight_type can be : [normal, normalv7, lite, litev7, heavy, bittubev2, artocash, alloy, marketcash, b2n, stellitev4, haven, fast, italo]

[RuMiner]

Great job! Thanks, Doc!
1.6.2 was worse for me, now 1.6.3 is better than 1.6.1, especially on Vega64. 8 cards give 17320 vs 17076 on 1.6.1, CNv7

[doktor83]

If you have a Vega, and want to mine V7 algos, use Adrenaline 18.6.1 drivers, if you want to mine Heavy algos, use Adrenaline 18.3.4

[doktor83]

Great job! Thanks, Doc!
1.6.2 was worse for me, now 1.6.3 is better than 1.6.1, especially on Vega64. 8 cards give 17320 vs 17076 on 1.6.1, CNv7

Yep, i managed to squeeze out more hash on Vegas, i will try to do even more 

[zeef]

If you have a Vega, and want to mine V7 algos, use Adrenaline 18.6.1 drivers, if you want to mine Heavy algos, use Adrenaline 18.3.4

will try

[LeonManveli]

If you have a Vega, and want to mine V7 algos, use Adrenaline 18.6.1 drivers, if you want to mine Heavy algos, use Adrenaline 18.3.4

And what about RX580 driver version to mine heavy and V7? Thank you Doktor83 for your work

[fellowbeing]

vega got more hash on 1.6.2, rx series maybe a little less, but next version should get things back on track.

I have a rig of 5 Vega Frontier Editions. I have not tried on heavy algo but the CN7 is giving me about 100 less. Intensity is set to 120 with double threads. I am very happy with SRB Miner.

with XMR STAK, if I set intensity to 2048 for each thread then I would get closer to the hashrate of SRB but still 100 or so less. With SRB 1.6.1 I am getting a stable 10.9 KH/s, if I added another Vega FE card it would be 13 KH/s.


You can see photos of my rig and its making here: https://postimg.cc/gallery/ivrtjh24/

As you know Vega FE cards have 16 GB of HBM2. Is it possible to run more threads on these GPUs ? It would be awesome if there was a way to utilize this memory. I would appreciate if you could provide an answer to this question, as you are writing the code for SRB Miner and you know better how much memory can be used.

Thanks for the awesome SRB Miner

I don't have FE vega, but try this for V7 algo:

"gpu_conf" :
[
   { "id" : 0, "intensity" : 126, "worksize" : 8, "threads" : 2}
]

Thanks, but it appears that doesn't help Vega FE cards. I tried 1.6.3 but it is the same so I am using 1.6.1 again. I'll mess around with it again and will update.

Edited: If I have to give up on Blockchain drivers for Adrenalin 18.6.1 to see these changes, then it will be a dead end for me, Vega FEs can't be overclocked on those drivers.

[doktor83]

Thanks, but it appears that doesn't help Vega FE cards. I tried 1.6.3 but it is the same so I am using 1.6.1 again. I'll mess around with it again and will update.

if you copied newer version of miner over old one, make sure you delete .srb files before running the miner it has to create new .srb files.

[UnclWish]

If you have a Vega, and want to mine V7 algos, use Adrenaline 18.6.1 drivers, if you want to mine Heavy algos, use Adrenaline 18.3.4

And what about RX580 driver version to mine heavy and V7? Thank you Doktor83 for your work

On latest 18.6.1 Windows 10 driver cn-v7 and heavy on my RX 580 works good! Speed is not lower than on 18.3.4.

[henri2018]

V1.6.3
- More speed on Vega56 cards on every algo (probably on Vega64 and VegaFE too, i don't have them to test)
- Added parameter --enableduplicategpuid (used in start.bat), that enables to use the same gpu id in gpu_conf multiple times
- Removed useless kernels 3 and 4
- Removed useless CPU opencl support
...
Don't copy new version of miner over old one, or if you do make sure you delete .srb files before running it!

Great update, dok.
I am mining bittube.
Vega56 increased minimum 30 h/s, some got 40 and 50 h/s.
RX580 increased about 5-10 h/s.

Need to monitor it within few hours to get it stable hashrate.

[mrsimit]

Great job! Thanks, Doc!
1.6.2 was worse for me, now 1.6.3 is better than 1.6.1, especially on Vega64. 8 cards give 17320 vs 17076 on 1.6.1, CNv7

Can you please share your Overdrivetool settings for V7 algo and your intensity

thanks a lot

[duyquang06]

Anyone can share the devcon file config reset for vega, please.

 

[livada]

Anyone can share the devcon file config reset for vega, please.

 

.reset.bat

cd %~dp0
timeout /t 10
devcon.exe disable "PCI\VEN_1002&DEV_687F"
timeout /t 10
devcon.exe enable "PCI\VEN_1002&DEV_687F"
timeout /t 10