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

[GKumaran]

V1.7.0

+ Performance boost for BitTube/Haven/Heavy/Italo algos up to ~10%, some cards gained 1% some 10%, i don't have exact numbers, you should try it out and see.
Vegas gained the most, also i hope that i fixed the Ellesmere 8Gb cards problem, not reaching max hashing speed. If you had 1k+ on some earlier versions, you should be able to reach it now again but you MUST set fragments to 0 if you are setting it by hand, otherwise it probably won't reach max speeds.
And no fee increase because of the extra performance, it stays at ~0.85%

+ Some small optimisation gained around +10hs on my test Vega56 on V8 algo

Spot on:

Vega 64 :

Algo	1.6.9	1.7.0
Heavy	1550	1685
V8	2018	2035

Almost same power consumption


Thx doc !~!

[1713534833]

1713534833

[1713534833]

1713534833

[1713534833]

1713534833

[doktor83]

480 8gb samsung (strap Ubermax 3.1) 1300/2000/1.00v
Bittube
config:
{ "id" : 0, "intensity" : 59, "worksize" : 8, "threads" : 2}

srb 1.6.9 1169h  
srb 1.7.0 1100h  

Speed increase on BitTube/Haven/Heavy/Italo algos up to ~10% (depends of gpu)

Houston we have a problem!

Good for you if you had that speed on 1.6.9, continue using that version then

Hello

I found a bug with the new startup monitor option
If you run benchmark-algos.bat, the startup monitor will kick in and restart the miner every 120seconds and the benchmark never finishes
The .bat file needs "--disablestartupmonitor" OR make sure that "--benchmarkalgos" always disables all watchdogs
edit: one more! benchmark-algos doesn't benchmark "dark" nor "festival"

Thanks for reporting these, i will check it out

So... My RX 574 + 100H/s (v.1.6.9 = 750-770   v.1.7.0= 850-870) - nice! But RX 564 lower on new version (v.1.6.9 = 500   v.1.7.0= 450) bad! All cards same options at two versions.

distinguished developer
one question
do you still support old pitcairn video cards ?
After the test of your latest version 1.7.0 I am at a loss?
srb 1.6.9 pitcairn 370 2 gb stelite v4 --- 485h / s
srb 1.7.0 ---- 226h / s
total failure

Confirming the problem On the RX560 cards there is big hashrate fall... (measured now on haven algo)
v1.6.9 = 613H/s
v1.7.0 = 501H/s
The settings, the driver version - all the same.
If I copy the kernel files, compiled on v1.6.9, to the .\Cache dir of v1.7.0, the RX560's hashrate is about 520-530H/s. But anyway - the problem is present

P.S. RX570/580 have some hashrate increase, is about 30H/s

Could you guys try out other configurations, like change the intensity and fragments parameter?
Yes i know you would all like if always the same settings would work, but that is not possible unfortunately, one code has different effects on different gpu types.

For 560 4g try setting fragments in gpu_conf to 32 for haven algo (that was the value in the previous version), leave the intensity same for this test.
and clear the cache folder, don't copy binaries from older versions etc.
also please leave it work for minimum 5 minutes to get the 5 min average

EDIT:

Ok, my tests show this on Haven :

RX550 4G no mod :

Intensity : 15
Worksize : 12
Threads  : 2

5 minute average result

1.6.9
Fragments 1   : 292 h/s
Fragments 32 : 270 h/s

1.7.0
Fragments 1   : 322 h/s
Fragments 32 : 304 h/s

Later i will do this same test on my Rx560 4G and report back.

[doktor83]

what to do if you get compute errors?

also getting compute errors, can't shake them, on vega FEs, 18.q3.1-oct1 driver, Heavy

Try lowering you memory clock by 10mhs, and see if that helps

[Juniormkl]

SRBMiner 1.7.0 already added to https://www.minermonitoring.com/ https://www.ethmonitoring.com/ Join our discord https://discord.gg/aGGaeqq

[MinersRus]

V1.6.9 vs V1.7.0 V8 comparison on my 7x Vega 56's


Vega #1 1891 1910 + 19 H/s
Vega #2 1871 1893 + 22 H/s
Vega #3 1884 1904 + 20 H/s
Vega #4 1852 1869 + 17 H/s
Vega #5 1844 1865 + 21 H/s
Vega #6 1855 1874 + 19 H/s
Vega #7 1840 1863 + 23 H/s
---------------------------------
Totals   13037 13178 +141 H/s or about +20 H/s for each Vega 56


Thx doc !~!

[Ultrasonik]

-------------------------STATS-------------------------

Miner version:  1.7.0

1 min. average speed:   956 H/s
5 min. average speed:   956 H/s   


-------------------------STATS-------------------------

Miner version:  1.6.7



Pool: bittube.miner.rocks:5555

1 min. average speed:   1002 H/s
5 min. average speed:   1001 H/s

Max. speed:             1004 H/s     

-------------------------------------------------------
Rx470 8GB   Algo: bittubev2   Adrenalin 18.10.2   for some reason the old version is the fastest

[Kgonla]

My firsts tests with RX 580 cards V8 algo:

v1.6.8 4gb = 1000 H/s
v1.6.8 8gb = 935 H/s

v1.6.9 4gb = 985 H/s 
v1.6.9 8gb = 975 H/s 

v1.7.0 4gb = 1010 H/s 
v1.7.0 8gb = 975 H/s 

for 1.6.8 & previous best intensity for 4gb was 61 & 116 for 8gb
for 1.6.9 & 1.7.0 best intensity for 4gb is 54 & 100 for 8gb

Also say that with newer versions the card connected to monitor is much more sensitive to background programs loosing more than 200 H/s in some cases when with 1.6.8 & previous was loosing at most 30 H/s

[dumada]

Looking at profitability charts something seems wrong, I went on whatomine and it suggested to switch back to ethereum , due to the power/hash combination of cnv8 vs ethhash.

Something is really off becase when i moved my rigs to ethereum from cnv8 my power use went up by 4 amps

that basically an extra amp per rig nearly.  that is around 1000 watts at 240V.

someone needs to correct the power use calculation used by the profit calcs . V8 uses more power than v7 but no where near close to ether as the calcs have it

I think you need to use your own power consumption to calculate. For me, ETH and V8 has similar profitability.

[sergbolshoy]

distinguished developer
one question
do you still support old pitcairn video cards ?
After the test of your latest version 1.7.0 I am at a loss?
srb 1.6.9 pitcairn 370 2 gb stelite v4 --- 485h / s
srb 1.7.0 ---- 226h / s
total failure

Confirming the problem On the RX560 cards there is big hashrate fall... (measured now on haven algo)
v1.6.9 = 613H/s
v1.7.0 = 501H/s
The settings, the driver version - all the same.
If I copy the kernel files, compiled on v1.6.9, to the .\Cache dir of v1.7.0, the RX560's hashrate is about 520-530H/s. But anyway - the problem is present

P.S. RX570/580 have some hashrate increase, is about 30H/s

Bro, can you tell me you settings about RX560 - v1.6.9 = 613H/s, my cards only 550H/s

[doktor83]

Okay, so some of you said that the hashrate on Rx560 4g is lower on 1.7.0 vs 1.6.9 on heavy algos (said Haven) - HERE
I took some time and made some tests :

Both cards are without bios mod.
Before every test i used the 'reset' button in ODT , just to be sure.
5 minute average result taken, same card, same clocks same driver same machine same everything.


RX550 4G:

Intensity : 15
Worksize : 12
Threads  : 2

1.6.9
Fragments 1   : 292 h/s
Fragments 32 : 270 h/s

1.7.0
Fragments 1   : 322 h/s
Fragments 32 : 304 h/s


RX560 4G:

Intensity : 15
Worksize : 8
Threads  : 2

1.6.9
Fragments 1   : 349 h/s
Fragments 32 : 340 h/s

1.7.0
Fragments 1   : 354 h/s
Fragments 32 : 356 h/s


So as you can see i can't reproduce the huge -100hs drop you said. As a matter of fact, i can only produce + hash, not -  

[doktor83]

My firsts tests with RX 580 cards V8 algo:

v1.6.8 4gb = 1000 H/s
v1.6.8 8gb = 935 H/s

v1.6.9 4gb = 985 H/s 
v1.6.9 8gb = 975 H/s 

v1.7.0 4gb = 1010 H/s 
v1.7.0 8gb = 975 H/s 

for 1.6.8 & previous best intensity for 4gb was 61 & 116 for 8gb
for 1.6.9 & 1.7.0 best intensity for 4gb is 54 & 100 for 8gb

Also say that with newer versions the card connected to monitor is much more sensitive to background programs loosing more than 200 H/s in some cases when with 1.6.8 & previous was loosing at most 30 H/s

Ok so obviously your 8G has some issue, maybe the strap, maybe just a bad card.
Anyways. You lowered the intensities and get same or better results. That's what i call a win 
Probably lower power consumption with lower intensities.

[goran89]

On 1.7.0 I am getting better hashrate
my rx 580 4gb get about 10 H/s more on v8 (from 990 to 1000)
but my rx 570 4gb get about 40 H/s more on v8 (from 945 to 985)

For 5 cards (3x570 and 2x580) my hashrate went up for about 140 H/s compared to 1.6.9 with same settings but power consumption also increased for about 30W (from 630W to 660W).
Also my 570's are getting to much memory errors now, so I will need to lower intensity on v8.
For fast algo I am getting about 200H/s less compared to 1.6.9. with same settings.

[doktor83]

On 1.7.0 I am getting better hashrate
my rx 580 4gb get about 10 H/s more on v8 (from 990 to 1000)
but my rx 570 4gb get about 40 H/s more on v8 (from 945 to 985)

For 5 cards (3x570 and 2x580) my hashrate went up for about 140 H/s compared to 1.6.9 with same settings but power consumption also increased for about 30W (from 630W to 660W).
Also my 570's are getting to much memory errors now, so I will need to lower intensity on v8.
For fast algo I am getting about 200H/s less compared to 1.6.9. with same settings.

play with fragments (its enough to try 0,1,2,4,8,16) and see which one gives the best results for that particular card.
also try aes 0,1 or 2 .
Fast algo is V7 variant nothing has changed in code for v7 and variants
my guess is that the default fragment value isn't optimal for your cards.

also in the next version i will also return the 'old' code for heavy, that was in 1.6.7, and some of you say it was better for you than the one now.
So with a new parameter (yes, again), you can choose which mode for heavy you want to use.

That way everyone will be happy 

[sergbolshoy]

Okay, so some of you said that the hashrate on Rx560 4g is lower on 1.7.0 vs 1.6.9 on heavy algos (said Haven) - HERE
I took some time and made some tests :

Both cards are without bios mod.
Before every test i used the 'reset' button in ODT , just to be sure.
5 minute average result taken, same card, same clocks same driver same machine same everything.


RX550 4G:

Intensity : 15
Worksize : 12
Threads  : 2

1.6.9
Fragments 1   : 292 h/s
Fragments 32 : 270 h/s

1.7.0
Fragments 1   : 322 h/s
Fragments 32 : 304 h/s


RX560 4G:

Intensity : 15
Worksize : 8
Threads  : 2

1.6.9
Fragments 1   : 349 h/s
Fragments 32 : 340 h/s

1.7.0
Fragments 1   : 354 h/s
Fragments 32 : 356 h/s


So as you can see i can't reproduce the huge -100hs drop you said. As a matter of fact, i can only produce + hash, not -  

So, I tried to change different settings (i,w,f,k,t) and nothing helped (algo Haven). SRB v.1.7.0 for RX564 (with mod bios) much slower than SRB v.1.6.9    
But for RX574 SRB v.1.7.0  (with mod bios) much upper than SRB v.1.6.9    

[djim123]

V1.6.9 vs V1.7.0 V8 comparison on my 7x Vega 56's


Vega #1 1891 1910 + 19 H/s
Vega #2 1871 1893 + 22 H/s
Vega #3 1884 1904 + 20 H/s
Vega #4 1852 1869 + 17 H/s
Vega #5 1844 1865 + 21 H/s
Vega #6 1855 1874 + 19 H/s
Vega #7 1840 1863 + 23 H/s
---------------------------------
Totals   13037 13178 +141 H/s or about +20 H/s for each Vega 56


Thx doc !~!

Can You share ,settings and configuration,please.

[livada]

6*vega 56
CN lite V7

srbminer 1.6.4 = 25500HR
srbminer 1.6.9 = 25650HR
srbimner 1.7.0 = 25580HR

[-manki-]

My Vega 64 gained 12H/s vs 1.6.9 on v8. Stable at 2134H/s, power consumption remained the same.

[dingdongtobias]

6*vega 56
CN lite V7

srbminer 1.6.4 = 25500HR
srbminer 1.6.9 = 25650HR
srbimner 1.7.0 = 25580HR

11hs less for 1 card that lot of money lost lol

[Rabotaet]

What are the settings for rx570 4Gb for v1.7.0 and v8?

[MinersRus]

V1.6.9 vs V1.7.0 V8 comparison on my 7x Vega 56's


Vega #1 1891 1910 + 19 H/s
Vega #2 1871 1893 + 22 H/s
Vega #3 1884 1904 + 20 H/s
Vega #4 1852 1869 + 17 H/s
Vega #5 1844 1865 + 21 H/s
Vega #6 1855 1874 + 19 H/s
Vega #7 1840 1863 + 23 H/s
---------------------------------
Totals   13037 13178 +141 H/s or about +20 H/s for each Vega 56


Thx doc !~!

Can You share ,settings and configuration,please.

All settings for SRBMiner are the same for each Vega 56 and done in the V8 config file:

intensity: 112
worksize: 16
threads: 2

Vega 56's #1, #2 and #3 are in one system that has Radeon Adrenalin Edition 18.8.1 installed which actually produces high hash rates than the recommended Radeon Adrenalin Edition 18.5.1 installed in the system with Vega 56's #4, #5, #6 and #7.

These two systems are on a Biostar H81A motherboard and are using a single HP 1200 Watt Server Power Supply (87% Efficient) with a 17x 6-pin Breakout board. The motherboard is powered by an 180 watt Pico PSU that gets it's 12 volt input from a 6-pin connector on the breakout board. I like my systems only running on a single power supply.


System one is hashing on an  i3-4340 at 73 H/s and system two is hashing on an  i5-4570 at 195 H/s.

All power measurements are done at the wall with a Power Usage Monitor. That way the entire system is measured (motherboard, processor, memory, ssd) and the Vega 56's and it is the real power not the incomplete ones that only look at the software measurements like HWINFO which does not consider the inefficiencies of power supply and internal power conversion circuits or the rest of the system.

Realize that V8 is much more power hungry than V7 was.

I do not use soft power play tables as I have found that V7 and V8 basically run between P6 & P7 states on the core and P3 on memory. The Vega 56's are reference designs of various vendors with stock BIOS.

So my Wattman settings for six of the Vega 56's are:

Core:
Min P6 1212 MHz @ 905 mv
Max P7 1452 MHz @ 925 mv

Memory:
Min/Max P3 935 MHz @ 900 mv

Fans:
Min: 2700
Target: 4900

Temperature:
Max: 75
Target: 65

EDIT: I forgot this setting. I needed to do this +10 as a few of the Vega 56's clocks were only in the high 1200's MHz and not the 1350+ MHz they should be and were flicking between P6 and P7 state.. It turned out those were being power limited. So be sure to do this setting also.
Power Limit +10

I do have one problematic Vega 56 that I have to run at:

Core:
Min P6 1212 MHz @ 960 mv
Max P7 1452 MHz @ 985 mv

Memory:
Min/Max P3 900 MHz @ 950 mv

With all that out of the way here are the final numbers:

System one 3x Vega 56's:

Vega #1 1911 H/s
Vega #2 1893 H/s
Vega #3 1904 H/s
--------------------
Total  Vega's 5708 H/s
i3-4340 adds 73 H/s
-------------------------
System Total:  5781 H/s

Total Power at the wall: 675 watts


System two 4x Vega 56's:

Vega #4 1869 H/s
Vega #5 1865 H/s
Vega #6 1874 H/s
Vega #7 1863 H/s
--------------------
Total  Vega's 7471 H/s
i5-4570 adds 195 H/s
-------------------------
System Total:  7666 H/s

Total Power at the wall: 898 watts