[ANN] TeamRedMiner v0.10.10 - Ironfish/Kaspa/ZIL/Kawpow/Etchash and More - page 115.

gerixusyu

newbie

Activity: 24

Merit: 0

RX580 8gb NITRO+

What the problem is only 920mh/s on V8 XMR

Driver version is 18.6.1 mclock 1200/920mV gpuclock 2100/900mV cn_config 7+7, also try 8+8 and 6+6 same mh/s per card

Here a picture of my miner and OverDriveNtool

https://i.imgur.com/EGnTOT4.png TeamRedMiner 0.3.8
https://i.imgur.com/SZbyVPH.png OverDriveNtool

Can someone Help woth my cards and 4 rigs ,King regard's

JuanHungLo

hero member

Activity: 935

Merit: 1001

I don't always drink...

Quote from: dragonmike on January 01, 2019, 10:19:05 AM

You tell him Kerney!
Everybody loves a know-it-all noob coming out of nowhere to dish out insults and claim he knows everything better in every aspect than the devs! Grin

Ditto.

And to add to the conversation, the "Ignore" button works great. For me, 'This user is currently ignored' shows up on nearly every other page of this thread.

dragonmike

hero member

Activity: 1274

Merit: 556

You tell him Kerney!
Everybody loves a know-it-all noob coming out of nowhere to dish out insults and claim he knows everything better in every aspect than the devs! Grin

kerney666

member

Activity: 658

Merit: 86

Aight Randy aka smeads, I'm normally all about courtesy and being polite, but that has to be a two-way street, no?

First you dump 200 lines of Powershell output because you failed to RTFM in the release README.md or the frontpage of this ANN thread, more specifically API: The miner includes a read-only api based on the sgminer-5.5 API. Both the json and text formats are supported. For more details, we refer to the sgminer api documentation.. I assume you're not familiar with cgminer/sgminer/ccminer, and a google search for "sgminer api" was way too much to ask?

Next, when explained that this is a plain TCP/IP socket API, you whine about it with the following nugget...

Quote from: smeads on December 31, 2018, 04:43:41 PM

Yuck. Ok.... I'm passing on this miner. Sad

...because writing json over a plain TCP/IP socket connection and parsing the json response in a Powershell script is too much to ask for a devops engineer running "the largest Cryptonote calc site on the planet"? It should be a 10-15 line PS function. Anyway, I respect any decision you make in this regard, it's none of my business really.

However, what I don't respect is that you then proceed to do some 40 line advertisement post for your "Shitcoin Profit Manager" (sic) and of course have to top it off with another insult:

Quote from: smeads on December 31, 2018, 05:55:46 PM

Also.....1 more thing. Fastest is debatable. I use SRB for my AMD GPU's, and this miner was only a few hashes faster than SRB on 580's. Factor in a high % dev fee, and the earnings do not look as good.

Yeah... So let me ask you this: maybe the biggest CN profit calc site on the planet should include power draw and power cost when estimating the most profitable coin/miner? This miner can't do more than being the absolute tightest CNv2 kernel in all latency-critical parts of the algo. For the users that don't work very hard on their straps and clocks, they will "only" benefit from this by a lower power draw, not a higher hashrate. For Polaris cards, all miners are also very close to saturating the theoretical bandwidth, again meaning that the pros of this miner will only show if you care about wattage. For Vegas, let me know when another miner can produce 2400+ h/s reads on Vega 64 LCs or FEs at the same power draw and we can talk about the fastest CN miner.

Fwiw, we might add a built-in http adapter for the api at some point. We don't like pulling in external dependencies, even if the open source license is permissive. We might make an exception for this though.

evlo

jr. member

Activity: 155

Merit: 4

TRM has about 200W lower consumption on my 12gpu rig and about 1% better REAL hashrate, other miner cast, stak - almost same hashrate and consumption, cast - little better hashrate, almost unnoticeable, jce - worst hashrate and compsution of all, i did not bother to test other like ig

pbfarmer

member

Activity: 340

Merit: 29

Quote from: smeads on December 31, 2018, 05:55:46 PM

I also write Python, among a bunch of others. It's not a matter of me being able to do it, unfortunately. It's a matter of someone downloading, using my automated mining software, and having to set all of that up -- not to mention me being stuck supporting it. I'm trying to cater to the newbie... which means simplicity, allowing them to get up and running quickly and easily. My software also allows an advanced user to config almost anything.

If you haven't been, go to my site https://www.profitbotpro.com . I own and run the largest Cryptonote calc site on the planet (132 CN currencies). I support coins listed on an exchange, as well as 38 newer coins, not yet listed. If someone uses my mining software, it pairs with the site, and mines whatever is the most profitable. Users can also elect to use my cloud monitoring for free. I connect to more than 500 API's to pull in data. Look around, I have graphs, charts, and other valuable information. It's a 1 stop shop.

My calc form allows users to drill down, and only see what they want to see. Off of this form, they can easily (1 click) put those values in the miner, and it will profit mine based on that search criteria.

You basically set it up and walk away. I run it on all 8 of my rigs. Once a day I count my earnings.

Within the miner (and also on the cloud monitor), they see: the hashrate, accepted shares, rejected shares, est rewards in that currency, and est USD profit. I have screenshots and videos on the page, to show you how it all works.

I'm getting about 50k hits on my API daily... as well as 400-800 unique web hits per day. Trust me, I'm driving traffic to exchanges and pools. I can drive traffic to this miner too.... but I need them to use httpd versus installing yet another open source piece of software.

Hopefully this makes more sense. I'm a 1 man show on the dev side, I have 2 helpers to keep the coins current. So with writing several products, and also working a full time devops engineer job... my time is spread really thin. Happy New Year, BTW

Also.....1 more thing. Fastest is debatable. I use SRB for my AMD GPU's, and this miner was only a few hashes faster than SRB on 580's. Factor in a high % dev fee, and the earnings do not look as good.

I definitely wasn't advocating adding another step for your users to manage. I was simply suggesting incorporating the API comms into your software - and that while different, it shouldn't be much harder than interfacing w/ an HTTP API (or you can just wrap my whole interface into your software.)

When first released, TRM was significantly faster than SRB. Most people focused on the reduced power consumption, but this really is kind of missing the bigger picture, as h/r and power consumption are two sides of the same coin. Anyway, for similar power consumption, TRM gave >6% better hashrate vs SRB, easily covering the fee. Other miners including SRB and JCE have since taken clues from TRM and improved their performance recently, but I have yet to benchmark them again.

smeads

newbie

Activity: 17

Merit: 0

I also write Python, among a bunch of others. It's not a matter of me being able to do it, unfortunately. It's a matter of someone downloading, using my automated mining software, and having to set all of that up -- not to mention me being stuck supporting it. I'm trying to cater to the newbie... which means simplicity, allowing them to get up and running quickly and easily. My software also allows an advanced user to config almost anything.

If you haven't been, go to my site https://www.profitbotpro.com . I own and run the largest Cryptonote calc site on the planet (132 CN currencies). I support coins listed on an exchange, as well as 38 newer coins, not yet listed. If someone uses my mining software, it pairs with the site, and mines whatever is the most profitable. Users can also elect to use my cloud monitoring for free. I connect to more than 500 API's to pull in data. Look around, I have graphs, charts, and other valuable information. It's a 1 stop shop.

My calc form allows users to drill down, and only see what they want to see. Off of this form, they can easily (1 click) put those values in the miner, and it will profit mine based on that search criteria.

You basically set it up and walk away. I run it on all 8 of my rigs. Once a day I count my earnings.

Within the miner (and also on the cloud monitor), they see: the hashrate, accepted shares, rejected shares, est rewards in that currency, and est USD profit. I have screenshots and videos on the page, to show you how it all works.

I'm getting about 50k hits on my API daily... as well as 400-800 unique web hits per day. Trust me, I'm driving traffic to exchanges and pools. I can drive traffic to this miner too.... but I need them to use httpd versus installing yet another open source piece of software.

Hopefully this makes more sense. I'm a 1 man show on the dev side, I have 2 helpers to keep the coins current. So with writing several products, and also working a full time devops engineer job... my time is spread really thin. Happy New Year, BTW

Also.....1 more thing. Fastest is debatable. I use SRB for my AMD GPU's, and this miner was only a few hashes faster than SRB on 580's. Factor in a high % dev fee, and the earnings do not look as good.

pbfarmer

member

Activity: 340

Merit: 29

Quote from: smeads on December 31, 2018, 04:43:41 PM

Yuck. Ok.... I'm passing on this miner. Sad

The rest of them (JCE,SRB,XMRig,XMR-Stack) allow me to interface using built in httpd. I'm not collecting any money off my products, so I have to pick and choose what I spend my time on. I'm implementing some other new miners today, which also use httpd.

Please let me know if you guys change the way this is handled.

I get it, but i'd suggest you take a look before passing, considering this is the fastest cnv2 miner out there. I wrote the HTTP interface in a few hours in python - never having written python before. It's really pretty simple stuff. Btw, I have no affiliation w/ the miner/devs.

smeads

newbie

Activity: 17

Merit: 0

Yuck. Ok.... I'm passing on this miner. Sad

The rest of them (JCE,SRB,XMRig,XMR-Stack) allow me to interface using built in httpd. I'm not collecting any money off my products, so I have to pick and choose what I spend my time on. I'm implementing some other new miners today, which also use httpd.

Please let me know if you guys change the way this is handled.

pbfarmer

member

Activity: 340

Merit: 29

Quote from: smeads on December 31, 2018, 07:44:34 AM

Quote from: kerney666 on December 29, 2018, 12:15:39 PM

Quote from: smeads on December 29, 2018, 10:13:00 AM

I've been asked to include your miner, in my automated mining software (www.profitbotpro.com); however, I have a few questions:

#1 When will your miner support other CN algo's? The 1st page mentions CN8, but not any of the others.
#2 When will you have a release-ready version? It seems you folks may still be in beta?

Of course, there's 35 pages of forums.... so maybe I missed some information.

Have a good day.

Randy

Hi Randy!

Yeah, we probably had expected to be including more variants at this point, but right now CNv8 is the only CN family algo. I believe we'll make a push in Jan though.

The beta status is more due to missing features rather than stability issues etc. I'd say the major remaining feature is pool failover support. With that added, I think we might drop the beta tag, it's almost a little silly that it's still around

.

-- K

Thanks for the reply. I have your miner, mining... however, it's not letting me connect to the API.

Miner says: [2018-12-31 06:36:50] API initialized on 127.0.0.1:8080

Webpage says: (http://127.0.0.1:8080/)

This page isn’t working 127.0.0.1 sent an invalid response.
ERR_INVALID_HTTP_RESPONSE

Here's what I need from the API:

total hashrate, accepted shares, rejected shares, total shares (I can extrapolate rejected if you just give me total & accepted)

*** To add to this further.... if I go to powershell and run:

$api = Invoke-RestMethod -Uri "http://127.0.0.1:8080" -Method Get

PS C:\Users\randy>
$api = Invoke-RestMethod -Uri "http://127.0.0.1:8080" -Method Get
Invoke-RestMethod : The server committed a protocol violation. Section=ResponseStatusLine
At line:2 char:8
+ $api = Invoke-RestMethod -Uri "http://127.0.0.1:8080" -Method Get
+ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ CategoryInfo : InvalidOperation: (System.Net.HttpWebRequest:HttpWebRequest) [Invoke-RestMethod], WebException
+ FullyQualifiedErrorId : WebCmdletWebResponseException,Microsoft.PowerShell.Commands.InvokeRestMethodCommand

With the miner running:

PS C:\Users\randy> netstat

Active Connections

Proto Local Address Foreign Address State
TCP 127.0.0.1:6870 MR04:61890 ESTABLISHED
TCP 127.0.0.1:8080 MR04:62798 TIME_WAIT
TCP 127.0.0.1:8080 MR04:62799 TIME_WAIT
TCP 127.0.0.1:8080 MR04:62808 TIME_WAIT

Miner not running (nothing else using 8080)

PS C:\Users\randy> netstat

Active Connections

Proto Local Address Foreign Address State
TCP 127.0.0.1:6870 MR04:61890 ESTABLISHED
TCP 127.0.0.1:11331 MR04:57470 ESTABLISHED
TCP 127.0.0.1:11331 MR04:57471 ESTABLISHED
TCP 127.0.0.1:56657 MR04:56658 ESTABLISHED
TCP 127.0.0.1:56658 MR04:56657 ESTABLISHED

Thanks.

They exposed an RPC API, not an HTTP/REST one. You can use https://github.com/rdugan/cgminerhttpinterface to convert to an HTTP interface, or (prob more appropriately) just pull bits and pieces out to create your own API comms module for your platform.

smeads

newbie

Activity: 17

Merit: 0

Quote from: kerney666 on December 29, 2018, 12:15:39 PM

Quote from: smeads on December 29, 2018, 10:13:00 AM

I've been asked to include your miner, in my automated mining software (www.profitbotpro.com); however, I have a few questions:

#1 When will your miner support other CN algo's? The 1st page mentions CN8, but not any of the others.
#2 When will you have a release-ready version? It seems you folks may still be in beta?

Of course, there's 35 pages of forums.... so maybe I missed some information.

Have a good day.

Randy

Hi Randy!

Yeah, we probably had expected to be including more variants at this point, but right now CNv8 is the only CN family algo. I believe we'll make a push in Jan though.

The beta status is more due to missing features rather than stability issues etc. I'd say the major remaining feature is pool failover support. With that added, I think we might drop the beta tag, it's almost a little silly that it's still around

.

-- K

Thanks for the reply. I have your miner, mining... however, it's not letting me connect to the API.

Miner says: [2018-12-31 06:36:50] API initialized on 127.0.0.1:8080

Webpage says: (http://127.0.0.1:8080/)

This page isn’t working 127.0.0.1 sent an invalid response.
ERR_INVALID_HTTP_RESPONSE

Here's what I need from the API:

total hashrate, accepted shares, rejected shares, total shares (I can extrapolate rejected if you just give me total & accepted)

*** To add to this further.... if I go to powershell and run:

$api = Invoke-RestMethod -Uri "http://127.0.0.1:8080" -Method Get

PS C:\Users\randy>
$api = Invoke-RestMethod -Uri "http://127.0.0.1:8080" -Method Get
Invoke-RestMethod : The server committed a protocol violation. Section=ResponseStatusLine
At line:2 char:8
+ $api = Invoke-RestMethod -Uri "http://127.0.0.1:8080" -Method Get
+ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ CategoryInfo : InvalidOperation: (System.Net.HttpWebRequest:HttpWebRequest) [Invoke-RestMethod], WebException
+ FullyQualifiedErrorId : WebCmdletWebResponseException,Microsoft.PowerShell.Commands.InvokeRestMethodCommand

With the miner running:

PS C:\Users\randy> netstat

Active Connections

Proto Local Address Foreign Address State
TCP 127.0.0.1:6870 MR04:61890 ESTABLISHED
TCP 127.0.0.1:8080 MR04:62798 TIME_WAIT
TCP 127.0.0.1:8080 MR04:62799 TIME_WAIT
TCP 127.0.0.1:8080 MR04:62808 TIME_WAIT

Miner not running (nothing else using 8080)

PS C:\Users\randy> netstat

Active Connections

Proto Local Address Foreign Address State
TCP 127.0.0.1:6870 MR04:61890 ESTABLISHED
TCP 127.0.0.1:11331 MR04:57470 ESTABLISHED
TCP 127.0.0.1:11331 MR04:57471 ESTABLISHED
TCP 127.0.0.1:56657 MR04:56658 ESTABLISHED
TCP 127.0.0.1:56658 MR04:56657 ESTABLISHED

Thanks.

carlosmonaco

newbie

Activity: 105

Merit: 0

Quote from: livada on December 30, 2018, 04:46:10 PM

Quote from: carlosmonaco on December 30, 2018, 01:28:30 PM

Hello ,
I try to understand your conversation but my english is bad and my understanding about PPT are worst.
I'm trying to make my RX Vega 56 and 64 stable ( reference card with blower ).
I'm using these for now,
[Profile_0]
Name=Vega64
GPU_P0=852;900;0
GPU_P1=991;900;0
GPU_P2=1084;900;0
GPU_P3=1138;900;0
GPU_P4=1150;900;0
GPU_P5=1202;900;0
GPU_P6=1212;900;0
GPU_P7=1395;950
Mem_P0=167;900;0
Mem_P1=500;900;0
Mem_P2=800;900;0
Mem_P3=1090;900
Fan_Min=3000
Fan_Max=4900
Fan_Target=70
Fan_Acoustic=2400
Power_Temp=85
Power_Target=0

[Profile_1]
Name=Vega56
GPU_P0=852;900;0
GPU_P1=991;900;0
GPU_P2=1084;900;0
GPU_P3=1138;900;0
GPU_P4=1150;900;0
GPU_P5=1202;900;0
GPU_P6=1212;900;0
GPU_P7=1395;950
Mem_P0=167;900;0
Mem_P1=500;900;0
Mem_P2=800;900;0
Mem_P3=925;930
Fan_Min=3000
Fan_Max=4900
Fan_Target=70
Fan_Acoustic=2400
Power_Temp=85
Power_Target=0

Some Vega 56 are running around mem 900 others around 945, core is always 1395 .
I get between 1850 and 1950 for vega 56 and 1900-2000 for 64.
If I understand mem voltage will be the same as core is this one is higher ? so no need to set up this one ?
What are the best setting to get stability and if possible the best hashrate, ( I don't care about consumption it's winter here )

Best regards

vega 64 go up- 1100mem 1500+ gpu - 2100+HR
vega 56 950mem ok but for my nitro(hynix) i use only 1400mhz mem

Thanks ! So 1500 core and 1100 mem for 64 and 1500 950 for 56.
And what about voltage ?

MaxHa$h

newbie

Activity: 37

Merit: 0

Quote from: MoneroCrusher on December 31, 2018, 01:30:22 AM

Did you define hashrate print interval? This prints it all 15s "-l 15"

No I did not, didnt even know there was one as the miner states it auto does it every 30s. where are the commands for this miner listed? I randomly found the watchdog cmd on page 29?

Also, side note. this one a friends miner ive been trying to get running. anyone have this issue ever?
Miner will mine on blockchain driver, but may display hashrate once or none at all. this is on win10 x64 build 1709 or 1809 (doesnt like 1803 at all)
if i load any of the more current driver ie: 18.3.4,18.5.1,18.6.1. miner just hangs when ran, last line seen on the screen is teamredminer.exe, it does the same with other miners. if I DDU and go back to blockchain drivers it still wont mine. I have to either reinstall windows or update windows then recover. this miner was running fine for months, its a dell inspirion 3847 turned 3gpu miner, the pc updated to 1803 and ever since no matter what i do, complete format, delete partitions, start fresh results are identical. ive mixed a matched the riser cables, pluged the gpu directly on the board, updated the bios to latest. ive never had this kind of issue where a fresh install wouldnt fix.

Also i took an image of the PC to revert back to 1709 faster, but i also downloaded the complete 1709 update 3GB file but when ran it auto skips and puts 1809, anyone know how to force an update from a microsoft direct file?

MoneroCrusher

jr. member

Activity: 88

Merit: 1

Did you define hashrate print interval? This prints it all 15s "-l 15"

MaxHa$h

newbie

Activity: 37

Merit: 0

What would cause the miner to mine but not display hashrate? but the other info

Code:

[2018-12-30 21:20:00] Successfully initialized GPU 0: Polaris with 32 CU (PCIe 04:00.0) (CN 6+6)
[2018-12-30 21:20:00] Successfully initialized GPU 1: Polaris with 8 CU (PCIe 03:00.0) (CN 2+2)
[2018-12-30 21:20:00] Successfully initialized GPU 2: Polaris with 8 CU (PCIe 01:00.0) (CN 2+2)
[2018-12-30 21:20:28] Stats Uptime: 0 days, 00:00:30
[2018-12-30 21:20:47] Pool pool.supportxmr.com share accepted. (GPU2) (a:1 r:0) (61 ms)
[2018-12-30 21:20:49] Pool pool.supportxmr.com received new job. (job_id: HhaeGXAYpOmrRWSpyxkJYWIgo3KA)
[2018-12-30 21:20:49] Pool pool.supportxmr.com set difficulty to 15000
[2018-12-30 21:20:50] Pool pool.supportxmr.com received new job. (job_id: U4RnBdP4IzjmNkaMYTXT0TQyl3s6)
[2018-12-30 21:21:21] Pool pool.supportxmr.com received new job. (job_id: yK6sFGt0UXNz++YfnkkqGr2a8dff)
[2018-12-30 21:21:40] Pool pool.supportxmr.com share accepted. (GPU1) (a:2 r:0) (60 ms)
[2018-12-30 21:21:49] Pool pool.supportxmr.com received new job. (job_id: d/Dg4FzOw+SaKCaq+kRJlOqUXcZd)
[2018-12-30 21:21:49] Pool pool.supportxmr.com set difficulty to 10890
[2018-12-30 21:21:57] Pool pool.supportxmr.com share accepted. (GPU2) (a:3 r:0) (45 ms)
[2018-12-30 21:22:03] Pool pool.supportxmr.com received new job. (job_id: 8+i3/ALt048429smHDNxuJBJpn31)
[2018-12-30 21:22:13] Pool pool.supportxmr.com share accepted. (GPU1) (a:4 r:0) (57 ms)
[2018-12-30 21:22:14] Pool pool.supportxmr.com share accepted. (GPU1) (a:5 r:0) (60 ms)
[2018-12-30 21:22:16] Pool pool.supportxmr.com share accepted. (GPU2) (a:6 r:0) (20 ms)
[2018-12-30 21:22:22] Pool pool.supportxmr.com share accepted. (GPU2) (a:7 r:0) (43 ms)
[2018-12-30 21:22:31] Pool pool.supportxmr.com share accepted. (GPU1) (a:8 r:0) (36 ms)
[2018-12-30 21:22:49] Pool pool.supportxmr.com received new job. (job_id: vQLZX3tF7LiF7vddE4KFuMPwUCRR)
[2018-12-30 21:22:49] Pool pool.supportxmr.com set difficulty to 18570
[2018-12-30 21:23:03] Pool pool.supportxmr.com share accepted. (GPU1) (a:9 r:0) (40 ms)
[2018-12-30 21:23:08] Pool pool.supportxmr.com received new job. (job_id: 5yaYk7VzGn0xzOEpHQoxM7D0g9T8)
[2018-12-30 21:23:23] Pool pool.supportxmr.com share accepted. (GPU1) (a:10 r:0) (14 ms)
[2018-12-30 21:23:33] Pool pool.supportxmr.com received new job. (job_id: Ibdsi9d8MF5WN4jEIdxh4HQnYCPS)
[2018-12-30 21:23:34] Pool pool.supportxmr.com share accepted. (GPU2) (a:11 r:0) (33 ms)

livada

newbie

Activity: 417

Merit: 0

Quote from: carlosmonaco on December 30, 2018, 01:28:30 PM

Hello ,
I try to understand your conversation but my english is bad and my understanding about PPT are worst.
I'm trying to make my RX Vega 56 and 64 stable ( reference card with blower ).
I'm using these for now,
[Profile_0]
Name=Vega64
GPU_P0=852;900;0
GPU_P1=991;900;0
GPU_P2=1084;900;0
GPU_P3=1138;900;0
GPU_P4=1150;900;0
GPU_P5=1202;900;0
GPU_P6=1212;900;0
GPU_P7=1395;950
Mem_P0=167;900;0
Mem_P1=500;900;0
Mem_P2=800;900;0
Mem_P3=1090;900
Fan_Min=3000
Fan_Max=4900
Fan_Target=70
Fan_Acoustic=2400
Power_Temp=85
Power_Target=0

[Profile_1]
Name=Vega56
GPU_P0=852;900;0
GPU_P1=991;900;0
GPU_P2=1084;900;0
GPU_P3=1138;900;0
GPU_P4=1150;900;0
GPU_P5=1202;900;0
GPU_P6=1212;900;0
GPU_P7=1395;950
Mem_P0=167;900;0
Mem_P1=500;900;0
Mem_P2=800;900;0
Mem_P3=925;930
Fan_Min=3000
Fan_Max=4900
Fan_Target=70
Fan_Acoustic=2400
Power_Temp=85
Power_Target=0

Some Vega 56 are running around mem 900 others around 945, core is always 1395 .
I get between 1850 and 1950 for vega 56 and 1900-2000 for 64.
If I understand mem voltage will be the same as core is this one is higher ? so no need to set up this one ?
What are the best setting to get stability and if possible the best hashrate, ( I don't care about consumption it's winter here )

Best regards

vega 64 go up- 1100mem 1500+ gpu - 2100+HR
vega 56 950mem ok but for my nitro(hynix) i use only 1400mhz mem

carlosmonaco

newbie

Activity: 105

Merit: 0

Quote from: pbfarmer on December 29, 2018, 07:45:23 PM

Quote from: livada on December 29, 2018, 06:32:00 PM

Quote from: pbfarmer on December 29, 2018, 05:43:12 PM

Quote from: nordmann666 on December 29, 2018, 10:15:02 AM

Quote from: pbfarmer on December 27, 2018, 06:08:20 PM

mem P3 will always be set to P4 core in PPT. You can override this in ODNT w/ a profile later *if necessary*, though it shouldn't really ever be necessary. 'Mem voltage' in PPT and profiling tools (ODNT, wattman, etc) is simply a floor for core (and soc) voltage. If it's set below your core voltage, it's pretty much inconsequential (though I'm not certain if it also sets a floor for AVFS feature, so a reasonable number may still be desireable.) On the other hand, if your Mem P3 voltage setting is above your core P7, then you will run core at that voltage instead. Your actual mem voltage is fixed in the BIOS.

regarding stability - are you trying to get to effective 1450 @ 850, or your setting is 1450? If it's the former, that's likely just not possible. If it's the latter, i wouldn't focus so much on the setting, but rather look at the effective. I have some cards that drop 100Mhz under load (and therefore need to be set ~1500,) while others only drop only 25Mhz (and can be set closer to your 1430 number.)

so if i understand you...after i have to set P7 Core to 860 (everything lower will result in instant crash or core speed 850MHz) i can also set Mem P3 to 860 cause it will not affect the power consumption?
and i tried P7 1400 with 830,840,850mv but nothing to get stable - after 860mv i can push P7 until 1460-1480 (effective 1375-1395 depends on card)

it´s most important for me to reach stable hs around 1900-1950 with lowest possible power consumption. with getting back to vega 56 bios i save 10W/card - thats good..but the cards need 160-170W - to much

Yes, you can set mem-p3 to anything <= core-p7 and it will not affect power consumption. Or you can just leave it set to whatever you have in the PPT, as you will always be using your core-p7 mv while mining.

As for your stability issues - i would guess you're running up against SOC/controller power needs, rather than core. A 5-10mv change in voltage probably wouldn't give you an extra 80+Mhz headroom for core, so it's more likely the 860mv is required for the 1107Mhz SOC (56 still runs 1107 for the top state despite the much lower mem clock iirc.) If you really want lower power, you can try playing some tricks - like locking out mem-p4 and using the mem-p3 as your top state, which seems to use a tiered SOC clock, allowing something closer to the necessary <=950Mhz, which should further reduce power needs. I've only taken a cursory look at this functionality tho, so can't say for sure whether it will work.

How you chg p4 mem in ODNT? only p3 in Sad

How you change SOC in ODNT?
i use PPT generator but question is:
in ODNT i chg p3 mem etc 1100/860mv . what p4 mem use in ppt table? more = less ?
SOC on my vega56/64bios is 1200. Must change this?
http://imgbox.com/sRzuqn07

put your ppt table here for 56 and 64 .

now i use 940mhz/900mv 1500/900mv and i have 6*vega nitro+ hynix- 11600hr with 1080W from the wall or 11950Hr with 1150W for the wall.
ref card 56 samsung use 64 bios and 900mv and have 2030hr and vega 64 LE with 900mv have 2145hr

Sorry, meant p3 - 0-indexing got me... (fixed in OP)

You can't necessarily change SOC, it's auto-scaled based on mem clock. From my tests it seems p3 has a floor at SOC pN-1 (1107), while p2 scales lower, which is why i suggested locking out p3 and using p2 as the top state, since SOC clock only needs to be >= mem clock. SOC clock is in BIOS, and as you showed, is actually extracted to PPT, but I haven't had any luck changing it via PPT w/o crashing the GPU.

Hello ,
I try to understand your conversation but my english is bad and my understanding about PPT are worst.
I'm trying to make my RX Vega 56 and 64 stable ( reference card with blower ).
I'm using these for now,
[Profile_0]
Name=Vega64
GPU_P0=852;900;0
GPU_P1=991;900;0
GPU_P2=1084;900;0
GPU_P3=1138;900;0
GPU_P4=1150;900;0
GPU_P5=1202;900;0
GPU_P6=1212;900;0
GPU_P7=1395;950
Mem_P0=167;900;0
Mem_P1=500;900;0
Mem_P2=800;900;0
Mem_P3=1090;900
Fan_Min=3000
Fan_Max=4900
Fan_Target=70
Fan_Acoustic=2400
Power_Temp=85
Power_Target=0

[Profile_1]
Name=Vega56
GPU_P0=852;900;0
GPU_P1=991;900;0
GPU_P2=1084;900;0
GPU_P3=1138;900;0
GPU_P4=1150;900;0
GPU_P5=1202;900;0
GPU_P6=1212;900;0
GPU_P7=1395;950
Mem_P0=167;900;0
Mem_P1=500;900;0
Mem_P2=800;900;0
Mem_P3=925;930
Fan_Min=3000
Fan_Max=4900
Fan_Target=70
Fan_Acoustic=2400
Power_Temp=85
Power_Target=0

Some Vega 56 are running around mem 900 others around 945, core is always 1395 .
I get between 1850 and 1950 for vega 56 and 1900-2000 for 64.
If I understand mem voltage will be the same as core is this one is higher ? so no need to set up this one ?
What are the best setting to get stability and if possible the best hashrate, ( I don't care about consumption it's winter here )

Best regards

pbfarmer

member

Activity: 340

Merit: 29

Quote from: livada on December 29, 2018, 06:32:00 PM

Quote from: pbfarmer on December 29, 2018, 05:43:12 PM

Quote from: nordmann666 on December 29, 2018, 10:15:02 AM

Quote from: pbfarmer on December 27, 2018, 06:08:20 PM

mem P3 will always be set to P4 core in PPT. You can override this in ODNT w/ a profile later *if necessary*, though it shouldn't really ever be necessary. 'Mem voltage' in PPT and profiling tools (ODNT, wattman, etc) is simply a floor for core (and soc) voltage. If it's set below your core voltage, it's pretty much inconsequential (though I'm not certain if it also sets a floor for AVFS feature, so a reasonable number may still be desireable.) On the other hand, if your Mem P3 voltage setting is above your core P7, then you will run core at that voltage instead. Your actual mem voltage is fixed in the BIOS.

regarding stability - are you trying to get to effective 1450 @ 850, or your setting is 1450? If it's the former, that's likely just not possible. If it's the latter, i wouldn't focus so much on the setting, but rather look at the effective. I have some cards that drop 100Mhz under load (and therefore need to be set ~1500,) while others only drop only 25Mhz (and can be set closer to your 1430 number.)

so if i understand you...after i have to set P7 Core to 860 (everything lower will result in instant crash or core speed 850MHz) i can also set Mem P3 to 860 cause it will not affect the power consumption?
and i tried P7 1400 with 830,840,850mv but nothing to get stable - after 860mv i can push P7 until 1460-1480 (effective 1375-1395 depends on card)

it´s most important for me to reach stable hs around 1900-1950 with lowest possible power consumption. with getting back to vega 56 bios i save 10W/card - thats good..but the cards need 160-170W - to much

Yes, you can set mem-p3 to anything <= core-p7 and it will not affect power consumption. Or you can just leave it set to whatever you have in the PPT, as you will always be using your core-p7 mv while mining.

As for your stability issues - i would guess you're running up against SOC/controller power needs, rather than core. A 5-10mv change in voltage probably wouldn't give you an extra 80+Mhz headroom for core, so it's more likely the 860mv is required for the 1107Mhz SOC (56 still runs 1107 for the top state despite the much lower mem clock iirc.) If you really want lower power, you can try playing some tricks - like locking out mem-p4 and using the mem-p3 as your top state, which seems to use a tiered SOC clock, allowing something closer to the necessary <=950Mhz, which should further reduce power needs. I've only taken a cursory look at this functionality tho, so can't say for sure whether it will work.

How you chg p4 mem in ODNT? only p3 in Sad

How you change SOC in ODNT?
i use PPT generator but question is:
in ODNT i chg p3 mem etc 1100/860mv . what p4 mem use in ppt table? more = less ?
SOC on my vega56/64bios is 1200. Must change this?
http://imgbox.com/sRzuqn07

put your ppt table here for 56 and 64 .

now i use 940mhz/900mv 1500/900mv and i have 6*vega nitro+ hynix- 11600hr with 1080W from the wall or 11950Hr with 1150W for the wall.
ref card 56 samsung use 64 bios and 900mv and have 2030hr and vega 64 LE with 900mv have 2145hr

Sorry, meant p3 - 0-indexing got me... (fixed in OP)

You can't necessarily change SOC, it's auto-scaled based on mem clock. From my tests it seems p3 has a floor at SOC pN-1 (1107), while p2 scales lower, which is why i suggested locking out p3 and using p2 as the top state, since SOC clock only needs to be >= mem clock. SOC clock is in BIOS, and as you showed, is actually extracted to PPT, but I haven't had any luck changing it via PPT w/o crashing the GPU.

livada

newbie

Activity: 417

Merit: 0

Quote from: pbfarmer on December 29, 2018, 05:43:12 PM

Quote from: nordmann666 on December 29, 2018, 10:15:02 AM

Quote from: pbfarmer on December 27, 2018, 06:08:20 PM

mem P3 will always be set to P4 core in PPT. You can override this in ODNT w/ a profile later *if necessary*, though it shouldn't really ever be necessary. 'Mem voltage' in PPT and profiling tools (ODNT, wattman, etc) is simply a floor for core (and soc) voltage. If it's set below your core voltage, it's pretty much inconsequential (though I'm not certain if it also sets a floor for AVFS feature, so a reasonable number may still be desireable.) On the other hand, if your Mem P3 voltage setting is above your core P7, then you will run core at that voltage instead. Your actual mem voltage is fixed in the BIOS.

regarding stability - are you trying to get to effective 1450 @ 850, or your setting is 1450? If it's the former, that's likely just not possible. If it's the latter, i wouldn't focus so much on the setting, but rather look at the effective. I have some cards that drop 100Mhz under load (and therefore need to be set ~1500,) while others only drop only 25Mhz (and can be set closer to your 1430 number.)

so if i understand you...after i have to set P7 Core to 860 (everything lower will result in instant crash or core speed 850MHz) i can also set Mem P3 to 860 cause it will not affect the power consumption?
and i tried P7 1400 with 830,840,850mv but nothing to get stable - after 860mv i can push P7 until 1460-1480 (effective 1375-1395 depends on card)

it´s most important for me to reach stable hs around 1900-1950 with lowest possible power consumption. with getting back to vega 56 bios i save 10W/card - thats good..but the cards need 160-170W - to much

Yes, you can set mem-p3 to anything <= core-p7 and it will not affect power consumption. Or you can just leave it set to whatever you have in the PPT, as you will always be using your core-p7 mv while mining.

As for your stability issues - i would guess you're running up against SOC/controller power needs, rather than core. A 5-10mv change in voltage probably wouldn't give you an extra 80+Mhz headroom for core, so it's more likely the 860mv is required for the 1107Mhz SOC (56 still runs 1107 for the top state despite the much lower mem clock iirc.) If you really want lower power, you can try playing some tricks - like locking out mem-p4 and using the mem-p3 as your top state, which seems to use a tiered SOC clock, allowing something closer to the necessary <=950Mhz, which should further reduce power needs. I've only taken a cursory look at this functionality tho, so can't say for sure whether it will work.

How you chg p4 mem in ODNT? only p3 in Sad

How you change SOC in ODNT?
i use PPT generator but question is:
in ODNT i chg p3 mem etc 1100/860mv . what p4 mem use in ppt table? more = less ?
SOC on my vega56/64bios is 1200. Must change this?
http://imgbox.com/sRzuqn07

put your ppt table here for 56 and 64 .

now i use 940mhz/900mv 1500/900mv and i have 6*vega nitro+ hynix- 11600hr with 1080W from the wall or 11950Hr with 1150W for the wall.
ref card 56 samsung use 64 bios and 900mv and have 2030hr and vega 64 LE with 900mv have 2145hr

pbfarmer

member

Activity: 340

Merit: 29

Quote from: nordmann666 on December 29, 2018, 10:15:02 AM

Quote from: pbfarmer on December 27, 2018, 06:08:20 PM

mem P3 will always be set to P4 core in PPT. You can override this in ODNT w/ a profile later *if necessary*, though it shouldn't really ever be necessary. 'Mem voltage' in PPT and profiling tools (ODNT, wattman, etc) is simply a floor for core (and soc) voltage. If it's set below your core voltage, it's pretty much inconsequential (though I'm not certain if it also sets a floor for AVFS feature, so a reasonable number may still be desireable.) On the other hand, if your Mem P3 voltage setting is above your core P7, then you will run core at that voltage instead. Your actual mem voltage is fixed in the BIOS.

regarding stability - are you trying to get to effective 1450 @ 850, or your setting is 1450? If it's the former, that's likely just not possible. If it's the latter, i wouldn't focus so much on the setting, but rather look at the effective. I have some cards that drop 100Mhz under load (and therefore need to be set ~1500,) while others only drop only 25Mhz (and can be set closer to your 1430 number.)

so if i understand you...after i have to set P7 Core to 860 (everything lower will result in instant crash or core speed 850MHz) i can also set Mem P3 to 860 cause it will not affect the power consumption?
and i tried P7 1400 with 830,840,850mv but nothing to get stable - after 860mv i can push P7 until 1460-1480 (effective 1375-1395 depends on card)

it´s most important for me to reach stable hs around 1900-1950 with lowest possible power consumption. with getting back to vega 56 bios i save 10W/card - thats good..but the cards need 160-170W - to much

Yes, you can set mem-p3 to anything <= core-p7 and it will not affect power consumption. Or you can just leave it set to whatever you have in the PPT, as you will always be using your core-p7 mv while mining.

As for your stability issues - i would guess you're running up against SOC/controller power needs, rather than core. A 5-10mv change in voltage probably wouldn't give you an extra 80+Mhz headroom for core, so it's more likely the 860mv is required for the 1107Mhz SOC (56 still runs 1107 for the top state despite the much lower mem clock iirc.) If you really want lower power, you can try playing some tricks - like locking out mem-p3 and using the mem-p2 as your top state, which seems to use a tiered SOC clock, allowing something closer to the necessary <=950Mhz, which should further reduce power needs. I've only taken a cursory look at this functionality tho, so can't say for sure whether it will work.

Topic: [ANN] TeamRedMiner v0.10.10 - Ironfish/Kaspa/ZIL/Kawpow/Etchash and More - page 115. (Read 211877 times)