[ANN] TeamRedMiner v0.10.10 - Ironfish/Kaspa/ZIL/Kawpow/Etchash and More - page 59.

todxx

member

Activity: 176

Merit: 76

Team Red Miner v0.5.9 released

https://github.com/todxx/teamredminer/releases

Changes in v0.5.9

Added x16rv2 for the upcoming Ravencoin fork.
Optimization work on x16r: +8-10% hashrate depending on clocks.
Optimization work on x16r: mem clock no longer as important.
Issue fix: kernels split into multiple binaries to fix linux amdgpu-pro driver issues.

While the biggest change in this release was the addition of x16rv2, there are also some fixes for CN algos to work around driver bugs/crashes.
If anyone was having issues with CN algos resulting in crashes in v0.5.8, we suggest trying v0.5.9.

kerney666

member

Activity: 658

Merit: 86

Quote from: sudhsrx on September 24, 2019, 01:37:38 PM

Quote from: todxx on September 20, 2019, 12:57:49 PM

Team Red Miner v0.5.8 released

https://github.com/todxx/teamredminer/releases

Changes in v0.5.8

Added Chukwa-512 algo For Turtlecoin (trtl_chukwa).
Issue fix: kernels not loaded properly for Conceal.
Issue fix: added logic for pool reconnect on N rejected shares in a row (see --pool_max_rejects=N).

The main change in this release is that we've added support for the Chukwa algo for Turtle Coin!
This algo runs particularly well on the VIIs, so if you have some I suggest giving it a spin

hey todxx,

Got some issues over here,

*Well my system consists of Vega 64 STRIX OC running over HIVE OS (latest version 0.6-73@190922)
now here's the problem
-I tested Chukwa via TRM 0.5.8 ,posted my test results here before too.
-After a day , I tried running CNR as usual on TRM 0.5.8,well cards won't even initialize even after 10 mins.
-So Went back to TRM 0.5.7 (had CNR running over this about a week with great stability)
-Well now on TRM 0.5.7 one of my Gpu goes dead in 5 mins.
-Downgraded hive os to 0.6-69@190922
-Again TRM 0.5.7 one of my Gpu goes dead in 15 mins.
-Downgraded linux amd pro drivers from 19.40 to 18.40
-Again TRM 0.5.7 one of my Gpu goes dead in 15 mins.
>> well in short TRM 0.5.8 messed my system (not a accusation! I'm very much happy using TRM since inception Grin

) just need to find what went wrong. <<
as I have different OC profiles for CNR (used for a month) & chukwa, theres no question of memory timings.
could there be issues of card stuck with voltage?

kindly help Cry

Hi!

We're aware of issues on HiveOS Vega images for the 0.5.8 release for CN + variants, got notified today. We should soon be releasing 0.5.9 that includes x16rv2 and we'll make sure that runs ok with HiveOS on our own test rigs. What is strange is that you had issues when you rolled back to TRM 0.5.7 for cnr, I'm sorry I can't really say what's going on there. Given a clean reboot of the rig and applying the same clocks as you used before for cnr, nothing should have changed really? I assume you're still seeing the issues?

sudhsrx

newbie

Activity: 11

Merit: 0

Quote from: todxx on September 20, 2019, 12:57:49 PM

Team Red Miner v0.5.8 released

https://github.com/todxx/teamredminer/releases

Changes in v0.5.8

Added Chukwa-512 algo For Turtlecoin (trtl_chukwa).
Issue fix: kernels not loaded properly for Conceal.
Issue fix: added logic for pool reconnect on N rejected shares in a row (see --pool_max_rejects=N).

The main change in this release is that we've added support for the Chukwa algo for Turtle Coin!
This algo runs particularly well on the VIIs, so if you have some I suggest giving it a spin

hey todxx,

Got some issues over here,

*Well my system consists of Vega 64 STRIX OC running over HIVE OS (latest version 0.6-73@190922)
now here's the problem
-I tested Chukwa via TRM 0.5.8 ,posted my test results here before too.
-After a day , I tried running CNR as usual on TRM 0.5.8,well cards won't even initialize even after 10 mins.
-So Went back to TRM 0.5.7 (had CNR running over this about a week with great stability)
-Well now on TRM 0.5.7 one of my Gpu goes dead in 5 mins.
-Downgraded hive os to 0.6-69@190922
-Again TRM 0.5.7 one of my Gpu goes dead in 15 mins.
-Downgraded linux amd pro drivers from 19.40 to 18.40
-Again TRM 0.5.7 one of my Gpu goes dead in 15 mins.
>> well in short TRM 0.5.8 messed my system (not a accusation! I'm very much happy using TRM since inception Grin

) just need to find what went wrong. <<
as I have different OC profiles for CNR (used for a month) & chukwa, theres no question of memory timings.
could there be issues of card stuck with voltage?

kindly help Cry

kerney666

member

Activity: 658

Merit: 86

Quote from: livada on September 24, 2019, 05:18:13 AM

Quote from: kerney666 on September 24, 2019, 05:09:04 AM

Quote from: livada on September 24, 2019, 02:46:12 AM

nice job with chukwa algo. thx.

vega56@64bios(samsung) has the same HR as vega64Le.hmm..
i try all but vega 64le not give more than 127k HR.

3*vega56(samsung) + 1*veg64LE = 440000hr 755W(gpu=850mhz)

3*vega56(samsung) = 382000hr 710W(gpu=1440mhz)

6*vega56(hynix) = 632000 1090W(gpu=1100mhz)

3*vega56(hynix) = 326000hr(gpu=1440mhz)

vega56(samsung)=127000
vega56(hynix)=108800
vega64LE(samsung)=127500

my power supply is to week and not have power for max hr with all card.

Hi! Yeah, memory bandwidth is everything here, most Vegas with Samsung mem are capable of reaching the same hashrate, so it's not too surprising you're not seeing a boost with the V64s.

thx for this version.

You see my image? i see in your post invalid image???
i chg link for ss mybe now ok...

It's fine, the images are ok in your first post, they go invalid when I quote and reply.

livada

newbie

Activity: 417

Merit: 0

Quote from: kerney666 on September 24, 2019, 05:09:04 AM

Quote from: livada on September 24, 2019, 02:46:12 AM

nice job with chukwa algo. thx.

vega56@64bios(samsung) has the same HR as vega64Le.hmm..
i try all but vega 64le not give more than 127k HR.

3*vega56(samsung) + 1*veg64LE = 440000hr 755W(gpu=850mhz)
https://images2.imgbox.com/05/7f/aZqeszcv_o.jpg

3*vega56(samsung) = 382000hr 710W(gpu=1440mhz)
https://images2.imgbox.com/a2/87/JStrE6Xu_o.jpg

6*vega56(hynix) = 632000 1090W(gpu=1100mhz)
https://images2.imgbox.com/ef/5d/V6yMqOZG_o.jpg

3*vega56(hynix) = 326000hr(gpu=1440mhz)
https://images2.imgbox.com/82/12/OIPFnWFZ_o.jpg

vega56(samsung)=127000
vega56(hynix)=108800
vega64LE(samsung)=127500

my power supply is to week and not have power for max hr with all card.

Hi! Yeah, memory bandwidth is everything here, most Vegas with Samsung mem are capable of reaching the same hashrate, so it's not too surprising you're not seeing a boost with the V64s.

thx for this version.

You see my image? i see in your post invalid image???
i chg link for ss mybe now ok...

kerney666

member

Activity: 658

Merit: 86

Quote from: livada on September 24, 2019, 02:46:12 AM

nice job with chukwa algo. thx.

vega56@64bios(samsung) has the same HR as vega64Le.hmm..
i try all but vega 64le not give more than 127k HR.

3*vega56(samsung) + 1*veg64LE = 440000hr 755W(gpu=850mhz)

3*vega56(samsung) = 382000hr 710W(gpu=1440mhz)

6*vega56(hynix) = 632000 1090W(gpu=1100mhz)

3*vega56(hynix) = 326000hr(gpu=1440mhz)

vega56(samsung)=127000
vega56(hynix)=108800
vega64LE(samsung)=127500

my power supply is to week and not have power for max hr with all card.

Hi! Yeah, memory bandwidth is everything here, most Vegas with Samsung mem are capable of reaching the same hashrate, so it's not too surprising you're not seeing a boost with the V64s.

livada

newbie

Activity: 417

Merit: 0

nice job with chukwa algo. thx.

vega56@64bios(samsung) has the same HR as vega64Le.hmm..
i try all but vega 64le not give more than 127k HR.

3*vega56(samsung) + 1*veg64LE = 440000hr 755W(gpu=850mhz)
https://images2.imgbox.com/05/7f/aZqeszcv_o.jpg

3*vega56(samsung) = 382000hr 710W(gpu=1440mhz)
https://images2.imgbox.com/a2/87/JStrE6Xu_o.jpg

6*vega56(hynix) = 632000 1090W(gpu=1100mhz)
https://images2.imgbox.com/ef/5d/V6yMqOZG_o.jpg

3*vega56(hynix) = 326000hr(gpu=1440mhz)
https://images2.imgbox.com/82/12/OIPFnWFZ_o.jpg

vega56(samsung)=127000
vega56(hynix)=108800
vega64LE(samsung)=127500

my power supply is to week and not have power for max hr with all card.

sudhsrx

newbie

Activity: 11

Merit: 0

Quote from: todxx on September 20, 2019, 12:57:49 PM

Team Red Miner v0.5.8 released

https://github.com/todxx/teamredminer/releases

Changes in v0.5.8

Added Chukwa-512 algo For Turtlecoin (trtl_chukwa).
Issue fix: kernels not loaded properly for Conceal.
Issue fix: added logic for pool reconnect on N rejected shares in a row (see --pool_max_rejects=N).

The main change in this release is that we've added support for the Chukwa algo for Turtle Coin!
This algo runs particularly well on the VIIs, so if you have some I suggest giving it a spin

AWESOME OP,

Hashrate achieved on chukwa as below.

VEGA 64 STRIX OC : 119 KHS (Samsung + memtweak)
VEGA 64 LC (Sapphire) : 125 KHS (Samsung + memtweak)

PIOUPIOU99

copper member

Activity: 293

Merit: 11

.RX 470/480/570/580-4/8G = 55kh/59kh for 1050/810-1750/800

.Rx 550/560 2G = 25/27/28 for 1050/840-1700/750 or 1950/820

.Vega 56/64 (samsung & hynix) stock bios /AMD tweak = 105-110kh first test .

edit: Chukwa (trtl)

Elder III

sr. member

Activity: 1246

Merit: 274

Nice work on the Radeon VII, I can confirm getting around 250 KH/s at 250 watts.

kerney666

member

Activity: 658

Merit: 86

Quote from: KL0nLutiy1 on September 20, 2019, 01:53:18 PM

Hello, @todxx, could you add Nimiq support?

Well, we can look into it. I already know our numbers, we can do approx 515 kh/s on a Vega 64 at 1107 MHz mem clk and aggressive mem timings, this with ok efficiency, power draw will be on the high side but not insane.

However, first we need to implement special stratum support for their dumb mode, we for sure won't spend time porting their nano or full node setups from node.js to C/C++. This might mean we'll support a single pool, namely icemining.ca (they already run 80% of the network anyway). Also, all this work would be for a single coin, so a huge overhead to make a few bucks/day in dev fee.

Some algos and coins are just better suited for open source setups where you can pull in code and get going easily. Or, you're the type that doesn't care about open source licenses, but we do. Regardless, I'll look into it.

-- K

tale01

newbie

Activity: 121

Merit: 0

increase hashrate from ninjaminer, GREAT

heavyarms1912

full member

Activity: 729

Merit: 114

Radeon VII mining chukwa

Code:

245 KHs at 1550 core, 1250 hbm2 850mV
230 KHs at 1440 core, 1250 hbm2 830mV 
192 Khs at 1200 core, 1200 hbm2 800mV

KL0nLutiy1

newbie

Activity: 88

Merit: 0

Hello, @todxx, could you add Nimiq support?

todxx

member

Activity: 176

Merit: 76

Team Red Miner v0.5.8 released

https://github.com/todxx/teamredminer/releases

Changes in v0.5.8

Added Chukwa-512 algo For Turtlecoin (trtl_chukwa).
Issue fix: kernels not loaded properly for Conceal.
Issue fix: added logic for pool reconnect on N rejected shares in a row (see --pool_max_rejects=N).

The main change in this release is that we've added support for the Chukwa algo for Turtle Coin!
This algo runs particularly well on the VIIs, so if you have some I suggest giving it a spin

PIOUPIOU99

copper member

Activity: 293

Merit: 11

hello team ,

rx 550 on rig after reboot does not want to work anymore,
riser changed by an ok, tested a hdmi ok, but no longer works on trm, blue screen or error com, it tells you something ?

thx .

pepe000

newbie

Activity: 17

Merit: 0

Quote from: kerney666 on September 06, 2019, 10:51:49 AM

Quote from: pepe000 on September 06, 2019, 05:24:41 AM

Quote from: ku4eto on September 06, 2019, 12:45:48 AM

Quote from: pepe000 on September 05, 2019, 06:15:19 PM

Quote from: kerney666 on September 05, 2019, 03:13:52 PM

Quote from: pepe000 on September 05, 2019, 09:41:30 AM

I get error when starting TRM on cnr algo
- Failed to initialize device idx 5 (-31)

Well, I can't say you overdid it providing details about your setup Grin

.

The error is a buffer allocation error, so I'm guessing you're running win10 and have too little swap available to support >= 6 cards? If so, is it possible to increase it to at least >= 48GB?

Another wild guess is that your 6th is a smaller gpu, but you have specified a cn_config that tries to allocate more scratchpads than the gpu can support.

48GB did the job, but i am getting around 800 h/s on each rx 570 x 6 (1250/2100mhz)

You know, Ethash timings are not the same as CN timings.

I know, i have timings as recommended from TRM for rx570 8gb

I'm guessing you are running 570 8GBs with Hynix mem, the most problematic Polaris gpu to get good CN perf out of.

Do you know what mem type you have? If I'm correct, check this recent post by a fellow user: https://bitcointalksearch.org/topic/m.52275435, using those straps should give you a nice little bump at least and maybe hit 950 h/s or so.

Yup, it`s Hynix mem
Thank You all for the reply

kerney666

member

Activity: 658

Merit: 86

Quote from: pepe000 on September 06, 2019, 05:24:41 AM

Quote from: ku4eto on September 06, 2019, 12:45:48 AM

Quote from: pepe000 on September 05, 2019, 06:15:19 PM

Quote from: kerney666 on September 05, 2019, 03:13:52 PM

Quote from: pepe000 on September 05, 2019, 09:41:30 AM

I get error when starting TRM on cnr algo
- Failed to initialize device idx 5 (-31)

Well, I can't say you overdid it providing details about your setup Grin

.

The error is a buffer allocation error, so I'm guessing you're running win10 and have too little swap available to support >= 6 cards? If so, is it possible to increase it to at least >= 48GB?

Another wild guess is that your 6th is a smaller gpu, but you have specified a cn_config that tries to allocate more scratchpads than the gpu can support.

48GB did the job, but i am getting around 800 h/s on each rx 570 x 6 (1250/2100mhz)

You know, Ethash timings are not the same as CN timings.

I know, i have timings as recommended from TRM for rx570 8gb

I'm guessing you are running 570 8GBs with Hynix mem, the most problematic Polaris gpu to get good CN perf out of.

Do you know what mem type you have? If I'm correct, check this recent post by a fellow user: https://bitcointalksearch.org/topic/m.52275435, using those straps should give you a nice little bump at least and maybe hit 950 h/s or so.

pepe000

newbie

Activity: 17

Merit: 0

Quote from: ku4eto on September 06, 2019, 12:45:48 AM

Quote from: pepe000 on September 05, 2019, 06:15:19 PM

Quote from: kerney666 on September 05, 2019, 03:13:52 PM

Quote from: pepe000 on September 05, 2019, 09:41:30 AM

I get error when starting TRM on cnr algo
- Failed to initialize device idx 5 (-31)

Well, I can't say you overdid it providing details about your setup Grin

.

The error is a buffer allocation error, so I'm guessing you're running win10 and have too little swap available to support >= 6 cards? If so, is it possible to increase it to at least >= 48GB?

Another wild guess is that your 6th is a smaller gpu, but you have specified a cn_config that tries to allocate more scratchpads than the gpu can support.

48GB did the job, but i am getting around 800 h/s on each rx 570 x 6 (1250/2100mhz)

You know, Ethash timings are not the same as CN timings.

I know, i have timings as recommended from TRM for rx570 8gb

ku4eto

jr. member

Activity: 194

Merit: 4

Quote from: pepe000 on September 05, 2019, 06:15:19 PM

Quote from: kerney666 on September 05, 2019, 03:13:52 PM

Quote from: pepe000 on September 05, 2019, 09:41:30 AM

I get error when starting TRM on cnr algo
- Failed to initialize device idx 5 (-31)

Well, I can't say you overdid it providing details about your setup Grin

.

The error is a buffer allocation error, so I'm guessing you're running win10 and have too little swap available to support >= 6 cards? If so, is it possible to increase it to at least >= 48GB?

Another wild guess is that your 6th is a smaller gpu, but you have specified a cn_config that tries to allocate more scratchpads than the gpu can support.

48GB did the job, but i am getting around 800 h/s on each rx 570 x 6 (1250/2100mhz)

You know, Ethash timings are not the same as CN timings.

Topic: [ANN] TeamRedMiner v0.10.10 - Ironfish/Kaspa/ZIL/Kawpow/Etchash and More - page 59. (Read 211877 times)