playfast
Jr. Member
Offline
Activity: 131
Merit: 3
|
|
March 14, 2019, 12:14:47 PM |
|
I had stability issues with 0.41 and got it fixed, it looks like my version of win 10 wasn't properly starting the default set variables, changing them to setx and removing the = from each line worked for me.
|
|
|
|
NCarter84
Jr. Member
Offline
Activity: 195
Merit: 4
|
|
March 14, 2019, 03:20:48 PM |
|
I have a power color RX570 4gb card that just dies in after about 10 mins.... Running 1200 core with 900-1000mv and 2025 memory with 925-1000mv... setting TRM at 8+6.
Any thoughts?
|
|
|
|
john1010
|
|
March 14, 2019, 03:33:32 PM |
|
At last somebody here was creating a miner that exclusive for AMD's, I hope this miner has its advance feature against rival mining software, I will going to try this one and i wanna see the improvement.
|
|
|
|
P00P135
|
|
March 14, 2019, 04:01:54 PM |
|
I have a power color RX570 4gb card that just dies in after about 10 mins.... Running 1200 core with 900-1000mv and 2025 memory with 925-1000mv... setting TRM at 8+6.
Any thoughts?
Try 2000mem 8+8 or 8+7
|
|
|
|
NCarter84
Jr. Member
Offline
Activity: 195
Merit: 4
|
|
March 14, 2019, 05:06:12 PM |
|
I have a power color RX570 4gb card that just dies in after about 10 mins.... Running 1200 core with 900-1000mv and 2025 memory with 925-1000mv... setting TRM at 8+6.
Any thoughts?
Try 2000mem 8+8 or 8+7 All my 4gb cards seem to crash on 8+8 or 8+7, so been running them really well (940-975hs) on 8+6.
|
|
|
|
cecilia1
Newbie
Offline
Activity: 180
Merit: 0
|
|
March 14, 2019, 05:22:45 PM |
|
Hello,
Does anybody know what is the pool hashrate showed along with the GPU and average hashrate report?
|
|
|
|
evlo
Jr. Member
Offline
Activity: 155
Merit: 4
|
|
March 14, 2019, 07:56:19 PM |
|
are you planning to do low power for current cryptonight algo too? I'm getting about 1650w per rig, while i remember that with "cn8" it was much less with your miner
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
March 14, 2019, 08:41:29 PM |
|
Hello,
Does anybody know what is the pool hashrate showed along with the GPU and average hashrate report?
The pool hashrate is calculated in the miner from accepted shares. I would argue it is the most important metric to look at since this is effectively what you're getting paid for. Since it's only based on accepted shares at the user's pool, it automatically deducts dev fee, hw errors, rejected shares by the pool, any network downtime for your rig, etc. Your pool should mirror this value over time. We display the value since you started the miner though, pools will normally use a shorter window, so they won't match 100% at any point in time.
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
March 14, 2019, 08:47:29 PM |
|
are you planning to do low power for current cryptonight algo too? I'm getting about 1650w per rig, while i remember that with "cn8" it was much less with your miner
Well, for the time being this is pretty much it. The power draw difference compared to other miners is still significant. Our benchmarks shows a +2-3% increase in power draw, but it varies with gpu type and clocks, of course. You'll have to thank the Monero algo constructors for this, the random ops they chose for the inner loop are heavy for the gpu. The main culprit is the choice of 37.5% multiplications among the random instructions. However, all those choices were made for certain reasons, so it is what it is. For comparison, my 8 x Vega 56 rig pulls 1600W at the wall doing 16 kh/s, so 1650W is still a big consumption. May I ask what you're running in your rig?
|
|
|
|
evlo
Jr. Member
Offline
Activity: 155
Merit: 4
|
|
March 14, 2019, 11:08:03 PM |
|
11 570 nitro+ 1 580 xfx something IIRC 850 core 12.5khs reported
|
|
|
|
todxx (OP)
Member
Offline
Activity: 176
Merit: 76
|
|
March 15, 2019, 02:53:12 AM |
|
Team Red Miner v0.4.2 releasedhttps://github.com/todxx/teamredminer/releasesChanges in v0.4.2 - Added cryptonight v8 half (--algo cnv8_half) algo for coins such as stellite and masari.
- Added cryptonight v8 double (--algo cnv8_dbl) algo for coins such as x-cash.
- Added cryptonight v8 reverse waltz (--algo cnv8_rwz) algo for coins such as graft.
- Added support for running devices on multiple OpenCL platforms.
- Fixed more issues with console colors on older windows versions.
- Added more cpu verification optimization for CN/R. CN/R cpu usage should decrease ~70%.
In this release we've added some more CN variants! We've also fixed a couple bugs, and sped up our CN/R cpu verification code which should help rigs with low end celeron cpus.
|
|
|
|
pbfarmer
Member
Offline
Activity: 340
Merit: 29
|
|
March 15, 2019, 06:45:38 AM |
|
May I ask about the efficiency compared to CNv8, assuming that's what you have tuned for?
Finally got the vega 64 rig running - looking at 16.82kh/s @ 1375w ATW, or 2100h/s @ 167-168w avg (ex. 35w platform idle) - shooting for ~1370-1400 effective cclock between 825-837mv on avg. I had 2100 @ 165 in WTM for cnv2, so pretty much looking at a 2% power bump for same h/r like you suggest. Full disclosure - my cnv2 #s may have been slightly cherry-picked, but i've been more diligent on tuning cnv4, so i think it's a wash. Here's a screeny of my 64 rig, as well as a 580-8 rig (~995h/s @ 105w avg. - w/o the two POS hynix GPUS bringing down the avg, would be more like 1010.): Btw - I updated to 0.4.1 (right before your next release, of course.) Was waaay more stable - kept losing 50+h/r from a random card at startup every time on 0.4.0 (paired w/ frequent crashes,) so I guess your init updates were quite helpful!
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
March 15, 2019, 09:41:20 AM |
|
May I ask about the efficiency compared to CNv8, assuming that's what you have tuned for?
Finally got the vega 64 rig running - looking at 16.82kh/s @ 1375w ATW, or 2100h/s @ 167-168w avg (ex. 35w platform idle) - shooting for ~1370-1400 effective cclock between 825-837mv on avg. I had 2100 @ 165 in WTM for cnv2, so pretty much looking at a 2% power bump for same h/r like you suggest. Full disclosure - my cnv2 #s may have been slightly cherry-picked, but i've been more diligent on tuning cnv4, so i think it's a wash. Here's a screeny of my 64 rig, as well as a 580-8 rig (~995h/s @ 105w avg. - w/o the two POS hynix GPUS bringing down the avg, would be more like 1010.): Btw - I updated to 0.4.1 (right before your next release, of course.) Was waaay more stable - kept losing 50+h/r from a random card at startup every time on 0.4.0 (paired w/ frequent crashes,) so I guess your init updates were quite helpful! Looking good indeed! Hmm. Are we stupid enough to write the ANSI color codes to disk when --log=miner.log is used, or do you pipe the miner output into your log file? I guess you'd need --disable_colors to reg rid of them regardless .
|
|
|
|
pbfarmer
Member
Offline
Activity: 340
Merit: 29
|
|
March 15, 2019, 09:56:45 AM |
|
May I ask about the efficiency compared to CNv8, assuming that's what you have tuned for?
Finally got the vega 64 rig running - looking at 16.82kh/s @ 1375w ATW, or 2100h/s @ 167-168w avg (ex. 35w platform idle) - shooting for ~1370-1400 effective cclock between 825-837mv on avg. I had 2100 @ 165 in WTM for cnv2, so pretty much looking at a 2% power bump for same h/r like you suggest. Full disclosure - my cnv2 #s may have been slightly cherry-picked, but i've been more diligent on tuning cnv4, so i think it's a wash. Here's a screeny of my 64 rig, as well as a 580-8 rig (~995h/s @ 105w avg. - w/o the two POS hynix GPUS bringing down the avg, would be more like 1010.): Btw - I updated to 0.4.1 (right before your next release, of course.) Was waaay more stable - kept losing 50+h/r from a random card at startup every time on 0.4.0 (paired w/ frequent crashes,) so I guess your init updates were quite helpful! Looking good indeed! Hmm. Are we stupid enough to write the ANSI color codes to disk when --log=miner.log is used, or do you pipe the miner output into your log file? I guess you'd need --disable_colors to reg rid of them regardless . Ah yeah - got it set up as a systemd service w/ output redirected for universal logging - for those rare occasions I mine algos you don't support. Hadn't noticed that flag - thanks... Also, love the multi-platform support in 0.4.2 for my mixed rigs!
|
|
|
|
MaxHa$h
Newbie
Offline
Activity: 37
Merit: 0
|
|
March 15, 2019, 11:51:39 AM |
|
I have a power color RX570 4gb card that just dies in after about 10 mins.... Running 1200 core with 900-1000mv and 2025 memory with 925-1000mv... setting TRM at 8+6.
Any thoughts?
Try 2000mem 8+8 or 8+7 All my 4gb cards seem to crash on 8+8 or 8+7, so been running them really well (940-975hs) on 8+6. Ironically i get over 1KH/s on all 4gb 570/80's with 6+6 and over 900 H/s on 8gb with 8+8 (Zero crashes or restarts) https://ibb.co/17FN2xf
|
|
|
|
heavyarms1912
|
|
March 15, 2019, 01:32:20 PM |
|
I have a power color RX570 4gb card that just dies in after about 10 mins.... Running 1200 core with 900-1000mv and 2025 memory with 925-1000mv... setting TRM at 8+6.
Any thoughts?
Try 2000mem 8+8 or 8+7 All my 4gb cards seem to crash on 8+8 or 8+7, so been running them really well (940-975hs) on 8+6. Ironically i get over 1KH/s on all 4gb 570/80's with 6+6 and over 900 H/s on 8gb with 8+8 (Zero crashes or restarts) 8gb won't always hash higher than 4gb card on CNR. All depends on type of memory straps, clocks, temps.
|
|
|
|
WhyMe
|
|
March 15, 2019, 02:16:38 PM |
|
Hi, anyone has hashrates on last version for new algos with RX 580 8G ?
|
|
|
|
MaxHa$h
Newbie
Offline
Activity: 37
Merit: 0
|
|
March 15, 2019, 03:08:04 PM |
|
Hi, anyone has hashrates on last version for new algos with RX 580 8G ?
I see between 920-960 on 8GB click my image link in the above post
|
|
|
|
vmozara
Member
Offline
Activity: 190
Merit: 59
|
|
March 15, 2019, 04:43:31 PM |
|
Is there any chance of heavy variant miner of TRM?
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
March 15, 2019, 06:02:42 PM |
|
Is there any chance of heavy variant miner of TRM?
Absolutely, could very well happen next. I'm not sure about the future though, feels like there is some forking away from Heavy-derived algos going on.
|
|
|
|
|