SigiSinatra
Newbie
Offline
Activity: 50
Merit: 0
|
|
March 09, 2019, 10:22:32 PM |
|
cnr won't start...lyra2rev3 works though. https://i.imgur.com/H8oLL3i.pngwin10, driver ver. 18.6.1 // 1408-905 | 1100-900 What could be the problem?
|
|
|
|
pbfarmer
Member
Offline
Activity: 340
Merit: 29
|
|
March 09, 2019, 10:27:49 PM Merited by dragonmike (1) |
|
Yep, absolutely. Now also confirmed in HWINFO64. GPU Clock = 1360-ish (vs 1408 set, so 50-ish droop) GPU Memory Clock = 1100.0 MHz (solid) GPU SOC Clock = 1199.0 MHz (solid)
EDIT: same accross all 6 cards.
EDIT 2: I've now been playing with the lower mem pstates and noticed that whatever clocks I enter there are being ignored and the cards are using presets instead. Very weird. SOC then sets itself to different freqs as well depending on which mem p_states are active...
Ok, I see what happened... Looks like your PPT has an edit changing p7 SOC to 1200 from 1107. Look for "C0,D4,01,00,07" (only first occurrence) and try changing it to "6C,B0,01,00,07". Still not sure why that would require 1v. When I run 1200 SOC, i generally see voltage requirements around 900mv +/- 20mv. So the 1200 is hardcoded in the reg, yeah. I can change it down to 1107, it'll spare me more headaches. Now why are the mem p_states hard coded too? Whatever clocks I enter there are being ignored and the cards are using presets instead. Currently I run in mem_p2 and it forces 800MHz clock. That gives me 1800h/s per card (power draw for 6 at the wall is 1120W) and is actually more efficient than running 2000 at total power draw of 1400W! That drop from 1200 to 1107 SOC should save you at least 50mv, from my experience. Drivers don't allow changing mem states p0-p2 at runtime. You should be able to edit them in PPT if you want, but i've never bothered. In the past, you could not edit core p0-p5 either (the whole reason everyone uses PPTs,) but this was changed w/ 19.x drivers. Unfortunately, they didn't make the same change for mem states (at least in early vers of 19.x - haven't tested latest versions.)
|
|
|
|
philipma1957
Legendary
Offline
Activity: 4326
Merit: 8983
'The right to privacy matters'
|
|
March 09, 2019, 10:31:17 PM |
|
oh this looks like I have some for to do here.
I have been pointing the vegas to BCI
have not mined this coin in a while. other then with my thread ripper.
I will need to check out how to get the threadripper to mine this.
will a thread ripper work with this software or is it all cpu bound.
|
|
|
|
dragonmike
|
|
March 09, 2019, 10:37:32 PM |
|
Yep, absolutely. Now also confirmed in HWINFO64. GPU Clock = 1360-ish (vs 1408 set, so 50-ish droop) GPU Memory Clock = 1100.0 MHz (solid) GPU SOC Clock = 1199.0 MHz (solid)
EDIT: same accross all 6 cards.
EDIT 2: I've now been playing with the lower mem pstates and noticed that whatever clocks I enter there are being ignored and the cards are using presets instead. Very weird. SOC then sets itself to different freqs as well depending on which mem p_states are active...
Ok, I see what happened... Looks like your PPT has an edit changing p7 SOC to 1200 from 1107. Look for "C0,D4,01,00,07" (only first occurrence) and try changing it to "6C,B0,01,00,07". Still not sure why that would require 1v. When I run 1200 SOC, i generally see voltage requirements around 900mv +/- 20mv. So the 1200 is hardcoded in the reg, yeah. I can change it down to 1107, it'll spare me more headaches. Now why are the mem p_states hard coded too? Whatever clocks I enter there are being ignored and the cards are using presets instead. Currently I run in mem_p2 and it forces 800MHz clock. That gives me 1800h/s per card (power draw for 6 at the wall is 1120W) and is actually more efficient than running 2000 at total power draw of 1400W! That drop from 1200 to 1107 SOC should save you at least 50mv, from my experience. Drivers don't allow changing mem states p0-p2 at runtime. You should be able to edit them in PPT if you want, but i've never bothered. In the past, you could not edit core p0-p5 either (the whole reason everyone uses PPTs,) but this was changed w/ 19.x drivers. Unfortunately, they didn't make the same change for mem states (at least in early vers of 19.x - haven't tested latest versions.) I can confirm that hardcoding back down to 1107 solved everything. I'm now running 1408/1100 @ 850mV, SOC clock @ 1107 MHz. 12Kh/s, 1200W at the wall. Of course this means I will not push the mem clocks higher, but tbh it was never really worth it. Especially as it completely buggered up the power states/delivery. Thanks WillJ7 and pbfarmer you've been legends, as usual. If I only had merit left to spare I'd dish it all out to you!
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
March 09, 2019, 10:45:26 PM |
|
any chance of getting higher numbers than 16 for cn_config.... vega VII only gets 2.5Kh with TeamRed. I get 2.9Kh with SRBminer for cnr.
Hi alu! Oh for sure. I wrote about it earlier today, we have done zero optimizations for VII. We even debated not allowing Radeon VII on CN just because it would look bad and be far from the quality standard we're striving for. However, we left it in there, I believe it could still be interesting to choose TRM from an efficiency perspective in some cases. Other miners have higher perf though for sure. That said, I have gotten ~2750 h/s with TRM on my VII using 16+16 or 16-16, can't remember which was the better one. We have done some initial tests, and we know we'll provide a solid hashrate+efficiency on the VII as soon as we've had the proper time for some more R&D.
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
March 09, 2019, 10:47:20 PM |
|
cnr won't start...lyra2rev3 works though. win10, driver ver. 18.6.1 // 1408-905 | 1100-900 What could be the problem? Looking into a similar issue right now, a few Vega rigs are having init issues it seems. Just to collect some more data for me, what cpu do you have in that rig? Also, I assume you know you have enough swap enabled?
|
|
|
|
SigiSinatra
Newbie
Offline
Activity: 50
Merit: 0
|
|
March 09, 2019, 10:56:55 PM |
|
Looking into a similar issue right now, a few Vega rigs are having init issues it seems. Just to collect some more data for me, what cpu do you have in that rig? Also, I assume you know you have enough swap enabled? I have Intel Celeron G3930 @ 2.90GHz. I'm not sure what swap is, but if it means virtual memory then yes. I set virtual memory to 60000 MB. SRBminer cnr works with the same settings.
|
|
|
|
kaiokenx10
Newbie
Offline
Activity: 21
Merit: 0
|
|
March 09, 2019, 11:02:33 PM Last edit: March 09, 2019, 11:20:18 PM by kaiokenx10 |
|
Looking into a similar issue right now, a few Vega rigs are having init issues it seems. Just to collect some more data for me, what cpu do you have in that rig? Also, I assume you know you have enough swap enabled? I am having the exact same error, this has been my favorite miner for CN, but i can't get it running. When i do 12+12, the avg hashrate on my flashed vegas 56 says 1450 h/s, and the current hashrate shows 3000 h/s. I am not sure what is the problem. SRBminer works fine, but i preffer this one. Please give us the solution to this ASAP. P.S. I tried it both on 19.x drivers, and 18.6... same things happen
|
|
|
|
pbfarmer
Member
Offline
Activity: 340
Merit: 29
|
|
March 09, 2019, 11:40:58 PM |
|
Yep, absolutely. Now also confirmed in HWINFO64. GPU Clock = 1360-ish (vs 1408 set, so 50-ish droop) GPU Memory Clock = 1100.0 MHz (solid) GPU SOC Clock = 1199.0 MHz (solid)
EDIT: same accross all 6 cards.
EDIT 2: I've now been playing with the lower mem pstates and noticed that whatever clocks I enter there are being ignored and the cards are using presets instead. Very weird. SOC then sets itself to different freqs as well depending on which mem p_states are active...
Ok, I see what happened... Looks like your PPT has an edit changing p7 SOC to 1200 from 1107. Look for "C0,D4,01,00,07" (only first occurrence) and try changing it to "6C,B0,01,00,07". Still not sure why that would require 1v. When I run 1200 SOC, i generally see voltage requirements around 900mv +/- 20mv. So the 1200 is hardcoded in the reg, yeah. I can change it down to 1107, it'll spare me more headaches. Now why are the mem p_states hard coded too? Whatever clocks I enter there are being ignored and the cards are using presets instead. Currently I run in mem_p2 and it forces 800MHz clock. That gives me 1800h/s per card (power draw for 6 at the wall is 1120W) and is actually more efficient than running 2000 at total power draw of 1400W! That drop from 1200 to 1107 SOC should save you at least 50mv, from my experience. Drivers don't allow changing mem states p0-p2 at runtime. You should be able to edit them in PPT if you want, but i've never bothered. In the past, you could not edit core p0-p5 either (the whole reason everyone uses PPTs,) but this was changed w/ 19.x drivers. Unfortunately, they didn't make the same change for mem states (at least in early vers of 19.x - haven't tested latest versions.) I can confirm that hardcoding back down to 1107 solved everything. I'm now running 1408/1100 @ 850mV, SOC clock @ 1107 MHz. 12Kh/s, 1200W at the wall. Of course this means I will not push the mem clocks higher, but tbh it was never really worth it. Especially as it completely buggered up the power states/delivery. Thanks WillJ7 and pbfarmer you've been legends, as usual. If I only had merit left to spare I'd dish it all out to you! Awesome! Glad you got it worked out. A couple last tweaks for you... 1. You can go up to 1107 on mem - tho obv won’t make much of a diff. 2. I tend to see 1400 effective possible w/ 850mv. W/ a 1408 setting, you’re prob running quite a bit lower at least on some of your GPUs due to acg - my 64 settings range from 1425-1550 to get to 1400.. Just keep bumping your clock setting til you see close to 1400 in hwinfo. 3. If you really want to shave the last bit of excess power, you might be able to get voltages on some cards down as low as 825mv. Every card is different tho, so like the core clock setting, you’ll have to tune each individually. And it’s an opaque relaionship between voltage and core clock - as you drop mv, core will need to be raised to maintain your target clock. It’s a bit of an art form finding the balance.
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
March 09, 2019, 11:57:57 PM |
|
cnr won't start...lyra2rev3 works though. win10, driver ver. 18.6.1 // 1408-905 | 1100-900 What could be the problem? Looking into a similar issue right now, a few Vega rigs are having init issues it seems. Just to collect some more data for me, what cpu do you have in that rig? Also, I assume you know you have enough swap enabled? I am having the exact same error, this has been my favorite miner for CN, but i can't get it running. When i do 12+12, the avg hashrate on my flashed vegas 56 says 1450 h/s, and the current hashrate shows 3000 h/s. I am not sure what is the problem. SRBminer works fine, but i preffer this one. Please give us the solution to this ASAP. P.S. I tried it both on 19.x drivers, and 18.6... same things happen Working on it, always messy when we can't reproduce it ourselves, heavyarms1912 is right now testing build after build to zoom in on the issue.
|
|
|
|
|
heavyarms1912
|
|
March 10, 2019, 12:37:01 AM |
|
Thanks to this and kerney. figured out the problem. Would let kerney provide an update.
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
March 10, 2019, 12:37:55 AM |
|
BOOM! THIS IS IT! We should probably hardcode all our desired local worksizes and this wouldn't have happened. FOR THE TIME BEING FOLKS, SEND DONATIONS TO HAMMUH AND MAKE SURE YOUR ENVIRONMENT VARIABLE GPU_MAX_WORKGROUP_SIZE IS EITHER SET TO 256 OR NOT SET AT ALL!I'll start working on making sure that isn't necessary in an upcoming release.
|
|
|
|
SigiSinatra
Newbie
Offline
Activity: 50
Merit: 0
|
|
March 10, 2019, 12:59:15 AM |
|
BOOM! THIS IS IT! We should probably hardcode all our desired local worksizes and this wouldn't have happened. FOR THE TIME BEING FOLKS, SEND DONATIONS TO HAMMUH AND MAKE SURE YOUR ENVIRONMENT VARIABLE GPU_MAX_WORKGROUP_SIZE IS EITHER SET TO 256 OR NOT SET AT ALL!I'll start working on making sure that isn't necessary in an upcoming release. It worked! Thank you for quick reply.
|
|
|
|
hammuh
Jr. Member
Offline
Activity: 41
Merit: 1
|
|
March 10, 2019, 01:00:11 AM |
|
setting the environment variable worked for me too
|
|
|
|
heavyarms1912
|
|
March 10, 2019, 01:35:26 AM |
|
but with this setting hashrate is only 2000h/s, not 3000h/s 3000 was never the correct value for hashrate on vegas.
|
|
|
|
D8V1D
|
|
March 10, 2019, 02:28:02 AM |
|
Was having the same problems with my rx vega. Adding this to the bat file fixed it.... set GPU_MAX_WORKGROUP_SIZE=256
|
|
|
|
entroxseven
Newbie
Offline
Activity: 22
Merit: 0
|
|
March 10, 2019, 02:38:51 AM |
|
Any idea why I have no status info or hashrates display for me with 0.4.0 ?
It just says Team Red miner Version 0.4.0 with a blinking cursor behind it. But mining is working normally in the backround as I can see with my power consumption and with the pool-monitor.
I was using 0.3.7 before that which worked fine.
|
|
|
|
Mashy81
Jr. Member
Offline
Activity: 225
Merit: 1
|
|
March 10, 2019, 06:41:26 AM |
|
Was having the same problems with my rx vega. Adding this to the bat file fixed it.... set GPU_MAX_WORKGROUP_SIZE=256
This fixed it for me also
|
|
|
|
Mashy81
Jr. Member
Offline
Activity: 225
Merit: 1
|
|
March 10, 2019, 06:42:15 AM |
|
Great work guys. My rx 550s. Are hashing at 535hs each
|
|
|
|
|