- Joined
- Apr 11, 2007
- Messages
- 33,266
Afterburner is working fine on my 390X's with the latest drivers. Though I'm overclocking only and my cards are voltage locked.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
Ick nicehash down
- Thread starter Elkwood
- Start date
Afterburner is working fine on my 390X's with the latest drivers. Though I'm overclocking only and my cards are voltage locked.
- Joined
- Oct 19, 2004
- Messages
- 11,933
Afterburner 4.4.0 didn’t work with AMD RX580 drivers for about a 2 month span. Is 4.4.2 working? I haven’t tried it yet.
- Joined
- Apr 11, 2007
- Messages
- 33,266
- Joined
- Oct 19, 2004
- Messages
- 11,933
Interesting. None of my friends with RX580 for mining have been able to do afterburner 4.4 for the last month or two. That’s four of us on various rigs.
- Joined
- Apr 11, 2007
- Messages
- 33,266
I keep getting the pop up to upgrade AB to 4.4.2, but since it's working I don't really want to fuck with it and screw myself.
390 =/= RX cards. RX cards have been problematic for the afterburner's latest versions.
- Joined
- Apr 11, 2007
- Messages
- 33,266
I have no voltage control. Mining Equihash that's about the best I can do. 418-424 H/s. If I'm mining Ethash I can crank the core up to 1140, temps are like 65c and I get about 31.5 - 32 Mh/s.
- Joined
- Oct 19, 2004
- Messages
- 11,933
Not worth the wear and tear on the laptop if you aim to keep it. Laptop fans and cooling systems aren’t robust enough to take the heat build up. Your laptop fan will be maxed out and the unit uncomfortably hot.
My opinion anyway
No way I’d mine on my alienware with 1060. I plan to keep that for at least five years.
well I paid $400 for the laptop at walmart open box special. The way I look at it, the return on investment will happen within a year, aka within warranty. I have the GPU underclocked and the CPU undervolted, and both max out at 70C. Not going to be 24/7 like my mining rig, but probably 12 hours a day when I'm not actually using it. I'll let you know if she breaks
Works for me but im using 4.4.2 it was a pain tho to get it to undervolt not even sure it will stick with a reboot.
- Joined
- Oct 19, 2004
- Messages
- 11,933
I haven’t tried 4.4.2 with my AMD cards. I have wattman in a stable spot right now. I’m not messing with it. I have 12 RX580 machines. In wattman you have to make all changes individually for each card. Talk about a pain! The next time wattman resets after a driver update then I’ll try wattman again. Right now wattman is working. Even in between reboots.
the HP omen? yeah I'm hoping to kill it before the HP warranty runs out then not mining on the replacement.well I paid $400 for the laptop at walmart open box special. The way I look at it, the return on investment will happen within a year, aka within warranty. I have the GPU underclocked and the CPU undervolted, and both max out at 70C. Not going to be 24/7 like my mining rig, but probably 12 hours a day when I'm not actually using it. I'll let you know if she breaks
Not a bad idea to check on your power cables guys. I was using a standard 3 prong computer cable for the rig I am mining on (which is drawing 400w from the wall) and decided now that it is set up, going to go headless and put the pc in another room. When I unplugged the cable, it was pretty warm. Power cables should not be warm! I cut the cable so I don't use it again, but it was rated for 10A 250V. NO WAY this cable could take 2500 watts. I compared the inner wires and the copper strands are no thicker than what I found in my chinese sata to pcie adapters. Just something that doesn't take a minute to check, check your power cords and adapter cables to make sure they aren't heating up. Swapped to a cable rated for 10A 120V, and it stays cool to the touch. Glad I found this before adding another GPU.
Parja
[H]F Junkie
- Joined
- Oct 4, 2002
- Messages
- 12,671
Irrelevant. He has a 1080Ti.
Elkwood
Limp Gawd
- Joined
- Sep 17, 2017
- Messages
- 178
Not bad advice at all !! Its not a bad idea to check all the ps cables from time to time. I also check my outlets. Had to replace a few in recent years.
My poor house is over 65 years old. Only has a 100 amp service. So i am very careful about the loads.
With AMD cards, I've found registry editing and the overdrive tool to be the most stable/efficient to use. For my 5xVega 64 machine, I don't even have the control center or wattman installed. I'm using the information from here https://www.amdxmrminer.com/reach-stable-2000-hs-mining-xmr-on-vegas/ and overdrive tool to set the fan speed. 10.8Kh/s (cpu counts for about 650h/s) from the rig currently. does about .1 XMR a day.
MSI Afterburner works fine, I was able to undervolt the card, she now runs 1658MHz core and 12006MHz VRAM with 840mv +/- 15mv, Claymore 10.2 gets +38MH/sec when the screen is blanked out, when I type on it it is 37.5MH/sec +/- a bit. I can have the kids play minecraft and it only goes down by 2-3MH/sec whereas the game runs full max at 60-65fps with no issues. That is amazing. That is with the undervolting to max 875mV. Actually, when I keep RAM at 12GHz but also increase clocks to 2GHz all graphs show it holds it but the MH/sec go down by 2-4. The card so far runs fastest with the core at around 1600-1800 and the VRAM as fast as it goes. Temps lowered considerably, the thing got quiet again 43-45°C with 22°C room temp.
Nanopool says, with the current hash rate it's ~100€/month. Electricity not incl.. company expense. Would it make sense to switch between coins or better steadily mine eth. There is the dual mining option in Claymore that I havent yet used, to keep it simple.
I installed Awesome Miner but need to play wit it, so far it fetches my CDM and I can control it, install new miners etc..all good. what I miss and need to set up are the right pools and wallets for the coins that make sense. Until I get that sorted, a simple Command Prompt is OK, dont need no GUI yet with 1 coin and 1 miner.
I am not sure if I leave the VRAM at 12G for 24/7 mining, regardless the low voltage and watercooling, that is ~1GHz above specs. Micron, I love you
43-45°C with 22°C room temp.Nanopool says, with the current hash rate it's ~100€/month. Electricity not incl.. company expense
. Would it make sense to switch between coins or better steadily mine eth. There is the dual mining option in Claymore that I havent yet used, to keep it simple.I installed Awesome Miner but need to play wit it, so far it fetches my CDM and I can control it, install new miners etc..all good. what I miss and need to set up are the right pools and wallets for the coins that make sense. Until I get that sorted, a simple Command Prompt is OK, dont need no GUI yet with 1 coin and 1 miner.
I am not sure if I leave the VRAM at 12G for 24/7 mining, regardless the low voltage and watercooling, that is ~1GHz above specs. Micron, I love you
Parja
[H]F Junkie
- Joined
- Oct 4, 2002
- Messages
- 12,671
MSI Afterburner works fine, I was able to undervolt the card, she now runs 1658MHz core and 12006MHz VRAM with 840mv +/- 15mv, Claymore 10.2 gets +38MH/sec when the screen is blanked out, when I type on it it is 37.5MH/sec +/- a bit. I can have the kids play minecraft and it only goes down by 2-3MH/sec whereas the game runs full max at 60-65fps with no issues. That is amazing. That is with the undervolting to max 875mV. Actually, when I keep RAM at 12GHz but also increase clocks to 2GHz all graphs show it holds it but the MH/sec go down by 2-4. The card so far runs fastest with the core at around 1600-1800 and the VRAM as fast as it goes. Temps lowered considerably, the thing got quiet again
Nanopool says, with the current hash rate it's ~100€/month. Electricity not incl.. company expense
I installed Awesome Miner but need to play wit it, so far it fetches my CDM and I can control it, install new miners etc..all good. what I miss and need to set up are the right pools and wallets for the coins that make sense. Until I get that sorted, a simple Command Prompt is OK, dont need no GUI yet with 1 coin and 1 miner.
I am not sure if I leave the VRAM at 12G for 24/7 mining, regardless the low voltage and watercooling, that is ~1GHz above specs. Micron, I love you
You could be making 2.5-3x that much if you mined something like Sibcoin or Zclassic. As I said before, there's a lot better algorithm options for NVIDIA than Ethash.
With my one Vega 64, Sapphire Trixx works well, mining Electroneum using AwesomeMiner with Cast-XMR miner (specifically made for Vega and block chain drivers. I had to update the miner to the most recent version for it to work as a note). Pulling over 2Kh/s. Found if I keep the clock speed more steady as in not fluctuating I can maintain the 2K+ speeds. Also keeping the GPU cool or more importantly the HBM cool was the key, at least for me to push the HBM speeds up. So I control the power by adjusting clockspeed and GPU voltage to keep it relatively smooth. Anyways in my case coolness led to very cool hashing.
With one card that would make sense, with 5 it's faster to run power shell and over drive tool scripts to disable and enable all 5 cards and set their fan speeds automatically. The cards on boot are already set to the OC settings I found to work best via registry edit. This let's me reboot the PC every other day and have it start mining right away.
Looking into it, wow..that would be 250-300€/month.....almost unbelievable
Thank you Parja
Well the Vega 64 is holding up after 6+ hours over 2kh/s for CryptoNight. That hash rate is higher then the my 1080 Ti and two 1070's combined. Seems like if you loose focus on the miner command window and just clicking on something else it will go down to less than 2Kh/s and won't go back up unless you stop and restart the miner. Anyways the Hashrate average is the average from the beginning of the run, so far this is about 7 hours ~2050h/s: I can see why Vega's are sought after.
Elkwood
Limp Gawd
- Joined
- Sep 17, 2017
- Messages
- 178
Wow that is some nice return for a single vega. I been doing ETN on my 1920x threadripper at around 1050ish h/s. On nano pool that is netting me around 40-50 coins a day. I am using ccminer 3.8.
have the chip running 3.9gigs at 1.250volts with cas 14 3200 mem speed. The vega must be a real heat monster for u to be seeing +50c temps on water. I love to see what type of returns i could
get with a set of them if i could find them at a reasonable price.
Pulling 380w from the wall, yes that could be better - I probably should do more experimentation there. I also did use two 1700x's which pulled around 900h/s combined, they are now mining using Nicehash. Anyways, Vega 64, Nano and a 290 in one day plus a few hours -> 250 coins pulled in at MPH.
Elkwood
Limp Gawd
- Joined
- Sep 17, 2017
- Messages
- 178
How are the 1700x doing on NH for u ? The 1920x was pretty good for me at around 4 to 6 dollars a day. I want to accrue ETN
or i have it pointed at NH. If your system is only pulling 380 watts from the wall and u are mining with the cpu/vega thats really not that bad.
If i pull the plug out of the wall on my ups. Its showing me a 700ish watt draw on it from my system which is 2 1080 ti's/1920x + 32" 4k monitor.
that's while mining of course. Not sure if that is true draw from the wall.
I been very impressed with the 1920x.
The 1700x at 4ghz with NH is around $2 and the other 1700x at 3.4ghz is around $1.80. That 1920x is indeed impressive, with less cores it is doing much better then two 1700x's!
Then on NiceHash.com it shows I am making $4.20 a day? So which is more accurate?
Then on NiceHash.com it shows I am making $4.20 a day? So which is more accurate?
- Joined
- Oct 19, 2004
- Messages
- 11,933
I’ve previously found nicehash estimates to be square on.
been playingwith that stuff, right now I use awesome miner with a profit-switcher setup...heck..its doing 8,50$/day for a single card, mostly Skein algo.
Thank you very much for kicking me the right direction !!!!!!!!
-----=-__[H]ard|OCP__-=---- gives you wings
JustReason
razor1 is my Lover
- Joined
- Oct 31, 2015
- Messages
- 2,483
running my LC vega64 @1290/1100 and 5cores of my 1800X @4.0Ghz, I'm pulling 260Watts at the wall. GPU voltage for core sets around .875V. Best I can approximate GPU Watts at is ~130W. GPU chip power shows an average of 102W, not sure what rest of card adds.
Finally got around to what you recommended, using XMR-Stak for both 1700x now which gives a combine rate around 1.1kh/s+. Interesting is the 3.4ghz 1700x is doing 560h/s and the 4ghz 1700x with 3466mhz memory is doing 545h/s
, same basic configuration of using 8 cores (no smt ones).What is you hash rate with the Vega? Are you using XMR-Stak as well for that? I tried but it would only pull around 1200h/s while Cast XMR does 2000h/s+. Ok I will try decreasing clock speed to 1300mhz and see what happens and decrease voltage as well.
JustReason
razor1 is my Lover
- Joined
- Oct 31, 2015
- Messages
- 2,483
It seems to be very memory sensitive more so than core. You have to run the 2 thread approach and I get 660H/s each so 1320H/s with both. I also have to run 17.12.2 because for some reason I cant get the block chain drivers to work, probably need a new win 10 install. but with everything working I don't want to mess with it.Finally got around to what you recommended, using XMR-Stak for both 1700x now which gives a combine rate around 1.1kh/s+. Interesting is the 3.4ghz 1700x is doing 560h/s and the 4ghz 1700x with 3466mhz memory is doing 545h/s
What is you hash rate with the Vega? Are you using XMR-Stak as well for that? I tried but it would only pull around 1200h/s while Cast XMR does 2000h/s+. Ok I will try decreasing clock speed to 1300mhz and see what happens and decrease voltage as well.
Took the Vega 64 down to 1300mhz and still maintained the same hash rate of 2.05kh/s, Hurray! took GPU voltage down but anything under 1v it crashes. It reduced about 50w of power in the end. XMR-Stak pushes 342h/s out of the 6700 which is also decent.
What is the configuration for the 2 thread approach? Can you post the CPU config file?
Also ordered a Vega FE as well
- seems like the only available Vega in the world at the moment. Vega FE you can load two different drivers (pro and a gaming driver, you maybe able to load three drivers but not sure on that one). Now if I can get the hashrate up with the 17.12.2 I will be rather happy with that. I think I will try again plus hoping the FE pro drivers can give some good hash rates as well.Tried one thread per ccx. But at 3.7GHz, no getting more than 79h/s per ccx that way.
Whole CPU efficiently loaded 1 thread per real core, I measure just 73h/s per core.
The best aggregate number I can get is about 588h/s. Maybe 600 right after launch,
but quickly settles down to 588 after a few minutes...
Two threads both loaded to Core 0 no smt, and that way too is also 39h/s+39h/s.
Four threads on Core 0 no smt, only 19.8h/s thread, still about 79h/s per core...
Even with no competition between cores for cache and memory, I don't see how
we get there from here with a two thread approach.
16 threads, two per even numbered core (without smt?), gives me 533h/s.
About same dimished performance as 16 threads distributed in the normal
SMT way.
Ohwait, maybe onto something. Four "low power: true" threads one core=133h/s.
And about same for three threads or two. But tanks when all real cores are given
the same load. One core per ccx, doubleloaded with lowpower thread seems OK.
But is still nowhere near what one regular thread per real core would add up to....
Interesting but is it useful? Four lowpower threads, one per SMT cores 0,1,8,9
give 82h/s each for 327h/s. Better than half performance for only the work of
two cores! But doesn't stack well with additional threads on other cores.
Maybe something else useful could be done with other cores, that wouldn't cause
so many cache misses? I'm just assuming its cache misses that are the problem
that prevents scaling, don't actually know. Measuring just two unleashed cores,
we could theorize 1308h/s might be possible on all eight. If CPU threads weren't
all stomping each other's epeen into the dirt. Maybe hope for later software.
Going back to one thread per real core. Allowing all threads to prefetch hurts
performance, but allowing just ONE thread to prefetch seems to bump up the
perfomance of all rest. My total now 605h/s. One prefetch per CCX no better
than one prefetch per CPU. Also not losing 12 hashes over time as before, or
the new rate of loss is enough slower, I havn't seen it take effect yet.
Seems I didn't waste my time on this experiment after all.
Up slightly from 588 to 605, I'll take that...
"cpu_threads_conf" :
[
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 0 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 2 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 4 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 6 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 8 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 10 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 12 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 14 },
],
Whole CPU efficiently loaded 1 thread per real core, I measure just 73h/s per core.
The best aggregate number I can get is about 588h/s. Maybe 600 right after launch,
but quickly settles down to 588 after a few minutes...
Two threads both loaded to Core 0 no smt, and that way too is also 39h/s+39h/s.
Four threads on Core 0 no smt, only 19.8h/s thread, still about 79h/s per core...
Even with no competition between cores for cache and memory, I don't see how
we get there from here with a two thread approach.
16 threads, two per even numbered core (without smt?), gives me 533h/s.
About same dimished performance as 16 threads distributed in the normal
SMT way.
Ohwait, maybe onto something. Four "low power: true" threads one core=133h/s.
And about same for three threads or two. But tanks when all real cores are given
the same load. One core per ccx, doubleloaded with lowpower thread seems OK.
But is still nowhere near what one regular thread per real core would add up to....
Interesting but is it useful? Four lowpower threads, one per SMT cores 0,1,8,9
give 82h/s each for 327h/s. Better than half performance for only the work of
two cores! But doesn't stack well with additional threads on other cores.
Maybe something else useful could be done with other cores, that wouldn't cause
so many cache misses? I'm just assuming its cache misses that are the problem
that prevents scaling, don't actually know. Measuring just two unleashed cores,
we could theorize 1308h/s might be possible on all eight. If CPU threads weren't
all stomping each other's epeen into the dirt. Maybe hope for later software.
Going back to one thread per real core. Allowing all threads to prefetch hurts
performance, but allowing just ONE thread to prefetch seems to bump up the
perfomance of all rest. My total now 605h/s. One prefetch per CCX no better
than one prefetch per CPU. Also not losing 12 hashes over time as before, or
the new rate of loss is enough slower, I havn't seen it take effect yet.
Seems I didn't waste my time on this experiment after all.
Up slightly from 588 to 605, I'll take that...
"cpu_threads_conf" :
[
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 0 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 2 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 4 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 6 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 8 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 10 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 12 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 14 },
],
Last edited:
Parja
[H]F Junkie
- Joined
- Oct 4, 2002
- Messages
- 12,671
2 threads per Vega card...
GPU config file should look like...
{ "index" : 0, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 0, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false },
...for each Vega card.
Elkwood
Limp Gawd
- Joined
- Sep 17, 2017
- Messages
- 178
https://www.nicehash.com/news/227
An interesting move. Wonder how long it will be before
the trans fee come back down to earth.
An interesting move. Wonder how long it will be before
the trans fee come back down to earth.
- Joined
- Apr 11, 2007
- Messages
- 33,266
Just trying to bring some customer base back. Jacking the rates and payout threshold scared a lot of miners off.
Parja
[H]F Junkie
- Joined
- Oct 4, 2002
- Messages
- 12,671
Sweet! Free way to get BTC to GDAX works for me!
- Joined
- Oct 19, 2004
- Messages
- 11,933
So it only works if you manually transfer money out from your internal nicehash wallet to your external coinbase wallet? Not if you are mining directly to your coinbase wallet via the nicehash miner?