Topic: EWBF's CUDA Zcash miner - page 114. (Read 2164327 times)

The versions of the risers normally change depending on the type of power connection that they accept.

Guys Hi. Those of you who has more then one1080ti cards in the rig, which risers are you using? Does riser version affect on the performance? I see now 006, 006c,007s, and even "new" 009 and can't understand what to take?

Hm, I lowered memory frequency and errors stopped. Strange coincidence that it was working 24 * 7 hours before and only started happening after driver update.

No more often than one time per day. I would also want to know, why 

Why not just tweak it and observe the hashrate changes?)
On my 1080 I noticed that disabling P2 CUDA downclock in driver brings back memory frequency from 9500 to stock 10500 (factory overclocked bios) and it makes a very noticeable impact on hashrate, overclocking further to 11-11.1 GHz gives almost no gain. GPU is easier, just overclock until it hits desirable temps/TDP, I limited it to 80% PL and got 580 sols @ 180 W @ 68°C. Unlimited stock goes to 90% PL and overclocked to 100%, which is about 610 sols, but it becomes hot (72+) and noisy

Ran into an issue, wondering if anyone's seen something similar. Have 5 1080TI's on a win 10 setup, mobo is a gigabyte z97x-gaming 5, it has 6 pcie slots. It has an x16 slot which runs at x16, and an x16 slot running at x8, which shares bandwidth with the full x16 slot. It also has an x16 slot running at x4 speeds, sharing bandwidth with 3 other x1 slots. Toggling the x4 slot to x1 speeds in the BIOS enables it to be used with the x1 slots concurrently.

I would like to move from 5 to 6 cards, and that is where I have a problem. I'm having an issue with the pcie x16/8 slots. Specifically, cannot get the miner to utilize 2 cards if they are in the x16 and x8 slot respectively. The x16 slot by itself works fine. The x8 slot by itself works fine. But if they both have a card, only the x16 slot card is picked up. So at the moment, I have to use the x16 slot, the x4 slot, and the 3 x1 slots, all of which work fine together. But again, want to add a last card to this rig. Curiously, it's not like the card in the x8 slot is not detected if there is something in the x16 slot. It shows up as a device in device manager, and various overclocking programs can see it as well. It just wont be picked up by the miner, only 4 cards get used. It only gets used if I pull the card out of the full x16 slot. I even tried putting in a pcie wifi card in the x8 slot with the 5 other cards running. That not only detected and installed, but it connected and was in use, meaning for sure there is no issue with the slot running when the full x16 slot is in use.

Has anyone seen something like this? Seems to be a problem with the miner, as again, the card is detected by the OS/various programs, and the slot can even be used by other pcie devices, it's just that the miner doesn't use it to mine if a card is present there at the same time as a card in the x16 slot. Wasn't able to find an alternative mining program to just test and see if it would make full use of the cards in the expected manners, the few I tried didn't work very well. If anyone has a solution to this issue, would be great, or at least a suggestion of another mining program to use which can support 5+ nvidia gpus, just so I can test and see if it works on that.

a bit of both work well in my testing.

Does overclocking memory clock in Afterburner make much difference to Zcash mining, or is this much more about core clock speed?

I'm on 5x ASUS DUAL OC GTX 1070s and the most I can seem to get stable is +162 MHz core clock. Memory is at +560 but not sure if it matters.

I'm getting about 2350 Sol/s, so about 470 Sol/s per card.


EDIT - Quick view of the api:

thanks for the info, i'm thinkning is the cpu not strong enought for windows 10 it's a g4400 and have 8 gig ram.

how do i check what you're saying with nvidia-smi .?

I think you should try to mine blake2s on 120% TDP and -i 31...

No, seriously, how could the miner kill the card and not some hardware fault? Does the software control your over/underclock? Do you set particular voltage, speed, fans, temp targets, etc.?

It's by far easier to just set voltage in AB and stop caring about Power limit. Why would you leave the card to more around frequencies, instead of locking it with lower voltage and set speed?

With AB your card would do around 750 at 1898 MHz, +500 Mem at 0.975V max, and this is some estimate I trow around, some tweaking would probably give you even better speeds.

- Open AB regular skin
- Press Ctrl+F
- Press on a dot over 0.975
- Slide the frequency slider is it's at 1898 or 1911 MHz
- Click on the dot w/o moving it
- Press Ctrl+L
- Move the mem slider to +500
- Move the Power level to 120%
- Open the temp target next to the power slider (like drop down arrow)
- Set 80 degrees C (I've no idea how much in F, never understood this weid system)
- Apply
- Don't allow applying the clock on start for now!
- Open the AB options
- Go to Fan
- Move the top dot so it's above 80 degrees C
- move around other dots so you get more fan speed at 60 degrees
- Start miner
- Try tweaking these in a way to get lowest possible voltage with highest frequency and best mem speed. Memory doesn't need much power. Core needs power but you can lower speed/voltage until you like the speed. After you find stable settings - allow AB to apply them on startup.

That is if you have the card set for mining, if you want gaming too... it would probably work but needs testing.

 Probably driver versions - some driver versions are just more efficient.
 Part of it though might be the MUCH higher overhead of a typical Windows system vs a typical LINUX system - but that should only account for a percent or 2 difference.

 Might also be the Windows tendancy to want to put stuff into low power mode at times EVEN WHEN YOU ARE USING IT (LINUX is a lot better about NOT doing that particular idiocy).
 Check if that's happening with nvidia-smi - if you have one or more cards that are in P8 state, that's your issue.

The strix fan GPS needs more power and they are hotter than the other cards.

I have checked the other web sites and nicehash. They say THE GTX1080 Ti give 620 Sol/s with EBWF miner http://1stminingrig.com/ccminer-2-0-rc2-tpruvot-gtx-1080-ti-mining-performance/

Could you tell me the hashrate valus with EBWF miner at stock settings please ?

And does anyone use founder edition GPU for mining ? How about the temparature ? One fan is enough for the GPU ?

I'm using a Asus GeForce ROG Strix Gaming GTX 1080Ti.

I put TDP  at 85% because i get around 725 Sols with 205W~. If I put TDP at 100% I would get around 750 Sols and 250W~ and the GPU gets high temp.

Why did you limit the power to %85 ?
You can get higher Sol/s with %100 power.

which asus 1080ti do you have. i am thinking to buy 2 of them, but temporary i am at aurus extreme and evga sc2. Any better option?
thank you

hello,

i am new to mining with ewbf.

how i do set up this api  in windows?

thank you for help.

fogot to say, i put ip in the local miner, and it say api blind eror

Also, how i need to change DHCP to be fixed. i am using tp link ruthar, i have enable DHCP. do i disable it to get working?

What are you configuration on that 1080 ti and what kind of power supplyes are you using?

Thank you.

That you, that will be great.

On a side note, does the miner check the .cfg file for changes periodically once it is running?  It would be nice to be able to edit the .cfg file with changes to the mining pools without having to restart the miners.

As I've said before, --api is not working with static ips in my LAN. Using --api x.x.x.x:42000 doesn't help either, api remains disabled.

EDIT: Ok, weird... I've changed port 42000 for 3333 and now --api 192.168.0.201:3333 is working fine. And for the first time using --api parameter a Windows 10 Firewall window popped up and asked for permission.




That would be awesome!

Does this happen often? And why?)
I was mining on 1080+970 for a week without any problems, but today I updated drivers from 382.19 to 382.53 and now I'm having these red warnings that GPUs were stopped.
I rolled back to 382.19, still errors. It can stop mining after 2 or 30-60 minutes.


Driver errors with ID 13:

Driver does not crash, frequencies stay the same. It would be handy if miner could recover from this by restarting itself, or I'll have to write similar program to kill it