Pages:
Author

Topic: GekkoScience has a new stickminer that does 300+GH - page 29. (Read 22193 times)

newbie
Activity: 13
Merit: 0
...
I see however, the problem rather be the CGminer itself, because it deregulates the stick for wathever reasons
...
'wathever reasons' is the problem.

Often as simple as a power source issue.

Hello Kano,

Thank you for your reply. Smiley

I've got a gekko usb hub with 120W PS and a fan running beside this miner. The power consumption of these can't be more than 5V/3A=15W + 5V/2,5A=12W, Sum: 27W, which is covered with the 120W PS abundandtly.

If I run two other miners + the fan there are no problems. They are running rock solid so the PS can't be a problem.

If I put this problematic miner in usb-port-pair with the others, it kicks the others out and they restarts automatically in about every 2-3 minutes.
If I run it beside them - not in a port-pair -, the two are running stable and this one is restarting with a reduced T until it gets to a point where the T is in the (-) minus range... Than it kicks the others out as well and they stop working.

I'm starting cgminer with the following line:

--gekko-compacf-freq 420 --gekko-start-freq 350 --gekko-mine2 --gekko-tune2 85

Now I lost nerves playing around with them days long without any success and let the 2 unproblematic without the other one running.  Roll Eyes
joz
newbie
Activity: 9
Merit: 1
FWIW I'm able to run a Compac-F natively off of an Intel NUC (6CAYB - older model) USB port, albeit at ⅔ optimal speed - that being at 350 MHz, where it runs stable at 222 GH/s and pulls 2.4 A
legendary
Activity: 4466
Merit: 1798
Linux since 1997 RedHat 4
...
I see however, the problem rather be the CGminer itself, because it deregulates the stick for wathever reasons
...
'wathever reasons' is the problem.

Often as simple as a power source issue.
newbie
Activity: 13
Merit: 0
I've been playing with it as you described for more than 2 weeks now. Sad I started with the same angle the others are in (17-18 min and 25 min). Then I've found this post (see below). The picture shows the factory Vcore. The NewPac is the same in this regard as the F, I suppose, therfore I was playing in this area to hit the sweet spot without success until now. I have a feeling as the miner would need a memory reset or so and then the problem would be solved.
I can't load the image right now, but please try dialing all the way back from what you set it to (apparently around 20-25 min) back to like 5 min and try to go from there.
These miners have no memory, so there's nothing to reset. They are extremely simple devices, consisting of essentially:
  • Crystal
  • ASIC
  • FTDI chip
  • Counter chip
  • Buck converter circuit (some passive components and a buck controller IC)
So nothing that really has a 'state' except of the set screw that is taken as an input from the buck controller IC and regulates the step-down buck converter circuit accordingly.

I've tought about the heat too, but if it was an issue why are the others performing well?! ...even at 500 MHz? I'm afraid if I set a higher MHz at this stick and turns out to be the heat the real problem, I'm melting down the whole stick in a blink of an eye...
Higher MHz doesn't really make it draw more power; your vcore screw is what really mostly makes the difference. Please try dialing all the way back and retry from there.

Anyway, I won't give up. I'm still trying to find the right position, but it's getting to be really annoying to waste so much time on this.
I thought so too; had a very bitchy stick (it needed just the right start frequency and vcore), but then I remembered the first page of this thread:

It is also made for overclocking. Hot-rod overclocking.

It's supposed to be like this.. Wink

You're right... Cheesy
I try other, maybe lower start frequencies and the dialing up method as well. Go-go-go!!!  Cool


Update on 13/05/2022

I've found the sweet spot. The T P numbers lines up at the same level again, both % are sitting on 100 and the GH/s can easily reach 400 at 420MHz. The spot was at 28 min instead of 30 or 26. I was playing in this intervall ever since. From 5-25 min the stick was unstable, couldn't even reach the 100% level.
I see however, the problem rather be the CGminer itself, because it deregulates the stick for wathever reasons (can't be heat) and pushes its performance drastically back to the 180-260 GH/s range (Avg. 230). Thus the F loses stability after a while due to forced underperformance and restarts. The other two sticks are running stable and happily in the 230-350GH/s range (Avg. 275) at 420MHz.
hero member
Activity: 882
Merit: 5818
not your keys, not your coins!
I've been playing with it as you described for more than 2 weeks now. Sad I started with the same angle the others are in (17-18 min and 25 min). Then I've found this post (see below). The picture shows the factory Vcore. The NewPac is the same in this regard as the F, I suppose, therfore I was playing in this area to hit the sweet spot without success until now. I have a feeling as the miner would need a memory reset or so and then the problem would be solved.
I can't load the image right now, but please try dialing all the way back from what you set it to (apparently around 20-25 min) back to like 5 min and try to go from there.
These miners have no memory, so there's nothing to reset. They are extremely simple devices, consisting of essentially:
  • Crystal
  • ASIC
  • FTDI chip
  • Counter chip
  • Buck converter circuit (some passive components and a buck controller IC)
So nothing that really has a 'state' except of the set screw that is taken as an input from the buck controller IC and regulates the step-down buck converter circuit accordingly.

I've tought about the heat too, but if it was an issue why are the others performing well?! ...even at 500 MHz? I'm afraid if I set a higher MHz at this stick and turns out to be the heat the real problem, I'm melting down the whole stick in a blink of an eye...
Higher MHz doesn't really make it draw more power; your vcore screw is what really mostly makes the difference. Please try dialing all the way back and retry from there.

Anyway, I won't give up. I'm still trying to find the right position, but it's getting to be really annoying to waste so much time on this.
I thought so too; had a very bitchy stick (it needed just the right start frequency and vcore), but then I remembered the first page of this thread:

It is also made for overclocking. Hot-rod overclocking.

It's supposed to be like this.. Wink
newbie
Activity: 13
Merit: 0
Quote
The flat side of the "knob" is almost in this position: _.
That's too high. Look at your other sticks and for a start, set the pot on problem one to the same angle.
Looking at the pot like a clock, the default is for the flat to be *just* past being in line with the long edge of the stick - call it 15-16 minute position. What you describe I would call the 30 min mark.
I run mine with it set to 20-25 min mark and the stick is quite happy at 500MHz pulling 2.8A.
Damn, then I'm actually slightly undervolting it. Now that I think about it, I believe I slightly reduced it until it got unstable to be on the safe side regarding temps and to lower power consumption a bit. Mine runs like this at roughly 300GH/s.

The reason he still only pulls roughly 2.4A might be due to cgminer downclocking it due to instability (which probably stems from overheating).

Thanks guys!
I've been playing with it as you described for more than 2 weeks now. Sad I started with the same angle the others are in (17-18 min and 25 min). Then I've found this post (see below). The picture shows the factory Vcore. The NewPac is the same in this regard as the F, I suppose, therfore I was playing in this area to hit the sweet spot without success until now. I have a feeling as the miner would need a memory reset or so and then the problem would be solved.

Return the stick to it's factory vcore to ease power draw at high frequency. 

http://23.108.83.14/images/NewPac-0006.png

Test them one at a time to see if you can push past 55GHs.

Start at 300MHz, unless you get 68GHs (0.228 x 300) don't go higher .

Try with and without the hub - single stick.   

I find my dell's usb port does pretty well with a single stick, fan, no powered hub.

Once you get 68GHs, bump the freq up by no more than 25-50MHz at a time, test for 24hrs...

One at a time first and you should be able to find where the bottleneck is.

I've tought about the heat too, but if it was an issue why are the others performing well?! ...even at 500 MHz? I'm afraid if I set a higher MHz at this stick and turns out to be the heat the real problem, I'm melting down the whole stick in a blink of an eye...

Anyway, I won't give up. I'm still trying to find the right position, but it's getting to be really annoying to waste so much time on this.
hero member
Activity: 882
Merit: 5818
not your keys, not your coins!
Quote
The flat side of the "knob" is almost in this position: _.
That's too high. Look at your other sticks and for a start, set the pot on problem one to the same angle.
Looking at the pot like a clock, the default is for the flat to be *just* past being in line with the long edge of the stick - call it 15-16 minute position. What you describe I would call the 30 min mark.
I run mine with it set to 20-25 min mark and the stick is quite happy at 500MHz pulling 2.8A.
Damn, then I'm actually slightly undervolting it. Now that I think about it, I believe I slightly reduced it until it got unstable to be on the safe side regarding temps and to lower power consumption a bit. Mine runs like this at roughly 300GH/s.

The reason he still only pulls roughly 2.4A might be due to cgminer downclocking it due to instability (which probably stems from overheating).
legendary
Activity: 3612
Merit: 2506
Evil beware: We have waffles!
Quote
The flat side of the "knob" is almost in this position: _.
That's too high. Look at your other sticks and for a start, set the pot on problem one to the same angle.
Looking at the pot like a clock, the default is for the flat to be *just* past being in line with the long edge of the stick - call it 15-16 minute position. What you describe I would call the 30 min mark.
I run mine with it set to 20-25 min mark and the stick is quite happy at 500MHz pulling 2.8A.
hero member
Activity: 882
Merit: 5818
not your keys, not your coins!
I have a current tester and it pulls at 420MHz ~2,4A. The flat side of the "knob" is almost in this position: _. If I turn it like / or | can't reach 100%, gets sometimes super hot and stops around 76% max 88%, than "collapses", restarts at a lower frequence. I tried more than 420MHz i.a. 450, 470, 500MHz without success. I think it should work at 420 just fine, because it is definitely above the 200-300 "instability" threshold.
Okay, weird; so to be honest, mine pulls almost 3A as I said, so feel free to increase maybe to a target of 2.6A or 2.8A; I'm not entirely sure right now what position that would visually equal (depends on stick to stick as well). My stick is roughly between \ and |, if I look at the stick the way it sits in the hub (USB port to the bottom, vertically down pointing towards planet earth's centre). Starting at 350 to 400 sounds good, that's what I do too. You might have actually set it way too high, maybe try \ setting and then slightly turn clockwise until it gets stable.
newbie
Activity: 13
Merit: 0
~ snip ~
It's not clear if the stick shows this behaviour when ran on its own, as well. Does the issue only happen when run together with other sticks?
Does every stick run on its own '6A rail' (connected with a line)?

Do you have current testers for your sticks? Would be great if you can measure what each stick is pulling. If the problematic stick is below 3A, feel free to increase core voltage a bit and maybe point a fan directly on it to make sure these ~15W can be cooled off reliably.

Also play around with your start speed. I have a stick that needs a pretty high start frequency to really get going.
But first and foremost, label your sticks and figure out the settings required for each stick; they can definitely vary. Then you can set them per-stick as needed via cgminer (e.g. by running multiple instances).

I labeled the sticks and I know their numbers because I had to adjust one already, which is running fine now.

The problematic stick shows this behaviour either way: running alone in a 6A-double-slot, running in pair with another stick in a 6A-double-slot, or running all alone in the hub (no other sticks). I tried every single slot and variation (alone or with the other sticks) as well, without any improvement.

I have a current tester and it pulls at 420MHz ~2,4A. The flat side of the "knob" is almost in this position: _. If I turn it like / or | can't reach 100%, gets sometimes super hot and stops around 76% max 88%, than "collapses", restarts at a lower frequence. I tried more than 420MHz i.a. 450, 470, 500MHz without success. I think it should work at 420 just fine, because it is definitely above the 200-300 "instability" threshold.

I'm starting it at 350MHz, but tried 380 or 400 with the same outcome.

@NotFuzzyWarm: Thank you. I must have overseen it, because I've read the entire readme and looked explicitly for the "P", but found nothing. Plateau makes sense now. Smiley
I'm cooling the stick with the same 10 cm diameter fan, what I'm using for the other 2 sticks as well. The fan is placed directly above the sticks, if they run together. If the miner runs alone the fan is placed before the alu-heatsink so it blows the heat away and cools the heatsink down. The fan mustn't be a problem and its air-troughput is adjustable. The problematic stick isn't hotter than the others.
legendary
Activity: 3612
Merit: 2506
Evil beware: We have waffles!
Hello fellow miners Smiley...

Can you please tell me what "P" stands for and how could I get this value lined up with the others?

It really would be great if I could get this stick working again. Thanks so much for your suggestions in andvance.
If you read through the readme for cgminer you will find that "T" is Target speed and "P" is where the speed has Plateaued at. The plateau speed should be very close to your target speed. If it isn't either the stick is getting too hot or perhaps you just need to increase Vcore a tiny bit using the adjustment pot.
hero member
Activity: 882
Merit: 5818
not your keys, not your coins!
~ snip ~
It's not clear if the stick shows this behaviour when ran on its own, as well. Does the issue only happen when run together with other sticks?
Does every stick run on its own '6A rail' (connected with a line)?

Do you have current testers for your sticks? Would be great if you can measure what each stick is pulling. If the problematic stick is below 3A, feel free to increase core voltage a bit and maybe point a fan directly on it to make sure these ~15W can be cooled off reliably.

Also play around with your start speed. I have a stick that needs a pretty high start frequency to really get going.
But first and foremost, label your sticks and figure out the settings required for each stick; they can definitely vary. Then you can set them per-stick as needed via cgminer (e.g. by running multiple instances).
newbie
Activity: 13
Merit: 0
Hello fellow miners Smiley

I've got 3 miners, one of witch is acting weird. I have a gekko usb hub with 120W PS and active cooling of course. I run the miners with the following start line on cgminer's 4.12 version: --gekko-compacf-freq 420 --gekko-start-freq 350 --gekko-mine2 --gekko-tune2 85

Two of the miners running smoothly and stable on 420 MHz with an avg. of 275 GH/s.
The 3rd miner however was stable for a good while on this frequency but than it begun to destabilize itself.

Example line from CGminer for the "healthy" sticks:
0: GSF 1005xxxx: BM1397:01+ 420.00MHz T:420 P:420 (30:15) | 100% WU: 100% | 313.1G / 273.7Gh/s WU:3823.6/m

Example line from CGminer for the problematic stick:
2: GSF 1005xxxx: BM1397:01+ 420.00MHz T:420 P:384 (30:15) | 100% WU: 100% | 274.1G / 237.5Gh/s WU:3274.1/m

The first Mhz value and the "T" (I assume it means target value) lines up with each other, but the "P" (I have no clue what it could be, maybe power or performance?) never reaches the first two anymore and remains always below them. The "P" comes to a halt either around at 384/386 or sometimes at 404/416. (The performace depends on this, therfore the reduced hashing power.) Both percentages are sitting on 100%.
The stick must be okay, because it can reach the "healthy" sticks' hashrate within the 5s time window, otherwise it couldn't get to it at all.

I've noticed, if I put this problematic miner in pair (gekko USB-port-wise) with an other, it may destablize the one I put it in pair with. Sometimes, after several restarts (hotplug-out and plug-in back after waiting 10+ sec/or hours, or complete cgminer restart) very-very rarely the "P" value reaches the "T" but some times later it restarts itself and never reaches the "T" again.

I've tried every single angle with the slightest turns of the little knob in the right bottom of the stick, every single usb-port or variation but nothing helps. I fresh and re-installed, both the CG-miner and RPi op.sys several times, no changes there. I tried different pools, just to see whether the pool's algorythm has any effect on the performance but this wasn't the answer either.

This isn't rocket science, even so cannot solve the problem. Sad

Can you please tell me what "P" stands for and how could I get this value lined up with the others?

It really would be great if I could get this stick working again. Thanks so much for your suggestions in andvance.
hero member
Activity: 952
Merit: 938
What is the best way to confirm that the miners are hooked up to the pool correctly with my correct BTC address? Is it as simple as the console saying "Connected to solo.ckpool.org diff 442 with stratum as user ***btc address***"?

Was curious if I could easily join a shared pool and how that works. Can I simply join a shared pool and they payout my share? Would love to see confirmation of payment to address from a pool whether it's solo or shared.

Really just trying to make sure it's working properly in the super rare case a block is mined.

Yes, you can see the output of your miner, it will say where you are connected, and what's happening. That should be enough.

Also, every pool will offer you some kind of dashboard that you can connect to, and see what's happening to confirm this.

For example, with solo ck pool you just need to put your BTC address in this page and hit enter: https://solo.ckpool.org

Different pools have different rules to join them, some will require an account, others you can just point your miner there with your pay address as the username like ckpool.
hero member
Activity: 882
Merit: 5818
not your keys, not your coins!
What is the best way to confirm that the miners are hooked up to the pool correctly with my correct BTC address? Is it as simple as the console saying "Connected to solo.ckpool.org diff 442 with stratum as user ***btc address***"?

Was curious if I could easily join a shared pool and how that works. Can I simply join a shared pool and they payout my share? Would love to see confirmation of payment to address from a pool whether it's solo or shared.

Really just trying to make sure it's working properly in the super rare case a block is mined.
Yes, that should suffice! You will also see some sats in your pool's dashboard coming in after a while.. Smiley

You can only be paid out after reaching some threshold. This threshold differs between pools, but I don't think you can reach it with a single Compac F on most pools in the span of a few months.
One exception might be NiceHash; since they also offer payout via LN, you can withdraw at a much lower balance (around an order of magnitude lower than other pools, ~10x lower, if I remember right).
Additionally, if you just need 'a bit more' to push you over the withdrawal threshold, I believe you might be able to just deposit some BTC over LN or fire up your GPU for a few hours.
newbie
Activity: 13
Merit: 1
What is the best way to confirm that the miners are hooked up to the pool correctly with my correct BTC address? Is it as simple as the console saying "Connected to solo.ckpool.org diff 442 with stratum as user ***btc address***"?

Was curious if I could easily join a shared pool and how that works. Can I simply join a shared pool and they payout my share? Would love to see confirmation of payment to address from a pool whether it's solo or shared.

Really just trying to make sure it's working properly in the super rare case a block is mined.
legendary
Activity: 3578
Merit: 1090
Think for yourself
What does the difficulty setting do? Should I be adjusting this or just leaving it as is? Mine is currently set to 442.

Pools adjust the difficulty for your miner automatically to receive around 18 shares per minute.  The difficulty you set is just the starting point.  No real need to mess with it.
newbie
Activity: 13
Merit: 1

I'm assuming you're using the BitcoinMerch USB hub:

https://i.imgur.com/tFIqfJi.jpg

That USB hub is quite alright, much better than most, but it has a limit of 2.4A per port. You're hitting that limit when getting towards 400MHz, so try setting it a bit lower, say 380MHz or so. Note that it might be the combined power that is tripping the hub as well, so check your power supply to see how much it is providing. Experiment with different values per miner. You should be able to get up to around 480MHz with a single miner connected.

You can confirm this by connecting one of those cheap USB measuring devices like this one:

https://i.imgur.com/X5haK6Y.jpg

Keep going up towards 400MHz and you'll see that when it reaches 2.4A it will reset.

It's really difficult to find a USB hub with data and power with more than 2.4A. The recommended one is the GekkoScience USB hub which is custom made for USB mining. It's pretty expensive and hard to find though.

https://i.imgur.com/C70ERqM.jpg

That one will allow you to go higher than 400MHz with no issues.

If you do manage to get that one, make sure to have excellent cooling as it will get really hot.


Yep I'm using that hub from Gekko. What does the difficulty setting do? Should I be adjusting this or just leaving it as is? Mine is currently set to 442.

Also, am I able to issue a command to the sticks while actively mining to increase the frequency and/or difficulty? I've been closing out my console to do it, but I'd like to keep the session open while issuing another command if possible.

Thanks
sr. member
Activity: 486
Merit: 262
rm -rf stupidity
I used this one: https://www.thingiverse.com/thing:4947599  that you linked.
hero member
Activity: 952
Merit: 938

Great setup!.

I've been looking at those noctuas to make my miners as silent as possible. How's the noise level with so many fans?

Also, where did you get that mount for the fan attached to the heatsink?. I haven't found a proper solution for that yet.

Almost completely silent.  I have 2 Octominers running upstairs in there, but earlier when I had both off I didn't even realize the fans were running (I sit RIGHT next to them lol).

The mounts for the heatsink were from Thingiverse that I 3dprinted.

Excellent, I'll get the noctuas then.

I think I found the thingiverse project: https://www.thingiverse.com/thing:4947599

There's also this one as well: https://www.thingiverse.com/thing:2694546
Pages:
Jump to: