user356546
Newbie
Offline
Activity: 38
Merit: 0
|
|
December 18, 2015, 11:32:57 PM |
|
What is considered to be an "acceptable" HW error rate?
|
|
|
|
VirosaGITS
Legendary
Offline
Activity: 1302
Merit: 1068
|
|
December 19, 2015, 02:10:27 AM |
|
What is considered to be an "acceptable" HW error rate?
Whatever floats your boat. I get one in 10 000, because i more or less did not wanted to raise the voltage by even a tiny bit for that. This stick can do 0HW~ easily, if you see frequent HW, then raising the voltage would fix it, but its not necessarily worthwhile to raise voltage over negligible HW rate, depending on how you look at it.
|
|
|
|
user356546
Newbie
Offline
Activity: 38
Merit: 0
|
|
December 19, 2015, 04:21:25 AM |
|
What is considered to be an "acceptable" HW error rate?
Whatever floats your boat. I get one in 10 000, because i more or less did not wanted to raise the voltage by even a tiny bit for that. This stick can do 0HW~ easily, if you see frequent HW, then raising the voltage would fix it, but its not necessarily worthwhile to raise voltage over negligible HW rate, depending on how you look at it. Thanks for the response; I am getting about 1.8% HW Error Rate at the moment with everything running at .71-.72V. I am going to let them run through the night (3 brand new sticks tonight) and see how it looks tomorrow morning. If it still looks like this I will up the voltage to see if I can knock down that rate to closer to .1% like I had with a single stick. Thanks again.
|
|
|
|
zOU
|
|
December 19, 2015, 07:24:14 AM |
|
ok i found out issue
Minera = piece of cra...
instead
1: Disable miner in interface
2: SSH to unit
3: compile your own cgminer latest... they use old stuff even if you run updates in menu
4: 0 errors
I run my 6 compacs on minera using a custom miner. I have absolutely no problem.
|
|
|
|
hedgy73
Legendary
Offline
Activity: 1414
Merit: 1077
|
|
December 19, 2015, 08:06:43 AM |
|
What is considered to be an "acceptable" HW error rate?
Whatever floats your boat. I get one in 10 000, because i more or less did not wanted to raise the voltage by even a tiny bit for that. This stick can do 0HW~ easily, if you see frequent HW, then raising the voltage would fix it, but its not necessarily worthwhile to raise voltage over negligible HW rate, depending on how you look at it. Thanks for the response; I am getting about 1.8% HW Error Rate at the moment with everything running at .71-.72V. I am going to let them run through the night (3 brand new sticks tonight) and see how it looks tomorrow morning. If it still looks like this I will up the voltage to see if I can knock down that rate to closer to .1% like I had with a single stick. Thanks again. That's very high, even 0.1% is too high. Just looking at my cgminer screen I've had 1 HW error since last re-start in 20 hours with 3 sticks @ 300MHz (16GH / stick). That's with 652k accepted shares and 60 rejected.
|
|
|
|
VirosaGITS
Legendary
Offline
Activity: 1302
Merit: 1068
|
|
December 19, 2015, 01:32:33 PM |
|
What is considered to be an "acceptable" HW error rate?
Whatever floats your boat. I get one in 10 000, because i more or less did not wanted to raise the voltage by even a tiny bit for that. This stick can do 0HW~ easily, if you see frequent HW, then raising the voltage would fix it, but its not necessarily worthwhile to raise voltage over negligible HW rate, depending on how you look at it. Thanks for the response; I am getting about 1.8% HW Error Rate at the moment with everything running at .71-.72V. I am going to let them run through the night (3 brand new sticks tonight) and see how it looks tomorrow morning. If it still looks like this I will up the voltage to see if I can knock down that rate to closer to .1% like I had with a single stick. Thanks again. That's very high, even 0.1% is too high. Just looking at my cgminer screen I've had 1 HW error since last re-start in 20 hours with 3 sticks @ 300MHz (16GH / stick). That's with 652k accepted shares and 60 rejected. Yes, i meant i have about 0.0472 % which i consider "a bit high" but near negligible as its not really affecting my hashrate, but it could still be under 0.0001%. Near 2% is very high and you're losing a "significant portion of your hashrate" and it would not take much voltage raise to bring it down back to 0%~
|
|
|
|
user356546
Newbie
Offline
Activity: 38
Merit: 0
|
|
December 19, 2015, 03:25:22 PM |
|
That's very high, even 0.1% is too high. Just looking at my cgminer screen I've had 1 HW error since last re-start in 20 hours with 3 sticks @ 300MHz (16GH / stick). That's with 652k accepted shares and 60 rejected.
Yes, i meant i have about 0.0472 % which i consider "a bit high" but near negligible as its not really affecting my hashrate, but it could still be under 0.0001%. Near 2% is very high and you're losing a "significant portion of your hashrate" and it would not take much voltage raise to bring it down back to 0%~ I bumped all my sticks up to .75V and now I am running at .04%; thanks for the advice. It had a negligible effect on the power (increase by 2 watts at the wall) so I may end up trying .76V instead.
|
|
|
|
fire27
Newbie
Offline
Activity: 12
Merit: 0
|
|
December 20, 2015, 02:16:25 AM |
|
so i tore apart some old pc's took the heatsink/fan from the CPU and replaced the stock ones. Running 1 stick at 325Mhz and .75v. going to drop the voltage down 1-2v. if i had a couple of smaller heatsinks i would run them at 400Mhz. The fans are running off an old AT power supply. i left the switch attached so i can turn off the fans with a button push.
|
|
|
|
VirosaGITS
Legendary
Offline
Activity: 1302
Merit: 1068
|
|
December 20, 2015, 02:34:09 AM |
|
so i tore apart some old pc's took the heatsink/fan from the CPU and replaced the stock ones. Running 1 stick at 325Mhz and .75v. going to drop the voltage down 1-2v. if i had a couple of smaller heatsinks i would run them at 400Mhz. The fans are running off an old AT power supply. i left the switch attached so i can turn off the fans with a button push.
Interesting... can i/we get a screenshot of that? I'm kind of picturing a fat ass cpu heatsink ducttaped to the stick, with the stick upside down and a USB cord extender. Also if you put such a big heatsink on it, i don't think the chip could generate enough heat to warm up the heatsink much, so passive would work great, right?
|
|
|
|
|
VirosaGITS
Legendary
Offline
Activity: 1302
Merit: 1068
|
|
December 20, 2015, 03:14:07 AM |
|
Oh, this is much smaller than i had imagined. Well i suppose the heattsink probably dissipate more heat more easily. If you can run that fan super quiet, i think you would still be able to handle 400mhz on it. A pretty nice performance for the stick at that point.
|
|
|
|
earlmo
Jr. Member
Offline
Activity: 39
Merit: 1
|
|
December 20, 2015, 04:21:14 AM |
|
the other one that i am going to put on was on a pentium 75 or 133. slightly bulkier than stock. The heatsink/fan i took off an AMD is too bulky to be efficient. It wouldn't even touch the chip and its like 3x the size of these other 2
Yeah, so I did this today as well: The fan is 12V, but I've got it wired to 5V and cannot hear it at all. The sticks are currently at 337.5 Mhz, vcore of 0.74V and can almost feel warmth in the heat sink. Won't go much higher, as my input is down to 4.90V already. It fits in my PC where the cat won't try sleeping on it--nice to have options. .earl
|
|
|
|
sidehack
Legendary
Offline
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
|
|
December 20, 2015, 04:29:51 AM |
|
When I test sticks, the input routinely drops to something like 4.3V and they keep going. I think the test hub I have going right now with about 20 sticks at 200MHz (an Eyeboot 49er courtesy of Klintay, I should post pictures one of these days) is measuring about 4.6 volts. As long as you're not smoking out your 5V source by drawing too much current (for a given power use, the current will increase as voltage decreases), the sticks don't really mind a voltage sag too much. The only time I've had a stick give me problems from low input voltage it was modified to about 900mV and trying to break past 485MHz on a crappy hub.
|
|
|
|
Biodom
Legendary
Offline
Activity: 3892
Merit: 4332
|
|
December 20, 2015, 04:59:00 AM Last edit: December 20, 2015, 05:47:59 PM by Biodom |
|
First batch five sticks chugging along on 7 slot usb 3.0 Superbpag connected through usb 2.0 to pi zero. Second batch of four sticks doesn't work so far on 7+3 Anker usb 3.0 either connected through non powered usb 2.0 or directly to rasp pi 2. Did anybody made this Anker work?
EDIT: 4 new sticks work fine in Superbpag on pi zero and do nothing on Anker (USB 3.0) on pi 2 next step, I will connect Superpag to pi 2, but this should theoretically work as this definitely worked for others. a bummer- i don't want to buy yet another hub (a third one).
EDIT2: well, as i expected, four new sticks on Superbpag (usb 3.0) and pi 2 work just fine. The problem is that 10 slot Anker (usb 3.0). Oh, well, will dedicate it to phone charging.
|
|
|
|
hurricandave
Legendary
Offline
Activity: 966
Merit: 1003
|
|
December 20, 2015, 05:48:20 AM |
|
Is novak trying to push something to CK on Github currently or is someone else try to help by posting some changes there?
|
|
|
|
user356546
Newbie
Offline
Activity: 38
Merit: 0
|
|
December 20, 2015, 12:46:42 PM |
|
the other one that i am going to put on was on a pentium 75 or 133. slightly bulkier than stock. The heatsink/fan i took off an AMD is too bulky to be efficient. It wouldn't even touch the chip and its like 3x the size of these other 2
Yeah, so I did this today as well: https://farm1.staticflickr.com/619/23491442279_bc850cf275_o_d.jpghttps://farm1.staticflickr.com/717/23776864461_e98a7ba44b_o_d.jpgThe fan is 12V, but I've got it wired to 5V and cannot hear it at all. The sticks are currently at 337.5 Mhz, vcore of 0.74V and can almost feel warmth in the heat sink. Won't go much higher, as my input is down to 4.90V already. It fits in my PC where the cat won't try sleeping on it--nice to have options. .earl Are the Orange and Yellow/Orange and Black wires in the first picture soldered directly to your sticks? Are those monitoring your vCore? Another question, why kind of thermal interface material are you using? Thanks for posting the pics, that is a great idea to put these inside the PC case.
|
|
|
|
sidehack
Legendary
Offline
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
|
|
December 20, 2015, 02:54:01 PM |
|
Is novak trying to push something to CK on Github currently or is someone else try to help by posting some changes there?
I kinda doubt he's pushing anything to cgminer git, but he was going to be doing a bit of work on cgminer-gekko this weekend. Adding bitshopper stick support, if nothing else. Until some of the Icarus overlap stuff is taken care of, there's no way driver-gekko would be offered to the main build.
|
|
|
|
earlmo
Jr. Member
Offline
Activity: 39
Merit: 1
|
|
December 20, 2015, 05:07:34 PM |
|
Are the Orange and Yellow/Orange and Black wires in the first picture soldered directly to your sticks? Are those monitoring your vCore?
Another question, why kind of thermal interface material are you using? Thanks for posting the pics, that is a great idea to put these inside the PC case.
Yes, those wires are for vcore readings. They don't actually connect to the motherboard though, just my voltmeter. I still had some thermal paste (cooler master and arctic silver) left over from when I built my pc. Those bitmain chips are tiny compared to a cpu, you don't need much. I was also thinking of getting one of those fan controllers/displays that fit in your drive bay to monitor temp and fan speed, or one with a custom display that would also display the hash rate and best share, but those will have to wait until one of our 'lottery stickets' comes through. .earl
|
|
|
|
hedgy73
Legendary
Offline
Activity: 1414
Merit: 1077
|
|
December 20, 2015, 06:13:18 PM |
|
First batch five sticks chugging along on 7 slot usb 3.0 Superbpag connected through usb 2.0 to pi zero. Second batch of four sticks doesn't work so far on 7+3 Anker usb 3.0 either connected through non powered usb 2.0 or directly to rasp pi 2. Did anybody made this Anker work?
EDIT: 4 new sticks work fine in Superbpag on pi zero and do nothing on Anker (USB 3.0) on pi 2 next step, I will connect Superpag to pi 2, but this should theoretically work as this definitely worked for others. a bummer- i don't want to buy yet another hub (a third one).
EDIT2: well, as i expected, four new sticks on Superbpag (usb 3.0) and pi 2 work just fine. The problem is that 10 slot Anker (usb 3.0). Oh, well, will dedicate it to phone charging.
I have the 10 slot Ankers running sticks from a pc but not a pi. Perhaps this is one of the many USB 3.0 hubs the pi's dont like .
|
|
|
|
fire27
Newbie
Offline
Activity: 12
Merit: 0
|
|
December 20, 2015, 09:03:34 PM |
|
Are the Orange and Yellow/Orange and Black wires in the first picture soldered directly to your sticks? Are those monitoring your vCore?
Another question, why kind of thermal interface material are you using? Thanks for posting the pics, that is a great idea to put these inside the PC case.
I just used some leftover TG-7. overkill but i had it on hand
|
|
|
|
|