Hello all,
I have a weird problem with Equihash when mining with 6 GPU. It crashes my rig completely, no matter the software, intensity, or overclock. (Tried EWBF, DTSM, Excavator, ccminer-tpruvot) All other algorithms works 100% fine. The rig runs fine when mining Equihash with 5 GPU.
Any thoughts?
GPU 6 * 1070 Ti Gigabyte
Motherboard Gigabyte GA-H110-D3A (rev. 1.0)
PSU Cooler Master V1200 Platinum
OS Windows 10 + Awesome Miner
Risers 006C
Nvidia driver 390.65
Did you modify the pci-e speed to 2x or 1x? I suggest 1x.
It was set to X2. I have set it to X1 and the same still applies. 
You are out of power
Although the power rating of the 1070 Ti is set at 180W, the power consumption can some times spike as high as 200W on stock settings.
previous post with SS of the power usage
Hey Citric, I donât want to rule that possibility out, but I question it.
How come other power-hungry algorithms like Skein, Lyra2RE2 or Neoscrypt work just fine on the 6 GPU? To test the claim I have tried setting my cardsâ power TDP to just 50%, and the same effect still happens. I intend to get a power meter soon to test the actual usage.
Regards,
all I can tell you are the facts as I know them
1070ti stock is 180w
riser 50w
mobo/cpu/fan/HD/ram around 100w to 125w
GPU 5 * 180w = 900w
Riser 5 * 50 = 250w (risers take anywhere from 25w to 50w and if powered by sata should be one riser per cable)
Mobo = 100w
With 5 GPUâs (stock) you are running 1,125w to 1,250w, again with a 1200w PSU you are out of power
Want to get a rough idea of the watts from your GPUâs with Zcash, Run 5 gpuâs with EWBF and make sure you have the --pec in your config / bat file⌠run for 10 minutes and what are your gpuâs power consumption? It will show it on screen
Risers do not consume any power themselves. They simply feed power to the GPU. What power output the GPU reports to the software is essentially the GPU power + riser power (what is fed through the PCIe slot). All my risers are powered with 6 pin power cables, NOT SATA.
Again, motherboards also do not consume power on their own, only the hardware that is plugged into them⌠Such as, the Chipset (5W), CPU (10W) , RAM (5W), HDD (3W)âŚ
I ran EWBF with 5 cards and the --pec flag. The output on the console read 175 W per card on average, with my settings. It actually crashed after about 20 minutes, guess 5 cards isnât stable after allâŚ
Again, what is weird is that I can run algorithms like Skein for more than 24 hours easily, which draw OVER 190 W per card on average (as reported by HWinfo), on all 6 cards at once. If I was out of power, it wouldnât just be Equihash that would be complaining!
It is also foolish to believe a 1200 W PSU can support JUST 1200 W. In fact, this PSU was tested to comfortably sustain 1500 W without getting too hot, or shutting down.
Male Bovine fecal matter!!!
The only time the risers ONLY feed power to a gpu is when they do not have an external power plugin, Ie a 1050 / 1050ti. Risers consume power, its a fact, documented, tested, proven with pictures and volt meters, white papers, and videoâs.
as for your comment that Motherboards themselves donât consume power, Motherboards alone consume 20w to 65w (depends on the board itself, not including what is plugged into them)
But wait⌠So I just proved to you with the --pec flag that your system is out of power, you admit that it isnât stable with 5 gpus and yet you still tell me Iâm wrong?
You came here asking for help with Zcash, Iâm was attempting to do that. I donât deal with the other algorithms on a daily basis, zec I do.
There is absolutely nothing I can say about that line that will not get a ban hammer dropped on me
I wish you the best of luck
It makes no sense for me to be out of power, mate.
I tested it again with the 5 cards and 50% power limit and the cards read all <100 W and it still crashed.
Now if youâre telling me this PSU can not sustain 5 cards running at <100 W for 10 minutes.
But can sustain 6 cards running at >190 for indefinite, you must be tripping.
Could you send some screenshots that proves what you are claiming? (all 6 cards running @ 190W).
If you are 100% sure itâs equihash related than I would try ewbf or dstm directly without going through awesome miner. If still having problem then I would try a linux distro.
It makes perfect sense. Your denial on the other handâŚ
Just the fact that you think itâs âfoolish to believe a 1200 W PSU can support JUST 1200 Wâ, speaks volumes.
Hello johnwisdom, thank you for your reply. I have purchased a power meter to be 100% sure, but I am starting a fresh log from Skein mining (the most power-hungry algorithm) in HWinfo right now. I will report back after an hour. (Although, I have been running Skein for over 2 weeks, no problem)
I have also tried bypassing Awesome Miner and trying miners directly, the same unfortunately still applies. I have yet to try a Linux distro, but it isnât feasible for me to change environment like that, especially since Equihash mining is such a small part of my bussiness.
(Also a bit about how a 1200W PSU supports more than 1200W⌠This PSU has 100A on the singular 12V rail for a total of 1200W. It also has 25A on the 3.3V and 5V rail for a total of 125W. That is a total of 1325W⌠A PSU wont shut down immediately as soon as its rails get overloaded. The ripple will increase, which is bad of course, but not a disaster. As soon as it gets too hot, it will shut down.)
You can see here the HWinfo log for mining Skein for 1 hour. The average wattage per card was ~197 W, with peaks up to 208 W. Please explain to me how the PSU could handle this while crashing running 5 cards Equihash on 50% power limit (~100 W per card).
My power meter will arrive tomorrow. I will take some measurements at the wall.
First of all, stop comparing entirely different algorithms. Second, you havenât actually shown that youâre running any equihash miners at 100W per card, all youâve shown is youâre taxing your PSU over itâs limit on another algorithm (which youâve not actually shown either, just HWInfoâs statistical output that tells us nothing but the stats for your hardware).
I feel like this is the conversation thatâs taking place:
You: Please tell me why this apple tastes like an orange?
Everyone else: Because itâs not actually an apple, itâs an orange.
You: No! Itâs an apple!
Everyone else: But itâs not.
You: Well Iâm going to continue calling it an apple.
Uhhh. Read the column GPU Power on the HWinfo screenshot⌠Do I need to circle it for you? It clearly states wattage. You can find it under the section for each GPU, the minimum, average and maximum for the logged hour.
If you want to see a screenshot of 5 cards running at 100 W on Equihash then here you go⌠donât know what it adds or why I need to prove myself for everything I say.
Guess what? It crashes shortly after this screenshot. Why are we still bringing power into this? I would edit the title now into Equihash crashes my rig. Since the GPU count does not seem to matter after all.
You donât need to circle anything. I can read it quite clearly. Just because Iâm looking at the Gas gauge in my vehicle, doesnât mean I know what type of gas is being used. Itâs a pretty simple concept. Youâre showing the stats of the hardware out of context. It doesnât show anything other than itâs being used and itâs power consumption. Not what itâs being used for (A different algorithm I might add).
Power is still a factor because youâre admittedly over taxing your PSU and wondering why itâs misbehaving. Iâd expect that youâd be able to follow along in your own thread.
That being said, Iâm tapping out at this point. Itâs clear you already know the answer to your problem. Otherwise you wouldnât be here asking for help, right?
No, I do not know the answer, hence why I am coming here⌠Iâm sorry if I seem frustrated, itâs because I am. However, I do not intend to be hostile.
Could you please elaborate how power can be different like a type of gas? I do not understand the concept. If itâs being taxed 190 W itâs being taxes 190 W, right? No matter what it is actually doing. The miner has been running for weeks mining tens of different algorithms, how is Equihash different? I find that hard to grasp.
I will download a Linux distro to the rig and run a miner on Linux and see if that makes a difference. And again, tomorrow I will take power measurements from the wall with a power meter.
first off all u should realize that electronics has some life span in case u push it hard the lifespan decreases with each life decrease you will got a slightly worst result from your PSU some hard pushed PSU can come to a time when they wont even hold the proper voltage and in the power load they reboot the system or win used to disable a device in case it failing often due to power drops. OH BTW ower sucking your PSU in spikes is considered as a HARD PUSH on the PSU electronics a lot of PSU wont survive this for of long period even when it only spiking for a split of second
OH btw its advised to calculate that the max load on PSU to be 8-10% lover than the max PSU rate
The PSU has 100 A on the 12 V rail for a total of 1200 W. The only thing running on the 12 V rail are GPU, and they stay on average 180 W (when submitting my Skein screenshot earlier I just wanted to prove it wasnât a power issue, I raised the power limit a bit), for a total of 1080 W on the 12 V rail. That is exactly 10% below spec, and well in safe limits. The other hardware is ran on the 3.3 and 5 V rail which support a combined 25A for a total of 125 W. Again running CPU, RAM, HDD on these is well within spec.
I will mess with the power limit of my cards to ensure the spikes do not âtire outâ the PSU in the future.
That being said, it is clear that with even 5 GPU on 50% power limit my system still crashes. Yhsy is not a power issue⌠Anyone got an explanation for that? Thatâs what I want to focus on.
LMFU u write it by your self " with peaks up to 208 W" this is the value u need to calculate not average 180W âŚthe average is good for you when you want calculate how much you cards eat from your utility bill
