Post
Topic
Board Mining (Altcoins)
Re: Who ordered the AMD Radeon VII? Lets make a list of Miners / Hashrates / Coins
by
solominer247
on 11/05/2019, 20:59:39 UTC
Depends on what GPU you are using, and what coin you intend to mine. As an example, a GTX 1060 can consume up to 120 W. A Radeon VII can consume up to 300 W. For Ethereum, if you tune it properly, you can usually cut 25%+ off the power usage, but you don't want to use that as a basis, you always want to leave an overhead so you can mine other coins as needed. Also, as a rule of thumb you don't want to use your PSU at more than 80% since it's not efficient that way. So a reasonable expectation is that a 1600W power supply will probably provide about 1200W power efficiently. That will be able to power 4x Radeon VII's (1200W/300W), or 10x GTX 1060's (1200W/120W).

If you have miners inside, noise can be a problem. If it's outside, it's usually not that noticeable.

Hey mate, thanks for your reply.

I've looking at : Sapphire AMD Radeon VII. I've had Sapphire products in the past and work like a charm, hardly encounter issues, plus its a respective brand in my opinion. I would have gone for Asus or Gigabyte but their price difference is around £60 when it would do the same.

When it comes to coin, I will be leaving it with Nicehash as I've used their service ever since I started mining. I know others had issues and will have negative feedback but it's been working without an issue for me. Or do you think with AMD Radeon VII, using Nicehash would be bad option?
 
Right, understand. I'm best to have only 4 x AMD Radeon VII. But as you mentioned I can tweak and reduce the wattage.

At the moment my mining rig is left indoors, in fact in my bedroom and I currently have 4x Asus GTX 1070 OC 8GB, connected to 5 fans that run at 100% . But I have plans to move the mining rig to a different location in the house. Do you think 4x AMD Radeon VII will create a lot of noise? 

Mining wattages and efficiency calculations for Ethereum can be found here:
https://docs.google.com/spreadsheets/d/1w6qQtpwzlKRPSx3hYIpTrO1d29SwAIhOV4CNGi3Jk6Y/edit?usp=sharing

Let me know if it needs corrections. Thanks.

Noted!

Higher end PSUs tend to have a reasonably flat efficiency curve up to 95+% load.  Here's a measurement of the P2 for example:

https://img.purch.com/evga-1600-p2-psu-cross-load/w/600/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9ML0ovNDc3NzAzL29yaWdpbmFsL0NMX2VmZmljaWVuY3kuanBn

And a curve for the T2:

https://tpucdn.com/reviews/EVGA/SuperNOVA_T2_1600/images/efficiency.jpg

Sure, you lose a couple percent from the peak around 33% load, but it's certainly not worth handicapping your capacity as a result.  I run my 1600 P2s close to full cap in certain situations w/ 8 vega 64s on them.

Also, as I mentioned elsewhere, VIIs can do well over 300w.  The TDP rating is stock - not really something miners (or most gamers spending $700 on a GPU) care about.  That being said, even 300w is most likely pretty inefficient for mining on almost any algorithm - 200-250w is probably the target area for decent efficiency on a VII.  I run ethash at 215w for 82MH/s. 

I would say you could easily plan 5 VIIs on a 1600w PSU - possibly even 6 if you're good about power management.  Mining hashrates, as well as power use, are not fixed quantities.  You can tune your performance of any algorithm however it suits you.  Just because everyone else runs ETH @ 90MHs for 260w, doesn't mean it's the only, or even best, option.

Finally, in addition to noise considerations (and they are loud, though not quite as bad as blowers,) you should think about heat management as well. 1600w is basically a space heater - one you'll be running 24/7, I assume.


Hey mate, thanks for your reply.

Nice work on the image it helps a lot.

Yes, I totally agree with you about "everyone else runs ETH @ 90MHs for 260w". It all about finding the sweet spot lols, makes changes, making the tweaks, monitor and then applying everything. I suppose it will take some time to find the best setting, but I was really hoping to fit 6 gpus onto my rig.