Topic: Who ordered the AMD Radeon VII? Lets make a list of Miners / Hashrates / Coins - page 9. (Read 12978 times)

Different brands, different connectors But some brands are using the same manufacture, like EVGA and SuperFlower for example.

But there are good cable mods on the market. If you don't have enough original cables, just order a few of them. But remember, don't buy the cheap ones. Someday you will regret that decision. Always buy from brands with reputation and read carefully the comments on like Amazon. If you see pictures with smoked or burned cables, skip them!

@badbart: ...until your home burns down my friend.

Hey mate, I purchased the PSU brand new only to be used for Radeon VII
You expect EVGA top range products such as G2 to have spare cables...

Can't I use corsair cables instead?

I have those adapters and one did burn.  I just buy another psu, they are cheap.

Hey mate, thanks for reply man.

"Edit: Yes, just connect 1 of these double cables (6+2 and 6) into 1 slot in your GPU, forget the 6 Pin."

The RADEON VII requires 2x 8-pin power. You're saying to connect only 1 of the 6+2 and 6 into 1 slot in the GPU, will the GPU still work, because 2x 8-pin power is required?

"DON'T split the remaining 6 Pin with an adapter into 8 Pin and connect the second slot of that same GPU!"

Ouch! I was going to do that tomorrow when the adapter arrived. Thanks man, I will give it a miss now.

I still have my Corsiar PSU, that has the exact cables that are required for RADEON VII to work, do you think its OK to use corsair cables?

I got two Radeon VII from AMD the Red 50 Year Anniversary Gold Edition. https://www.amd.com/en/products/graphics/amd-radeon-vii

When I run only 1 of them at stock settings, I get about 2900 h/s.

When I run both of them.  I only get about 3800 h/s.

I would expect about 5800 h/s.  What do you think is going on?

I am using XMRig AMD 2.14.1 with Windows 10.

It automatically compiles the config file. This is what it gives me.   See below:

"threads": [
        {
            "index": 0,
            "intensity": 960,
            "worksize": 8,
            "strided_index": 2,
            "mem_chunk": 2,
            "unroll": 8,
            "comp_mode": false,
            "affine_to_cpu": false
        },
        {
            "index": 0,
            "intensity": 960,
            "worksize": 8,
            "strided_index": 2,
            "mem_chunk": 2,
            "unroll": 8,
            "comp_mode": false,
            "affine_to_cpu": false
        },
        {
            "index": 1,
            "intensity": 960,
            "worksize": 8,
            "strided_index": 2,
            "mem_chunk": 2,
            "unroll": 8,
            "comp_mode": false,
            "affine_to_cpu": false
        },
        {
            "index": 1,
            "intensity": 960,
            "worksize": 8,
            "strided_index": 2,
            "mem_chunk": 2,
            "unroll": 8,
            "comp_mode": false,
            "affine_to_cpu": false
        }
    ],


I'm tempted to just take one out and run it in another PC.  But, there must be a way to get them to work at full mining speed together.

My junction temp is 85° now with two cards.  And before with only one it was around 95°.  So, with one, it was definitely working harder.

I also just noticed that GPU-Z reports my memory as Samsung.  Do you think that it is being reported correctly?  See Below:

http://foxautomation.com/gpuz/Capture.gif

Never touch any adapter if you don't want to burn your house down!

1 slot in your PSU for 1 slot in your GPU, if you want to make it correct. So you can fire up 4 GPU's. A 5th card if you use the last 9th PCIe and a good (not cheap!) CPU/PCIe cable.

Many ppl will say "Hey, just use 1 cable for 1 GPU. Get any cheap chinese adapter and its fine."... Never do this! These cables were made cheaply and get very hot and if you are unlucky, it can burn.

Edit: Yes, just connect 1 of these double cables (6+2 and 6) into 1 slot in your GPU, forget the 6 Pin.

DON'T split the remaining 6 Pin with an adapter into 8 Pin and connect the second slot of that same GPU!

Sorry, I'm getting confused with the cables, apologies.

I've managed to connect 2x Radeon VII with 4 x 8-pin pcie cable.

Now I have 5 of these : https://images.evga.com/products/gallery/100-CK-1300-B9_XL_4.jpg (I think its 6+2pin + 6pin connector)

Wouldn't four double PCIe cables mean you can connect four GPUs? Then use four of the remaining single ones for GPU 5 and 6?

Hey, yes the 4 double PCIe cables work fine and can power 2 GPUs.
Issue is with the 5x single PCie, do I need to purchase an adapter for those now?

Something like this : https://www.amazon.co.uk/Startech-Express-Power-Adapter-Cable/dp/B001TK3TJY or https://www.amazon.co.uk/dp/B07GJ6GB2B/ref=pe_3187911_185740111_TE_item

Nicehash keeps on crashing. I've used MSI Afterburner settings as follows :
Core 950 mV, Power limit -10%, Core Clock 1500 MHz, Mem 900 MHz, Fan Speed auto.

Miner runs for less than 5 minutes and stops few seconds later restarts...

What can should I do?

---

Hey, yes thanks for the heads up.
Both of my rigs are open air.

1st frame : Veddha V3C 6 GPU Mining Rig Aluminum Alloy Stackable Case Open Air Frame Rack (6x Asus GTX 1070 OC 8GB) with 120mm fans x5
2nd frame : https://www.ebay.co.uk/itm/6-7-GPU-crypto-mining-rig-frame-aluminum-case-Crypto-Monster/222333360848?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m1438.l2649 (Radeon VII) with 140mm fans x5

Niiice neat setup. Looks like it took you some time to plan everything out?

My experience with Radeon VII is that it is highly compute bounded. It has so much free mem bandwidth that neither timings nor memory clock matters

It's very possible you can get higher h/r by updating timings, but you certainly can't drop power use in doing so.  Besides the fact the mem package itself will use more power, core clock is already required to be reasonably high to take advantage of all the mem bandwidth.  Increasing mem performance will only require the same or higher core clock to take advantage.


Depends on your timeline i guess...  The standard depreciation window for computer hardware is generally 3-5 years.  Even at the low end of that range, using todays revenue rate (which obviously isn't great,) the VII is the better buy - at least at my power cost (USD 0.124/kwh).  If prices continue to go up - the comparison only tilts more in favor of the VII.  OTOH, if you expect the market to collapse (or just want to hedge) then I suppose the 470 would be the better buy - but maybe there's a better investment for your hard-earned money than mining in that scenario anyway.

Btw - I assume you're talking about buying 470s on the second-hand market...  Personally, based on some of the stuff I've seen in these forums/chatrooms, I wouldn't trust used mining equipment further than I could throw it (nor would I buy anything less than 8gb - meaning even higher cost.)

Maybe use the strap tool then you can get better hash per watt at 100 MHz and say 180 watt Vega 7 would be
Interesting to me.  220 watts for 85 MHz is poor compared to 470 at 29.8 MHz at 90 watts and $80 per card

That PSU as new has 5x single PCIe and 4 double PCIe cables.

Hey, guy's I managed to install my 1st AMD Radeon VII today but there is issue.

I have EVGA SuperNOVA 1600 G2. The connectors on the GPU require 2 x 8-pin PCIe power connectors.
With the PSU, only 1 x ; 2 x 8-pin PCIe power connectors were provided.

How on earth do I connect the rest of GPUs? Do I need to buy adapters now?

Someone, please share some light

I'm seeing 82+Mhs for 215w atw (including idle) w/ 1575 cclock, 935 mclock @ ~858mv.  On Vega 56/64, I think every incremental voltage adjustment (6.25mv) resulted in ~ +/- 3w, so you’re prob at ~210w+ including idle.

Iirc, I was running sub 850 while tuning, but crashing too frequently, so settled at 858.  That was a while ago tho, so could be mistaken.

Get Anyone managed to run this beast lower than 850mv gpu voltage? It has 850mv mvddi so I think it getting me from going down the voltage. Now I have 1540/940@850mv with my new Powercolor Radeon VII (Samsung mem) and 81mhs eth. Have no from the wall measurement but I think it is around 200watt (150watt gpu asic power according to hwinfo)

The point was simply to consider moving cooling solution power load off a nearly maxed PSU (and that sufficient spacing can make a huge difference.)  I offered my setup as an example, not the one-and-only design.

There are different solutions for each case. For example in my garage im using 2 pipes - one for remove hot air and one as inflow. Also, i often see rigs in boxes with 120mm fans both side.
So your solution is nice but for your conditions.

Remember to consider your fans power draw when calculating headroom on the PSU.  I think 120mm fans run around .33a on avg - so maybe about 4w each?  If you’re really pushing the limits of the PSU, your best bet is probably an open air rig w/ good spacing, and an external cooling solution if possible.  Here’s my setup for example - the 8x64 rig is on the left.  The spacing alone is generally enough to keep these reasonably cool (40-45c) even at 70-80f/21-27c ambient.  The cabinet is more for noise suppression and dust control (this is my garage/workshop too!) but the airflow helps on those summer days when temps get up significantly higher.  Many people just point a box fan at their rigs. (btw, the fans on the one rig aren’t necessary, but I didn’t bother removing them as I had the power headroom):

It all comes down to efficiency in the coin you want to mine. Especially if you have higher power costs. 1070's are pretty efficient in Ethereum though. Someone ought to do testing and comparison with X16R or CN-R. I can test 1060 and VII (those are all I got). I've already created a efficiency comparison on Ethash, but I could probably cook up a table on the other algos.