CryptoCurrency Mining with GPUs *spawn*

thanks for the advice. I have to figure out exactly where the ram is on my zotac.

My vega 56 is the card I mostly mine on 24/7 as my wife only plays civ and its typically once every few weeks with me. I think I'm going to repast it as its a launch unit blower model. Just trying to get confirmation on if i need to get new pads for the vrms. Right now its giving me 36mh but i have it way under clocked because otherwise it over heats. I'm hoping repasting it will allow me to get closer to 40mh/s . I know some people have them getting 50mh/s but they have the fan really loud on them and typically in a room set up for mining in an open bench.

I'm going easy on my 3080.. I'm getting about 82mh/s with the board drawing 271.3w. I don't want to kill the card mining. Just make a bit of a profit. Right now nanopool is saying i should make 380-400 a month mining at its current rate but since i shut off the mining to game a few hours a day its going to be less than that. I'm okay with that however.

AFAIK most Zotac 3080s are fine, they don't have low quality thermalpads and Zotac actually applied pads between back of the PCB and backplate. So your backplate acts as a heatspreader. I.E, MSI uses graphen infused plastic backplate on Trio X/Ventus lineup, and Gigabyte didn't even bother to thermalpads and constant throttling at 60MH/s for entire 3080 lineup. I think you can decrease your power consumption to 220W levels by using my afterburner settings. GPU is running at 1200MHz@725mV. This would help not only increase your profit by but also cooler and less stress on your card.
I think short burst of cool-hot sessions during gaming is more stressful on the card as it expands and shrinks often, than steady temperature on an undervolted card during mining.
 
Last edited:
This is interesting graph of ethereum network hash rate:
View attachment 5272

https://ycharts.com/indicators/ethereum_network_hash_rate#:~:text=Ethereum Network Hash Rate is,122.1% from one year ago.

Last time there was crypto boom and bust at least one dude made a ton of money by figuring in out advance the bubble had bursted. If I remember right he short sold stock. I could remember wrong. This 2 part podcast has the details. Basically he observed hash rates and did some leg work to understand situation better than others.

https://ark-invest.com/podcast/nvidia-and-the-crypto-implosion-of-2018-with-akrams-razor-part-1/



I looked for patterns in the relation between hashrate and price, to try and see what he saw in Fev-2018. In February of 2018, you could see the bubble burst pattern. Its when the Hashrate crosses with the price in its way up and on its way down.

NoFsZX4.png



The price crossed the hashrate again on its way upwards a few days ago. Its insanity but it if the pattern repeats, suggests a rise up to ~2500$ followed by a correction down to ~1800$ and finally a run up to ~5000$ in May.
 
AFAIK most Zotac 3080s are fine, they don't have low quality thermalpads and Zotac actually applied pads between back of the PCB and backplate. So your backplate acts as a heatspreader. I.E, MSI uses graphen infused plastic backplate on Trio X/Ventus lineup, and Gigabyte didn't even bother to thermalpads and constant throttling at 60MH/s for entire 3080 lineup. I think you can decrease your power consumption to 220W levels by using my afterburner settings. GPU is running at 1200MHz@725mV. This would help not only increase your profit by but also cooler and less stress on your card.
I think short burst of cool-hot sessions during gaming is more stressful on the card as it expands and shrinks often, than steady temperature on an undervolted card during mining.

I found a tear down of the zotac I have
https://www.einfoldtech.com/2021/01/zotac-gaming-geforce-rtx-3080-amp-holo-review/
https://www.einfoldtech.com/wp-content/uploads/2021/01/DSF1870-2048x1210.jpg

It does look like there is some black thermal padding.

I wonder if I would still get a benefit by spending $10 or $20 on some heatsinks still.
 
I found a tear down of the zotac I have
https://www.einfoldtech.com/2021/01/zotac-gaming-geforce-rtx-3080-amp-holo-review/
https://www.einfoldtech.com/wp-content/uploads/2021/01/DSF1870-2048x1210.jpg

It does look like there is some black thermal padding.

I wonder if I would still get a benefit by spending $10 or $20 on some heatsinks still.
If your backplate is scorching hot where the vrams line ups during mining it means it's working as intended and putting heatsinks on top of that directing some airflow to them might be helpful. What's your vram Tjunction during mining? Nvidia sets 110C Tj as throttle point.
 
If your backplate is scorching hot where the vrams line ups during mining it means it's working as intended and putting heatsinks on top of that directing some airflow to them might be helpful. What's your vram Tjunction during mining? Nvidia sets 110C Tj as throttle point.

what do you use to see the vram tjunction ? I haven't check the gpu. I will check it out tonight when I get home
 
what do you use to see the vram tjunction ? I haven't check the gpu. I will check it out tonight when I get home

Looks like my gpu temp on that is 56c across the board (current, min , max , av) and the memory junction temp is 98,96,98,97.9. I guess if the limit is 110c I can push the over clock on the ram a bit more. Its at 133mhz right now (the gpu is -400) . I have to get behind my desk and check out the back of the card later on got to run out now.
 
If you're concerned about memory temps scratching on the three-digit-limit, you should also keep other G6 cards under surveillance, some of them - reference designs as well - can easily reach 100°C even in open benchtable setups.
 
If you're concerned about memory temps scratching on the three-digit-limit, you should also keep other G6 cards under surveillance, some of them - reference designs as well - can easily reach 100°C even in open benchtable setups.

Apparently those temps are chip internal junction temps for gddr6x. Or that's at least my impression. This is similar to how amd chips junction temperature can be high but it's not really dangerous.

https://www.tomshardware.com/news/hwinfo64-adds-gddr6x-temp-monitoring-rtx30series
 
Apparently those temps are chip internal junction temps for gddr6x. Or that's at least my impression. This is similar to how amd chips junction temperature can be high but it's not really dangerous.

https://www.tomshardware.com/news/hwinfo64-adds-gddr6x-temp-monitoring-rtx30series

Yes, I've been aware of the addition to HWInfo. It says Junction Temperature and for G6X it's at 110 °C, where Nvidia throws in the brakes. All I'm saying is, that also on cards with regular G6-Chips, you can reach 100 °C easily (didn't try more, because I didn't want to risk my card). Sometimes, the reason apparently is, that with ETH mining, you undervolt the core and often also lower the clocks, so core temp is nicely between 55 and 70 °C and fans spin slow. That's where the G6(X) rise, because they're mostly disregarded by the GPU's fan controllers, so the low air flow does not cool them down enough, they keep heating up and eventually overheat, without the fan spinning up at all.

You only need to be aware of this so you can put your fan on to some extra rpm. At 1700 U/min instead of the regular 1200 U/min on my 5700 XT AE, memory temp did not break 90 °C anymore. With the lower (automatic) fan speed, it went up to 100 °C in less than 15 minutes on an open benchtable.
 
Yes, I've been aware of the addition to HWInfo. It says Junction Temperature and for G6X it's at 110 °C, where Nvidia throws in the brakes. All I'm saying is, that also on cards with regular G6-Chips, you can reach 100 °C easily (didn't try more, because I didn't want to risk my card). Sometimes, the reason apparently is, that with ETH mining, you undervolt the core and often also lower the clocks, so core temp is nicely between 55 and 70 °C and fans spin slow. That's where the G6(X) rise, because they're mostly disregarded by the GPU's fan controllers, so the low air flow does not cool them down enough, they keep heating up and eventually overheat, without the fan spinning up at all.

You only need to be aware of this so you can put your fan on to some extra rpm. At 1700 U/min instead of the regular 1200 U/min on my 5700 XT AE, memory temp did not break 90 °C anymore. With the lower (automatic) fan speed, it went up to 100 °C in less than 15 minutes on an open benchtable.

As I understand it the interesting temperature one should worry about is something like 15-18C lower than junction temp. At least based on that gddr6x is still within manufacturer specified limits when mining assuming junction temp is 110C. This is based on some amd info on junction temperatures. Could be gddr6x is somewhat different though.

I wouldn't mine to begin with if I was interested in trying to make cards lifetime as long as possible. Or at least I would do some downclocks and not run the card as hard as it can be run. or mine until the next crypto bust happens and sell the card.
 
I just ordered Radeon Pro W5700 from domestic Amazon, funnily enough cheaper than a RX 580 8GB, estimated delivery time March 20 :runaway:

I don't have physical space for the new card on the motherboard but I think there's enough space If I hang it at top front of the case and hooking up with a USB riser. It has a blower type cooler which is great, especially since it will dump all the heat to outside directly without effecting much of the case temperature. Regardless, I will try swapping whether it's better to hang RTX2060 or W5700 for lower temps. Expecting 60MH/s@120W.

My question is do I have to flash an unlocked bios in order to undervolt the card? Is there anybody here with mixed AMD Nvidia rig on Win10? It's my first AMD graphics card since 9800XT times, anything should I be careful?

img_20210214_223430dfn6jvy.jpg
 
Internet cafes are reinventing themselves into cryptocurrency mining farms (guru3d.com)
February 16, 2021
They are placing together all the graphics cards from the local computers and mounting them on various platforms to have them mine all day. One of the most famous cybercafes in Vietnam announced on Facebook, “By transforming the business in this pandemic, the benefits are higher, making it clear that mining is being more than profitable for this class of companies.

The owner of this particular cybercafé also owns a computer store, which explains how they had access to so many GeForce RTX 3080 graphics cards to mount them on more powerful farms that complement the rest of the systems to pay off the investment faster.
 
My Mining Rig mining ETH on seven Vega 56's - 334.72 MH/s

I had seven Vega 56's laying around from the end of 2018 and with ETH price exploding I decided to start mining again. When I last mined I was using TeamRedMiner on Windows 10 and I had a very hard time getting those Vega's stable, not crashing or even running etc.

But with MMPOS (Linux version) I have had major success. Highly recommend this mining OS.
https://app.mmpos.eu

MMPOS gives 8 free credits per day so my 7 Vega's will cost 7 credits per day so I can use MMPOS for free.

I have two rigs. One with 4x Vega 56's the other with 3x Vega 56's.

334.7 MH/s total for both rigs at 1.2 KW power at the wall.

Vega 56 power ranges from 96 to 116 watts. The one at 116 watts I had to up the core voltage to 900 mv on the core and 900 MHz memory speed. The other six all are running at Core 815mv, Memory 910 MHz. The Core frequency is set to 991 MHz.

To help others here is my setup options in MMPOS.

In MMPOS I selected the latest TeamRedMiner v0.8.1 for mining ETH

In the Rig Miner Configuration - default I have these settings:

Fan Management: agent (temperature based)
Mininum fan speed: 45%
Target temperature: 65c
Target Memory temperature: 75c

In the Rig Worker Configuration- default I have these settings:

Core: 991 MHz
Mem: 910 MHz
Volt: 815 mv
Tune: med (ethash)
Controls Fan by: agent

with these settings I am getting 47.9 MH/s for each of the six Vega 56's and 47.2 MH/s for the one Vega 56 that has memory at 900 MHz.

I am hashing to Nanopool and the Calculator shows: Very approximated earnings of $42.748 per day, $299.242 per week and $1,282.468 per month.

-----------------

Edit: 2/22/2021 and 2/24/2021

I did some more individual tweaking:


on four of the Vega 56's (great ones) I have these settings:


Core: 995 MHz
Mem: 950 MHz
Core Volt: 815 mv
Mem Volt: 875 mv
Tune: med (ethash)
Controls Fan by: agent

The results are: 49.33 MH/s each at 100, 113, 115 and 118 watts.


on two of the Vega 56's (good ones) I have these settings:


Core: 995 MHz
Mem: 930 MHz
Core Volt: 815 mv
Mem Volt: 875 mv
Tune: med (ethash)
Controls Fan by: agent

The result is: 48.63 MH/s at 109 watts.

Core: 995 MHz
Mem: 920 MHz
Core Volt: 850 mv
Mem Volt: 900 mv
Tune: med (ethash)
Controls Fan by: agent

The result is: 47.92 MH/s at 114 watts.


and the final stubborn one I have these settings:


Core: 995 MHz
Mem: 900 MHz
Core Volt: 900 mv
Mem Volt: 900 mv
Tune: med (ethash)
Controls Fan by: agent

The results are: 47.21 MH/s at 112 watts.

341.1 MH/s at Total Rig power of 1222 watts at the wall.
 
Last edited:
Back
Top