CryptoCurrency Mining with GPUs *spawn*

Naive as it sounds, having multiple parties (the more the better) involved stregthens the security of the currency. Having most of the hash power distributed between 2-3 entities on the other hand means it's much easier for them to agree on forgeries or to change all sorts of things n the currency's implementation (such as block sizes)

That's exactly it. The more parties involved in securing the blockchain the harder it is for an individual or even a group to compromise it. Making mining possible for and appealing to the largest possible group is very beneficial to a cryptocurrency that uses PoW for block validation.
 
Why would it be considered a disaster unless you are deeply invested in GPU mining farms though? It is the natural progression.
Edit: Related - why shouldn't AMD design a chip explicitly for mining, and either keep it in house for profitable mining, or sell it openly?

From my limited understanding, the algorithms were designed with increasing difficulty in mind, thus preventing too steep inflation. Additionally, when literally overnight THashes could be achieved, where before it was MH or even GH, the whole point of distributed security becomes moot. AFAIK, when any one party controls more than 50% of the mining power, they could validate every transaction, legit or not.

Now imagine a non-ASIC proof cryptocurrency where a party secretly develops said ASICS and deploys them across their mining farm(s). Hence disaster. :)
 
Where is the 130W comming from? HWINFO has 102W/ 43W and 105W/ 44W for GPU/ Memory respectively, for each card. Would those figures even include board power?
 
System crunching pulling 406W (measured at the PSU) = 148W system idling + 129W GPU 0 + 129W GPU 1

How hard is it to look at the pictures mentioned in the reddit post and where did that 385W number even came from?




nke4xoP.png
 
Also, I do hope that CPU is overclocked (or is an ancient Vishera system like mine :) )? 148W system idle seems like such a huge number
 
System crunching pulling 406W (measured at the PSU) = 148W system idling + 129W GPU 0 + 129W GPU 1

How hard is it to look at the pictures mentioned in the reddit post and where did that 385W number even came from?




nke4xoP.png
Surely some of that 148W idle is the GPU's? Also HWINFO clearly states over 145W for each GPU (GPU+VRAM), so it can't be 130W while hashing.

Also that text file in the middle of the screen just looks like the owners own man maths figures, not the output from any program.
 
Gpus idle very very low nowadays. Both my amd and nvidia rigs with 8 gpus each only draw 50 W after login.

Vega could be different of course, but the few posts here say that it's not the case
 
Surely some of that 148W idle is the GPU's? Also HWINFO clearly states over 145W for each GPU (GPU+VRAM), so it can't be 130W while hashing.

Also that text file in the middle of the screen just looks like the owners own man maths figures, not the output from any program.

He's using a HX1000i PSU which sends the figures to HWinfo through USB. The only thing that he's using to calculate the cards' power consumption are total system figures, since HWinfo doesn't detect Vega's power numbers correctly:



Idle power numbers with Vega 64 are ridiculously low at least since 17.8.2. The card that's not connected to a monitor should be in deep sleep and the card that is connected to a monitor has its core clocks going down to sub-30MHz (my own goes down to 28MHz).
 
Well in that case that is truely remarkable, do you know what's responsible for the increased hashrate; new drivers or a new version of Claymore?
 
Don't know about claymore but he's using the regular 17.8.2 gaming drivers because the others can't change voltage and clocks correctly.

He's simply maintaining the HBM2 at a solid 1100MHz while downclocking the core to ~950MHz at ~1V.
 
In the post he mentions clock tuning only the things he did.

The lastest claymore is version 9.8, which just added Vega support some 21 days ago ( https://github.com/nanopool/Claymore-Dual-Miner , see readme)

He also seems to be backing away from the 130W number :
"3rd Edit. Thanks to Mumak for clearing some information up. My total power looks to be closer to 141.052w. Taking data from the current HWiNFO beta for core power of 0.449w * 256 + HBM power of 26.108w."
 
Smells fishy to me.
If these numbers are achievable (43MH/s @<180w) there will be dozens on confirmations and how to videos forthcoming.

Hwinfo is meaningless. Measure at the wall.
 
Back
Top