Profile cover photo
Profile photo
Marc Bevand
Marc's posts

Post has attachment
With all the Ryzen leaks these last few months it was becoming increasingly obvious that AMD was going to kick Intel's butt. But now it's official, wow!: AMD Ryzen 7 1800X ($500) with turbo disabled outperforms the 8-core Intel Broadwell-E i7 6900K ($1100) in 6 out of 8 benchmarks:

Post has attachment
"If you think you know what the state of the payments system will be 10 years out you’re in a state of delusion."

Quote from Charlie Munger, one of American Express’s largest investors, and also long-time business partner of Warren Buffett. Source:

Bush's 9/11 speech at the Islamic Center of Washington DC is perhaps even more relevant in present times, than it was 16 years ago.

Post has attachment
AMD Zen core is 10% smaller than Intel's current gen, and has 2x larger L2

Post has attachment

Post has attachment

Post has attachment
+John Lilic I replied to your post at but for some reason my comment is not showing up. So here it is:

You quote VICE Motherboard and the BECI but both publish grossly inaccurate metrics. VICE completely failed to account for a 33x(!) improvement in Bitcoin mining ASIC efficiency and BECI overestimates power consumption by ~6x.

Read my critic of the VICE Motherboard article in section 5 of my post at

Read my critic of BECI at I am in touch with the guy working on BECI and he will fix his numbers, see our Twitter discussion:

Finally, you said the cost of sending a transaction is "the amount of electricity to power ~47 Hong Kong homes for 1 day" but this is very wrong. You are off by a factor 35x(!). The "average household electricity consumption in Hong Kong is about 400 kWh per month" or 13 kWh per day ( The energy cost per Bitcoin transaction was $1.80 as of Jan 2016 ( which is equivalent to 18 kWh assuming $0.10/kWh. So one Bitcoin transaction uses the daily energy of ~1.4 Hong Kong houses, not 47. (And it is rapidly falling over time.)

Post has attachment is riddled with errors.

1. The current break-even Watts per GH/s is estimated at 0.589 W per GH/s. Taking this metric to compute Bitcoin's energy consumption is completely unrealistic as the efficiency of ASICs currently online is around 1/4th this value. As I documented in as of January 2016 the average energy of Bitcoin mining ASICs was around 0.15 J/GH (or W per GH/s). Nowadays it is even lower because miners continuously upgrade to more recent hardware that has better W per GH/s efficiency.

2. The page assumes miners pay average electricity costs of 0.06 USD per kWH. I was personally running a GPU farm in Washington state in Douglas County in 2011 and was paying 0.02 USD per kWH. Multiple utilities offer very low rates in various counties of Washington State: Chelan County, Grant County, etc. This is why other professional miners also set up data centers in this state, like MegaBigPower. The current rate offered by the Douglas County PUD is 0.0233 USD per kWH ( I would be surprised if the largest professional miners who make up most of the mining power were paying more than 0.04 (which is already TWICE more expensive than the Douglas County rates.)

3. The page states Bitcoin's estimated annual energy consumption is 11.14 TWh and that this represents 0.05% of the world's energy consumption. This implies the world's energy consumption is 22300 TWh, but this is wrong. According to IEA statistics the world's consumption for year 2014 was 9425 Mtoe, or 109613 TWh. The page seems to confuse energy consumption with electricity consumption. The numbers would line up assuming the latter. Probably all mentions of "energy" should be replaced with "electricity".

Post has attachment

Blockchain technology is listed in China's National 13th Five-Year Plan - Informatization Strategy by State Council: 
Wait while more posts are being loaded