[ANN] AEON: Scalable, private, mobile-friendly cryptocurrency

[smooth]

Any literature on the PoW algorithm? Runs surprisingly low temp on my FX-8350

How many threads?

With enough threads most CPUs get hot. Unless your cooling is really good I guess.

As far as literature the best source are the block comments in slow-hash.c.

7 threads. 62C, stock heatsink/fan. Cryptonight was pushing my temps into the 70s.

That is interesting. My guess is something like the lower cache use (CN-L is 1/2 of CN) fits entirely in L2 (4x2MB) and allows the L3 to go into a lower power mode. Likely that is also faster.

Sounds like you have identified a good part for AEON mining potentially.

[Nik4691]

Cryptonight don't scale above 6 threads in this processor. Cryptonight-lite scales to all 8 threads. That is 350 h/s for Cryptonight and 850 h/s for Cryptonight-lite. 8core AMD FX at 4GHz that is.
It is very light in power consumption. For a measure, the Lyra2RE algo ups my temperature to more than 70 while Cryptonight-lite don't go above 60 degrees  Celsius.

[myagui]

Looking at this exchange about mining performance reminds me of something that I wanted to ask a while ago:
With the change into Cryptonight-Lite, we have improved performance on processors with smaller cache. However, I understand that for an equal number of cores and similar cache sizes, the factor most impacting mining performance, is the presence (or absence) of AES-NI capability.

If we were to ever aim that a mobile device could be a mining contributor to the AEON network, I think we'd do wonders if we managed to bridge the performance gap between the popular x86-64 desktop, and the usually ARM-based mobile and lightweight platforms. I would imagine that this would require switching the PoW internals that are favored by AES-NI, and replacing these cryptographic operations with ones that would be better matched in ARM performance, all the while not introducing a significant advantage for the GPU realm.

Any thoughts, inputs, crazy(er) ideas? I realize we just recently had a hard fork to implement Cryptonight-Lite, but hopefully most will agree that open minded discussions usually bring about the healthiest development environment. 

[smooth]

Looking at this exchange about mining performance reminds me of something that I wanted to ask a while ago:
With the change into Cryptonight-Lite, we have improved performance on processors with smaller cache. However, I understand that for an equal number of cores and similar cache sizes, the factor most impacting mining performance, is the presence (or absence) of AES-NI capability.

If we were to ever aim that a mobile device could be a mining contributor to the AEON network, I think we'd do wonders if we managed to bridge the performance gap between the popular x86-64 desktop, and the usually ARM-based mobile and lightweight platforms. I would imagine that this would require switching the PoW internals that are favored by AES-NI, and replacing these cryptographic operations with ones that would be better matched in ARM performance, all the while not introducing a significant advantage for the GPU realm.

Any thoughts, inputs, crazy(er) ideas? I realize we just recently had a hard fork to implement Cryptonight-Lite, but hopefully most will agree that open minded discussions usually bring about the healthiest development environment. 

New ARMs have AES extensions that are equivalent to AES-NI, but in any case it isn't that that essential for verifying a chain, only for mining.

[myagui]

New ARMs have AES extensions that are equivalent to AES-NI, but in any case it isn't that that essential for verifying a chain, only for mining.

Ah thanks smooth, then perhaps some specific optimizations for ARM will then bridge the performance somewhat? I though that ARM was still lacking in this respect, had only found a couple of specialty processor models that are specifically used in security appliances and feature AES-NI, I did not look into similar/equivalent instruction sets native to ARM.

I understand how this is just significant towards mining and not particularly about chain verification.
My suggestion was thinking in lines of having mobile or low-power ARM devices as contributing miners and not just nodes. Say, a tiny ARM PC running a low power solo-mining node (with meaningful hashrate and returns), or likewise a mobile device.

[smooth]

New ARMs have AES extensions that are equivalent to AES-NI, but in any case it isn't that that essential for verifying a chain, only for mining.

Ah thanks smooth, then perhaps some specific optimizations for ARM will then bridge the performance somewhat? I though that ARM was still lacking in this respect, had only found a couple of specialty processor models that are specifically used in security appliances and feature AES-NI, I did not look into similar/equivalent instruction sets native to ARM.

It's in most ARMv8 (64 bit) designs. However, last I checked the instruction set specs were NDA, so it isn't possible to implement in open source, at least not without someone reverse engineering it. Once the specs become available one way or another I will add it to the ARM support we already have.

EDIT: I just found that openssl has added support for the instructions in their (open source) library so I will look into adapting that for cryptonight(-lite) mining.

I understand how this is just significant towards mining and not particularly about chain verification.
My suggestion was thinking in lines of having mobile or low-power ARM devices as contributing miners and not just nodes. Say, a tiny ARM PC running a low power solo-mining node (with meaningful hashrate and returns), or likewise a mobile device.

Interestingly even without AES hardware the RPi2 is not a terrible low-hash-rate mining node. The extreme efficiency of the processor in terms of performance-per-watt (one of the best in the industry) makes that possible. There may be other hardware variants that are similar. Also, there are some low end Intel chips that do have AES-NI with very low power usage.

[myagui]

[...] EDIT: I just found that openssl has added support for the instructions in their (open source) library so I will look into adapting that for cryptonight(-lite) mining.

Awesome! I'll venture getting a low power ARM box to run as a permanent local node for AEON, if this works out favorably       

Interestingly even without AES hardware the RPi2 is not a terrible low-hash-rate mining node. The extreme efficiency of the processor in terms of performance-per-watt (one of the best in the industry) makes that possible. There may be other hardware variants that are similar. Also, there are some low end Intel chips that do have AES-NI with very low power usage.

Agree. The Intel stuff just seems to bump the budget by quite a bit, but there are definitely some really nice low power CPUs that would do very well with Cryptonight & -Lite.
My dream box for a low power node that would run multiple coin daemons as well as solo mine, would be something using the Atom C2750, or for even less power, Pentium N3700 or Celeron N3150. Pair that with an onboard mobile Nvidia Maxwell chip, and I'm all in! 

[smooth]

Agree. The Intel stuff just seems to bump the budget by quite a bit, but there are definitely some really nice low power CPUs that would do very well with Cryptonight & -Lite.
My dream box for a low power node that would run multiple coin daemons as well as solo mine, would be something using the Atom C2750, or for even less power, Pentium N3700 or Celeron N3150. Pair that with an onboard mobile Nvidia Maxwell chip, and I'm all in! 

It's a bit more than a RPi2 but the N3150s are still not too expensive. You can get MB+CPU for <$90. Add some RAM and a small USB drive and you're good to go. I'm pretty sure this would run well under $150 total.

[myagui]

Yep. That's the model I've been looking at most closely. I've been forever waiting for a low cost mobo with the Atom C2750 (way more powerful than the N3150), but it does not look like one will ever exist...

The only reason I haven't thrown some money at the N3150 or N3700 yet, is that I'm hopeful that some vendor will offer a barebones version that includes some mobile Nvidia chip onboard. That would be the perfect lightweight node for me. A low cost, compact and efficient platform, with both some CPU & some GPU compute capability.
FWIW, I think that there are plenty such devices based on AMD silicon, as well as mixes of Intel CPU w/onboard AMD GPUs. For some reason, I really dislike AMD's GPU parts, so there's that.

If anyone comes across such a device, please post about it. The only other thing on my wishlist, is that the onboard Nvidia be compute 5.0 or higher, no matter how powerful. Cheers! 

[adhitthana]

I notice in the OP we have.

Lead developer: smooth
Release engineering, Q/A, support: Arux
Other roles: open (PM smooth)
Original developer (as Monero fork): anonymous

Has anything been happening on this front?

[smooth]

I notice in the OP we have.

Lead developer: smooth
Release engineering, Q/A, support: Arux
Other roles: open (PM smooth)
Original developer (as Monero fork): anonymous

Has anything been happening on this front?

Which front? The other roles? Not specifically. There are a few people helping with social media. Anyone who wants to be more involved in any role is welcome to contact me about it.

[generalizethis]

Too much blue? https://www.reddit.com/r/aeoncoin

[aerbax]

Too much blue? https://www.reddit.com/r/aeoncoin

Way too much blue. 

[americanpegasus]

Too much blue? https://www.reddit.com/r/aeoncoin

 
  
 


Holy shit, you literally need to reverse the entire color scheme.  A+ for effort, but try the color-reversed version.  A little goes a long way.

[smooth]

Too much blue? https://www.reddit.com/r/aeoncoin

Way too much blue. 

Somebody spilled a can of paint?

[smooth]

New security update version (0.9.5.0)

Recommended for all users

Changes in this release

SECURITY: Fix upnp buffer overflow
Fix compile error with gcc 5.1.0
Fix node unblocking message formatting
Code clean ups

https://github.com/aeonix/aeon/releases/tag/v0.9.5.0

[smooth]

New security update version (0.9.5.0)

I have also updated the experimental pruning branch

[myagui]

Thank you for the updates Smooth, very neatly timed as I finally have my (light) node back online, yay!

Unofficial binaries for the brave (or impatient?) Windows folks:

Aeon-Win-x64-0.9.5.0
Aeon-Win-x64-0.9.5.0-light-node

Ran a brief smoke test on both versions and all looks well.
I'm staying on the pruning node for long term testing.

[smooth]

I have accepted a bounty bid from Wolf`to release open source AMD mining software for AEON for the same bounty as the earlier NV miner, which is 9000 AEON.

The requirement is that it be released as open source on github and tested by some independent community members.

If the bounty is not claimed in some reasonable period of time I will open it up to others. For now it is assigned to Wolf`

EDIT: to be released on github

[smooth]

I have accepted a bounty bid from Wolf`to release open source AMD mining software for AEON for the same bounty as the earlier NV miner, which is 9000 AEON.

The requirement is that it be released as open source and tested by some independent community members.

If the bounty is not claimed in some reasonable period of time I will open it up to others. For now it is assigned to Wolf`

Done: https://ottrbutt.com/tmp/wolf-aeon-miner.tar.xz

I edited above that it should be released on github, as the earlier miner was. There is no requirement that you continue to maintain it, others can fork it and take over.