Author

Topic: [ANN] AEON [2019-09-27: Upgrade to version 0.13.0.0 ASAP HF@1146200 Oct 25] - page 211. (Read 625666 times)

legendary
Activity: 1154
Merit: 1001
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!  Cool
legendary
Activity: 2968
Merit: 1198
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 inGrin

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.

legendary
Activity: 1154
Merit: 1001
[...] 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  Smiley     

Quote
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 inGrin
legendary
Activity: 2968
Merit: 1198
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.

Quote
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.
legendary
Activity: 1154
Merit: 1001
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.
legendary
Activity: 2968
Merit: 1198
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.  Smiley

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.

legendary
Activity: 1154
Merit: 1001
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.  Smiley
sr. member
Activity: 352
Merit: 250
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.
legendary
Activity: 2968
Merit: 1198
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.

hero member
Activity: 606
Merit: 500
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.
legendary
Activity: 2968
Merit: 1198
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.
hero member
Activity: 606
Merit: 500
Any literature on the PoW algorithm? Runs surprisingly low temp on my FX-8350
hero member
Activity: 896
Merit: 1000
Should have stuck to duckNote. If not for the continuity and inertia, then surely for the seemingly genial brand.  

duck is still in our pockets  Wink
But with evolution of blockchain and new features, digital is a way to go. ducknote is still one of XDNs alter ego. All in this story will fall into place over time - all duck/Dark/DigitalNote will be recognizable in different kind of clients, applications and    
use of XDN blockchain*. We will implement our vision in source code.

*actually block tape with transactions trees growing on it.

Sorry for interruption your AEON discussion here.
sr. member
Activity: 450
Merit: 250
Good news from the Monero thread. Wolf0 is going to work on an open source AMD miner for CryptoNight. I hope someone is then able to tweak it for the CryptoNight-Lite algo.

Wolf0 has taken up working on an AMD open source miner!

https://forum.getmonero.org/7/open-tasks/2400/open-source-amd-miner-by-wolf0

Please check out the getmonero.org forum thread and add your comments or suggestions.
hero member
Activity: 714
Merit: 504
I don't want to fuel any further this off-topic venture into XDN, but...

Can of worms. Can of worms...

Quote
The reality is, that ducknote, had none of the appeal, or the image/presentation that digitalnote claims today. For that reason, myself, and certainly many others, dismissed ducknote mining entirely, during what was - by far - the highest reward period.

It's kinda interesting to hear that viewpoint. I'm not a miner but an investor. When duckNote was launched with its name and low-fidelity logo I became very involved OTC, precisely because it was called duckNote. I remember picking up bags and bags of the stuff at around 0.00000006. It had, in my opinion, a massively radical down-to-earth and creative brand that, had the developer stuck to it, could have easily turned into something of an credible pandemic - kind of like a cryptonote version of Doge.

In the end the developer was a dick: practically uncontactable, no sense of community strategy, and then he goes and throws it all away...not just by rebranding once... but twice... eventually settling on digitalNote, which has to be the blandest, most sterile, most risable attempt at seriousness ever conceived. It's reminiscent of a student project that's desperately attempting to emulate the superficialities of a business world they have very little knowledge or experience of.

Should have stuck to duckNote. If not for the continuity and inertia, then surely for the seemingly genial brand.  
hero member
Activity: 770
Merit: 504
Yes, let's get back on topic please: 
 
Discussion of Aeon.  My fault for taking us into the weird; if you want to sling mud and argue about other currencies, there's plenty of other places to do it.
legendary
Activity: 1154
Merit: 1001
I don't want to fuel any further this off-topic venture into XDN, but personally, I will never dismiss the name choice at launch as just a creative form of scam. Maybe it was not intentional, maybe it was, I don't see that there will ever be a way to be certain.

The reality is, that ducknote, had none of the appeal, or the image/presentation that digitalnote claims today. For that reason, myself, and certainly many others, dismissed ducknote mining entirely, during what was - by far - the highest reward period.

So at best: very poor choice of naming will have most definitely contributed to an inferior initial distribution, and this was further aggravated by the quick drop in mining returns.
At worse: it was a creative scam, posing as just another meme coin void of any worth throughout the most important/rewarding emission period, only to become all serious and meaningful by the time that rewards have dropped significantly.

Some of the recent ideas or developments might be worthwhile. I haven't been much available, but would like to look into how PoS is being approached there. I'd love to one day see a proper cryptonote implementation with both PoW and PoS, but one that does not affect the privacy features/specification in any substantial measure (unless it improves upon them).
sr. member
Activity: 450
Merit: 250

Good development of XDN won't undo the poor choice for the emission curve.
full member
Activity: 177
Merit: 100
lol, XDN is on top of my shortlist. http://digitalnote.org/ is more than just an 'interesting' CryptoNote with 'interesting ideas'. @smooth has spread fud in the past and it makes me sick when he does do it. He has recently started to see DigitalNote for what it is so yes, do your research. See that XDN has had one of a kind CryptoNote features implemented long ago with improvements and new features being added at a steady pace. Spend some time going through the https://bitcointalksearch.org/topic/ann-digitalnote-xdn-icco-announce-news-1082745 forum to see all of the planning and development. Buy cheap now (58 satoshi atm) and hold for the future. Use the wallet for your encrypted messaging. I was fast mining at the beginning and have quite a lot but I will say one important fact when it comes to @smooth's more than questionable fud; there are a lot of coins in the free market for cheap. For example, there are 160,000,000 available on hitBTC, 5,000,000 on Bittrex and over 16,000,000 on Poloniex all for under 101sat. I have already sold a large percentage on all three just to pay my hydro for when I was mining at the start of XDN.

Any cryptonote with continuing development has a good future.

By the way @smooth, I do commend you for your countless hours of 'positive' work in the cryptocurrency field.
Jump to: