[ANN] CureCoin 2.0 is live - Mandatory Update is available now - DEC 2018

[ComputerGenie]

...Only the rig cost came from me.

Which was the part he was off by.

[bardacuda]

Right just as I explained. Don't see where I'm going off course here.

[cryptohunter]

By your own figures, you've spent $3200 on the 1070s but $3600 on the 1080s.

 (the $400 and $600 are close enough I'm not going to quibble over the actual pricing, which is a very little less in both cases to go with the lowest-cost cards on Newegg on any given day).

 *HOWEVER* your figure of 800K PPD for a 1080 appears to be HIGHLY optimistic based on the database at https://docs.google.com/spreadsheets/d/1vcVoSVtamcoGj5sFfvKF_XlvuviWWveJIg_iZ8U2bf0/pub?output=html# (I don't have any 1080s so I can't compare based on my own figures), while that 600k figure is in line with my 1070 farm of VERY MILDLY overclocked 1070 cards but very PESSAMISTIC vs the database.

 I think you're trying to make a case here based on BAD DATA.

• The pricing in that comment is using is his figures, not mine

Actually those were your figures. My figures were $550 and $800 CAD respectively (even though it's actually more like $510 vs. $780 for the cheapest dual or triple fan design but I was being conservative). I knew if I used my own figures you would nitpick everything.

@QuintLeo, the difference comes from the fact that you need an extra $400 for the rest of the components of each rig (One really good PSU like an EVGA 1300W G2 and then cheap/used everything else, milk crate or self-built wood/aluminum or whatever for case/chassis etc.). You only need one rig for 6 cards and two rigs for 8. I wanted to use his own example of the worst case for the 1070s where you need to pay for an extra rig and only get 7 or 8 cards to show that even in that worst case and with his overly optimistic 1080 PPD numbers, the 1070s still come out ahead. But of course it's still wrong because FLDC only pays out once a month lol.

But you're completely ignoring the fact that the guy with lower per-rig income would have 50% more of those rigs lol

Now, I'm sure you're still stuck on the "more cards" end of your thought so, remember this...
Cards 7 & 8 will need another full set of hardware to run.

Ignoring the  electric difference (~$15.77 USD per year for *total system* @ $0.01 per kWh):

GTX 1070
Cost: ~$400 USD
PPD: 600,000
FLDC per month: 12,000 - ~$14.72 USD
CURE per month: 787 - ~$35.43 USD
Purchase ROI: ~8 months "raw cost" per card
Yearly Revenue: ~$599 USD

GTX 1080
Cost: ~$600 USD
PPD: 800,000
FLDC per month: 16,000 - ~$19.29 USD
CURE per month: 1050 - ~$47.25 USD
Purchase ROI: ~9 months "raw cost" per card
Yearly Revenue: ~$798 USD

...

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

what happened to the other coin they were releasing with curecoin?

they did an alpha or beta release and it went wrong?

What is happening with that coin and when is the snap shot on the cure coin wallet?

It was some quantum resistant new design?

Is it shelved for good or is it still worth holding this cure coins in my wallet awaiting some snapshot date to quality for the new coin?

I have seen no futher mention of it in months now?

I think Vorksholk was waiting on some legal or copyright stuff or something like that but it was more or less finished...that was my understanding anyway.

Thanks for reply I will seek him out and ask him directly again.

[bardacuda]

Perhaps reply to this topic in the curecoin forums:

https://www.curecoin.net/forum/viewtopic.php?f=31&t=3898

It hasn't been updated, but I think those forums are followed more closely by Cygnus and Vorksholk than this thread is, which is understandable with the amount of irrelevant posts one has to read through to get caught up if you've been away from this thread for a while (I'm clearly guilty of adding to this number. See above  ). Also it was posted by Cygnus so he may get a notification in his e-mail or something if you reply to it.

I've chatted with whoever runs the CureCoin account on Facebook and they seem to respond pretty quickly too. I don't think it's either of the devs but I'm sure they could get in touch with them if you ask.

[QuintLeo]

By your own figures, you've spent $3200 on the 1070s but $3600 on the 1080s.

 (the $400 and $600 are close enough I'm not going to quibble over the actual pricing, which is a very little less in both cases to go with the lowest-cost cards on Newegg on any given day).

 *HOWEVER* your figure of 800K PPD for a 1080 appears to be HIGHLY optimistic based on the database at https://docs.google.com/spreadsheets/d/1vcVoSVtamcoGj5sFfvKF_XlvuviWWveJIg_iZ8U2bf0/pub?output=html# (I don't have any 1080s so I can't compare based on my own figures), while that 600k figure is in line with my 1070 farm of VERY MILDLY overclocked 1070 cards but very PESSAMISTIC vs the database.

 I think you're trying to make a case here based on BAD DATA.

• The pricing in that comment is using is his figures, not mine

Actually those were your figures. My figures were $550 and $800 CAD respectively (even though it's actually more like $510 vs. $780 for the cheapest dual or triple fan design but I was being conservative). I knew if I used my own figures you would nitpick everything.

@QuintLeo, the difference comes from the fact that you need an extra $400 for the rest of the components of each rig (One really good PSU like an EVGA 1300W G2 and then cheap/used everything else, milk crate or self-built wood/aluminum or whatever for case/chassis etc.). You only need one rig for 6 cards and two rigs for 8. I wanted to use his own example of the worst case for the 1070s where you need to pay for an extra rig and only get 7 or 8 cards to show that even in that worst case and with his overly optimistic 1080 PPD numbers, the 1070s still come out ahead. But of course it's still wrong because FLDC only pays out once a month lol.

But you're completely ignoring the fact that the guy with lower per-rig income would have 50% more of those rigs lol

Now, I'm sure you're still stuck on the "more cards" end of your thought so, remember this...
Cards 7 & 8 will need another full set of hardware to run.

Ignoring the  electric difference (~$15.77 USD per year for *total system* @ $0.01 per kWh):

GTX 1070
Cost: ~$400 USD
PPD: 600,000
FLDC per month: 12,000 - ~$14.72 USD
CURE per month: 787 - ~$35.43 USD
Purchase ROI: ~8 months "raw cost" per card
Yearly Revenue: ~$599 USD

GTX 1080
Cost: ~$600 USD
PPD: 800,000
FLDC per month: 16,000 - ~$19.29 USD
CURE per month: 1050 - ~$47.25 USD
Purchase ROI: ~9 months "raw cost" per card
Yearly Revenue: ~$798 USD

...

 My figures weren't $400 and $600 - someone else came up with those, I just agreed that they were reasonable for comparison.
 I never stated any figures of my own.

 I wasn't looking at one WU - I was looking at the AVERAGES page - which points to your 800k PPD figure for a 1080 as being quite a bit on the high side of a FAIR estimate, and your 600k PPD probably being a bit on the low side, both cards being equally overclocked.

 $400 seems quite a bit on the high side for the "additional base system" to me, but that's likely due to the fact I don't build riser systems.

 Even if that figure is included, that would give you a second system that 3 cards could be added to before you need to do it again on the 1070 side, vs having to add a second base system on the 1080 side before you could add more cards - which would allow the 1070 based setup to pull away from the 1080 using reinvested profits.

 Let's look at $8000 - 2 full 1080 rigs, 12 cards at 9,600,000 PPD vs 3 full 1070 rigs 18 cards at 10,800,000 PPD.
 Interesting how a straight up FULL RIG comparison favors the 1070.

 You see this same result at all price points that do not compare a "optimal price point for exactly full systems" 1080 setup to a non-full-rigs 1070 setup.

 If the 1080 didn't carry a premium price compared to it's additional performance, it would win - but paying 50% more per card for 33% additional performance (or LESS using posted real world figures, probably due to the memory system not being all that much faster) overcomes the savings on non-GPU components in a FAIR comparison.


 It IS interesting that the $4000 point was chosen for the comparison though, since it seems to be THE most favorable point for the 1080 - and the 1080 STILL ends up losing past the first month or two.

[ComputerGenie]

... I wasn't looking at one WU - I was looking at the AVERAGES page...

This database was created to compare frame times with other users in order to see whether your PPD is where it should be or if there is something wrong with the system (instability, incorrect client configuration, throttling etc.)

The entire sheet is/was meant to be able to look at a given WU, and see how your results for that WU compare to someone else that had already submitted their info for that WU on the overclock.net forum.
The "averages" don't come out the same when you compare PRCG to PRCG.
When you look at the guys that have multiple rigs, the "average" answer you'll get is that the "average" 1070 will pump an "average" of about 600kPPD and the "average" 1080 will pump an "average" of about 800kPPD. This is their experience over time, not 1 time over 1 WU.
The "database" contains 17 numbers for a 1080, which is 17 single WUs; that is less than a single full day of folding for a single 1080.

[bardacuda]

My figures weren't $400 and $600 - someone else came up with those, I just agreed that they were reasonable for comparison.
 I never stated any figures of my own.

Right, I was talking to ComputerGenie when I said that, because he said they were MY figures.

• The pricing in that comment is using is his figures, not mine

Actually those were your figures...

...

$400 seems quite a bit on the high side for the "additional base system" to me, but that's likely due to the fact I don't build riser systems.

Right, like I say I was being conservative with all of my numbers and erring in favor of the 1080s in order to do a comparison in the worst case scenario for the 1070s...and to not have to endure having everything I said being nitpicked to death and implications of me being mentally challenged. (To no avail concerning that latter part)

It IS interesting that the $4000 point was chosen for the comparison though, since it seems to be THE most favorable point for the 1080 - and the 1080 STILL ends up losing past the first month or two.

That's why I chose it   (And presumably why Mr. Genie chose it to counter an earlier post of mine)

Now, I'm sure you're still stuck on the "more cards" end of your thought so, remember this...
Cards 7 & 8 will need another full set of hardware to run.

[ryohazuki89]

Just for fun speculation, it sure is unusual to see curecoin hold a price this high for this long, appears to be climbing too.

4 day chart: http://puu.sh/tHzgE/0cfd0578dd.png

It's flatlined now, I'm panic holding as usual but I expect a steady drop from here.. surprise me curecoin!

[ComputerGenie]

$400 seems quite a bit on the high side for the "additional base system" to me, but that's likely due to the fact I don't build riser systems.

Where are you getting your motherboards? For the rest of us, $400 USD is cheap for a setup with a mb that has 6* x PCI Express 3.0 x16 Slots.





*less than PCI Express 3.0 x16 and both cards (1070 & 1080) suffer PPD loss from 2% to 50% depending on the slot type

[QuintLeo]

$400 seems quite a bit on the high side for the "additional base system" to me, but that's likely due to the fact I don't build riser systems.

Where are you getting your motherboards? For the rest of us, $400 USD is cheap for a setup with a mb that has 6* x PCI Express 3.0 x16 Slots.

*less than PCI Express 3.0 x16 and both cards (1070 & 1080) suffer PPD loss from 2% to 50% depending on the slot type

 AS I SAID, I don't build riser systems.
 I don't think I own a motherboard that has 6x PCI-E slots (I know some have 5), but I don't care enough to bother going to check.

 I also don't build "optimised specifically for Folding" systems - my recent builds have been 3x PCI-E slot AMD FM2 based systems, with an A10 APU that uses the CPU cores to support the folding GPUs and the APU's GPU to run Dnet RC5 via MooWrapper/BOINC and generating some Gridcoin on the side - I'm a VERY long term participant in the Distributed.net project, going back to DES days.

 My "base systems" probably run $250-$300 but that's with a much higher cost CPU than a "Folding-specific" system needs, could easily cut over $100 of the price by going with a bottom-end A4 or switching to an AM3+ solution with a Sempron CPU instead. Plenty of well-under $100 motherboards with 3x PCI-E 16-bit slots on the AMD side, and I tend to overkill some on the RAM too.

 The Seasonic X850 power supplies I use are mild overkill, but I wouldn't run a 6 card rig of 1070s on a X1250 (or an EVGA G2 1300) anyway.

[Vorksholk]

Just thought I'd check in, project is still active. Someone mentioned SigmaX, that is still planned to launch along with cc2.0.

If you have any specific questions, please PM me, as I don't have time to read through this thread on a regular basis.

[artoar_11]

Soon we will begin internal testing with Core_22. It's CUDA! It is expected that NVidia GPU's to get approximately 30% more performance. I participate in internal tests. For me it is a pleasure.

Postby JohnChodera » 30 Jan 2017, 21:56

> I hope so, but I'm not optimistic. It seems that EITHER the hot-fix or the V.18 FAHCore_21 eliminates the errors but, at the same time, performance is reduced.

I believe the driver hotfix is responsible for the performance degradation since it enforces greater thread synchronization than the workaround we have in 0.0.18. I expect you will see a slight uptick in performance, but once we get core 22 out in a few weeks, you should see a big increase in performance---especially with CUDA.

https://foldingforum.org/viewtopic.php?f=66&t=29618&start=15#p292677

[ComputerGenie]

... I also don't build "optimised specifically for Folding" systems... My "base systems" probably run $250-$300 but that's with a much higher cost CPU than a "Folding-specific" system needs, could easily cut over $100 of the price by going with a bottom-end A4 or switching to an AM3+ solution with a Sempron CPU instead. Plenty of well-under $100 motherboards with 3x PCI-E 16-bit slots on the AMD side...

Then, perhaps, you shouldn't try to chime in on what is "better". You're building a system that's at a loss in bandwidth to start and you're pretending to have some valid response about production values?
Yes, I agree, there are boards with 3 slots; however, that option spends $300 to "save" $200...

As you admit, folding isn't mining, and as such things are different. Do future folders a service and stop giving a technical opinion that is based on something that's non-relevant.

[bardacuda]

@QuintLeo
Is there any particular 1070 model that you recommend for longevity / price / cooling, etc.? I was thinking the EVGA ACX 3.0 08G-P4-5171. I know they had an issue with missing thermal pads but they'll send a free kit to fix that and honour their warranty. Seems to be one of the cheapest and also its EVGA with dual fans, so it seems like the one to go with but I wanted to get your opinion first.

[QuintLeo]

... I also don't build "optimised specifically for Folding" systems... My "base systems" probably run $250-$300 but that's with a much higher cost CPU than a "Folding-specific" system needs, could easily cut over $100 of the price by going with a bottom-end A4 or switching to an AM3+ solution with a Sempron CPU instead. Plenty of well-under $100 motherboards with 3x PCI-E 16-bit slots on the AMD side...

Then, perhaps, you shouldn't try to chime in on what is "better". You're building a system that's at a loss in bandwidth to start and you're pretending to have some valid response about production values?
Yes, I agree, there are boards with 3 slots; however, that option spends $300 to "save" $200...

As you admit, folding isn't mining, and as such things are different. Do future folders a service and stop giving a technical opinion that is based on something that's non-relevant.

 Nope - I don't have a loss in bandwidth, the 6 card systems are the ones trying to shove twice the bandwidth through the SAME CPU bandwidth and usually same number of PCI-E lanes.
 I also didn't specify anything about "better" on the system build - I was just pointing out that the numbers on the "economics" for the 1070 vs. the 1080 were skewed and incorrect.
 I also don't "spend $300 to save $200" - I could match the total system cost PER CARD pretty much exactly with non-riser systems if I chose to do so in large part because I don't spend the INFLATED PRICING on super-high-end PS, plus I don't have to pay for the risers themselves, and I would end up with systems that are AT LEAST as reliable and stable and likely MORE reliable.

 Just because YOUR errors or outright lies have been exposed doesn't mean my points are invalid - just because I don't CHOOSE to build "optimised for folding" systems doesn't mean I don't know the ECONOMICS or HOW to do so.

[QuintLeo]

@QuintLeo
Is there any particular 1070 model that you recommend for longevity / price / cooling, etc.? I was thinking the EVGA ACX 3.0 08G-P4-5171. I know they had an issue with missing thermal pads but they'll send a free kit to fix that and honour their warranty. Seems to be one of the cheapest and also its EVGA with dual fans, so it seems like the one to go with but I wanted to get your opinion first.

 The EVGA cards I have are the "Black" 08G-P4-5173-KR model.
 Most of my Gigabytes are the GV-N1070WF2OC 2-fan "Windforce" model
 The rest of my Gigabytes are the GV-N1070IXOC "mini-ITX" model

 All are 8GB (but I think all 1070s released to date are 8GB)


 The Gigabyte 2-fan Windforce models I run seem to stay just as cool as the 2-fan EVGA cards I run, despite being a little higher "factory" clocked.
 The Gigabyte MiniITX stay cool as well but they have less obstruction in their airflow.

 Given a choice at the same price point I'd go with he Gigabytes due to the higher clocks.
 
 It's more about gettng ANY GPU set up with a SANE fan speed / curve than any other factor when it comes to keeping 1070s cool.

 I don't have a preference at all on reliability so far, haven't had issues with a 1070 failing yet.

 The Gigabytes seem to have a little more headroom on overclocking, despite their higher "factory" clock, at the cost of having a higher TDP (151 watts vs 180 watts using Nvidia-SMI to monitor with Afterburner set at the "+0 TDP" point).

[bardacuda]

Thanks for your input!

[xandry]

The EVGA cards I have are the "Black" 08G-P4-5173-KR model.
 Most of my Gigabytes are the GV-N1070WF2OC 2-fan "Windforce" model
 The rest of my Gigabytes are the GV-N1070IXOC "mini-ITX" model

Please share with us their PPD results.

[bardacuda]

That should depend only on clocks and not the model (unless there was throttling due to poor cooler design). I think he said he was getting ~600K PPD with "mild" overclocks.

[QuintLeo]

The EVGA cards I have are the "Black" 08G-P4-5173-KR model.
 Most of my Gigabytes are the GV-N1070WF2OC 2-fan "Windforce" model
 The rest of my Gigabytes are the GV-N1070IXOC "mini-ITX" model

Please share with us their PPD results.

 I don't keep track by individual card - and it varies with the specific work unit anyway.

 My primary criteria is:
(1) reputable maker (EVGA tried to mess that up, but it was more about the TOO LOW FAN CURVE STOCK which I've never put up with)
(2) cost
(3) in the case of the mini-ITX cards, allowing for good cooling of the card next to it.
(4) "factory" clock rate.