[XPM] Working on a GPU miner for Primecoin

[funnow]

How about Xeon Phi for XPM mining? Xeon Phi is OpenCL device and multi core x86 CPU in one. Does somebody try it? Is it more profitable, than CPU or GPU?

He he You was faster than me. I think Xeon Phi is a good solution.

[1714938214]

1714938214

[1714938214]

1714938214

[1714938214]

1714938214

[1714938214]

1714938214

[1714938214]

1714938214

[Krusher33]

So after you give out round 1 to beta testers are you going to release early to others who donate after?

I don't think its too crazy to ask for some results before we donate.  I bet there are lots of people on the fence with donations ready once some proof of work and or results can be seen.

I wouldn't do it.  Some folks will donate to get early beta and then just share with public.  And then the public will just tear programmers apart saying it's a p.o.s.  Keep it within the "inner" circle until it's actually ready for public.

[azwccc]

Okay, the first GPU-enabled version is ready. Right now it only does the sieving on the GPU. Primality tests are still done on the CPU. The code hasn't been optimized at all yet; the performance is equivalent (+- a few %) on these setups:
1) Phenom x6 1055T, all 6 cores
2) Phenom x6 1055T, all 6 cores, plus a HD6990 using both cores

I don't think the current version is worth releasing. It still needs a bit of work, I want it to be twice as fast. I estimate there is room for a five-fold improvement, especially when I get the primality tests done on the GPU as well.

Tomorrow I will do tests using a slow CPU with lots of GPU power (Sempron 140 with 2x6990).

I think it worth a release... At least we can have pool set up based on the miner.

[OnlyC]

Okay, the first GPU-enabled version is ready. Right now it only does the sieving on the GPU. Primality tests are still done on the CPU. The code hasn't been optimized at all yet; the performance is equivalent (+- a few %) on these setups:
1) Phenom x6 1055T, all 6 cores
2) Phenom x6 1055T, all 6 cores, plus a HD6990 using both cores

I don't think the current version is worth releasing. It still needs a bit of work, I want it to be twice as fast. I estimate there is room for a five-fold improvement, especially when I get the primality tests done on the GPU as well.

Tomorrow I will do tests using a slow CPU with lots of GPU power (Sempron 140 with 2x6990).

I think it worth a release... At least we can have pool set up based on the miner.

Yes,  I think pool is really important, more than a GPUminer.

[paulthetafy]

Okay, the first GPU-enabled version is ready. Right now it only does the sieving on the GPU. Primality tests are still done on the CPU. The code hasn't been optimized at all yet; the performance is equivalent (+- a few %) on these setups:
1) Phenom x6 1055T, all 6 cores
2) Phenom x6 1055T, all 6 cores, plus a HD6990 using both cores

I don't think the current version is worth releasing. It still needs a bit of work, I want it to be twice as fast. I estimate there is room for a five-fold improvement, especially when I get the primality tests done on the GPU as well.

Tomorrow I will do tests using a slow CPU with lots of GPU power (Sempron 140 with 2x6990).

I think it worth a release... At least we can have pool set up based on the miner.

Yes,  I think pool is really important, more than a GPUminer.

I'm not sure that i agree that a pool is more important, but regardless of that, I don't think releasing the code now is a good idea.  A fair number of people have donated to this project expecting a GPU miner and releasing this code now will mean opening the source code up to other developers.  If they happen to release that code before mtrlt completes the GPU implementation then it negates the donations by those people expecting an early release.  In fact, I would go as far as to say that I don't think mtrlt should release anything until the GPU implementation has been optimized up to his anticipated 5 fold increase over CPU, for the same reason.  BTW I donated.

[dudeguy]

Okay, the first GPU-enabled version is ready. Right now it only does the sieving on the GPU. Primality tests are still done on the CPU. The code hasn't been optimized at all yet; the performance is equivalent (+- a few %) on these setups:
1) Phenom x6 1055T, all 6 cores
2) Phenom x6 1055T, all 6 cores, plus a HD6990 using both cores

I don't think the current version is worth releasing. It still needs a bit of work, I want it to be twice as fast. I estimate there is room for a five-fold improvement, especially when I get the primality tests done on the GPU as well.

Tomorrow I will do tests using a slow CPU with lots of GPU power (Sempron 140 with 2x6990).

I think it worth a release... At least we can have pool set up based on the miner.

Yes,  I think pool is really important, more than a GPUminer.

I'm not sure that i agree that a pool is more important, but regardless of that, I don't think releasing the code now is a good idea.  A fair number of people have donated to this project expecting a GPU miner and releasing this code now will mean opening the source code up to other developers.  If they happen to release that code before mtrlt completes the GPU implementation then it negates the donations by those people expecting an early release.  In fact, I would go as far as to say that I don't think mtrlt should release anything until the GPU implementation has been optimized up to his anticipated 5 fold increase over CPU, for the same reason.  BTW I donated.

What speeds would 2 7870s get combined with an i3 3225 and is it worth it to pay 1BTC for early access?

[refer_2_me]

Okay, the first GPU-enabled version is ready. Right now it only does the sieving on the GPU. Primality tests are still done on the CPU. The code hasn't been optimized at all yet; the performance is equivalent (+- a few %) on these setups:
1) Phenom x6 1055T, all 6 cores
2) Phenom x6 1055T, all 6 cores, plus a HD6990 using both cores

I don't think the current version is worth releasing. It still needs a bit of work, I want it to be twice as fast. I estimate there is room for a five-fold improvement, especially when I get the primality tests done on the GPU as well.

Tomorrow I will do tests using a slow CPU with lots of GPU power (Sempron 140 with 2x6990).

I think it worth a release... At least we can have pool set up based on the miner.

Yes,  I think pool is really important, more than a GPUminer.

I'm not sure that i agree that a pool is more important, but regardless of that, I don't think releasing the code now is a good idea.  A fair number of people have donated to this project expecting a GPU miner and releasing this code now will mean opening the source code up to other developers.  If they happen to release that code before mtrlt completes the GPU implementation then it negates the donations by those people expecting an early release.  In fact, I would go as far as to say that I don't think mtrlt should release anything until the GPU implementation has been optimized up to his anticipated 5 fold increase over CPU, for the same reason.  BTW I donated.

What speeds would 2 7870s get combined with an i3 3225 and is it worth it to pay 1BTC for early access?

No one knows at this point. It's your gamble. You can buy a lot of time on a VPS for $100, to me a week of early access isn't worth it.

[hendo420]

How much has everyone paid in all together?

If its as high as I'm thinking we could have paid a professional programmers salary for a few months at least.

[forsetifox]

Everyone that payed 1 BTC for a early access to the miner is going to lose money IMO. Once news of the miner gets out the price is going to drop by ~40-60%.

I personally believe it will bottom out at 0.002 - 0.003.

[refer_2_me]

How much has everyone paid in all together?

If its as high as I'm thinking we could have paid a professional programmers salary for a few months at least.

About $6000.  56.6 BTC @ $105/BTC

https://blockchain.info/address/1FKfgFQeZCpjGDn7ocgj26nk4K2TNiVJyD

[hendo420]

How much has everyone paid in all together?

If its as high as I'm thinking we could have paid a professional programmers salary for a few months at least.

About $6000.  56.6 BTC @ $105/BTC

https://blockchain.info/address/1FKfgFQeZCpjGDn7ocgj26nk4K2TNiVJyD

I was right

[96redformula]

That is a nice chunk of change!

[blastbob]

You guys... pro coders costs 50 an hour if you want quality.

I donated 12BTC to this project, and i dont care much about the ROI. I made my money already

[refer_2_me]

You guys... pro coders costs 50 an hour if you want quality.

I donated 12BTC to this project, and i dont care much about the ROI. I made my money already

Well then, I am working on some primecoin ASICs, so you can donate here: 1Cxub82PuuVRR2fTXwPup2e8WJUaj4e6HH

[96redformula]

Once news of the miner gets out the price is going to drop

This makes no sense whatsoever.
If any, the rising difficulty will make XPM scarcer.

Difficulty != price

Time and time again we are shown that a higher difficulty does not equal a higher price.

[wetroof]

for PPC and XPC increase difficulty means decrease block reward. not so for BTC or LTC.

[erk]

Once news of the miner gets out the price is going to drop

This makes no sense whatsoever.
If any, the rising difficulty will make XPM scarcer.

Difficulty != price

Time and time again we are shown that a higher difficulty does not equal a higher price.

But difficulty does set the minimum price you can afford to sell a coin you mined, due to the cost of power consumption.

[notme]

Once news of the miner gets out the price is going to drop

This makes no sense whatsoever.
If any, the rising difficulty will make XPM scarcer.

Difficulty != price

Time and time again we are shown that a higher difficulty does not equal a higher price.

But difficulty does set the minimum price you can afford to sell a coin you mined, due to the cost of power consumption.

Sure, if you assume people would rather risk an complete loss rather than take a small loss.

[masterOfDisaster]

Once news of the miner gets out the price is going to drop

This makes no sense whatsoever.
If any, the rising difficulty will make XPM scarcer.

Difficulty != price

Time and time again we are shown that a higher difficulty does not equal a higher price.

From what I experienced with the difficulty and the price development, I dare to say that huge difficulty increases can be originated by people throwing lots of calculation power to a network, gaining lots of coins in a short time frame and when they sell those coins, prices may drop. In the end, if the difficulty stays high (because others gain interest in those coins and start to mine some on their own), the price starts to rise because it becomes more attractive to buy them than to mine them.
But I think there are a lot more aspects that have to be taken into account.
Its just a try to explain the early price drop and late price rise in case of a raising difficulty.

[fluffypony]

How much has everyone paid in all together?

If its as high as I'm thinking we could have paid a professional programmers salary for a few months at least.

About $6000.  56.6 BTC @ $105/BTC

https://blockchain.info/address/1FKfgFQeZCpjGDn7ocgj26nk4K2TNiVJyD

I was right

HAHA HHAHAHAHA HAH AHAAHAH AHA HAHAHAH AH AHA.

"a few months"

Mid-level C/C++ developers, particularly in a niche space like OpenCL/CUDA development, don't work for less than $4000/month. That does not include additional costs an employer would incur - equipment, unemployment insurance, consumables, administrative costs, office space, furniture for that developer, heat/lighting/water/electricity, and so on. Even amortised over many developers, those costs can easily add $1500 to the raw cost of a developer's salary.

Thus, you can understand why freelancers who are worth their salt don't work for under $50/hour (they have to cover those overheads on their own). In the initial intense phases of a new project like this, a developer can EASILY spend 10 hours a day on it - many days they can clock 16 hour work days. Even at 10 hours a day, the $6000 would only pay for 12 work days - that's a little over 2 weeks not including weekends (ie. quite a bit under 2 weeks if we factor in long work days and the occasional piece of weekend work). $6000 is a pittance, a bit of piss in a bucket for the level of work that the (sole!) developer will have to put in now and the amount of effort he'll have to put in on an on-going basis to support this. Do you think people will donate any substantial amount once they're using it and raking in tens of thousands of Dollars? Not a chance. People are greedy as fuck.

So no, we would not be able to pay "a professional programmers salary for a few months". We would possibly get a mid-level engineer for two solid weeks, or a senior dev for a bit less time.

Also, if there's any doubt as to my credentials on this subject: not only have I been a developer (professionally) for just over 14 years, but for the last 6 years I've moved beyond a team leader role and have been the senior manager of a major (local) dev house until I quit my job last year October to do my own thing. I know a thing or two about the financial implications of building or extending software.