Pages:
Author

Topic: [XPM] Working on a GPU miner for Primecoin - page 13. (Read 102789 times)

member
Activity: 104
Merit: 10
Okay the new GPU sieve is done, and using a single core of an HD6990 it's around 25x faster than a single CPU core of an x6 1055T. Next up: debugging and optimizing the primality tests.
legendary
Activity: 1022
Merit: 1001
looking forward to this
hero member
Activity: 756
Merit: 500
Can't wait. Smiley
legendary
Activity: 2674
Merit: 2965
Terminated.
Nice, let the mining begin very soon  Cheesy
sr. member
Activity: 464
Merit: 252
I'm working on a major optimization of the GPU sieve. Should be ready today, and the sieve should run 5-10x faster.
Today - it's good!
legendary
Activity: 1764
Merit: 1000
Hoping for the best!!! Wink

member
Activity: 104
Merit: 10
I'm working on a major optimization of the GPU sieve. Should be ready today, and the sieve should run 5-10x faster.
legendary
Activity: 2674
Merit: 2965
Terminated.
Any news on release?
hero member
Activity: 675
Merit: 507
Freedom to choose
thank you for devoting time to making a GPU miner
legendary
Activity: 1498
Merit: 1000
Watching this Awesomeness  Cool
hero member
Activity: 756
Merit: 500
Nice. Smiley

Just putting it out here for public record.

1BTC sent on 25 July. (7458d3f1730e99764bc3f39b5ee5650f0b587a1603d88155564331520ffe3302)
legendary
Activity: 1764
Merit: 1000
This is great news... Smiley

Hope we can do the beta testing soon Smiley


Thanks
member
Activity: 104
Merit: 10
You could try commenting out the bodies of functions, checking if compiling with optimization works and repeating until you find the code it's falling over on.

Yep, that's what I ended up doing and I did find and fix the problem. Smiley
donator
Activity: 1274
Merit: 1060
GetMonero.org / MyMonero.com
I agree with that to an extent. But I also think that one should weigh in quite a few factors before taking such a risk, IE reputation of founder, profitability (in this instance), eagerness to obtain the product, etc. If all of those factors (most importantly reputation) come to a positive conclusion, then I think you should be willing to pay what it's worth. If it's not worth it to you, then let the project be, but I don't really see a reason to bash others for supporting.

I paid my donation, so I'm definitely not bashing anyone;) I was replying to refer_2_me's comment that there should be a level of accountability, when in this case there actually isn't. mtrlt can take our money and run, and that's his prerogative. Maybe some will pursue him legally, he'll definitely be scorned, but there's virtually no accountability. We have sent our donation, and we trust that he will deliver, whether that trust is misplaced or not.
sr. member
Activity: 840
Merit: 251
While I agree with you about developers salaries, it's not the point of the discussion. In your jobs, there is a level of accountability which isn't present here.

There's no "accountability" in open source software - and let's not kid ourselves, this is FOSS. That the developer is taking donations for a prerelease does not change the fact that it will be released under GPL or similar licensing, so there's no "company" to shout at. Everyone needs to disconnect from their pseudoreality where they equate sending a donation to a FOSS developer to paying a company for a piece of software. As KickStarter and Indiegogo have demonstrated, even the legal framework around donating to an incomplete piece of software/hardware is murky, and backers may find themselves fresh out of luck with no legal recourse to reclaim their funds.

I agree with that to an extent. But I also think that one should weigh in quite a few factors before taking such a risk, IE reputation of founder, profitability (in this instance), eagerness to obtain the product, etc. If all of those factors (most importantly reputation) come to a positive conclusion, then I think you should be willing to pay what it's worth. If it's not worth it to you, then let the project be, but I don't really see a reason to bash others for supporting.
full member
Activity: 122
Merit: 100
I'm now working on getting the Fermat primality test done on the GPU. Everything but the bignum modulo operation is already done, but there's a problem: The OpenCL compiler crashes when I try to compile with optimizations turned on. Without optimizations the code compiles, but that's obviously not a good solution. I will try updating my graphics drivers.

I haven't forgotten about testing on a slow CPU + overkill GPU rig, but I still have some problems using multiple GPUs.

You could try commenting out the bodies of functions, checking if compiling with optimization works and repeating until you find the code it's falling over on. Could also try older or beta versions of the compiler.
full member
Activity: 347
Merit: 100
If it would be possible "hack" GPU that is recognize like multiple CPU it would be a solution, but It's impossible. So better use Intel Xeon PHI, but is not x64 architecture as I know.
newbie
Activity: 31
Merit: 0
sr. member
Activity: 282
Merit: 250
watching this thread ;p
donator
Activity: 1274
Merit: 1060
GetMonero.org / MyMonero.com
While I agree with you about developers salaries, it's not the point of the discussion. In your jobs, there is a level of accountability which isn't present here.

There's no "accountability" in open source software - and let's not kid ourselves, this is FOSS. That the developer is taking donations for a prerelease does not change the fact that it will be released under GPL or similar licensing, so there's no "company" to shout at. Everyone needs to disconnect from their pseudoreality where they equate sending a donation to a FOSS developer to paying a company for a piece of software. As KickStarter and Indiegogo have demonstrated, even the legal framework around donating to an incomplete piece of software/hardware is murky, and backers may find themselves fresh out of luck with no legal recourse to reclaim their funds.
Pages:
Jump to: