[XPM] Working on a GPU miner for Primecoin

[matt4054]

Thanks for the update!

[1715642774]

1715642774

[1715642774]

1715642774

[1715642774]

1715642774

[1715642774]

1715642774

[1715642774]

1715642774

[Dsfyu]

Alright, I have written the GPU code for the sieve. It compiles, but I haven't gotten to test it yet. I have to code the OpenCL interface first. I can't really use my existing code since all the other coins are "GPU does all the work", but the Primecoin miner is currently "CPU and GPU work together". The ultimate plan is, of course, to do everything on the GPU, but that's not gonna be the case in the first betas.

Thanks for the update! It's really refreshing to see this much communication when some other bitcoin merchants and individuals barely communicate at all. I know a few people that got money tied up in bitfloor back in april and they went up until about two weeks ago with no updates on when they might see their money again. You on the other hand have been giving regular updates about your work that people aren't even going to pay for when it goes live. I can't wait for this to go live just to see the results and what it means for this new growing cryptocurrency. If I had the BTC to get into the beta I would definitely donate it. With that said, when this gets out of beta I would be happy to support your work by donating the fractions of a bitcoin I have from selling my primecoins.

Gread job so far - I can't wait to see what you have when the beta is released tomorrow!

[mhps]

Alright, I have written the GPU code for the sieve. It compiles, but I haven't gotten to test it yet. I have to code the OpenCL interface first. I can't really use my existing code since all the other coins are "GPU does all the work", but the Primecoin miner is currently "CPU and GPU work together". The ultimate plan is, of course, to do everything on the GPU, but that's not gonna be the case in the first betas.

The GPU doesn't have to do everything. Cudaminer for LTC uses both CPU and GPU and works out nicely.  https://bitcointalk.org/index.php?topic=167229.0

[22naru]

good news. cheers.

[Bogart]

Sounds like good progress.  Donated. 

Do you think any of this work can be used towards an FPGA solution?

[colinistheman]

How much faster will GPU mining be than CPU mining for primecoin?

Same ratios of speed improvement in bitcoin or not as much?

[A L I E N]

How much faster will GPU mining be than CPU mining for primecoin?

Same ratios of speed improvement in bitcoin or not as much?

I'm sure he will know after beta testing

[pavimus]

Alright, I have written the GPU code for the sieve. It compiles, but I haven't gotten to test it yet. I have to code the OpenCL interface first. I can't really use my existing code since all the other coins are "GPU does all the work", but the Primecoin miner is currently "CPU and GPU work together". The ultimate plan is, of course, to do everything on the GPU, but that's not gonna be the case in the first betas.

so we must upgrade cpus on our rigs? or 2 core amd cpu is enough to load by 100% 4*7970?

[bcp19]

Alright, I have written the GPU code for the sieve. It compiles, but I haven't gotten to test it yet. I have to code the OpenCL interface first. I can't really use my existing code since all the other coins are "GPU does all the work", but the Primecoin miner is currently "CPU and GPU work together". The ultimate plan is, of course, to do everything on the GPU, but that's not gonna be the case in the first betas.

so we must upgrade cpus on our rigs? or 2 core amd cpu is enough to load by 100% 4*7970?

If you have a 2 core system, I'd say stick to other alt coin mining.  If this port is like other math-heavy projects I've seen go to GPU, it'll take 2-4 cores to load a single 7970.

[miningatnight]

All estimates would really be a guess on what CPU is needed till we know more.

[mtrlt]

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).

[atavacron]

Sweeeeeeeeeeeeeeeeeeeet!!!!!!!!!!!!!

[craslovell]

Good stuff, look forward to the pre-release. My rig is ready 

[Dsfyu]

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).

That's better than I expected - don't most projects like this have worse performance when they first start working than the equivalent? I remember valve talking about optimizing left 4 dead for linux and saying they were getting 2-3 fps on linux when it first started running compared to 200 fps on the same machine in windows.

Maybe it would help to have someone test this on a machine with an intel processor too - sometimes the proc brand can make a pretty big difference even when two procs are equivalent in performance overall.

[eCoinomist]

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).

So will this work on my Pentium dual core with 3x 7970? Can it handle 3x GPUs?

Or will I gain no mining increase until you have a full GPU version?

[96redformula]

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.

[relm9]

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.

He just posted some results on the other page...

[forsetifox]

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.

He just posted some results on the other page...

Where? I don't see any testing data any where in this thread?

[Dsfyu]

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.

He just posted some results on the other page...

Where? I don't see any testing data any where in this thread?

While it isn't concrete data he said that the performance of his 6990 and phenom II x6 together were equal to the cpu performance alone. The fact that it's working with part being done on the gpu is a result in itself. As he said, it has yet to be optimized and he hopes to have it run twice as fast. All we can do is be patient and wait to see what works.

[pyromaniac]

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?