[XPM] Working on a GPU miner for Primecoin - page 12.

oxfeeefeee

member

Activity: 73

Merit: 10

Quote from: bcp19 on August 03, 2013, 11:28:36 PM

Quote from: oxfeeefeee on August 03, 2013, 10:31:11 PM

Quote from: bcp19 on August 03, 2013, 09:53:32 PM

Depends on speed. We know sieving is running about 25x, but how much of the CPU time was used for sieving and how much for fermat testing? Let x = CPU sieve time and y = CPU Fermat time. If x=y (50/50 split) then GPU = 2% for 52% of normal time = 92.3% speed increase. If x=9y (90/10 split) then GPU = 3.6% for 13.6% of normal time = 635% speed increase. if 9x=y (10/90 split) GPU = .4% fo 90.4% of normal time = 10.6% speed increase. These are extremely rough figures, communication times between GPU/CPU will also apply. Maybe he'll solve Fermat testing on the GPU and get a full 2500% speed increase. Short answer: too many current unknowns.

I did a similar calculation, but then Koooooj@reddit pointed out that I was wrong. Because with a much powerful sieve, CPU gets less test to do for fixed number of output.

for example (with made up numbers):
before: 10000Numbers ---sieve--> 100Numbers ---test---> 1Number
after : 10000Numbers ---sieve--> 10Numbers ---test---> 1Number

this way, even we don't speed up the execution of Fermat test, the test time is down to 10%.

But he's comparing apples to oranges. If sieveing speed is increased 25x, then you need different parameters to sieve deeper. Also, where does he get these numbers? 10000 seems extremely small, 100 is 1%, so we are to believe the CPU sieve is 99% effective? The GPU sieve is 99.9% effective? There's a lot of questions that need to be asked here:
What is he using for h₀?
What is he using for c?
What is he using for e?
What is the first sieving prime?
What is he using for P_s?

I'm sorry, but those number he's tossing out look made up, with no proof to back them up.

And I made up those number just to explain the idea, not him.
This is just another possibility, GPU could still be as slow as 2x

bcp19

hero member

Activity: 532

Merit: 500

Quote from: oxfeeefeee on August 03, 2013, 10:31:11 PM

Quote from: bcp19 on August 03, 2013, 09:53:32 PM

Depends on speed. We know sieving is running about 25x, but how much of the CPU time was used for sieving and how much for fermat testing? Let x = CPU sieve time and y = CPU Fermat time. If x=y (50/50 split) then GPU = 2% for 52% of normal time = 92.3% speed increase. If x=9y (90/10 split) then GPU = 3.6% for 13.6% of normal time = 635% speed increase. if 9x=y (10/90 split) GPU = .4% fo 90.4% of normal time = 10.6% speed increase. These are extremely rough figures, communication times between GPU/CPU will also apply. Maybe he'll solve Fermat testing on the GPU and get a full 2500% speed increase. Short answer: too many current unknowns.

I did a similar calculation, but then Koooooj@reddit pointed out that I was wrong. Because with a much powerful sieve, CPU gets less test to do for fixed number of output.

for example (with made up numbers):
before: 10000Numbers ---sieve--> 100Numbers ---test---> 1Number
after : 10000Numbers ---sieve--> 10Numbers ---test---> 1Number

this way, even we don't speed up the execution of Fermat test, the test time is down to 10%.

But he's comparing apples to oranges. If sieveing speed is increased 25x, then you need different parameters to sieve deeper. Also, where does he get these numbers? 10000 seems extremely small, 100 is 1%, so we are to believe the CPU sieve is 99% effective? The GPU sieve is 99.9% effective? There's a lot of questions that need to be asked here:
What is he using for h₀?
What is he using for c?
What is he using for e?
What is the first sieving prime?
What is he using for P_s?

I'm sorry, but those number he's tossing out look made up, with no proof to back them up.

oxfeeefeee

member

Activity: 73

Merit: 10

Quote from: bcp19 on August 03, 2013, 09:53:32 PM

Depends on speed. We know sieving is running about 25x, but how much of the CPU time was used for sieving and how much for fermat testing? Let x = CPU sieve time and y = CPU Fermat time. If x=y (50/50 split) then GPU = 2% for 52% of normal time = 92.3% speed increase. If x=9y (90/10 split) then GPU = 3.6% for 13.6% of normal time = 635% speed increase. if 9x=y (10/90 split) GPU = .4% fo 90.4% of normal time = 10.6% speed increase. These are extremely rough figures, communication times between GPU/CPU will also apply. Maybe he'll solve Fermat testing on the GPU and get a full 2500% speed increase. Short answer: too many current unknowns.

I did a similar calculation, but then Koooooj@reddit pointed out that I was wrong. Because with a much powerful sieve, CPU gets less test to do for fixed number of output.

for example (with made up numbers):
before: 10000Numbers ---sieve--> 100Numbers ---test---> 1Number
after : 10000Numbers ---sieve--> 10Numbers ---test---> 1Number

this way, even we don't speed up the execution of Fermat test, the test time is down to 10%.

redphlegm

sr. member

Activity: 246

Merit: 250

My spoon is too big!

Block reward and difficulty are recalculated every block but there is a limit on the speed at which reward changes happen. We saw this with the release (and subsequent quick release) of hpx. I personally expect that there will be a short-term flood of coins for perhaps the first week (with block reward honing in on the new difficulty level) and then an eventual stabilization and rise from the low. The question is, though, if current prices are actually rational. Any depression due to the release of GPU may translate to long-term effects simply because it "should" have been lower than .006x anyway.

The short answer is "nobody really knows but some people are probably going to make a good bit of dough on this".

bcp19

hero member

Activity: 532

Merit: 500

Quote from: LlamaMaster on August 03, 2013, 09:30:42 PM

I'm having a hard time figuring out which direction price is going to shoot when this GPU miner is released. Will the price tank because there will be an abundance of new coins on the market (which somebody will surely dump for a shitty low price), or will price increase with the new interest in the coin that this release will generate?

This is probably a crystal ball question, but I'm sure somebody in here has some insight. Either way, it would most likely be completely useless for me to donate if I only have one 7950, right?

Depends on speed. We know sieving is running about 25x, but how much of the CPU time was used for sieving and how much for fermat testing? Let x = CPU sieve time and y = CPU Fermat time. If x=y (50/50 split) then GPU = 2% for 52% of normal time = 92.3% speed increase. If x=9y (90/10 split) then GPU = 3.6% for 13.6% of normal time = 635% speed increase. if 9x=y (10/90 split) GPU = .4% fo 90.4% of normal time = 10.6% speed increase. These are extremely rough figures, communication times between GPU/CPU will also apply. Maybe he'll solve Fermat testing on the GPU and get a full 2500% speed increase. Short answer: too many current unknowns.

Dsfyu

member

Activity: 75

Merit: 10

Quote from: LlamaMaster on August 03, 2013, 09:30:42 PM

I'm having a hard time figuring out which direction price is going to shoot when this GPU miner is released. Will the price tank because there will be an abundance of new coins on the market (which somebody will surely dump for a shitty low price), or will price increase with the new interest in the coin that this release will generate?

This is probably a crystal ball question, but I'm sure somebody in here has some insight. Either way, it would most likely be completely useless for me to donate if I only have one 7950, right?

People are speculating that the price won't tank in the same way as expected since the difficulty is recalculated with every block. It also changes the reward with difficulty since the reward is 999/(difficulty)^2 or something like that. That would mean if the difficulty was 15 the reward would be something like 4.4/block instead of the 11.6 or so we are getting right now. At the same time there will be people that will dump for low prices on the exchanges just to get cash quickly which will really drive the prices down. But the way I see this right now a gpu miner would help the difficulty rise fairly quickly which would drop the block rewards and make each individual coin gain more value over time with a lower inflation rate. With that said, the inflation rate will probably temporarily rise when the gpu miner is released since the difficulty isn't able to catch up instantly, meaning there will be some excessively high inflation for a short period of time. It's honestly anybody's guess where it will go since it's a new market without a lot of support yet.

LlamaMaster

newbie

Activity: 51

Merit: 0

I'm having a hard time figuring out which direction price is going to shoot when this GPU miner is released. Will the price tank because there will be an abundance of new coins on the market (which somebody will surely dump for a shitty low price), or will price increase with the new interest in the coin that this release will generate?

This is probably a crystal ball question, but I'm sure somebody in here has some insight. Either way, it would most likely be completely useless for me to donate if I only have one 7950, right?

DonChate

member

Activity: 61

Merit: 10

Keep it coming. Looking forward to beta.

akspa

newbie

Activity: 53

Merit: 0

Here's my transaction id, for public transparency -
89f9d224090d4918cffcc8366bfcd1bb414313416a145cecf42b06c7ccd10c2b

Here's to hoping for the best Cheesy

litecoin_messiah

sr. member

Activity: 356

Merit: 268

Anyone who mines on GPU, will they PM me if they want to help out testing a new pool for gpu's which in the works.

I can't personally test because i can't donate

ReCat

sr. member

Activity: 406

Merit: 250

Quote from: hasle2 on August 03, 2013, 11:19:20 AM

Quote from: mtrlt on August 03, 2013, 08:57:08 AM

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.

Is there a chance you will be sending out the beta today then?

lol.

did you donate?

if not, forget it. You'll get the miner long after the difficulty has soared through the roof.

FreedomCoin

hero member

Activity: 675

Merit: 507

Freedom to choose

Quote from: mtrlt on August 03, 2013, 08:57:08 AM

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.

Thats great news!

hasle2

full member

Activity: 122

Merit: 100

Quote from: mtrlt on August 03, 2013, 08:57:08 AM

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.

Is there a chance you will be sending out the beta today then?

pyromaniac

hero member

Activity: 639

Merit: 500

Please, let us use it at least week, but better month, before it goes public. I'm one of the investors.

bcp19

hero member

Activity: 532

Merit: 500

Quote from: Lauda on August 03, 2013, 09:37:00 AM

Quote from: mtrlt on August 03, 2013, 08:57:08 AM

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.

Wow nice, at first people thought gpu's will have 2x or 3x performance at most Shocked

That's only the sieve though, the fermat testing is still on CPU (from his earlier reports). It will be interesting to see how it performs overall tho.

Lauda

legendary

Activity: 2674

Merit: 2965

Terminated.

Quote from: mtrlt on August 03, 2013, 08:57:08 AM

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.

Wow nice, at first people thought gpu's will have 2x or 3x performance at most Shocked

dabster

newbie

Activity: 16

Merit: 0

Looking forward to testing it out.

blarg

full member

Activity: 170

Merit: 100

cant wait to mine xpm!

techbytes

legendary

Activity: 1694

Merit: 1054

Point. Click. Blockchain

Quote from: mtrlt on August 03, 2013, 08:57:08 AM

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.

Sweet!!

-tb-

hathmill

full member

Activity: 186

Merit: 100

Quote from: mtrlt on August 03, 2013, 08:57:08 AM

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.

Thanks for doings this! If u did not spend time doing it, I would and I would probably have failed and wasted a lot of time! Sending 1 BTC to you was probably the best money spent soso far this year

Topic: [XPM] Working on a GPU miner for Primecoin - page 12. (Read 102805 times)