[ANN] MMCPool.com - GPU,CPU(yam) - The Awesomest Pool!

[glitchboy]

yvg1900, thank you!

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

Hey all, I'm happy to announce the release of the GPU miner for MMCPool.com!
Download now: http://dl.mmcpool.com/gpu/mmcpool-gpu-miner-windows.zip (Windows, BETA!)

Please, share your feedback.

[tomorrow]

M7j version of yam miner available. It supports fast CPU mining for PTS, MMC, more mining pools and protocols.

To mine MemoryCoin with yam M7j use provided miner config template named yam-mmc.cfg as a base, check readme.txt for more details.
yam M7j optimised for mining MMC on CPUs with AES-NI support and has 5 different algo variations for HT, non-HT and GPU friendly setups.

yam M7j miner now supports PTS mining at beeeeer and includes fix for Win2003 and other older 64bit systems. Check readme.txt for details.

Download at:
https://www.dropbox.com/sh/jvp4wwek8jpohj7/RlW6hzYqTz
https://mega.co.nz/#F!h0tkXSxZ!f62uoUXogkxQmP2xO8Ib-g
http://sdrv.ms/18wfjYX
http://pan.baidu.com/s/1xOOWa

yvg1900

A big WOW and a big THANKS!

[yvg1900]

Easiest way to see if this is real HPM number is to point one instance to separate address at mmcpool and look at HPM estimation.

SPM is Solutions Per Minute, which shall converge to the HPM on the long round.

Actually, HPM is not Hash Per Minute, but Hashing Round Per Minute.

HPM numbers are measured only over COMPLETE (non-interrupted) rounds, which is the WORST case. They change only when number of complete rounds change.

Anyway, from your statistics number of rounds is still very small.

yvg1900

[sceneboss]

MMC Agg. SPM: 7.278, HPM: 10.406; Rnds C/I: 41/16, Don. C/I: 15/0; Cfg/Thr SPM: 7.278/0.9757, Cfg/Thr HPM: 10.407/1.3009 56 rnds AV=1, ART=46121

is this really over 10HPM first cpu miner 3.4 HPM then 4.5 HPM and now over 10? what is happening in here

https://www.dropbox.com/s/pnsjzmxh6o35n0j/Screenshot%202014-01-12%2023.10.52.png why those HPM numbers are same all the time? what is SPM? what numbers i should watch now?

[yvg1900]

M7j version of yam miner available. It supports fast CPU mining for PTS, MMC, more mining pools and protocols.

Works it with stratum too?
I have pool, but no clients yet
But if M7j works, I don't must create it.

At the moment not, but I will check if I can support that fast.

Any technical specifics related you this particular pool?

Tbh, I prefer to have a reference miner that uses stratum protocol to see what you feel proper client side support, but I have stratum support on the list for other coins, so it can be in line with that.

yvg1900

[yvg1900]

FT,

There are several different algo variations.

av=1 is HT-friendly. On 4-core with HT you shall run it 8 threads and 1Gb RAM per thread.
av=3 is for non-HT. If you plan to run 4 threads on HT mach, ensure that you select av=3. The best is if you can pin your threads to cores to unlock full potential of av=3.
av=2 is mediocre. In between 1 and 3, but you may give it a try.
av=4 is for older AMD CPUs, ppl reported this was the best in some cases.
av=5 is targeting Atoms in 64-bit mode, has completely different mem usage. Very slow one, but still faster than original one.

As of GPUs, I will evaluate possibilities, but as soon as GPUs are in the game, I will definitely jump in to be on par with this.

This is equally probable that GPUs will have some bottlenecks, but we shall see.

There are still some more optimizations that are possible for CPU, I even did not collect enough statistics to optimize memory access patterns in full, just added some basic stuff. There is equal chance that doubling memory per thread may also bring performance gain, besides it looks completely strange.

yvg1900

[Atrides]

M7j version of yam miner available. It supports fast CPU mining for PTS, MMC, more mining pools and protocols.

Works it with stratum too?
I have pool, but no clients yet
But if M7j works, I don't must create it.

[FreeTrade]

Amazing - I've got an i7 3700 that was seeing 4 hpm at 100% with 1GB

Now, it's running with 6 cores - 75% with 6GB and it's nearly 9HPM.

I imagine that if switch off hyperthreading and run it with 4cores at 4GB I'll get the best results and upto 12HPM?

I'm baffled, amazed, grateful and awed. What did you do? How is this possible? Can the same optimization be made for GPUs or have you put CPUs back on an equal footing long term?

[yvg1900]

M7j version of yam miner available. It supports fast CPU mining for PTS, MMC, more mining pools and protocols.

To mine MemoryCoin with yam M7j use provided miner config template named yam-mmc.cfg as a base, check readme.txt for more details.
yam M7j optimised for mining MMC on CPUs with AES-NI support and has 5 different algo variations for HT, non-HT and GPU friendly setups.

yam M7j miner now supports PTS mining at beeeeer and includes fix for Win2003 and other older 64bit systems. Check readme.txt for details.

Download at:
https://www.dropbox.com/sh/jvp4wwek8jpohj7/RlW6hzYqTz
https://mega.co.nz/#F!h0tkXSxZ!f62uoUXogkxQmP2xO8Ib-g
http://sdrv.ms/18wfjYX
http://pan.baidu.com/s/1xOOWa

yvg1900

[earntodie]

Please, make auto payouts limit lower. 0.25 MMC.

[glitchboy]

*** ATTENTION

Maintenance.. The pool will be back soon. Sorry!

DONE!

[yvg1900]

Does that also mean that pool is strict on the value of nTime in mined header? (means if I want for example every mining thread to run its own work shall I obey rule of getting work from server for every individual thread, or I am free to select nonce/nTime on miner side?)

yvg1900

[glitchboy]

Your pool advertising X-Roll-NTime support in responses. Does it support it as per specification? Does X-Roll-NTime value of 500 seconds set with a purpose?

I am asking that because of block target is 6 minutes, thus 360 seconds, and setting up rollntime to 500 seem a bit too high for me. Am I wrong with this assumption?

Current implementation of MMCMiner doesn't use it. I forgot to remove it from the server's response. Removed.

[yvg1900]

Great to know about longpool, PM me please when you have it ready for testing.

Another technical question.

Your pool advertising X-Roll-NTime support in responses. Does it support it as per specification? Does X-Roll-NTime value of 500 seconds set with a purpose?

I am asking that because of block target is 6 minutes, thus 360 seconds, and setting up rollntime to 500 seem a bit too high for me. Am I wrong with this assumption?

yvg1900

[glitchboy]

Is longpoll supported on this pool?

Not yet. I'll implement it soon.