Large Bitcoin Collider (Collision Finders Pool) - page 21.

rico666

legendary

Activity: 1120

Merit: 1037

฿ → ∞

Quote from: Vyrez on April 04, 2017, 07:28:56 AM

Pleiades Prototype alpha

Is this the marking of the KNL or another form of "cluster"?

Hmmmm?

Just yesterday I have thrown the pleiades source code from my Sandbox (I have it still in the repository though), because
with the arulbero-EC arithmetics and my GPU hash160 implementation we have something better.

Having said that, both the KNL as well as the Skylake-EX with their AVX512 engines are a hell of a machine for EC arithmetics.
If anyone has these processors, I can provide a generator binary.
So if anyone is looking for a dedicated machine with quite some key generation performance:

KNL + GPUs (probably too expensive) or Skylake-EX + GPUs (less expensive, but still some bucks)

With these we're getting in the region of 100s of Mkeys/s per machine.

Rico

Vyrez

newbie

Activity: 14

Merit: 0

Pleiades Prototype alpha

Is this the marking of the KNL or another form of "cluster"?

Hmmmm?

arulbero

legendary

Activity: 1932

Merit: 2077

Quote from: rico666 on April 04, 2017, 12:29:07 AM

Please always remember

LBC: 100 Mkeys/s = 200 Maddresses/s
oclvanitygen: 270 Mkeys/s = 270 Maddresses/s (only compressed)

Rico

I think that if oclvanitygen computes 270 Maddress/s, that means that it computes only 135M of x-coordinates (x = coordinates of points of the curve that represent the public keys) and not even a single y-coordinate.
Then the strings: "02x" and "03x" produce 2 addresses ( 2 for each x):

"02x" --> compressed public key
"03x" --> compressed public key

Current LBC generator computes 100 Mkey/s, that means that it computes 100M of x-coordinates + 100M of y-coordinates, then

"04xy" --> uncompressed public key
"02 or 03x" --> compressed public key

Total: 200 Maddresses/s.

Potentially our generator could computes other 2 addresses (not in the same time) from the same x/y, because there are other 2 points related to the same x,y coordinates:

"04x(-y)" --> uncompressed public key
"03x / 02x" --> compressed public key

Note: sha256 should take half time if only applied to strings like "02x" or "03x" (only 256bit) respect of "04xy" (more than 512 bit).
That's why oclvanitygen seems faster than LBC.

Essentially the cost of compute an uncompressed public key is higher than that related to compressed key (because of y and because of sha256)

hodlcoins

legendary

Activity: 1100

Merit: 1058

Just out of interest: what are your costs per day (or so)?

unknownhostname

member

Activity: 62

Merit: 10

Quote from: rico666 on April 03, 2017, 01:33:13 PM

In the past hour we were 1107156445 keys/s
(at this speed we will have #51 in 2 days)

Rico

Speed is even higher right now ...

Countdown to #51 ...

got more Gkeys generated than the whole pool combined

SlarkBoy

member

Activity: 114

Merit: 11

Quote from: rico666 on April 04, 2017, 12:29:07 AM

Very nice.
Please always remember

LBC: 100 Mkeys/s = 200 Maddresses/s
oclvanitygen: 270 Mkeys/s = 270 Maddresses/s (only compressed)

edit:

theoretically, you could distribute the load to your GPUs 3x 10 (or 10+11+11) and let one do other mining stuff.
4 x 64% = ~ 256% -> 144% spare capacity so one full GPU for sure.

I thought about exactly such a configuration (4 x 1080, but Skylake-EX core to feed the GPUs, should come close to 200 Mkeys/s)

Rico

Yeah. But I'm not interested with mining. That's why I came here Cheesy

We will found bounty or giveaway in a few days.

Janu$$

member

Activity: 86

Merit: 10

Quote from: rico666 on April 03, 2017, 02:54:49 PM

...
First off, welcome in the top30. Your GPUauth has been delivered.

...

Thx a lot!

I will use my Radeon GPU as soon as a Windows client or a LBC-Appliance with Radeon GPU support is released.

Have a nice day,
Janu$$

rico666

legendary

Activity: 1120

Merit: 1037

฿ → ∞

Holy Sh!t - if you look at the "Notable Dates" section of https://lbc.cryptoguru.org/about
you'll see the pool needed about 4 months to crunch its first 100tn keys.

The more we are into #51 search space (only 270 MBlocks to go), the more nervous I get. #51 is a mean bitch for sure - but should we get to the end of the search space without a hit I will have to shoot myself.

On the other hand - that's how I felt with #49 too. Cheesy

Rico

rico666

legendary

Activity: 1120

Merit: 1037

฿ → ∞

Quote from: SlarkBoy on April 03, 2017, 05:43:33 PM

test on new machine.
...

total 100 Mkeys/s
oclvanitygen speed around 270 Mkeys/s

Very nice.
Please always remember

LBC: 100 Mkeys/s = 200 Maddresses/s
oclvanitygen: 270 Mkeys/s = 270 Maddresses/s (only compressed)

edit:

theoretically, you could distribute the load to your GPUs 3x 10 (or 10+11+11) and let one do other mining stuff.
4 x 64% = ~ 256% -> 144% spare capacity so one full GPU for sure.

I thought about exactly such a configuration (4 x 1080, but Skylake-EX core to feed the GPUs, should come close to 200 Mkeys/s)

Rico

SlarkBoy

member

Activity: 114

Merit: 11

test on new machine.

Code:

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 375.39                 Driver Version: 375.39                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 1080    Off  | 0000:02:00.0     Off |                  N/A |
| 30%   45C    P2    59W / 180W |   5036MiB /  8114MiB |     63%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX 1080    Off  | 0000:03:00.0     Off |                  N/A |
| 27%   40C    P2    60W / 180W |   5036MiB /  8114MiB |     63%      Default |
+-------------------------------+----------------------+----------------------+
|   2  GeForce GTX 1080    Off  | 0000:81:00.0     Off |                  N/A |
| 29%   44C    P2    59W / 180W |   5036MiB /  8114MiB |     62%      Default |
+-------------------------------+----------------------+----------------------+
|   3  GeForce GTX 1080    Off  | 0000:82:00.0     Off |                  N/A |
| 27%   35C    P2    58W / 180W |   5036MiB /  8114MiB |     63%      Default |
+-------------------------------+----------------------+----------------------+

Architecture:          x86_64
CPU op-mode(s):        32-bit, 64-bit
Byte Order:            Little Endian
CPU(s):                32
On-line CPU(s) list:   0-31
Thread(s) per core:    2
Core(s) per socket:    8
Socket(s):             2
NUMA node(s):          2
Vendor ID:             GenuineIntel
CPU family:            6
Model:                 79
Model name:            Intel(R) Xeon(R) CPU E5-2667 v4 @ 3.20GHz
Stepping:              1
CPU MHz:               3499.875
CPU max MHz:           3600.0000
CPU min MHz:           1200.0000
BogoMIPS:              6401.29
Virtualization:        VT-x
L1d cache:             32K
L1i cache:             32K
L2 cache:              256K
L3 cache:              25600K
NUMA node0 CPU(s):     0-7,16-23
NUMA node1 CPU(s):     8-15,24-31

total 100 Mkeys/s
oclvanitygen speed around 270 Mkeys/s

rico666

legendary

Activity: 1120

Merit: 1037

฿ → ∞

Quote from: Janu$$ on April 03, 2017, 02:23:01 PM

what about https://www.nimbix.net/jarvice/ ?
Allegedly with Pascal GPU support:
"...Nimbix is the only public cloud provider featuring NVIDIA’s latest generation Pascal Tesla P100 GPUs with NVLink, a high-bandwidth, energy-efficient interconnect that allows data sharing at rates 5 to 12 times faster than the traditional PCIe interconnects...."

First off, welcome in the top30. Your GPUauth has been delivered.

Nimbix looks like similar business to the https://www.leadergpu.com/ link arulbero posted. I wasn't aware of all the "smaller" cloud providers (or let's say: hardware-on-demand) and with these offering newer GPU systems it is my hope that Amazon and Google will not take much longer to offer these too.

LBC doesn't need any fast interconnect between the GPUs it's perfectly parallelizable without any such inter-GPU communication, so NVlink is not a must.

Rico

Janu$$

member

Activity: 86

Merit: 10

Quote from: rico666 on April 03, 2017, 11:16:12 AM

Quote from: unknownhostname on April 03, 2017, 11:07:19 AM

ofc not ... lots and lots of CPU's ... no GPU for me Sad

Don't take it personally Wink

It's more like no (performant) client for K80. (yet)

I had a look if some cloud vendor offers any Maxwell (M60) or Pascal GPUs (P100). I'm pretty sure with these LBC would work and it would even mean a nice performance kick. Unfortunately so far only promises and plans:

https://blogs.nvidia.com/blog/2016/11/16/tesla-p100-on-google-cloud-platform/
observe how "now available" in the subject = "will be available" in the 1st line of the text

I checked today. Nada. Only K80 and only in some locations. Roll Eyes

I also am thinking about purchasing a dedicated 200 Mkeys/s machine...

Rico

Hi rico,
what about https://www.nimbix.net/jarvice/ ?
Allegedly with Pascal GPU support:
"...Nimbix is the only public cloud provider featuring NVIDIA’s latest generation Pascal Tesla P100 GPUs with NVLink, a high-bandwidth, energy-efficient interconnect that allows data sharing at rates 5 to 12 times faster than the traditional PCIe interconnects...."

Regards,
Janu$$

arulbero

legendary

Activity: 1932

Merit: 2077

Quote from: unknownhostname on April 03, 2017, 11:07:19 AM

Quote from: arulbero on April 03, 2017, 09:16:40 AM

Quote from: unknownhostname on April 03, 2017, 08:35:48 AM

1.25 Gkeys/sec

How? GPU-client on dedicated servers?

ofc not ... lots and lots of CPU's ... no GPU for me Sad

Could you share with us some informations about how do you got this speed? Is it expensive?

This pool could reach a very high speed if it were more people able to use a lot of CPU like you do. Wink

rico666

legendary

Activity: 1120

Merit: 1037

฿ → ∞

Quote from: unknownhostname on April 03, 2017, 12:19:32 PM

whats the current speed rico ?

Not the 2 day average?

In the past hour we were 1107156445 keys/s
(at this speed we will have #51 in 2 days)

Rico

unknownhostname

member

Activity: 62

Merit: 10

whats the current speed rico ?

Not the 2 day average?

arulbero

legendary

Activity: 1932

Merit: 2077

Something like that?

https://www.leadergpu.com/

rico666

legendary

Activity: 1120

Merit: 1037

฿ → ∞

Quote from: unknownhostname on April 03, 2017, 11:07:19 AM

ofc not ... lots and lots of CPU's ... no GPU for me Sad

Don't take it personally Wink

It's more like no (performant) client for K80. (yet)

I had a look if some cloud vendor offers any Maxwell (M60) or Pascal GPUs (P100). I'm pretty sure with these LBC would work and it would even mean a nice performance kick. Unfortunately so far only promises and plans:

https://blogs.nvidia.com/blog/2016/11/16/tesla-p100-on-google-cloud-platform/
observe how "now available" in the subject = "will be available" in the 1st line of the text

I checked today. Nada. Only K80 and only in some locations. Roll Eyes

I also am thinking about purchasing a dedicated 200 Mkeys/s machine...

Rico

unknownhostname

member

Activity: 62

Merit: 10

Quote from: arulbero on April 03, 2017, 09:16:40 AM

Quote from: unknownhostname on April 03, 2017, 08:35:48 AM

1.25 Gkeys/sec

How? GPU-client on dedicated servers?

ofc not ... lots and lots of CPU's ... no GPU for me Sad

arulbero

legendary

Activity: 1932

Merit: 2077

Quote from: unknownhostname on April 03, 2017, 08:35:48 AM

1.25 Gkeys/sec

How? GPU-client on dedicated servers?

unknownhostname

member

Activity: 62

Merit: 10

1.25 Gkeys/sec

Topic: Large Bitcoin Collider (Collision Finders Pool) - page 21. (Read 193420 times)