MQL5 Cloud Network: Are You Still Calculating? - page 5

 
The logs show that all the work in a million passes was completed and only one of the agents (Pr 141) was generating an Expert Advisor initialisation error. It could not even load the Expert Advisor.

We have specially expanded error messages to find problematic agents. Step by step we will get rid of them.
Документация по MQL5: Стандартные константы, перечисления и структуры / Коды ошибок и предупреждений / Ошибки компиляции
Документация по MQL5: Стандартные константы, перечисления и структуры / Коды ошибок и предупреждений / Ошибки компиляции
  • www.mql5.com
Стандартные константы, перечисления и структуры / Коды ошибок и предупреждений / Ошибки компиляции - Документация по MQL5
 
phi.nuts:

Since we are in trading business, maybe you've heard the term : Supply and Demand.

Maybe it's too much supply in agents and less demand for cloud.

how much commission is MQ taking from the supplier?
 
doshur:
how much commission is MQ taking from the supplier?
Read the FAQ : currently 10 %.
Questions Concerning Payment for Participation in the MQL5 Cloud Network
Questions Concerning Payment for Participation in the MQL5 Cloud Network
  • cloud.mql5.com
Questions concerning payment for participation in the MQL5 Cloud Network - distributed computing network
 

After upgrading to build 756 the login and accordingly the"sell resources" tick box was reset.

It turned out that when trying to restart services they do not restart....

I had to kill processes through Task Manager and then only restarting happened.


OpenCL also remains a problem:

NL 0 Startup 14:47:50 access rights to common directory successfully checked<br/ translate="no">PF 0 Startup 14:47:50 Service start initialized
RL 0 Startup 14:47:50 Create startup thread
FR 0 Startup 14:47:50 Thread successfully created
CE 0 Startup 14:47:50 MetaTester 5 x64 build 756 (25 Jan 2013)
JP 0 Server 14:47:50 MetaTester 5 started on 0.0.0.0:2001
GE 0 OpenCL 14:47:50 initialize
PD 2 Agent 14:47:50 OpenCL initialisation skipped due to critical error. Please update video drivers.
EK 2 Agent 14:47:50 OpenCL will be initialised during terminal's next launch.

DO 0 Startup 14:47:50 initialisation finished
NK 0 Startup 14:47:50 Startup thread finished successfully
LO 0 Network 14:49:50 Network 14:49:50 connected to 1.agents.mql5.com
 
LiveWizard:

After upgrading to build 756 the login and accordingly the "sell resources" tick box was reset.

It turned out that when trying to restart services they do not restart....

I had to kill processes through Task Manager and then only restarting happened.


There is also a problem with OpenCL:

OpenCL is not available on regular graphics cards in services mode (agents installed as services). This is a conceptual problem of the Windows driver model.
 

In the next build we will release improved support for OpenCL when working in service modes (we fixed some of our bugs), open access to statistics of OpenCL devices in the cloud and make it possible to actually use OpenCL in the network.

We won't predict the results yet, but we'll see them after the release.

OpenCL: Мост в параллельные миры
OpenCL: Мост в параллельные миры
  • 2012.05.16
  • Sceptic Philozoff
  • www.mql5.com
В конце января 2012 года компания-разработчик терминала MetaTrader 5 анонсировала нативную поддержку OpenCL в MQL5. В статье на конкретном примере изложены основы программирования на OpenCL в среде MQL5 и приведены несколько примеров "наивной" оптимизации программы по быстродействию.
 

Here is how OpenCL startup and initialisation looks like in service agents on the example of NVIDIA Tesla:

ML      0       Startup 07:18:59        access rights to common directory successfully checked
OF      0       Startup 07:18:59        Service start initialized
QL      0       Startup 07:18:59        Create startup thread
ER      0       Startup 07:18:59        Thread successfully created
DE      0       Startup 07:18:59        MetaTester 5 x64 build 756 (25 Jan 2013)
JP      0       Server  07:18:59        MetaTester 5 started on 0.0.0.0:2000
RK      0       OpenCL  07:18:59        Device #0: GPU NVIDIA Corporation Tesla T10 Processor with OpenCL 1.0 (30 units, 1440 MHz, 4095 Mb, version 307.45)
QJ      0       Startup 07:18:59        initialization finished
IO      0       Startup 07:18:59        Startup thread finished successfully
IK      0       192.168.1.132   07:19:12        login (build 756)
MG      0       Network 07:19:12        38680 bytes of group info loaded
NN      0       Network 07:19:12        1490 bytes of tester parameters loaded
CG      0       Network 07:19:12        708 bytes of input parameters loaded
EM      0       Network 07:19:12        817 bytes of symbols list loaded
MK      0       Tester  07:19:12        expert file added: Experts\OpenCLExpert.ex5. 8322 bytes loaded
MR      0       Network 07:19:12        224 bytes of optimized inputs info loaded
GG      0       Tester  07:19:12        successfully initialized
CQ      0       Network 07:19:12        22 Kb of total initialization data received
DJ      0       Tester  07:19:12        Intel Core i7-2600  @ 3.40 GHz, 16366 MB
CM      0       Tester  07:19:12        optimization pass 1 started
LD      0       Symbols 07:19:12        EURUSD: symbol to be synchronized
HO      0       Symbols 07:19:12        EURUSD: symbol synchronized, 3304 bytes of symbol info received
NJ      0       History 07:19:12        EURUSD: load 15 Kb of history data to synchronize
FR      0       History 07:19:12        EURUSD: history synchronized from 2012.01.02 to 2013.01.29
KP      0       History 07:19:12        EURUSD,H1: contains 6210 bars of beginning data from 2012.01.02 00:00 to 2012.12.31 19:00
DD      0       History 07:19:12        EURUSD,H1: history cache reserved for estimated 6754 bars
LE      0       History 07:19:12        EURUSD,H1: history begins from 2012.01.02 00:00
HM      0       Tester  07:19:12        1 : passed within 297 ms
KF      0       Tester  07:19:12        273 Mb memory used
Everyone will have it working from the next build.
 
Renat:

Here is how OpenCL startup and initialisation looks like in service agents on the example of NVIDIA Tesla:

Everyone will have it working from the next build.
Everyone == on GPUs dedicated to computing, or really everyone?
 
felidae:
Do all == on GPUs dedicated to computing, or indeed all?

Those GPUs that provide their API outside of video drivers. These are specialised compute cards like Tesla.

Some drivers (AMD/ATI) give you the ability to use CPUs as OpenCL devices. We are currently working on enabling them when working in agents. The performance of such devices may be low, but the main thing is that we will get wide availability of OpenCL.

 
Renat:

Those GPUs that provide their API outside of video drivers. These are specialised computing cards like Tesla.

They have no future for average users right now. I mean, at all.

Let's compare. Performance for double

7970 -- 0.95 TFlops

k20x -- 1.3 TFlops.

Now pay attention to the price )))

7970 -- 400

k20 -- 3200 (!) k20x costs unreal money.

Apparently your decision to organise agents as services was wrong.