Search Results for 'gpu'

[Ilya]

Hi Loonly.

 

I have 2 xeons of 8 cores (16 threads) each with 128GB RAM.

Yep I reserve one core for GUI, and I kept the GPU acceleration on (Honestly I didn’t even notice this setting).

Will try disabling it now and see how goes.

 

Ilya

[loonly]

hi Ilya,

i have got something similar, re-testing about 15K strategies.

i will try to re-download the SQ117 version and give a shot again.

what is your hardware config and sqx config? (all cores/reserved 1 gor GUI, GPU acceleration disabled?)

 

[ls3400219]

I found that the font on the interface of sq is very unfriendly and looks very vague. It feels that sq defines the various elements of the interface instead of using the definition of the system itself. I searched the forum and found that some people mentioned this before. It seems to be related to gpu, but the current version does not have this option, I hope the next update can solve this problem, thank you

[Fernando Morales]

Having to run multiple instances of SQ requires wasting lots of RAM for the GUI and program data. I find more effient the Metaquotes’ approach with MT5 using “tester agents” to distribute the calculations. If SQ were to have such agents it would easier to use computers in a LAN, or even GPUs

[Globegains]

Hi,

thanks for download page.

Yes 111 was working without problems. 111->112 was updated by updater not clean install if that helps.

My rig is Hyper-V windows 2016 without GPU card (SQX GPU UI disabled). Tried various combinations of settings (all variations). If you need any more data I will provide.

[mabi]

Turning of accelerate using GPU solved the problems.  Maybe some text should be added recomending turning it of if computor/VPS is remote controlled by third party software. (any)

[Mark Fric]

yourrodstaff wrote:

Mark, when the algo wizard become usable ? right now it can’t load the old file. hope next release?

yes, we are working on AlgoWizard, it should be much better in the new release

 

mabi wrote:

It seems SQx gets buggy if You remote control the workstation it is on. It is a lot of info on java problems under remote control on the web. Normally the java windows do not repaint and program hang. I had this problem for years with SQ3. It could not be remote controlled if computor had HDMI out but worked fine if it did not. Unfourtunately certain functions in SQX loads java windows and this seems to have other buggy effect on the program. Some programmers have a solution to this problem with this code -Dsun.java2d.d3d=false but i do not know if is applicable in case of SQx. One install of mine with SQx 108 is now creating alot of wierd bugs but another install that do not have HDMI out connection works fine in remote. This can be one reason to why some people are experience certain bugs and some are not.

what do you mean by buggy? have you opened a task about it?

It seems you are referring to some graphics problems when using remote desktop, I haven’t experienced it.

You can try to disable GPU rendering in Performance -> Browser tab, or try to add the following line to your StrategyQuantX.config file:
option -Dsun.java2d.d3d=false

But perhaps better way is to use SQ X running on VPS is through browser, just use the IP of your VPS and port 8080. If the port isn’t blocked you should be able to work with it.

Mark
StrategyQuant architect

[RNG]

The implementation over linux is a great idea, is written in java so i don’t think it’s very difficult recompile all for linux, anyway i still don’t understand why use this heavy slower UI, i think a lot of user (me included) run this software over dedicated server, where the gpu is something integrated with very slow power…
Can you implement a parallel version with a light GUI, not managed from chromium or something like that? It would be wonderful. Thanks for all your time and work.

[jmtc1230]

Mark for me I see that this is really a GPU or no GPU conversation. Because to buy a Tesla P 100 it would cost around six or $7000  where is to rent one from Google would cost around say eight dollars per hour.   This would offer around 300 times more processing power than our CPUs can do and I would plan to for small blocks of time. Imagine getting 300 days of processing done in 1 day

If we are talking about staying only in a CPU environment only then I agree with you it is not worth it  and would be better to build our own systems.

[jmtc1230]

JoJo, it could be but it does not have to be collective. It’s really just renting a virtual machine from Google with up to 96 Cores in it, and a Tesla P 100 GPU to be able to develop at very high speed with a windows environment.

[Mark Fric]

I said it few times already, the type of “real” trading engine that is implemented in new SQ4 is not suitable for computation using GPU.

But we are planning to make a simpler backtesting engine that can run on GPU, mainly for the purpose of training Neural Networks, or for very fats backtests using Selected Timeframe precision. So it is in middle or long term plan, but the exact architecture hasn’t been yet designed.

There are more important features for us to finish asap.

Mark
StrategyQuant architect

[Jojo]

jmtc1230 wrote:

Mark I know this has come up before but I just feel like I want to say it again. The idea of strategy quant being able to run on a GPU is probably our collectively largest possible gain. The nvidia P100 and V100 processors are literally hundreds of times faster than what we are all running. I just spent some time looking into google cloud processing and if this was possible we could all do batch processing for a few hours at a cost of around $8 per hour. This would be HUGE development!!!! What are your thought on this?

Hi jmtc1230, What do you mean “we could all” run batch processing at $8. Would it mean we share the cost by pooling our strats together for the purpose of processing?; have different logins, keeping our strats separate? I guess some of us have a lot more than others to process – how would we determine access/use/proportional cost? Sounds like a good idea – if only I can understand it better.

[jmtc1230]

Mark

I know this has come up before but I just feel like I want to say it again. The idea of strategy quant being able to run on a GPU is probably our collectively largest possible gain. The nvidia P100 and V100 processors are literally hundreds of times faster than what we are all running.

I just spent some time looking into google cloud processing and if this was possible we could all do batch processing for a few hours at a cost of around $8 per hour.  This would be HUGE development!!!!

What are your thought on this?

[mabi]

Testing different exits only on a strategy should be possible using GPU instead of CPU. Just generate a file with 1 billion predefined exit values and let a bunch of GPU’s figure it out..

[Mark Fric]

I’m sorry for a big number of issues in the last update. We had to make a lot of changes to support translation to other languages, and a lot of these problems were caused by this.

I’m thinking that maybe we’ll make a release of RC5 this Friday, to fix the most pressing issues.

 

Roberto wrote:

I can’t see in the roadmap the export of the strategies in easy language. Will sq4 support it? Thank you

yes, we will, but for the first release we focus only on MetaTarder 4/5. Support for EasyLanguage should be added within a month or so.

 

Dale Henry wrote:

custom indicators and GPU

this has been discussed few times, GPU cannot really be used if we want to allow users to programm their own indicators and other things and extend SQ4 with their own ideas. There still might be some possibilities how to use GPU in SQ4, we will focus on it when we have nothing more important to do.

Custom indicators – my original idea was that they will not be supported, because in SQ4 you will be able to programm your own indicator that doesn’t have any disadvantages of indicators defined only by importing data. But maybe we’ll add it, custom indicators in this sense will be just another data input, like an additional chart.

 

Jorn Conney wrote:

I think it will be great to have a feature in strategyquant to backtest compounding of profits…

it is already there, it is called Money Management. You can choose a different type of money management instead of fixed lots, and you can also write your own Money Management function.

 

Using CodeEditor and extending SQ4 with your own indicators, other blocks, money managements etc. is still not working 100%, but it will in the new version and we’ll prepare a lot of examples.

 

One more thing – please use Roadmap (https://roadmap.strategyquant.com) to submit your ideas or new feature requests, that is the right place to do it. It will be lost if you post it only on forum.

Mark
StrategyQuant architect