Evaluating CPU cores for optimisation - page 13
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Will it parallelize in the strategy tester? I.e. can I run N optimization threads on OpenCL on one CPU core?
I've got Ryzen 5 2600 now, so I'm thinking of getting something more powerful.
Memory 4 x 8GB CORSAIR Vengeance 3000
Main drive is NVMe Samsung 970 500GB, but I bought a 1Gb SSD with SATA-3 for multi-threaded tests, as my tester has been rapping it to death every time.
In the tester, the EA's internal logic can parallelise on OpenCL, but not the walkthroughs on the history.
Can someone dare to do strategy testing in OpenCL :) ?
You build your own tester for OpenCL and then make it possible to execute strategies in some format. It may be simplified, but it will be quick...
Can someone dare to do strategy testing in OpenCL :) ?
Make your own tester for OpenCL, then make it possible to execute strategies in some format. It may be simplified, but it will be quick...
You will have to rewrite all the logic of working with orders, balances and other things, in fact, the terminal and part of the server should be rewritten))))
You will have to rewrite all the logic of working with orders, balances and other things, in fact the terminal and part of the server will have to be re-written))))
There will be some brave people)).
There will be some brave people).
Everything is determined by the need... Only if a very complicated task arises, that it will be necessary to parallel testing. In this case the terminal and strategy tester won't be needed, the quotes history + TensorFlow will be enough).
I haven't encountered such problems so far. The last time I optimized a pass N times over with caching and hashmaps. It turned out to be much easier and faster
Tested Ryzen 3700x (frequency in turbo mode 4239 according to AIDA64; memory ddr4 3200 16 x 1) . Logs are attached. Immediately estimated what the table rows should look like
Ryzen 3700x 4239 8 8 8 39.15 60.42 18.13 735.63 476.66 1588.53
The average lag from Ryzen 3800x in six tests turns out to be 3.6%. (Scatter from 9.2% to -1.9%). True, the tester with the Ryzen 3800x has the memory running in dual channel mode, while I have a single card so far. I don't know if that can make a difference in this case. I can re-test it a bit later.
Andrey Pogoreltsev:
And I would also like to be able to write part of the EA in CUDA and then parallelize it on video cards, because there is a much more interesting perspective.
And if you remember that Google recently announced the achievement of quantum superiority...
By the way, why is the topic-starter's nickname crossed out in his posts? Is he no longer with us?
Can someone dare to do strategy testing in OpenCL :) ?
Make your own tester for OpenCL, then make it possible to execute strategies in some format. It may be simplified, but it will be quick...
You don't need to be brave; you just need to want and, to a lesser extent, need))
I did it, I got almost 4 times speedup on CPU (4 cores in CPU)... The GPU didn't get that kind of speedup - the data transfer costs eat up all the advantage of a huge number of cores.
Tested Ryzen 3700x (frequency in turbo mode 4239 according to AIDA64; memory ddr4 3200 16 x 1) . Logs are attached. Immediately estimated what the table rows should look like
Ryzen 3700x 4239 8 8 8 39.15 60.42 18.13 735.63 476.66 1588.53
The average lag from Ryzen 3800x in six tests turns out to be 3.6%. (Scatter from 9.2% to -1.9%). True, the tester with the Ryzen 3800x has the memory running in dual channel mode, while I have a single card so far. I don't know if that can make a difference in this case. I can re-test a bit later.
Thanks for the tests. I have compiled a table - it turns out that there is very little difference between Ryzen 3700x and Ryzen 3800x