Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
I'm now running i7 6 core PC optimizing one of my trading systems. It takes whole night for one optimization. I checked, all 12 logical cores are used by NT7 when doing the optimization. CPU time up to 80%.
My question is: If I buy, say 2 processors, total 2*12 cores workstation, will it improve the speed? My guess is a 48 logical cores, should improve the optimization speed 4 times comparing to my current 12 logical cores machine. However, I'm a little afraid that NT cannot fully utilize the 2 processors, maybe it sill uses only 12 logic cores. So before I buy the machine, I hope someone here can confirm me that NT7 can fully take advantage of the multiprocessors machine when doing the optimization.
Thanks.
Can you help answer these questions from other members on NexusFi?
Thank you for the answer. But I only want to improve the optimization speed, not the charting or trading speed. According to you test here: #65 of the link you give me (I cannot post link yet), it seems that NT7 does use all 48 cores when doing the optimization. Unless I misunderstood your screenshot, it's not from NT7 test, it's the other software?
OK, I tested via Amazon Cloud, it's 36 cores. The screenshot shows clearly NT7 can use all the cores, well, at least 36 cores.
It didn't take 100% for all cores simply because the more threads, the more syn is needed, so if design the trading system well, i.e. avoid calling outside indicators, making all simple indicators, such as SMA, EMA, MACD inline, then my guess is it'd take more CPU time.
Why not test it on the MACrossOver strategy included. It doesn't get much simpler. I believe you'll find the results disappointing with your 36 core test. What is the clock speed of these cores?
Like I mentioned earlier, everybody forgets the RAM/MMU/cache performance issues and would rather ignore them. Cores are no more of a Holy Grail for OO system performance than anything is in trading.
The only thing the task manager shows you is that there are active threads in relation to the system idle thread. It tells you nothing about the state of the cpu core on behalf of the active thread and it tells you nothing about the utility or quality of work being achieved. [Assuming the Cloud machine was not shared of course.]
Given that OS/Ninja OO/C#/etcetera are fixed, for these cases the only measure you need be interested in is elapsed time.
I will be looking for improvements in NT8 but not miracles.
Not quite, the same applies to any general object-oriented architecture and language.
You can take all cores to 100% with multi-threaded software under NT7 as it stands, but it depends what the threads are doing. If they need random access to OO structures then the MMU caches will be busted nearly all the time, whereas if they only need sequential/local access then the cores can fly without stalling.