A computer for testing and optimising the robot (Expert Advisor). - page 6
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
Optimisation on VPS - this is already from the category of perversions. Extremely low computing power per core, little RAM and haemorrhoids with configuration every time. In comparison even with laptops on 12800HX and Ryzen 9 7845HX at TDP 100 - 120 W per processor - overtake the standard VPS in several times and about one and a half times Xeon E-2336. Normal Dedicated server can be bought for 50 - 100 euros per month.
Oh and bollocks.... About Risen and Intel and the rest of the details.
What kind of overclocking?? It's the opposite of stifling.
What lasso process?
And who will wait for whom? What cores are lagging behind? Never write here again. storyteller.
it looks like they tested everything in their sleep))) and also "We", mde.
Oh and bollocks.... About Risen and Intel and in other details
What kind of overclocking?? It's the opposite of stifling.
What's the lasso process?
So who's waiting for who? What cores are lagging behind? Don't ever write here again. storyteller
in my sleep it looks like they tested everything))) and "We" too, mde.
Tested at a real 12800. At a minimum, the P and E cores have different frequencies. At most - three times fewer ports, fewer blocks on each port, wildly reduced instruction queue, no branch predictor (I don't know, maybe in 13). And you have in your picture, one P kernel outputs in 2 threads as much as 2 E kernels. So E cores are 2x slower than P cores. At a minimum, a 12 core ryzen will be equivalent to yours (actually faster, tested), and a 16 core is significantly faster. In case, with a run on a big story - meta under 10 - 15 threads will eat up all the RAM and on 12 - 16 core ryzen, each thread will have its own core. In your case, there will be 8 P cores and 2 - 7 E cores. Don't forget that if you leave 2 - 4 cores for work, Windows will immediately transfer these threads to small cores.
Now think about what latency is and what it is measured in. 20-20-20 at 3000 = 40-40-40 at 6000. That is, their latency in nanoseconds is the same and varies slightly from die to die. But the fact that the frame is divided into 2 ensures that your memory requests from each core will be more even (and memory access is a favourite of E cores). That is, DDR5 is at least not worse, and because of the bandwidth - in computational tasks, will give 10 - 15% increase.
What kind of overclocking? It's the opposite of stifling.
You want to strangle them because Intel could not get into the process technology and Intel processors with the same performance eat 1.5 times more. ) Plus, they can only use the whole processor in undervolting. If at least one core has lost the silicon lottery - there will be no undervolting. In razens you can do undervolting by cores and cool down the processor considerably. That said, I'm not a fan of razen. The 12 series was very good at the time of release, but with the release of 7xxxxx ryzen and real test - for me the choice has shifted towards the red ones.
I suggest to make a measurement, take a robot with codebase, if there is a wattmeter, it will be great in general
If I understand the problem correctly, then for the described effect of the younger cores you need to give up full brute force in favour of GA.
If I understand the problem correctly, for the described effect of the younger kernels, we need to abandon full brute force in favour of GA.
There are no problems with small cores in the test, the speed of a small core is equal to one thread of a large one, so the test always ends at the same time
A couple of hours ago I ran a half an hour test almost, all 24 threads passed exactly the same, small cores act as whole cores, they are not multi-threaded, so they are equal to one thread of a large core 100% plus minus.Optimisation on a VPS is already from the category of perversions. Extremely low processing power per core, little RAM and haemorrhoid with configuration every time. In comparison even with laptops on 12800HX and Ryzen 9 7845HX at TDP 100 - 120 W per processor - overtake the standard VPS in several times and about one and a half times Xeon E-2336. Normal Dedicated server can be bought for 50 - 100 euros per month.
Configuration is described above. Speed as on the dedik. You just need to find a normal VPN. I just outlined the method. If you like it, invest in the latest processors and memory. But if you buy the latest model of processor and memory and a shitload of memory just for optimisations, it's a senseless waste of money.