MetaTrader 5 on Intel Xeon Phi 7250 - 272 cores in one computer - page 7
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
Thanks for the review, it's all nicely described
for my part, these things are used in supercomputers... and they know what to put in there
the performance of such a thing is definitely better than a xeon server
of course, you can put something like HP superdome X - but the price is beyond the mind of the universe
when optimising EAs, it is not frequency that matters but number of threads ....
it is also true that the higher the frequency and the more cores the faster the calculation
https://aws.amazon.com/ru/ec2/spot/pricing/
The prices are more accurate in the LC. Payment is by the hour. You pay for your server only when you really need it, not for the whole month.
Configuration often varies by region.https://aws.amazon.com/ru/ec2/spot/pricing/
The prices are more accurate in the LC. Payment is by the hour. You pay for your server only when you really need it, not for the whole month.
Depending on the region, packages change often.how many hours is it? i only got a test one year on amazon
https://aws.amazon.com/ru/ec2/spot/pricing/
The prices are more accurate in the LC. Payment is by the hour. You pay for your server only when you really need it, not for the whole month.
Depending on the region, packages often vary.These are spot instances.
Spot instances are recommended:
They are not suitable for optimization. No matter how cheap they are.
Miracles don't happen, you have to pay for processing power.
0.6 quid per 1 hour m4.10xlarge. For spot prices rounded to the hour: 1:02:05 - the charge would be for 2 hours.
EBS is a virtual drive. It is charged separately, but is very cheap. Pay per GB/month.
These are spot instances.
They are not suitable for optimisation. No matter how cheap they are.
There are no miracles, you have to pay for processing power.
What is the basis of such a strange conclusion?
When selecting a spot instance, it is important to evaluate the tolerance of interruptions to the application and the target level of cost reduction. The lower the interrupt rate selected, the longer the spot instances will run.
The key here is the acceptability of application interruptions.
As soon as you run optimization at least once on such instance - share your impressions.
Probably a lot of people will be interested to see:
It happens.
Scared of interruptions? Actually, for optimisation, dropping agents is not critical.
The spot price is rarely updated. The price is fixed for the hour of use. If the price is updated by the next hour, an interruption will only occur when the requested price is exceeded. This is similar to a stop loss:)
For there to be no interruption it is enough to specify the rate price of the instance. My instance ran for 14 hours without interruption.
Scared of interruptions? Actually, for optimisation, agents dropping out is not critical.
The spot price is rarely updated. The price is fixed for the hour of use. If the price is updated by the next hour, an interruption will only occur when the requested price is exceeded. This is similar to a stop loss:)
For there to be no interruption it is enough to specify the rate price of the instance. My instance ran for 14 hours without interruption.
That's more fun now. But not quite yet. I have to calculate real savings.
With agents in the cloud everything is clear - you pay for agent's work time and no more. Optimization has begun, on pass 1 you have 256 (512) agents in a job, on pass 2 less agents will be employed, on pass 20 an order of magnitude less agents will be employed than on pass 1, etc.
That's not how it works with instance agents. You pay for time of work of all instance, and it doesn't matter how much time agents worked individually. I.e. it's necessary to calculate an optimal balance between computational power of instance and time spent on optimization so that you don't have to pay extra money and time.
And something in this regard seems to me that the instance will not be cheaper than a cloud from MQ, whatever it is. Even at discounted prices, i.e. spot prices.
And both of those options lose out to buying a pepelatsa cleanly :) Provided there is something to count, of course.