Discussion in 'Server Configuration and Hosting' started by RoldanLT, Sep 13, 2014.
Which one performs better?
I'm confuse on what server should I use :/
I would opt for 48GB RAM if you just plan to run websites on it. If games then Xeon E3 one because of higher clock. Ask the hosting company itself for there reviews?
IIRC L series has more cache but if you plan to run code with latest instruction set then E3 series might be good.
If you plan to run windows server then E3 could be consideration because of inbuilt GPU. So it all depends on what you are planning to use.
Just hosting A XenForo Forum, the one on my signature.
Will be using:
E3.. L5000 series always seemed to have some sort of hardware issue. E3's are more intended for game servers, so they also can handle a better pounding from MySQL as they are designed for faster thrashing. All my servers are E3's.
The dual 5630 benchmarks higher than 1245v2, but barely. Just depends on what you need really...more cores/threads or higher clock speed.
1245v2 performs better according to cpubenchmark.net
Just depends on what exactly they are doing to benchmark it. According to the same site (cpubenchmark.net), a single L5630 marks at 5089, so a dual system getting only 7033? Nah.
The dual setup is going to perform just fine (will outperform the 1245, in fact), in situations where you need more cores (i.e. VPS servers or such), or more than 32GB of RAM, since you can't add more to a 1245.
Not the right logic for dual cpu systems
It is when it comes to multi-threaded tasks. Using a number (or specialized benchmarks in the first place) to determine the better CPU is just silly in the first place. These figures are exactly why.
But if using a virtualizer (like ProxMox) the 5630 will be better than the 1245v2. You can use the extra cores for the virtualized VPS's.
Exactly the whole point I'm trying to make. It just depends on the task. The CPU numbers are perhaps a good overall indicator, but they really don't mean a whole lot.
Here's a very real world example: Our E5-1650's will spank our Dual E5-2620's all over the place in almost every single task (even though the dual system benchmarks slightly higher). Where the dual system excels is VPS servers (as @Tracy Perry mentioned) and multi-threaded tasks, and it isn't by the small margin the benchmarks will have you believe (i.e. less than 1000 point difference). The 2620's mop the floors up with the 1650 in that setting.
That's real world testing, not benchmarks using a very small criteria in a lab somewhere.
Allow me to explain further in how the real world works vs. how the lab works. You can run X amount of tasks on a given processor. You add a second processor and you double the amount of tasks. That to me is a doubling of the benchmark score. Now of course, a game or Excel spreadsheet, or whatever the hell they use to determine their scores are not going to double in performance, but when it comes to the sheer amount of threads and tasks that can be run, the performance is absolutely doubled. If these tests were truly 100% multi-threaded tasks, you would get roughly double the performance by adding the second processor.
Only running this:
PHP 5.6.x + Zend Opcache
Nginx 1.7.x + ngx_pagespeed, fastcgi_cache, spdy
For More core CPU I think only PHP-FPM and Nginx will benefited right?
@RoldanLT, for a single site use, the E3 1445 would probably be better. The only real limitation is the memory is maxxed at 32GB.
Thanks for your advice @Tracy Perry
Yeah I'm using it since May 2014
As soon as they get some more of the SoYouStart's in stock, I'm thinking about getting one to setup up cPanel on and start hosting some sites professionally. I've got cPanel setup on one of my VPS's on my ProxMox server for use by some people, but would really rather have it on a standalone system... and if I'm going to do that then I'll charge folks (other than the ones getting VPS's for free from me now) to use it. Can move a couple of business sites I host over to it and make what it costs for the server back just from them alone and the rest will be gravy.
Separate names with a comma.