STOP PROVING YOURSELF TO BE A DUMB HUMAN BEING.

Quote:

Originally Posted by JellyBitz I'm measuring computing resources in general, and I'm trying to put it at simple words; A program created at 1980, can run faster even single-thread into the newest processor these days. So, You're measuring the Ghz and some "technology"? You'll have to be specific for me, I'm clearly a noob. I'll have to ask again because I cannot found the answer there, I'll try to be more specific to you: You're saying one thread into a Pentium 4 3.8Ghz it's as fast as one thread into a Dual Core 3.8Ghz?

Again, that depends on frequency and other technologies, even a Core 2 Duo with lower frequency can be faster single-threaded than a Pentium 4 because of stuff like higher L2 Cache or higher FSB, for example, most of Core2 Duo processors of these years have lower frequency than Pentium 4 but they are far better single-threaded because they have higher L2 cache and FSB.

Also, and again, let stop comparing desktop PCs, a silkroad server will always be hosted in a server PC, so we should be comparing server CPUs, not desktop CPUs.

I made my first server with AMD Athlon 4200+ X2 cpu and it was handling 1500 online easily, so it seems CPU isn't that important.

But you should have an SSD, that's the most important thing, as soon as database gets overloaded which happens at arround 700 online players with 7200rpm disks, moving items in inventory and similar actions are starting to happen with delay. With higher RPM disks or rain you can handle a lot more players with HDD without lag but the best solution is SSD.

FK I MISSED THE FUN , poor old me :(

a lot of the people here never really witnessed 4k in a server (for real) that's the funny truth.

Quote:

Originally Posted by _SGA_ I made my first server with AMD Athlon 4200+ X2 cpu and it was handling 1500 online easily, so it seems CPU isn't that important. But you should have an SSD, that's the most important thing, as soon as database gets overloaded which happens at arround 700 online players with 7200rpm disks, moving items in inventory and similar actions are starting to happen with delay. With higher RPM disks or rain you can handle a lot more players with HDD without lag but the best solution is SSD.

O.o

Well...I guess I should search for an SSD. Thx.

Quote:

Originally Posted by OnlyRipper O.o Well...I guess I should search for an SSD. Thx.

Yup sro doesn't require that high CPU power. It's made to be ran on Pentium 3 level CPU's but i/o performance of the disk is important.
What I noticed is even after CPU usage hits 100% you can still add 400~500 more online players without any lag penalty. I think the game server is prioritizing some tasks over others when it doesn't have enough power to do all at once.
So any CPU with decent single thread performance will do the job especially when you think you'll run second gs for the second 1000 players. Just make sure you get the best i/o performance SSD. But you don't need a top noch EPYC or XEON CPU's, but still it's good to use server cpu with ecc ram for stability. Don't buy so called servers with i5 i7 cpus, they aren't servers.

Quote:

Originally Posted by _SGA_ Yup sro doesn't require that high CPU power. It's made to be ran on Pentium 3 level CPU's but i/o performance of the disk is important. What I noticed is even after CPU usage hits 100% you can still add 400~500 more online players without any lag penalty. I think the game server is prioritizing some tasks over others when it doesn't have enough power to do all at once. So any CPU with decent single thread performance will do the job especially when you think you'll run second gs for the second 1000 players. Just make sure you get the best i/o performance SSD. But you don't need a top noch EPYC or XEON CPU's, but still it's good to use server cpu with ecc ram for stability. Don't buy so called servers with i5 i7 cpus, they aren't servers.

Any Ryzen CPU with good MB thats support ECC Ram(even unregistred) can handle VSRO server, but anyway , for good pserver advertised on EPVPERS you need DDOS protection, its more important than HW configuration. SSD and enought RAM is obvios.

Quote:

Originally Posted by janicka Any Ryzen CPU with good MB thats support ECC Ram(even unregistred) can handle VSRO server, but anyway , for good pserver advertised on EPVPERS you need DDOS protection, its more important than HW configuration. SSD and enought RAM is obvios.

We second this. :)

Quote:

Originally Posted by hyperfilter We second this. :)

What is this?
Advertisement as posts now?
Hey Hyper-filter ad-guy: advertising here by replying to a thread which has NOTHING to do with your services makes me NOT to chose your service.
By the way... Business running bad? You make your product look rather unprofessional and not wanted by making these "advertisement posts" on these forums. Congrats: you gained at least 1 person who will NOT try out or rent your services :)
Sorry for the off-topic here. I really had to say this.

All in all, I agree with Mega. From what I've seen in the server hosting scene, these files simply have an outdated architecture, and there's a lot of overhead going on between the modules.
Also regarding threads or non-threads -> It might be that the threading itself improves after each CPU generation coming out. But think that the development of new processors and "technologies" focus absolutely on multi-threaded computing! Nobody builds single-threaded applications anymore - so why should Intel/AMD stick develop and research costs into this, if this is a dead horse?

Regarding "ECC-RAM" bullshitbingo here: yeah, kinda every rent-server in the net runs on ECC ram, which is a must for in-memory databases for example. Do you know any server hoster who offers non-ECC ram? never saw one. Anyway - let me quote something real quick regarding ECC-Ram:

Quote:

ECC VS Non-ECC: Speed According to tests, ECC RAM is slightly slower than non-ECC RAM. As claimed by many memory manufacturers, ECC memory is 2% slower than standard RAM because the additional time it takes for the system to check for any memory errors. After testing the comparable CPUs (Intel Core i7 4771 3.5GHz Quad Core 8MB versus Intel Xeon E3-1275 V3 3.5GHZ Quad Core 8MB), we found that the estimate value 2% is roughly correct.

... So with ECC, the performance is even worse :)
Don't get me wrong - I'm a fan of ECC and I use it everywhere where I can use it. But ECC is not really an argument worth here, as these files run on any type of ram, and there aren't any non-ECC servers (physical or virtualized), unless it's that old crappy P4 Box which you had in your basement during the last 8 years.

What suddenly pops into my mind: maybe virtualizing that 1-threaded application, and let multiple physical cores work on that virtualized "wrapper" could be feasible? This is the only thing that pops into my mind as a workaround for these old single-threaded applications.
Laugh if you want :) Just brainstorming here what one could give a try to increase the "single thread computing performance" and v-wrap it :)

In some specific scenarios, turning off Hyper Threading in Intel CPUs, could help improving overall performance of old applications.

Also you can always set the power profile of the server to "performance", this will ensure idle states are disabled or at least, reduced to a manner where you'll extract maximum performance.

Also, certain CPUs will boost up, when specific applications may be using only a few cores, instead of all them (For example, when you pin then, to specific cores only).

In fact theross is correct to the point that nowadays most manufacturers are focusing in multi-threading, and this is because the "raw clock speeds" aren't going further forward lately, you will notice that the CPU speeds aren't going in a 2x increment every year, but actually 10-15 or at most 20%, and this is because currently the silicon technology is on it's limit. And to overcome this, multi-core is the way to go, at least for now. Yes, you have CPUs which are faster than old generation CPUs if compared, but no, the latest CPUs aren't so much faster than their previous generation and that's it.

Moore law ( https://en.wikipedia.org/wiki/Moore%27s_law ) , was like that it wouldn't achieve such limit, but in fact, currently, this is what is going on.

Quote:

Industry experts have not reached a consensus on exactly when Moore's law will cease to apply. Microprocessor architects report that semiconductor advancement has slowed industry-wide since around 2010, below the pace predicted by Moore's law. However, as of 2018, leading semiconductor manufacturers have developed IC fabrication processes in mass production which are claimed to keep pace with Moore's law.

Quote:

As the cost of computer power to the consumer falls, the cost for producers to fulfill Moore's law follows an opposite trend: R&D, manufacturing, and test costs have increased steadily with each new generation of chips. Rising manufacturing costs are an important consideration for the sustaining of Moore's law.[45] This had led to the formulation of Moore's second law, also called Rock's law, which is that the capital cost of a semiconductor fab also increases exponentially over time.[46][47]

[Only registered and activated users can see links. Click Here To Register...]

Electus out there with no shame :D

Quote:

Originally Posted by MeGaMaX. What the hell is wrong with you people thinking your server can handle 4000+ players? What makes you think Silkroad can support more than 4000 players? Please put an end to the spread of this cancer. You're no better than Joymax, who stated in the documents that the game can only have 3500 legitimate players plus 500 premium players. Without delving too deeply into the explanation, I'll give you an example. Hopefully, you will stop using numbers greater than 4000 on your website. Silkroad makes use of a single Shard Manager. With 3500 players, the amount of packets sent from all modules to a single shard is already very high. Assume you are running 8x game servers, and you have 1500 players in parties on your server. The shard manager will try to send 1500 party refresh packets per second, plus whatever else is queued. Please stop making your website counter more than 4000, and at the very least be truthful about something. Consider the TRSRO grand opening with 3500 players. Keep in mind that TRSRO files are compiled by visual studio 2015 x64 modules, yet the grand opening is still lagging like hell with 3500 players. Compared to vsro which are compiled by visual studio 2005 x86 single thread. yet you people still put 10,000 capacity seriously? I hope this changes things, but I doubt it. Greetings, MeGaMaX.

challenge accepted.