Active Monster Buffer Setting (?)

General MajorMUD Discussion
User avatar
MiOw
Posts: 113
Joined: Tue Nov 15, 2016 7:40 pm

Active Monster Buffer Setting (?)

Post by MiOw »

I figure most buffer settings are less critical when the board is running entirely on SSD, but does the size of the active monster buffer impact regen in the realm? I noticed that the "Used" buffer count keeps itself slightly under but tight to the "Size" allowed when I raise or lower it. This made me wonder if the game is filling what it has access to, though living mobs generated as people run around could be a simple answer too. I remember hearing about ways to tweak regen back in the day and now think this is about the only option that would appear to relate if so. Does anybody have experience or thoughts on this? The general settings recommends 10x the number of playing users but I would be curious to know what kind of level other people are running.

Thanks!


ClassicMUD... lightly seasoned with no dupes!
www.mudinfo.net/viewtopic.php?f=64&t=2357
User avatar
BearFather
Posts: 652
Joined: Sun Feb 09, 2014 6:27 pm
Location: Portland, OR
Contact:

Re: Active Monster Buffer Setting (?)

Post by BearFather »

Never touched the buffer settings, left them default. Brad(CuckoRules) had played with them in the past maybe he will pipe up. In theory you raise them and mud will use the drive less to access stuff and provide a smoother game. In truth, an ssd or 7200 rpm drive is fast enough since mud is a slow game.


Shadoron
Posts: 22
Joined: Mon Oct 20, 2014 2:54 am

Re: Active Monster Buffer Setting (?)

Post by Shadoron »

In my limited findings with adjusting buffer settings, it's like Bear says. The higher you make them (up to a certain limit), the more of whatever you are buffering is dumped into RAM. So, for monsters, a higher buffer just means that your system is accessing the hard drive less, because more of the monsters that MMUD is "tracking" are in RAM. It shouldn't affect regen at all, unless you are using a version of MMUD that happens to have some kind of regen bug. Running MBBS / WG on a modern system means these settings really mean nothing. Back in the 'ole days, hard drives were extremely slow, so having 100 users all accessing data for MMUD would demand high buffers, otherwise everything would slow to a crawl trying to read data from the HD constantly. I'm running two versions of MMUD on each of three DMA servers, and I have to have the buffers set to their minimums to save on RAM to load up just the modules (the joy of DOS). I've had zero issues with slowdowns or regen.


User avatar
syntax
Site Admin
Posts: 517
Joined: Tue Jun 02, 2009 10:02 am

Re: Active Monster Buffer Setting (?)

Post by syntax »

Just for the record, SSD's aside, hard drive speed hasn't changed much in like 20 years.


cuckoorules
Posts: 88
Joined: Fri Mar 30, 2012 9:38 pm

Re: Active Monster Buffer Setting (?)

Post by cuckoorules »

The only place I could see this coming into play would maybe be with the special group Angel which is supposed to despawn when target is lost (for summoned spawn monsters like greater hellion that ideally would de-spawn if everyone wipes, ASD and others are set up this way also). I don't know a lot about the specifics, but it's possibly that by having a larger buffer, it may not ever need to free up that memory for other monsters, which could maybe cause the Angel types to not despawn. However, there's only a literal handful of stock examples using this behavior and I can't think of much else that would be impacted by it. Not even certain that this would be impacted, as I've had trouble even confirming the Angel behavior is true. It makes sense however, if you look at where it is used on stock monsters.


cuckoorules
Posts: 88
Joined: Fri Mar 30, 2012 9:38 pm

Re: Active Monster Buffer Setting (?)

Post by cuckoorules »

P.S. when I first launched my BBS I figured larger buffers are better so I think I upped or maxed pretty much everything and wasn't seeing any huge performance differences one way or the other and probably just left it that way, but I honestly don't know a lot of technical detail about what is happening.


User avatar
BearFather
Posts: 652
Joined: Sun Feb 09, 2014 6:27 pm
Location: Portland, OR
Contact:

Re: Active Monster Buffer Setting (?)

Post by BearFather »

syntax wrote:Just for the record, SSD's aside, hard drive speed hasn't changed much in like 20 years.
I would disagree here, the bus speeds are a lot faster then they used to be. 20 year's ago we had ata 66 5200 rpm drives with no cache or small amounts of cache on IDE. Now we have 15k rpm with 64m cache on sataIII, and that's not including SSD's.


User avatar
syntax
Site Admin
Posts: 517
Joined: Tue Jun 02, 2009 10:02 am

Re: Active Monster Buffer Setting (?)

Post by syntax »

BearFather wrote:
syntax wrote:Just for the record, SSD's aside, hard drive speed hasn't changed much in like 20 years.
I would disagree here, the bus speeds are a lot faster then they used to be. 20 year's ago we had ata 66 5200 rpm drives with no cache or small amounts of cache on IDE. Now we have 15k rpm with 64m cache on sataIII, and that's not including SSD's.
7200 rpm drives were released in 1992 and 5400 rpm drives are still considered standard with 7200 rpm usually considered an upgrade.

Thorough documentation here: https://www.micron.com/~/media/document ... age_wp.pdf

“over the past 20 years, microprocessor technology—which plays a key role in data storage ef ciency and function—has enabled CPU performance to nearly double every 18 months. Put another way, CPU performance has increased 16,800 times between 1988 and 2008, but HDD performance has increased by just 11 times.”


User avatar
BearFather
Posts: 652
Joined: Sun Feb 09, 2014 6:27 pm
Location: Portland, OR
Contact:

Re: Active Monster Buffer Setting (?)

Post by BearFather »

syntax wrote:7200 rpm drives were released in 1992 and 5400 rpm drives are still considered standard with 7200 rpm usually considered an upgrade.

Thorough documentation here: https://www.micron.com/~/media/document ... age_wp.pdf

“over the past 20 years, microprocessor technology—which plays a key role in data storage ef ciency and function—has enabled CPU performance to nearly double every 18 months. Put another way, CPU performance has increased 16,800 times between 1988 and 2008, but HDD performance has increased by just 11 times.”

So your saying 11 times is not faster? Yes it's not as far as a jump processor's have had, but memory hasn't made huge leaps either. And not sure about you but the avg guy in 92 could not afford a 7200 drive, the prices for one of those was fucking crazy. Hell I remember dropping $300 for a 1.7 gig drive in the late 90's and that was only a 5400 rpm drive.

Sure if you had some crazy money back in the 90's you could get yourself a nice pentium 166mx with a 500M 7200 drive and 6m of ram. But your looking to drop thousands on it back then. That's like saying we use holographic storage now. Just cause we have the tech doesn't mean it's cheap and reasonable.


User avatar
syntax
Site Admin
Posts: 517
Joined: Tue Jun 02, 2009 10:02 am

Re: Active Monster Buffer Setting (?)

Post by syntax »

BearFather wrote:So your saying 11 times is not faster? Yes it's not as far as a jump processor's have had, but memory hasn't made huge leaps either. And not sure about you but the avg guy in 92 could not afford a 7200 drive, the prices for one of those was fucking crazy. Hell I remember dropping $300 for a 1.7 gig drive in the late 90's and that was only a 5400 rpm drive.

Sure if you had some crazy money back in the 90's you could get yourself a nice pentium 166mx with a 500M 7200 drive and 6m of ram. But your looking to drop thousands on it back then. That's like saying we use holographic storage now. Just cause we have the tech doesn't mean it's cheap and reasonable.
If I say something is twice as fast, yea that sounds awesome, let alone 11 times faster. However, taking the whole picture in, where CPU has increased from 150MHz in 1996 to 4.4GHz (30x) and you also have the additional cores, threads, etc etc and RAM performance has increased from SDRAM at 100MHz in 1996 to DDR4 now running at 4,266MHz (42x) and also now utilizes multi-pathing and such, 11x doesn't sound like much.

It's also hard to compare performance gains between things that aren't of much relation to each other, so really all of these #x are kind of meaningless. But the point I'm trying to make is CPU and RAM (and video cards, actually) have made great strides in different areas to improve performance. Spinning disks haven't in any area other than capacity. There is "more" capacity, "more" cache, "more" speed, but nothing different. Until SSD's anyway.

I've been building computers on a annual basis, for the most part, since I was 15 (20 years ago) and I have always felt like hard drives have been the bottleneck in terms of "general computer usage" speed (e.g. boot time, opening programs, installing, copying/writing/moving files, etc etc). I'd build machines with 7200rpm high-cache dual-hard drives in raid-0 just to get the best performance, and it still sucked watching that blinking hard drive light time after time again waiting for shit to happen. It wasn't until the era of SSDs that the game really changed in the ways that it has changed for CPU, RAM, and video cards over the years.

Take a look at this graphic. The bottom rows are all spinning disks ranging from 5k drives to 15k drives. Look how little difference there is. A 7.2k drive has 35% more IOPs than the 5k, and the 15k has 64% more, but the SSD is over 10,000% more. So yea, I stand by my statements.
IOps_mean_comparison_EN.gif
edit: forgot to post graphic ><


Post Reply