ES - Hardware performace for RTW and MTW2

[Belisarius]

Hardware performace on RTW and what future upgrades to consider for MTW2

By - AlfonsoVIII

For my first article, I want to find out once and for all what makes Rome: Total War work well: the processor, the video card, or the memory and what to infer for M2TW, which is a popular subject. Truth set in stone the answer has always been the processor as the number one dependent. So, I decided to see for myself how the game performs using different processor speeds. Now my rig normally is as follows (stock):


Athlon 64 X2 4200 AM2 @2.209 Ghz
2GB PC800 DDR2 Memory (734Mhz actual)
BFG Tech 7900 GT OC 256MB PCI-E

Now, with the change in processor speeds, the memory, in Mhz, will change. I am not concerned with marking these numbers since the numbers are dependent to the processor speed. Now I will test my processor at the following speeds:

1.0 Ghz to 2.6 Ghz in .2 Ghz increments.

I will also run the same test by taking out one of the memory sticks to see if any correlation exists. My hope is to see whether anything over 1GB of memory is indeed overkill and how much performance is gained with the processor speed. The current price on my processor is around $180 USD and the next step up is the 4600, around $240. A big jump for just .2 Ghz, but is it worth it?

Now in order to test this, I will be using a custom battle using the Julii and the Brutii with max settings, except for unit size (large) and resolution (1024x768x32). The battle is using Darth Mod 8.1 and the battle involves 2440 men where I can get a good shot of all men in one screen for a good length of time to get a good fps number. For that I am using Fraps. I know that Fraps is used for video cards, but if the rumor is true that processor speed is indeed the driving force behind the game, then I plan on exposing it and making it the bottleneck in my system. Since processors are more expensive to upgrade than memory and video cards, and I do know there are exceptions, I will once and for all determine the truth since I never saw any benchmarks comparing RTW with different processor speeds. Now if this has been done before, then let this show in reality that RTW works with dual core and end that rumor at least.

Now the results:

Processor Speed (Ghz) Average FPS (1GB)

1.0 = 7 (7)
1.2 = 9 (9)
1.4 = 11 (11)
1.6 = 13 (13)
1.8 = 14 (14)
2.0 = 16 (16)
2.2 = 18 (18)
2.4 = n/a
2.6 = n/a

After tinkering with my processor, I was unable to safely overclock my processor, so I essentially gave up. Now my conclusion.

It is safe to say that processor speed is indeed the number one factor in the game. Now if you have more than 1GB RAM in your system, then adding memory is overkill. To those who need better performance, buying enough RAM to get to 1GB is a good choice only if you have a decent processor and want better performance. The graphics card did only add to the eye candy compared to my FX 5200, where I had everything on low. So if anyone is seriously asking for better performance, I would recommend putting things on medium to see how slow the system runs. Then take into account memory, and then decide that processor change may be required. I know that improving any one of the 3 aspects of a computer will improve performance but be sure to spend your money wisely. With M2TW coming out and with a 1.5 Ghz minimum, I feel comfortable that anyone who has less than a 2.0 Ghz should upgrade and / or overclock their processor to meet the demand of M2TW.

Since my findings are basic, I will happily to other testing for users who have different questions and / or holes with my findings. I know that the test is one dimensional and I can quickly complex it to get better answers, but I feel confident that my findings will provide a better solution for those looking for a better gaming experience and wish to buy a new computer or upgrade theirs when M2TW comes out. With that said, this test and any additions by the community will be put into place with the M2TW demo, when it comes out. This is AlfonsoVIII saying goodnight, good fight, and good upgrading.

[jaklei]

Great article!That was really something i needed now in the waiting time for m2tw
If i would like to run on highest settings and i got AMD Athlon 64 processor 4000+ 2,40GHz and an RAM memory of 2047mb
then do i have to overclock or could it run smoothly with huge unit settings with 1stack on 1stack encounters (2000-3000 men)?

And if not, do you know how i could overclock using windows vista?

I am only worried since ive been using this computer for quite some while now

[JuliusCaesar83]

great article!!

I have a new INtel E6700 Core2 Duo @~2,8 GhZ and i can`t play any mod with huge Unit size and best settings smoothly... avarage FPS 15-20 .... (with 1280x1024 or 1024x768-is not much better.)

INtel E6700 Core2 DUo @~2,8 GhZ
2048 DDR 2 @800Mhz OCZ Platinum Edition
XFX 7950 GX2 M520 Edition
1x74 GB Western Digital Raptor
2x. 200 GB Seagate
ASUS P5B Deluxe WIFI Mainboard

it depends highly on the processor.. my last PC was a AMD X2 4600 Ghz with a ASUS 7800 GTX and there is no difference in performance or quality...

[Spart]

Nice article.
I just did some experimenting myself to find out which parts to improve for Medieval 2.

The test

So, I did a custom battle in Rome, Julii vs. Brutii, approximately 3000 men on the field.
And system was: Athlon 64 3000+ @ 2,0 Ghz, Ati X800XL 256 mb and 1024 mb of RAM. Pretty balanced system, I think.
So, most of the graphics settings were set to low, but huge unit size. Looked pretty ugly.
Results:
-At the start, while setting my own troops to formation, I got constant 54-54 fps.
-When I started battle and the enemy appeared, fps dropped to 44. Still smooth and nice.

Okay, now I overclocked a bit. Cpu from 2,0 Ghz to 2,4 Ghz. Meaning it was now Athlon 64 3400+. Start the same battle again:
-At the beginning, again 54 fps! Exactly the same!
-Enemy appears, and... Fps is still 44 at this point... 400Mhz did nothing? Well, then I presumed that my graphics card was the bottleneck.(Which I know isn't possible) I set all setting from low to high, put on some AA, smoke/dust, shadows... Back to the game, but... What's this?!!! Fps is still 44. Didn't drop at all!

So, It doesen't seem to matter if my cpu is 2 or 2,4 Ghz or if I set graphics settings to low or high... How can this be explained? Have I achieved Rome: Total War's maximum performance?
Those framerates are really more than enough, 25 is usually very playable for me. So if I get even half of Rome's fps in Medieval 2, I'm satisfied. I just don't know if it's worth overclocking my cpu, cause it didn't give me any extra performance in Rome. Well, I hope it's more effective in M2. Maybe Rome just isn't able to perform any better.. So I'm asking you, what are your best fps reading while playing these bigger (2000 men) battles. Would like to know them to do some research. If system twice as powerful than mine can't do very much better, then it's the game's fault...

-Spartacus

[AlfonsoVIII]

Great article!That was really something i needed now in the waiting time for m2tw
If i would like to run on highest settings and i got AMD Athlon 64 processor 4000+ 2,40GHz and an RAM memory of 2047mb
then do i have to overclock or could it run smoothly with huge unit settings with 1stack on 1stack encounters (2000-3000 men)?

And if not, do you know how i could overclock using windows vista?

I am only worried since ive been using this computer for quite some while now

Well, you must have the 4600, which is at 2.4 Ghz, now what kind of video card do you have. It will matter, what kind you have as with mine, a 7900 GT OC, the test was on 1024x768x32 resolution with everything checked, every detail on the highest, and the unit size on large. The AA was on high I believe also, I am tweaking my video card settings, since I am having driver issues on the screen. As far as Vista, I am not touching it until DX10 comes out so I can test RTW or M2TW when it comes out. I am know expert on overclocking, but I need to do some more reading, because I safely OC my processor to 2.27 Ghz from 2.2, which is no real big deal. I will follow up this weekend with more information.

[Bonifacius]

I conducted a few tests so far myself and in my opinion. RTW does not adhere to traditional hardware rules. Here is my system:
X2 3800 (running at 2, 2.2, 2.3 ghz)
2GB of DDR (running at 400, 440, 460 mhz)
6800GS (at up to 525mhz GPu and 1050 mem)

The system I ran it against was
X2 4200 (2.2ghz)
2GB of DDR 400
7800GT

I dont have FPS data since I dont rely on that when considering upgrading. I judge everything on VISIBLE differences. In other words, can the better system make such a difference, that I am compelled to upgrade? Can the better system at least play the game on higher setting that my system simply cant? Traditional logic dictates that the system with the 7800GT will smoke my system on games like FEAR or quake. I was shocked. With maxed out settings and 6000 troops on the field, both systems were choppy at best. In my opinion, it makes no sense to buy a $320 graphics card (against my $170 card) so that the new system plays RTW just as crappy as my old one. When it came to FEAR however, the 7800GT system was able to bump up both resolution and AA/AF one step without a problem. While this is a good thing, some would argue that spending another $320 just to say you can play FEAR at 4X AA/8X AF is just not worth it. Was there a visible difference in quality? Yes. Could the 7800GT play FEAR at higher quality settings with playable frame rates whereas my 6800GS could not? Yes. Would it be worth $320? I say no. But I am getting off of the subject.

My next test was against my new HP 9000 laptop. Here are the specs:
Core2 duo @ 1.66 Ghz
2GB of DDR2 @667 mhz (CAS of 5 I am sure)
7600go

FYI, according to PCmark 05, the core 2 duo in the lappie is equal to a X2 3800 or perhaps a little less (1.95ghz or so). 3d mark 06 has the 7600go about 1000 points lower than the 6800GS. Simply put, when it comes to games, this lappie doesnt hold a candle to my desktop. I ran the systems maxed out with 2000 and then 4000 troops on the field. Again, I could not see any difference. I am sure the 6800GS would achieve more FPS, but I am confident in saying I could see no difference even if it did. What a shocker! Another tidbit. I changed my 7600go settings to max power saving (this achieved 654 3d marks vs 2449 marks when at full power) Even the strategy map was choppy when zoomed all the way out. Gameplay was terrible to say the least. In this test, the CPU had no impact on game play. It was all on the GPU. But in the other tests (when the CPUs are equal) the Gpu didnt matter. WTF?

I also played RTW on another laptop for a few months:
1ghz ULV centrino
512 MB of PC 2700 DDR
Intel extreme graphics 2

Even at the lowest settings, it was choppy when there was even 1000 troops on the field. What a waste of time.

Here is my theory: Up to a certain point, no matter what you buy, you will achieve nothing more with RTW (but not neccesarily Mtw:2) But if you are underpowered (as seen by my centrino lappie and extreme power saving settings on my 7600go) you will be sucking the proverbial tit.

My suggestion: If you can game now just fine, then wait to upgrade. Even better, wait for your friend to buy something new and see what it does. Then wait a few months and one-up him by buying something better and then hold it over his head and call him an idiot for buying too soon. If you have a weak system, then I would focus on a good GPU first, A HUGE MONITOR NEXT, then the CPU, and lastly the RAM.

I think the A64 single cores over 2.2 ghz are a really good deal at around $100. Unless you encode video or like to watch dirty movies while playing RTW, dual core is a waste.

Unless you overclock alot, any type of RAM or ASUS mobo will do.

I like the prices on the 7900GS and 7900GT. I would take the 7900GT any day. (if I didnt have a 6800GS)

You can go out at buy a $400 CPU and a $400 GPU if you want. Or even go SLI if you want. But I will go out on a limb and say that getting a 26"-31" LCD will have more of an impact on MTW:2 than expensive hardware.

Just my 10 cents.

[AlfonsoVIII]

I think the A64 single cores over 2.2 ghz are a really good deal at around $100. Unless you encode video or like to watch dirty movies while playing RTW, dual core is a waste.

I do not think it is a waste, lol. What do you think I do all day.

[Bonifacius]

You aint lyin bruddah. I got the TW on my projector and I do vids on both my 17" LCDs. I paid $300 for my X2. I am going to multi task until it blows up or my penile cord falls off....

[Oldgamer]

I've added a post over at the "Sega releases specifications" thread, and I thought I might add that I've talked to a lot of people who want to play M2TW, and didn't realize that they must have a DVD-ROM drive, simply to install the game. While the rest of us are considering CPUs, video cards, framerates, and so on, it appears to have escaped the attention of many folks that they're going to have to add a drive to their computer, in order to play the game.

The first couple of days after release are going to be interesting, to say the least. Could I suggest a sticky, starting now, on threads regarding M2TW, emphasizing the basic computer components necessary to play this game?

[AlfonsoVIII]

I've added a post over at the "Sega releases specifications" thread, and I thought I might add that I've talked to a lot of people who want to play M2TW, and didn't realize that they must have a DVD-ROM drive, simply to install the game. While the rest of us are considering CPUs, video cards, framerates, and so on, it appears to have escaped the attention of many folks that they're going to have to add a drive to their computer, in order to play the game.

The first couple of days after release are going to be interesting, to say the least. Could I suggest a sticky, starting now, on threads regarding M2TW, emphasizing the basic computer components necessary to play this game?

Well I think it might be common that people have DVD ROM drives, I had one since RTW release date, but that might be a common question on release date, so I will be ready to answer that. And most computers bought for atleast a year ago do come with a DVD drive in it.

[Oldgamer]

Well I think it might be common that people have DVD ROM drives, I had one since RTW release date, but that might be a common question on release date, so I will be ready to answer that. And most computers bought for atleast a year ago do come with a DVD drive in it.

I bought my current computer about 3.5 years, ago. It's a Hewlitt-Packard Pavilion 551w, with a Pentium 4 (no hyperthreading), and I was hot to have that microprocessor. Note that it came with a 3.5" floppy drive, and the CD-ROM drive.

I learned the hard way about the DVD-ROM drives, when I bought Silent Hunter III, and put the DVD in my CD drive! After about an hour, and endless frustration, I discovered the "DVD" with the words "ROM" and "Drive", on the game box. At that time, I didn't even know that DVD's were being introduced to the personal computer. I'm not technical.

Since then, I've found out that many of my real life and online friends don't have a DVD-ROM drive in their computer, and ALL of them are planning to play M2TW. SEGA's release of the game specs came as a nasty surprise to them.

I've been slowly upgrading my current computer to play M2 ... RAM, flatscreen monitor, and nVidea 6600, among other things. But if I didn't already have a DVD-ROM drive, because of SH3, I would have been in for a nasty surprise at 12:30am, on the day of the release (as I return from Wal-Mart). I suspect that there are others on the web who don't have this simple and inexpensive piece of hardware.

We should advise everyone about this.

[SPARTAN VI]

Nice article.
I just did some experimenting myself to find out which parts to improve for Medieval 2.

The test

So, I did a custom battle in Rome, Julii vs. Brutii, approximately 3000 men on the field.
And system was: Athlon 64 3000+ @ 2,0 Ghz, Ati X800XL 256 mb and 1024 mb of RAM. Pretty balanced system, I think.
So, most of the graphics settings were set to low, but huge unit size. Looked pretty ugly.
Results:
-At the start, while setting my own troops to formation, I got constant 54-54 fps.
-When I started battle and the enemy appeared, fps dropped to 44. Still smooth and nice.

Okay, now I overclocked a bit. Cpu from 2,0 Ghz to 2,4 Ghz. Meaning it was now Athlon 64 3400+. Start the same battle again:
-At the beginning, again 54 fps! Exactly the same!
-Enemy appears, and... Fps is still 44 at this point... 400Mhz did nothing? Well, then I presumed that my graphics card was the bottleneck.(Which I know isn't possible) I set all setting from low to high, put on some AA, smoke/dust, shadows... Back to the game, but... What's this?!!! Fps is still 44. Didn't drop at all!

So, It doesen't seem to matter if my cpu is 2 or 2,4 Ghz or if I set graphics settings to low or high... How can this be explained? Have I achieved Rome: Total War's maximum performance?
Those framerates are really more than enough, 25 is usually very playable for me. So if I get even half of Rome's fps in Medieval 2, I'm satisfied. I just don't know if it's worth overclocking my cpu, cause it didn't give me any extra performance in Rome. Well, I hope it's more effective in M2. Maybe Rome just isn't able to perform any better.. So I'm asking you, what are your best fps reading while playing these bigger (2000 men) battles. Would like to know them to do some research. If system twice as powerful than mine can't do very much better, then it's the game's fault...

-Spartacus

I used to own a X800XL, and did play Rome: TW on it. If you want to eliminate your GPU as the bottleneck, you'll need to run one demo with the settings high/medium, and then run another test with the same amount of units, only turn the settings all the way down. I did this and found my CPU, was in fact, bottlenecking me at around 5,000+ units. When I paused the game while all units were on screen, my framerate jumped up like mad. Unpausing slowed it back down to a crawl.

[SnArF|D]

I used to own a X800XL, and did play Rome: TW on it. If you want to eliminate your GPU as the bottleneck, you'll need to run one demo with the settings high/medium, and then run another test with the same amount of units, only turn the settings all the way down. I did this and found my CPU, was in fact, bottlenecking me at around 5,000+ units. When I paused the game while all units were on screen, my framerate jumped up like mad. Unpausing slowed it back down to a crawl.

Thats because when you pause the game, all the animations stop thus taking a massive load off your CPU.

[SnArF|D]

18FPS @ 2.2Ghz? thats pretty pathetic for a dual core. Thats actually horrid. 18FPS! RTW always has and will have issues with dual core. Its really a luck of the draw as to how dual core performs with RTW. Some need AMD's dual core optimizer to get at least decent performance in RTW while some get good performance right off the bat.
Dual core as of this very moment in gaming is completely useless.....NO games take full advantage of dual core and in fact it hurts performance in most games (i.e RTW)

[SnArF|D]

I ran some tests of my own.
System- ATI x800XL 256MB, 1GB PQI Power Series DDR400, AMD Athlon 64 2800+ 2.2GHz
-----
Settings- 1024x768-All Medium w/ low AA | 4x AF (AF was forced in ATI CCC)-
HUGE Units
Shadows ON
Reflections/Glints ON
Dust/Smoke ON
Desync Checked

Total soldier count - 2256

I found high quality shadows were the only performance killer. I went from 32FPS average to 54FPS average just by turning off "High quality" shadows and useing just normal quality shadows. Upon disabling high quality shadows my FPS never dropped below 50FPS, even when both armies were on the screen at the same time. At different zoom levels my FPS still never changed. Whether I was fully zoomed in, or half way, or all the way out.

If I turn off shadow completely I could definately push 3000+ soldiers.

My CPU while on the low end holds its own in RTW. Skt754 while obselete now, still performs admirably.

[Germanicus75]

I ran some tests of my own.
System- ATI x800XL 256MB, 1GB PQI Power Series DDR400, AMD Athlon 64 2800+ 2.2GHz
-----
Settings- 1024x768-All Medium w/ low AA | 4x AF (AF was forced in ATI CCC)-
HUGE Units
Shadows ON
Reflections/Glints ON
Dust/Smoke ON
Desync Checked

Total soldier count - 2256

I found high quality shadows were the only performance killer. I went from 32FPS average to 54FPS average just by turning off "High quality" shadows and useing just normal quality shadows. Upon disabling high quality shadows my FPS never dropped below 50FPS, even when both armies were on the screen at the same time. At different zoom levels my FPS still never changed. Whether I was fully zoomed in, or half way, or all the way out.

If I turn off shadow completely I could definately push 3000+ soldiers.

My CPU while on the low end holds its own in RTW. Skt754 while obselete now, still performs admirably.

Thanks Snarfd, that's interesting, especially since my current system is not so different from yours. What effects have you noticed from altering the AA?

[Incinerate_IV]

For those of you who are getting low FPS, do NOT turn the unit detail to max, since it disables the sprites completely. No video cards can handle rendering that much men on the screen at the same time.

And 15-20 FPS is considered smooth for RTW, this is a strategy game, not a FPS.

[Spart]

It doesen't seem to matter if I use max or low settings, I get the same fps... And I don't believe my gpu is a bottleneck, I can turn on even high AA and some AF and fps doesen't drop that much.. Although overclocking my cpu didn't have any effect either. But fps still so high that I can keep all settings on high and get like.. 4000 men.

I'll report here once I'll get to try Medieval 2 demo. I'll run the same test with that.

[SnArF|D]

My card handles AA admirably. I get no difference in FPS going from low to high AA.
I'm sure if I try all high w/ shadows OFF, I could still get the same 54FPS.

WHat me and you are both seeing is a CPU bottlneck and nothing more.

[Spart]

My card handles AA admirably. I get no difference in FPS going from low to high AA.
I'm sure if I try all high w/ shadows OFF, I could still get the same 54FPS.

WHat me and you are both seeing is a CPU bottlneck and nothing more.

Yup, looks like it... Although my overclocking didn't do any difference. As I said, I'll test with M2 demo, that hopefully solves it. =)