System Builder Marathon: Low Cost System

[muk]

http://www.tomshardware.com/2007/09/17/sbm_low_cost_system/index.html

We build the highest-performance PC that $1,000 can get, and prepare it to fight some upcoming system builder marathon competitors.

 

[zenmaster]

Much better than the $500 build in parts selection.

 

[onestar]

A balanced and intelligent selection of components yeilds a great blend of performance and value.

I too look forward to the comparison with the midrange system.

 

[Guest]

I fail to see how $1000 is a low cost system.

 

[cleeve]

That's easy.

$1000 is a low-cost system when it's getting compared against a $1500 and + systems... 😉

 

[No1sFanboy]

$1000 is more realistic for the kind of user reading Tom's. All the $500 builds were compromised and didn't include an OS. If someone asked me for a $500 build I'd direct them to Walmart for a prebuilt Dell.

Edit to add link:http://www.walmart.ca/wps-portal/st...pe=1&page=null&departmentId=70&categoryId=387

There for 497.88 CDN including Vista premium.

 

[sojrner]

agreed, at $500 you are hard pressed to build a better system (including OS and monitor) than what you can get from HP or Dell.

 

[Mega_Man_X]

my computer is ok and it was around 500

 

[fumuthabeotch]

I gotta say, I'm interested to see how the $1000 dollar rig will compare to the $1500 rig, since my budget for a new gaming rig is somewhere in between.

Another thing... try benchmarking with some newer games like maybe Bioshock?

 

[cleeve]

Nobody's saying your $500 computer isn't OK, Mega man. I made a $500 budget machine in our first marathon. It can still be done today.

However, alot of our readers commented that the $500 point was a little low, so we upped the ante this time around. We might lower it again next time. We'll see.

 

[Kazzi1]

18Amps per-rail for a total for 36Amps? That PSU really has 36Amps?
And a slight typo... top of page 6 you call the cooler an Arctic Cooling HyperTX 2.
Pretty good article though.

 

[cleeve]

Another thing... try benchmarking with some newer games like maybe Bioshock?

Yeah, we're in the middle of updating our benchmarking suite now so this is probably one of the last roundups using the old guard.

Still, Oblivion and FEAR aren't too easy on the hardware.

 

[Mega_Man_X]

ok just sayin

 

[cleeve]

18Amps per-rail for a total for 36Amps? That PSU really has 36Amps?

I believe if the PSU is being maxed out, it can handle a total of 29A between both of the 12v rails, although each rail is capable of 18A max - just not at the same time.

Still, 29A is very respectable on a budget PSU like this.

 

[shadowmaster625]

I just dont understand the point in comparing the new system to the old system when the old system has a E4300 that was clearly meant to be overclocked. And then NOT overclocking it. Do you guys have rocks in your brains? Why didnt you overclock them both and run at least one freakin test?

 

[cleeve]

*sigh*

We're comparing it because the price is similar, shadowmaster, as explained in the article.

As also explained, the overclocking tests are coming later in the week.

Please, read the article before insulting it. Throwing out an insult because you don't understand what's going on isn't flattering... trust me on that.

 

[jeremyrailton]

The results are rather underwhelming since both systems can be overclocked like crazy. That's when the real story will be told, when the GTX is allowed to really run with a 3.0ghz e4300 instead of the stock speed. How will the video card vs cpu choice look then? tune in next week, same bat-time, same bat-channel.

 

[Guest]

I'm satisfied with the system I built, would've been around $2000 if I bought the Samsung 226BW with it, but I priced it against Dell, Alienware, and Ibuypower and lesser systems from them all were of $3000.

That is a very good point to whoever said that it is difficult to make a cheaper/better system at $500 then Dell/HP at Walmart.

 

[PopsSoCal]

This is very interesting for the most part, And I am very anxious to find out the outcome of it all I just ordered a Antec P180b case and am in the process of building a good combination Media, Gaming, computer to have in the Living room to hook up to a 50" flat screen tv with surround sound and all he,he,he and this articale may help decide on which way to go..... Thanks

 

[extrasalty]

I really think it should be made clear- no matter how high the numbers are in games and even though it reflects some CPU bottlenecking, no mainstream monitor will show over 160 FPS (160 Hz)@1024x768. For that you need 130 kHz horizontal refresh which is rare even on professional CRTs. Same monitor will wind up at 120@1280x1024 and just over 100@1600x1200. For a normal LCD monitor 85kHz-92kHz is common which translates into maximum possible vertical refresh of 100 Hz@1024, and capped to 85 Hz on most.
So it may be fine and dandy game benchmarks proving it's better CPU but the results are useless in real life games. Unless extra eye candy is turned on there is hardly a difference since both GTS and GTX will be better than the monitor, not even speaking about what the human eye could possible notice over 60 FPS.
And when the extra eye candy is turned on, most games are bound to be GPU limited and GTX will win.
When different resolutions are compared on the same system in the Doom and F.E.A.R. benchmarks it's clear that the old system is CPU bound and the new system is GPU bound- so no matter the net increase , the slower GPU shows. IMHO Doom and F.E.A.R. are getting old to compare brand new hardware and not even get in high (1920 and above) resolutions. Oblivion is the only game on the test that shows the better scaling of the older system (GTX) as the resolution increases.
So to recap, the entire game section (almost) proves nothing. The choice of cpu is hardly a deciding factor in quality of game play, just as FPS is a poor quality indicator. Minimum FPS and FPS fluctuation has much greater impact on game play. In that light you managed to effectively lower the real game performance while increasing encoding and other CPU intensive application's performance. So it's not win-win situation, or balanced system- it's called trade off.
@randomizer
Do you even know what vsync is? Vertical refresh is how many times the screen is refreshed per second. If you have 100Hz, the monitor will not display 101 FPS but 100 with or without vsysnc. Period. Also you can't assume that if your top FPS are 120 everything else is fine- whats the point of comparing systems then? It's very possible that the GTX will have higher min FPS, which would be low enough on both cards to be noticable( below your monitor's maximum vertical refresh).
@zenmaster
I only pointed that the choice of games and resolutions doesn't allow fair comparison since both systems are over what normal monitor will display(in a way the monitor is bottleneck), and then there is a claim( at the end) this system is faster by 18%- yes it is, but on old games like Doom 3 and FEAR and on resolutions lower than 1600x. You don't buy a brand new computer and use over 3 year old game( Doom 3) to compare it.
I said both Doom and FEAR are CPU bound on the old system which is logical since the video is so powerful , but even then the FPS are above the capability of 99% of the monitors sold. The second point is that the current system doesn't scale well in the same game with different resolutions- meaning that if tomorrow I get HD 2560x1600 this system will suck. You can argue that hardly anyone will use that resolution, I can argue the opposite. My point is for gaming it's always more important to be well prepared- spend your money on a better video card- it costs more but it lasts longer, and you can't put a price on smooth game play. I do agree with you on the $500 system.
@cleeve
I only disagree with the final conclusion, which is that this system is 18% faster and I argue it is so because of the choice of games and resolutions. I use the benchmarks, assuming they are solid facts, to show you that the games chosen do not show the difference that well because both system exceed the video bandwidth of most monitors. The "almost" I left there for Oblivion. All in all my argument is that almost everyone falls into the hype "higher FPS is better, who cares you can't see it" which IMHO is pushed by the graphics chips makers.

 

[randomizer]

I really think it should be made clear- no matter how high the numbers are in games and even though it reflects some CPU bottlenecking, no mainstream monitor will show over 160 FPS (160 Hz)@1024x768. For that you need 130 kHz horizontal refresh which is rare even on professional CRTs. Same monitor will wind up at 120@1280x1024 and just over 100@1600x1200. For a normal LCD monitor 85kHz-92kHz is common which translates into maximum possible vertical refresh of 100 Hz@1024, and capped to 85 Hz on most.

What's your point? That's why you get tearing and you use VSync to fix that. Dude who cares what the monitor can and can not display, the framerate is higher, therefore the gameplay is smoother.

So it may be fine and dandy game benchmarks proving it's better CPU but the results are useless in real life games. Unless extra eye candy is turned on there is hardly a difference since both GTS and GTX will be better than the monitor, not even speaking about what the human eye could possible notice over 60 FPS.
And when the extra eye candy is turned on, most games are bound to be GPU limited and GTX will win.
When different resolutions are compared on the same system in the Doom and F.E.A.R. benchmarks it's clear that the old system is CPU bound and the new system is GPU bound- so no matter the net increase , the slower GPU shows. IMHO Doom and F.E.A.R. are getting old to compare brand new hardware and not even get in high (1920 and above) resolutions. Oblivion is the only game on the test that shows the better scaling of the older system (GTX) as the resolution increases.

Read what Cleeve said before. And the GTS/GTX are better than the monitor? I assume you are referring to what you said at the start, but then, you can say the same thing about many cards. My x1950 pro is also "better" than my monitor, if I play on low res and not-so-new games.

So to recap, the entire game section (almost) proves nothing. The choice of cpu is hardly a deciding factor in quality of game play, just as FPS is a poor quality indicator. Minimum FPS and FPS fluctuation has much greater impact on game play. In that light you managed to effectively lower the real game performance while increasing encoding and other CPU intensive application's performance. So it's not win-win situation, or balanced system- it's called trade off.

If you get an average framerate of 120+FPS you can be pretty certain that you aren't getting a low min FPS very often. If you are worried about FPS fluctuation, use VSync or another form of framerate cap. I don't remember where they said it was a win-win situation, but feel free to quote it.

 

[zenmaster]

I'm satisfied with the system I built, would've been around $2000 if I bought the Samsung 226BW with it, but I priced it against Dell, Alienware, and Ibuypower and lesser systems from them all were of $3000.

That is a very good point to whoever said that it is difficult to make a cheaper/better system at $500 then Dell/HP at Walmart.

The guy said a $500 system, not a $1500-$2000 system.
BIG difference.

The Retailers tend to add bigger and bigger margins to more expensive systems.
I don't want to start the discussion here, but read or post in the $500 section.
I know some people linked in Retail packages that were cheaper and included LCD and printer and OS.
These system also included faster processors, more ram, larger HDD, and niceties such as media readers, speakers, mice, keyboard, etc.. etc.. etc..

 

[zenmaster]

So to recap, the entire game section (almost) proves nothing. The choice of cpu is hardly a deciding factor in quality of game play, just as FPS is a poor quality indicator. Minimum FPS and FPS fluctuation has much greater impact on game play. In that light you managed to effectively lower the real game performance while increasing encoding and other CPU intensive application's performance. So it's not win-win situation, or balanced system- it's called trade off.

Your analysis is off more than a touch.
I would recommend rereading the article.

1) In your message you note how some games are limited by the CPU then you say it goes not matter.

2) There is not really a "trade-off". The first build was not really a $1000 build despite its claims. It went over by a large amount. This build could have fit in the GTX, it chose to not spend as much money as was noted in the article. A slightly cheaper Mobo could have even kept the prices the same and not much else would have changed.
Heck, the 2900XT would have fit in nicely without any other changes and still come in less.

So, yes it is a Win-Win. The system is better top to bottom as is with only losing a few benchmarks on people with very large monitors. And this could have been easily remedied if the article had kept the same budget, but it chose to live up to the title.

It is also balanced because money is spent appropriately.
An example of an unbalanced system was the recent $500 build which had a $100 power supply, but a GPU too weak to play any games. The argument that it would be good for upgrades did not hold water because the system was currently not usable as a "Gaming" system which is what the title of the article called it.

 

[cleeve]

So to recap, the entire game section (almost) proves nothing. The choice of cpu is hardly a deciding factor in quality of game play, just as FPS is a poor quality indicator. Minimum FPS and FPS fluctuation has much greater impact on game play. In that light you managed to effectively lower the real game performance while increasing encoding and other CPU intensive application's performance. So it's not win-win situation, or balanced system- it's called trade off.

I find it amusing that you say it proves (almost) nothing, and then immediately use the information in the benchmarks to draw a lengthy conclusion (something).

I therefore disagree with your suggestion that it proves (almost) nothing.

Maybe you wanted it to prove something different, but it most certainly proves something... 😀

 

[sojrner]

I find it amusing that you say it proves (almost) nothing, and then immediately use the information in the benchmarks to draw a lengthy conclusion (something).

I therefore disagree with your suggestion that it proves (almost) nothing.

Maybe you wanted it to prove something different, but it most certainly proves something... 😀

[:mousemonkey:5]


hmm, "I only knew that you'd know that I knew. Did you know THAT?"

[:mousemonkey:2]