Part 3: Building A Balanced Gaming PC

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
[citation][nom]pauldh[/nom]We have to draw the line somewhere in terms of additional hardware, as each CPU adds a substantial amount of lab-time. Not to mention these charts are about at capacity with 5 CPU's on the axis. Agree with you though, I'd like to bring an i3 into the series in place of a S775 chip.[/citation]

Hey man listen to me, why don't you guys include i3 as well as S775 chips so people could decipher that do they even need to upgrade from their overclocked E5x00/E6x00/E8x00 to these new i3s or not, so please don't exclude any of the S775 chip but just introduce i3 at same speeds as the S775 ships!Hope u will look forward to this.

-Thanks
 
You can allways check the ralative speed of some CPU that was not in this test and see, what other CPU's are nearest to it. Same to GPU's that are not in the test. This test shows how "generic" CPU power and GPU power are related to each other. And it does very job in that!
Very usefull article indeed! One upgrade would be a links to relative CPU and GPU charts, so that anyone can easily check out how his own cpu/gpu would have done in this test.
It has been interesting to see for example how very expensive GPU can benefit even when combined with low end CPU, but that you really need more CPU muscle to make it really usefull. (How too weak CPU can prevent the full power of really powerfull GPU and vice versa how little can be achieved by changing to more powerfull CPU if your GPU does not fit certain limits.
If and when Nvidia 400 series would have been in this test it would have startet above 295 (in low CPU end of chart) and would have needed hefty CPU to really start pull ahead (start pulling wider performance cap)
It's all about relative power of elements. We know how 480 behave compared to for example 5870 depending of the tested game. (A little bit higher) with steaper prize. As it has been said, it is all about balance!
 
I understand that it is hard to include all the hardware in the article, and people with some experience can get the general idea and fill the holes, but still whole segment of cards is missing.
 
A balanced PC is nice, but not always the best choice for long term value. If every component is balanced, there's no easy choice to which component to eventually upgrade.

I always find that when build a PC from scratch, to spend a little more on a cutting edge platform and CPU. Particularly if there are new interfaces to be had. Even if it is held back by the GPU, a new video card makes for an effective and affordable upgrade a year or two later.
In my opinion, if you want to "future proof" an component, it should be one that may not be so simple to replace or upgrade. While skimping on the parts that are easily replaced. Even if that means building a less balanced system, with less performance per dollar, in the short term.

In the flip side one could save some money and get a more affordable platform, using the money saved to get a more expensive GPU. That might mean better benches for your dollar in the short term. but a couple years later you'll be forced to replace the whole mobo and CPU, and any other components that may be no longer compatible with the new mobo.
 
[citation][nom]hixbot[/nom]A balanced PC is nice, but not always the best choice for long term value. If every component is balanced, there's no easy choice to which component to eventually upgrade.I always find that when build a PC from scratch, to spend a little more on a cutting edge platform and CPU. Particularly if there are new interfaces to be had. Even if it is held back by the GPU, a new video card makes for an effective and affordable upgrade a year or two later.In my opinion, if you want to "future proof" an component, it should be one that may not be so simple to replace or upgrade. While skimping on the parts that are easily replaced. Even if that means building a less balanced system, with less performance per dollar, in the short term.In the flip side one could save some money and get a more affordable platform, using the money saved to get a more expensive GPU. That might mean better benches for your dollar in the short term. but a couple years later you'll be forced to replace the whole mobo and CPU, and any other components that may be no longer compatible with the new mobo.[/citation]

It's best to switch back and forth. Ram and CPU are more often defined by the motherboard, a GPU has longer life. However, I find that it's worthwhile to go mainstream than to buy high end. You could buy a $400 GPU or two $150 ones instead...I tend to opt for the $150 ones being as its just a lot smarter, cheaper, and you end up with a better GPU in the long run anyway because you're buying more often. You also save on energy bills that way.
 
[citation][nom]hixbot[/nom]I always find that when build a PC from scratch, to spend a little more on a cutting edge platform and CPU. Particularly if there are new interfaces to be had. Even if it is held back by the GPU, a new video card makes for an effective and affordable upgrade a year or two later.[/citation]
A i7 940 have hardly more lifespan then a i7 920 for example. So it is almost always better not to buy the fastest cpu, but the one with the best price/performance ratio within that same architecture.
The money saved that way, which can be alot, can be used to invest in a much master videocard. So you don't have to struggle for a few years with a underpowered videocard. By the time the CPU becomes a bottleneck,
your overpriced CPU which only run's some Mhz faster will become just as obsolete as the slower CPU. Like a i7 920 for instance, can be overclocked just about as high as it's twice more expensive model.
Afcource, i also would not but a high-end, supreme priced videocard. Right now a HD5770/HD5850 have the "sweet spot".
 

Thanks. Yeah, a 1GB card gives you more options there. Admitedly I was new to the game, but tried to maximize IQ as much as possible without delving into advanced tweaking. Didn't want to bypass the frame buffer check and risk not comparing same details with each card (missing textures, etc.) at 2560x1600. Are you at 50% view distance? Shadow quality? Blur filter?
 

Appreciate the comments. This E8400 was fine at 4.5GHz also, but required a few bumps to 1.416V idle (1.384V load) for total stability. Didn't take the time to push past that. Fun/rewarding chips for sure. BTW, series is scheduled for 6 parts with a possibility of more. (Very time consuming and there are other stories to tackle)
 

Of course, earlier would have been better. We had similar issues with the HD 5000 series back in parts 1&2. A couple platforms were broken down off the bench prior to Fermi release, and Tom's early samples couldn't be dedicated to this series anyway. Honestly, you have to draw the line somewhere ... if we try to keep up with each hardware release, part 3 would never have gone live. This series can't compete there.... too time consuming. You'll need to refer to the launch reviews for the time being. Also keep in mind the (frustrating) state of the graphics market throughout the part 3 testing period. We can put many weeks into testing while availability/pricing can change almost overnight.

I3 and Fermi I can understand, it would be good to see how these scale in multiple configs. But not sure why you find midrange coolers or mobo selection an issue. IMO extreme cooling data would be less relevent to the majority of readers than with these fairly modest air coolers. There was still some room to push further anyway had we desired to push voltages further. An SSD would mainly affect load times. Other than smoothing out 2560x1600 in Crysis (still unplayable anyway) , I don't think swapping out the HDD would have altered performance numbers in the least.



 

This other hardware is outside the scope of comparison as this is mainly seeking CPU vs GPU balance. We explain our hardware selection was overkill, but chosen to make the whole series possible and trouble free. Example, you do not need an 850W PSU for an HD 5750 and E8400, but what about down the road when we overclock the i7 and explore multiple graphics cards? Gamers looking to build an upgradeable system do indeed take these factors into consideration. (mobo and PSU to support upgrades / second graphics card, etc.) Our forums are packed with people who could help you squeeze performance and relaibility out of your specific budget.

While our enthusiast class mobos may be capable of way more, we aren't dealing with extreme cooling and insane overclocks here. An inexpensive P45 could push our S775 CPUs also. The post right before yours complained why a newer Rampage III Extreme wasn't used and why such midrange coolers... showing two extremes among Tom's readers vs. yourself. We aim to reach a broad audience, but there is no way to keep everyone happy.

Anyway, my point being .... don't be hung up on the cost of other components as there was nothing extreme done here in terms of overclocking. Cost of an i7 platform and 6GB of RAM is something to take into consideration though.

Also, to be clear... Newegg had nothing to do with this (unlike the SBM series), but is commonly used as a basic reference for pricing. We chose the hardware we wanted to use, and are grateful for each manufactures' willingness to accommodate us and make this all possible.

 




I can now confirm two additional AMD processors (in transit) that will make it into Part 4.... one Athlon II X4 and one Phenom II X6. The PII X2 550 will remain the lone dual-core.

However, we are using the same game suite and likely the same 7 graphics cards in part 4, but will shoot to have the NVidia 400 series represented in parts 5 &6 at the latest.



 

Unfortunately, there is no possibility of adding OC'ed i3 results to the mix here; that's many days of benchmarking with an already loaded schedule. SBM testing is currently in progress followed by part 4 of this series containing all AMD processors. In part 5, I aim to include i3, but something(e6300?) would need to be pulled to make room/time. Too early to tell at this point, but adding GTX 400 to the mix, the various combos multiply quickly. (each CPU already adds 7 graphics cards).

If it's any consolation...today's feature pushes the i3-530 to 4.4GHz.
http://www.tomshardware.com/reviews/core-i3-530-overclock-lga-1156,2626.html
 
Just out of interest which processor would you guys say best approximates the E7200 in this benchmark, the E6300 or the E8400? Would the 3MB cache of it compared to the 2MB and 6MB of the other two be the biggest factor or something else?
 

Closer the the E6300 for sure both in L2 cache comparison and OC potential. It would vary some by game and core speed. Overall the stock 2.53GHz C2D E7200 is probably very close in games to the stock 2.8 GHz Pentium E6300 in part 1, I'd suspect able to pull ahead in some games due to the extra cache, and likely a tad slower in other games. Once you OC, the e7200 could scale better by core speed. I'd imagine if you hit 3.5-3.6 GHz or so on the E7200, results would be quite comparable to this higher clocked E6300 in part 3.
 
Ah right, thanks. Yeah, I've got an 8800GTS 320MB/E7200 @ 3.8Ghz combo at the moment and I'm thinking with an upgrade, getting purely say a HD 5870 is going to be far better value than also getting new mobo, CPU, and memory even given that it lags a fair bit behind the core i5/7s nowadays.
 
I have 2 5870's in crossfire with a Q6600 @ 3.6ghz (i Know!!!) playing games at 1920x1200. Looking at the graphs the core i5/7 processor advantage is less pronounced as the res climbs. I guess i am not gonna upgrade until buldozer or sandy bridge??

Nice artical...thanks
 
Amazing article, really looking forward to the review of AMD's product line. I think that article will really set the tone as far as budget builds are concerned. I know this may be asking a little much but what's the time frame as far as the next installment of this series?
 
ok why did they use GPUs from last year? I have an x3 720 unlocked and OC'd and am in need of a GPU. This article helps only a little. Why isn't 1920x1080 resolution used in benchmarks? Isn't this more popular than 1920x1200? And I want to see comparisons of the GTX 470 and the Radeon 5870 and 5850. Not the 4890 and GTX 260. Am I the only one who sees this as a severly flawed article for this reason?
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom