Part 1: Building A Balanced Gaming PC

[Summer Leigh Castle]

Very good article. I have a Phenom 965 with an aging 4850 and would like to see the gains from moving up to 5850.

 

[brockh]

I would like to read Quake3 / Quake live tests.

And from lower resolutions, and "pro" configs (almost with all minimun) to the unplayable raytraced http://graphics.cs.uni-sb.de/~sidapohl/egoshooter/.

Why?
1) Because is the best FPS ever made, the most played, and now that is free with quake live more.
2) Because is playable from old Nvidias TNTs to modern computers, and hard gamers still play in 640 x 480 120 Hz in CRT with a 120 fps (as maximun in config) for stable playing (ask any ciber athlete pro) with ultra low details and special configs for playing better, even in fast network mode.
3) Because is very sensible to CPU and GPU advances
4) Is multiplatform, can be tested in Linux, Mac OSX, and of course, MS WOS (opengl)
5) It is played and it will be payed for every FPS gamer in the world, even the childs that are not even in their parents mind, is a classic, and everybody knows it, at least for basic knowledgement of FPS games.
6) Is a real standard that everybody can test and compare from their computers 7) You can use ioquake that is open source to make a special test batch, and even an interface for uploading results and save time in making benchmarks.
8) It has gamma, from lower resolutions to high resolutions and raytracing
9) and it will last, because is fun, classic, and there is not possible to play raytraced version with actual computers.

The games tested are designed to put stress on the CPU & GPU. Quake 3 & Quake Live will not really ever stress the cards tested to the point where they would drop at or below 60 FPS; Quake Live might cause some problems, but that's only because it's distributed in Adobe Flash. That's not to say it's not a good game, but it's not relevant in this kind of testing. Also, looking at that "unplayable raytraced" (assuming you meant stressful on the system), it doesn't look like it would pose any threat to modern day computers. The test system was run on a cluster of 20 1.8 GHz "AMD XP1800" processors and alternatively a "powerful" (circa 2004) GPU. :/

Are you aware it's almost 2010? 😉

 

[donpacific2k]

It's about time for an article like this!! Suggestion: float an icon over the lines instead of those colored triangles and squares. Too hard to tell which one is which without constantly looking at the key

 

[JonnyDough]

But, in the end, the cheapest E6300/Radeon HD 4850 combination is still capable of dishing out playable performance on a 22” widescreen at its native resolution.

An FYI, so is my Opteron 185 (socket 939) running at 2.8+ghz with an OC'd 8800GTS 640mb. I average between 40-45 FPS @ 1680x1050, with a mere 2GBs of ram. I used NLite to strip XP though. 😛

 

[castilo]

I wasnt going to post because there were already 100+ posts but I just have to say THANK YOU! I liked this article so much I even disabled my script blocker on this site. I asked this exact question in the forums a few days ago and no one really had a definitive answer, now I know that I should bump up my gpu before bothering moving to i5/i7. This is so much more helpful than the usual product tests! Maybe one suggestion, is there still a point in running tests at 1680*1024? How many people who read this site are actually running that resolution, especially while running i7 and a 295? Maybe toms should have a hardware survey to see what its readers actually have in their systems?

Thanks!

 

[dark41]

Castillo,
I think you'd be amazed at how many TG readers are using less than top of the line components. As a custom system builder I very rarely get a request for top of the line CPU or GPU, even by avid gamers. Myself, I'm much more into bang for the buck, good performance for the best price, rather than the best performance for whatever manufacturers think they can squeeze out of me. 🙂

 

[dark41]

Sorry, Castilo*. Sure would be nice to have an edit option here. 😛

 

[Onus]

I agree, Dark41. While I am considering a 1080p monitor, there are probably quite a few people still happy with 1280x1024 and 1440x900; maybe even less. I think 1680x1050 will be around for a long time yet.

 

[pauldh]

Our original plans for this series were to test 3 resolutions. I added 1280x1024 for parts 1 & 2 but don't think we can keep testing that one. If we add hardware, something needs to give in order to keep the project manageable, cut down benching time and allow us to bring the stories to you a bit quicker.

 

[kmodeler]

I like the review. Clealry the GPUs still rely heavily on the CPUs with different levels of parallelization (ability to exploit multiple procs and hyperthreading). To that end, it might be useful to indicate % CPU load and/or proc utilization as additional axis/value which would have implications when running multiple processes/background tasks which you eluded to.

 

[LLJones]

Like many people out there, I was wondering what I could do for an upgrade to my rig. I decided on the X3 720 BE. Have not installed it yet. The video card was a much harder choice and it is to bad they left out the the 4770 Crossfire.

I decided to use the CF 4770 setup. I went with CF as the XFX 5770 is $200 and I lucked out and found a sale on the XFX 4770 @ $110 each. DX11 would have been nice, but there is nothing really out there to use it and I am going to do a serious upgrade next summer.

I have put the 4770's in and can say, Holy Schnikey!, this is Very nice. It replaced a tri fire 3870 setup. Absolutely no regrets whatsoever.

 

[sasquatch451]

Great article. Thanks. It would be nice to see SLI/Crossfire in part 2. And the Radeon 5870 too.

 

[fevzi]

I would like to make a humble request.
Nameley one of the most popular game out there is WOW I would actually like to see the chart on this game. According to many peoples view it's a CPU intense game, but still I would enjoy seein the chart.
Thanks for a great review.

 

[neiroatopelcc]

[citation][nom]fevzi[/nom]I would like to make a humble request. Nameley one of the most popular game out there is WOW I would actually like to see the chart on this game. According to many peoples view it's a CPU intense game, but still I would enjoy seein the chart. Thanks for a great review.[/citation]
It's been mentioned before, but unfortunately wow is very unpredictable.
players, monsters and npc's aren't always in the same places, it's not always the same time of day, and server lag along with a slow internet connection add to the lag. For instance a 2Mbit isn't sufficient to make dalaran feel lag free. In any event, most mediocre cards can run wow at native resolution on a 22"

 

[airgreek]

where is part II?

 

[Guest]

This is realy great stuff. Could be such work be done on MMO games like WoW/Aion/Warrhammer/AOC and so? This games have quite different needs than regular single player games and such work would be realy appreciated and more long living becouse you dont change MMO as often, so getting right rig is quite important. PPL say that these old engines dont need any new rig, but in real massive pvp it comes out as real problem in most of this games.

 

[cnox]

Hmm, Part 1 isn't in the featured list anymore, and everyone is waiting for Part 2. Let's push the button and get the article published, already.

And it would be interesting to see a MMO hardware setup because you don't have AI and cpu intensive tasks (theoretically) so it might favor more GPU than CPU. Also, those apps are more network heavy (such as Eve) so would be interesting what high network load would do to application performance.

 

[mapesdhs]

As others have said, excellent article, very nice graphing
structure; toms is back in the game for useful info!

A few questions though, if I may...

What was the rationale for testing with the i7 920? Previous
articles have shown this is not good value for money when it
comes to gaming as compared to the i7 860. Or was it merely
to show precisely the point that the high cost of the 920
doesn't help that much over an older quad-core? Either way,
I hope the 860 can be included later, and the 750 if possible
as I'm sure that will be a very popular choice for those buying
P55 systems, perhaps the most common option (especially oc'd).

What was the reason for not including the 5850/70 cards in the
first instance? Looking forward to seeing the results included
later if that can be done, but a tad surprised not to see them
on the graphs straight away, though I suppose it might clutter
things a bit. I agree with the suggestions of having a mouse-
over feature to show CPU/GPU combo and fps number when hovering
over each data point, that would be an excellent feature.

Can you please include at least the Athlon II X4 620 in your
next piece, alongside the Phenom IIs? There's been much discussion
as to whether L3 helps for gaming; the way you've done these
tests should finally put the issue to rest if you can test
the 620 aswell, _especially_ oc'd since IMO that's one of the
key attractions of the 620. I'm in the process of planning a
new rig for my brother using a 620 and 4890, so this article
series has come at the perfect time.

Lastly, I do understand your point about including too many
results in a single graph. The way you've presented this data
is certainly very good. But surely there's an easy way to be
able to show a lot more results on a particular graph: use a
much bigger image! Does toms *have* to have so much advert junk
all around each article? It's a bit nuts that people are using
so many 1600+ wide displays these days, yet the typical pixel
width of the main toms article column is only about 600 wide.
Why not simply show much larger/wider images with a lot more
data points? Or how about thumbnail with standard-type links to
a full size image? (I mean a basic HREF, not some fancy way of
popping up the larger image) And please grud do not use Flash,
it's a total PITA, eg. guru3d.com. For example, I used reduced
thumbnail/linked images on one of my SGI CPU comparison pages,
works rather well.

Overall though, fantastic job! Sincerely looking forward to Part 2.

Ian.

PS. Will there be 5850/70 CF results aswell?

 

[john boy]

Kudos to you for this and the upcoming series of articles. I had already purchased components for a new build, and I'm glad I caught this article. At the last part of Conclusion, you showed where the Radeon HD 4890 was at a high rate of performance and balance when paired with the Dual Core CPU's. The Radeon HD 4870 X 2 is often well balanced with the E8400 and Q9550 but at times it trailed behind the Radeon 4890 when paired with a weak CPU.

I'm curious to what the results would be with this test using an Intel E8600 3.33 GHz chip, which I just purchased would be. I also just purchased two new Asus EAH 4870 DK 1G Video Cards, and 4 GB's of GSKILL DDR-1200 PC2-9600,and Western Digital Caviar Black 1TB Sata 32MB 7200 HD, and the same exact Corsair CMPSU-850 HX850 watt power supply you tested. The only thing now I have to consider and purchase is the Motherboard. I was using a Asus P5Q on this setup, but wanted a better board that also supported two or more video cards, and i was stumped until I read your review. So now I have decided on the ASUS Rampage Formula 775 M B with the two PCI-e slots that can support the two Asus EAH 4870 DK 1GB cards I have, and the 4Gigs of GSKILL DDR-1200's.

I know the E8600 is two steps up from the E8400 you tested, but the only thing that now concerns me is after reading this article on the graphics cards, is how much of a difference is going to be using the two Asus EAH 4870's DK 1G Cards versus the Radeon HD 4870 X2 cards in this setup.

Before I ever read this article, and then after reading it, I was pretty close on this set up you folks selected when choosing components, with the exception of the Motherboard and Graphics Cards, and I know it will be better when I change over to the Asus 775 Rampage Formula, and the E8600 CPU, but are wondering if the two EAH 4870's I have will do the job compared to the Radeon H D 4870's x2.

I wouldn't have known about this on how to choose these products to build out a well balanced system and the tests you do help out so much in choosing components for a build. Excellent 1rst article, and am sure the rest are going to be informative and inpressive as well.

 

[terr281]

In reference to posts:

1. The article stated that Crossfire/SLI would be tested in the future. Patience is a virtue...
2. I am in the camp that would agree with the dropping of 1280x1024, as most new gaming systems at least run at 1680x1050 on a 22" WS monitor. ( http://en.wikipedia.org/wiki/Display_resolution , Steam survey, August 2009 ) Yes, 1280x1024 is still higher, but I am of the opinion that it is being slowly replaced.
3. The article stated new hardware, the ATI 5xxx series (and, with the edit, Intel I5s) would be included in time. Patience...
4. (To john boy, above) I believe previous reviews have shown that 2 times X card is generally better than a single card with 2 of the same card. (Example: 4870 X 2 versus 2 4870s) When dealing with your specific setup, I cannot... obviously... be for certain, but I can tell you I have 2 friends with near identical systems except for this exact difference. (Both are i7920s, 3 x 2 GB DDR3 1333 of same timings, etc...) The 2 x 4870 1 GB system has a PCMark score a bit higher than the 4870 X2 system. Different systems of identical parts will perform differently, but you shouldn't worry about the choice of 2 x cards versus a 1 x "2 cards in a single card" setup.
5. As for the choice of a i7920 for testing, I would recommend keeping it in the setup. Despite the recent i5 showing it performs "better" than the i7920, many users have a LGA 1366 system. Further, if this is to be a long running series of articles, then... a year or 3 from now... we may be at a point where the i5's limits versus the i7 show. (2 channel vs. 3 channel memory, 16x PCI-E bandwidth total limit, etc.) Further, with the pseudo-known release of the hexacore 1366 CPU next year, you will have the previous results available for comparison.

 

[mapesdhs]

Good points on the future inclusions, I was thinking more in
terms of whether the data would be added to the existing graphs.

I agree with keeping the 920 in the results. What I meant was,
from a value point of view, I reckon 860 results would be more
immediately useful. In other words, all three would be good -
750/860/920, but definitely not 870 and the ones above 920.

Ian.

 

[daveinphilly]

Perhaps there is something wrong with your testing? I see on this test that with the game Crysis @ 1680x1050 and a i7-920 processor and a HD 4890 card, you get 32 FPS with no AA.

But on your previous test, last month, the one called ""Radeon HD 5770 And 5750 Review: Gentlemen, Start Your HTPCs"", with the same HD 4890 card, the same Crysis game at the same rez, except with a slower i5 processor, you get a FPS of 49?

Huh? That makes no sense that a slower processor can get you a 50% increase in frame rates. Yes/No?

 

[pauldh]

Perhaps there is something wrong with your testing? I see on this test that with the game Crysis @ 1680x1050 and a i7-920 processor and a HD 4890 card, you get 32 FPS with no AA.

But on your previous test, last month, the one called ""Radeon HD 5770 And 5750 Review: Gentlemen, Start Your HTPCs"", with the same HD 4890 card, the same Crysis game at the same rez, except with a slower i5 processor, you get a FPS of 49?

Huh? That makes no sense that a slower processor can get you a 50% increase in frame rates. Yes/No?

Nothing wrong, just completely different test settings. That review tested Crysis at "High" details, this one cranked them up to "Very High".

 

[pauldh]

As others have said, excellent article, very nice graphing
structure; toms is back in the game for useful info!

A few questions though, if I may...

What was the rationale for testing with the i7 920?
...
What was the reason for not including the 5850/70 cards in the
first instance?
...
Can you please include at least the Athlon II X4 620 in your
next piece, alongside the Phenom IIs?
...
But surely there's an easy way to be
able to show a lot more results on a particular graph: use a
much bigger image!

1) Hardware gathering was prior to LGA 1156 (i5/i7) launch and testing had already begun. As in our SBM series, i7 920 was a clear choice at the high-end of recommended gaming CPUs.
2) Same reason, testing was finished at this time. We are looking to add i5/5800's for remainder of series.
3) Part 2 (AMD CPU's @ stock clocks) is already finished. We have considered adding that Athlon II,; if it does appear it wouldn't be until part 4 (AMD OC'ed). Each CPU adds a lot of bench time.
4) Unfortunately these charts are the maximum width we can use.

Thanks for the feedback. Have a great weekend!

 

[mapesdhs]

> 1) Hardware gathering was prior to LGA 1156 (i5/i7) launch and
> testing had already begun. As in our SBM series, i7 920 was a
> clear choice at the high-end of recommended gaming CPUs.

Oh I see! So the work began on this one quite some time ago. Ah well.


> 2) Same reason, testing was finished at this time. We are
> looking to add i5/5800's for remainder of series.

And the 860 I hope, as that's the really important one re
comparing to the 920.


> 3) Part 2 (AMD CPU's @ stock clocks) is already finished. We
> have considered adding that Athlon II,; if it does appear it
> wouldn't be until part 4 (AMD OC'ed). Each CPU adds a lot of
> bench time.

I understand; but if you can, I'm sure a lot of people would be grateful.


> 4) Unfortunately these charts are the maximum width we can use.

As in displayed on the page, yes, but surely one can make the
image act as a link to a much larger version which can then be
viewed just on its own? eg. open in new tab, Save Link As, whatever.


> Thanks for the feedback. Have a great weekend!

Well you too! 😀

Ian.