System Builder Marathon, Sept. 2010: $2000 Performance PC

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

notty22

Distinguished

Nope, I don't think the results are saying that. The 480's are performing much worse than the 470 because of the cpu/platform. You would see the same results with the gtx 460's as one member who has a thread going in the forums right now is.
There are always compromises, is gaming the MAIN focus of the rig, or encoding blu rays ?
 

Crashman

Polypheme
Former Staff
[citation][nom]brucek2[/nom]We need a solution for the "users love SSDs but they don't show up in benchmarks" issue. Are there folks who truly believe they are just placebos? If so, an article proving or disproving that theory would be very interesting. If we're past that, it ought to be possible to define some benchmark that quantifies the real benefits. Perhaps one focused purely on app start/exit, system boot, game level load time, game saves to mostly full/fragmented drives, import of full camera SD card into photo library application, etc.Bottom line: just too important an issue to not receive serious consideration and debate in performance oriented systems.[/citation]Past System Builder Marathons have proven that SSD's have little effect on this particular benchmark set. The benchmarks are chosen based on application popularity. Benchmarks are not picked to favor particular hardware, because that's not how people pick applications.[citation][nom]f-14[/nom]once request i will make, is add multiple application's at once performance, you've got multi cores, and one app running, i usually have a minimum of 4 apps going since pentium mmx 200mhz days, now days it's close to 6. most guys i know are downloading (torrents) stuff while playing online fps/mmo and playing music/video, some IM client/webcam/skype/voice chat. i'll burn a cd/dvd and encode cd/dvd's during all that also. it's an insane way to tax a quad core system and i'm curious if hexacore and octacore systems are going to benefit by 50 percent or more. don't look at me like that, i'd make one core run something for SETI, i'm not entirely evil. if you give me the power to do something, i'll find ways to push it to the limit and i'll show it's never enough.[/citation]That's a great idea! I know a lot of people game while they're encoding, and have requested such a test be formulated in the past. Unfortunately I probably put too little effort into the follow through, but I can try again!
 

brucek2

Distinguished
Dec 31, 2008
117
0
18,680
[citation][nom]Crashman[/nom]The benchmarks are chosen based on application popularity. Benchmarks are not picked to favor particular hardware, because that's not how people pick applications.[/citation]

Thanks for the reply. I agree that choosing popular applications comes first. But even just focusing on the popular applications you've already selected, say these specific games, there's been an editorial choice to focus on fps as the exclusive measure of performance while ignoring, for example, game start / level load time. So as a simple way to get started, how about also reporting either game load or level load times for each game on each system you are already providing fps results for?

I'm not sure I agree that the other "applications" I suggested aren't popular -- ie booting which seems about as universal as it gets -- but in the end the exact mix of these "user wait / system responsiveness" type tasks is less important than having at least something.
 

youssef 2010

Distinguished
Jan 1, 2009
1,263
0
19,360
[citation][nom]article[/nom]The complete system looks great, though some users might want to show off internal components by substituting in lighted fans.[/citation]

..or adding cold cathode light for far less hassle.

[citation][nom]Randomacts[/nom]They are only showing the $2000 computer? What about the $500 and $1000 one?[/citation]

Is this the first SBM for you?

[citation][nom]mosu[/nom]Just plain stupid choice of components, don't blame AMD.Why didn't you choose single Radeon 5970 and 1090T,even with 8 gigs of RAM.[/citation]

That would be a SIGNIFICANTLY slower configuration and I think you should behave yourself when talking about people who are definitely more experienced than you.

[citation][nom]sassan_88[/nom]I dont like systems Based on AMD's chipset[/citation]

Fanboys are really bad when they talk w/o logical reasons to back-up their theories

BTW Did you use afterburner 2.0? because it allows you to set the GPU voltage(couldn't help but ask)

[citation][nom]article[/nom] Stability was tested at an ambient temperature of 27° Celsius, and a cooler workspace would have allowed us to push for even higher clock speeds.[/citation]

What about the people living in 37C ambient temperature?

Don't mind toms.Nobody's perfect so, making mistakes isn't a big problem,but repeating them is the real problem. At least it will teach us to stay away from hexacore AMDs for gaming, especially when combined with Nvidia's chips.Thank you very much for the warning,toms
 

schwizer

Distinguished
Jan 14, 2010
121
0
18,690
Is there any way that the dropdown menu in each article could be about 1/4" wider. I keep clicking the white triangle on the right side of the dropdown menu and when i scroll straight down with the mouse, i'm outside of the drop down menu and it disappears again. And yes i know i can click it anywhere, but u just can't train myself to do it.

Nice review though. BTW do two 5890's and an overclocked X4 965 fit into the 2k budget these days? i'd like to see some numbers on that with some new drivers.
 

demonhorde665

Distinguished
Jul 13, 2008
1,492
0
19,280
love the fact you ghguys threw on teh 6 core goodness , here being that im in school for game art design i could that extra oompht for 3ds max. however i'm boggled over why in god's name did you throw nvidia gf 480's in the mix, they are extrememly hot , power hungry adn all around the MOST ineffecient vid cards on the market. not to forget that you even mention AMD Cross fire based chipset main baords are much more advanced now than the older nvidia sli ones.it just makes no since and i think your choice in main bord very well might be the biggest issue with why this system got toatlly trounced by the june build, in my exprience the main board chip set can make a huge difference even on games. seriously should have gone cross fire with 2 HD 5870's


now don't take me wrong i like nivida cards as well at amd vid cards. but seriously you whole video card section would have been better done with cross fire, you would have had a much better chipset , as well as less heat and less power draw on the video cards , lastly to top it off 2 5870's would have cut your price by 200 bucks when compared to 2 gf 480's
 

Shivetya

Distinguished
Jul 12, 2010
7
0
18,510
This reminds me of the quadfire build about a year ago... interesting and well executed idea, but not really enough processor muscle to allow the GPUs to really stretch their legs....which kills efficiency when you have powerhungry GPUS, as 2X GTX 480 or 4X 4850 are bound to be.
On a side note, I am curious how a 1055T SBM with 2 GTX460's in SLI would do. It would be difficult to squeeze into a $1000 build (at least without resorting to 768MB cards) but would probably be doable at $1100 or $1200. It seems like that might be more balanced but still pretty powerful at that price point, as I doubt a 1055T would bottleneck SLI'd 460s.
 

BSMonitor

Distinguished
Nov 19, 2007
167
0
18,680
Ummm, half way thru and the June system with Core i7 is superior to the overclocked Phenom X6... Pathetic. I understand Tom's is constantly pushing to put the best light on AMD, but the difference is obvious. Phenom II is a Core 2 competitor. Core i7 stands well above both. To include anything else in a $2000 "performance" system is ridiculous.
 
[citation][nom]BSMonitor[/nom]Ummm, half way thru and the June system with Core i7 is superior to the overclocked Phenom X6...To include anything else in a $2000 "performance" system is ridiculous.[/citation]
It was an experiment, whose failure very clearly showed that AMD is not for the high-dollar gaming performance segment; I hope this time the lesson is clear to even the most rabid AMD fanboys. That doesn't make this a "bad" machine (would slaughter my Q9450/HD5770), but far from the best use of $2K.
 

cknobman

Distinguished
May 2, 2006
1,123
267
19,660
Worst high performance build ever created by Toms.

I own all AMD/ATI parts but I know their place and it is definitely nowhere near high end.

Please dont waste time by using AMD procs in high end builds anymore. Also the 480 were a terrible choice. No one in their right mind would use those monsters in a realistic build. Too much power and too much money. Like you stated the best you can get right now would be 460's in SLI.
 

kilthas_th

Distinguished
Mar 31, 2010
40
0
18,530
These articles are great. Will read the $1000 guide here in a few minutes. I will say it was daring to go with the AMD chip, six cores or not. I think this article definitely proves why Intel can charge such a premium for its high-end chips, though. Here's hoping for a major realignment of competition with Bulldozer and Sandy Bridge, though it doesn't sound promising for AMD at this point.
 

Marcus52

Distinguished
Jun 11, 2008
619
0
19,010
The i7 970 is essentially a 12-core CPU (granted the hyperthreaded cores aren't quite full cores); the differences between it and the the 6-core Phenom II should be even more pronounced than the differences between the 4-core versions (assuming all cores are used, of course). I would tend to compare this build with an OC'd i7 930 rather than the the i7 970, which would allow you to keep the same video card setup and still fit the budget.

I am very happy to see the 1055 reach 4GHz in yet another builder's system, it looks like the 6-cores are actually more overclock friendly than the 4-cores were. Nice!

When it comes to video card frame rate capability - I've always found them shy of real capability, and that's still the case. I think there are a lot more people with higher end demands than Tomshardware allows for - and I think telling Nvidia and AMD they are doing a great job and producing products that are more than is needed now is sending the wrong message.

As long as I can't run whatever game I can play at max capability at least 60 fps at 2560x1600, the video cards aren't strong enough; That's not even considering things like a billion color capability, 120 Hz monitors needed for 3D, and 240 Hz televisions that exist today. Today's video cards fall flat on their faces with those demands individually, never mind actually combining them all in one monitor (which doesn't yet exist, but it will, barring a replacement technology).

;)


 

geok1ng

Distinguished
Jun 25, 2008
111
0
18,690
And a minor sugestion. This MSI borad may not be the best choice moneywise for a 980a SLI build. The ASUS M4N98TD EVO costs $140, you lose a third PCIE 16x slot and the integrated video but gain eSATA, SPDIF, 2 USB ports, a beefer power regulator.

I believe that the ASUS board could show better OC potential thanks to the 8+1 phases and lower power consumption, since it has no integrated video.

And the 980a chipset shines on the lower price range, with Geforce 460 SLI.

I i will agian ask for NB OC on these hexacore AMDs CPUs, a masure that Anandtech has show to offer at least 16% improvement on gamings benchmarks.
 

geok1ng

Distinguished
Jun 25, 2008
111
0
18,690
[citation][nom]Crashman[/nom]Yes, for AMD's "Northbridge" it is too late, because it was already overclocked by 44%:It's also "too late" to overclock the 980a, because it would require modified cooling for the motherboard. The 980a runs too hot on stock cooling to accept a moderate overclock.[/citation]
i dont get the message. Ypu pasted a BIOS screenshot that showed the NB running at 2016 Mhz, and saying it was a 44% overclock. the AT article makes extensive use of 3Ghz+ NB frequencies, and proof that a 3.2Ghz CPU/3GhzNB Thuban can easily beat a 4.0GHz CPU/2.0GHz NB overclock.

As i understand that BIOS screenshot you posted proves my point that these OC numbers were achievied using suboptimal CPU-NB settings.

Or is the fault on the 980a chipset all the time, that simply can not squeeze all the OC potential of the Thubans?
 

geok1ng

Distinguished
Jun 25, 2008
111
0
18,690
Once again on the subject of OC settings. I still dont get all the details of AM3 OCing, but the hexacore Thbans come in 2 flavor: the $300 16x200Mhz unlocked multiplier 1090T and the 14x200Mhz locked multiplier 1055T.

this article showed a BIOS screenshot of a maximum OC of 14x288=4032 Mhz for the CPU with 7x288=2016 Mhz for the NB.

http://www.anandtech.com/show/3877/asrock-890fx-deluxe-full-review-and-an-investigation-of-thuban-performance-scaling/7 is an article that shows a 1090T running at 16x250= 4000Mhz CPU / 12x250= 3000Mhz for the NB.

As i understand it the 1055T that runs at 14x288Mhz/7x288Mhz could run at 14x288/12x250. Or any NB multiplier that is not the "stock" 7x for that matter. Or does the locked multi implies that the 1055T run at most 14x/7x, are 14x/8x+ multi forbidden?
 

Crashman

Polypheme
Former Staff
[citation][nom]geok1ng[/nom]i dont get the message. Ypu pasted a BIOS screenshot that showed the NB running at 2016 Mhz, and saying it was a 44% overclock. the AT article makes extensive use of 3Ghz+ NB frequencies, and proof that a 3.2Ghz CPU/3GhzNB Thuban can easily beat a 4.0GHz CPU/2.0GHz NB overclock.As i understand that BIOS screenshot you posted proves my point that these OC numbers were achievied using suboptimal CPU-NB settings.Or is the fault on the 980a chipset all the time, that simply can not squeeze all the OC potential of the Thubans?[/citation]
Actually targeted 2200-2300 MHz there, but ended up using the lower multiplier to keep CPU temperatures down. Will keep that in mind and perhaps choose a lower lower core clock and higher CPU-NB multiplier next time that temperatures are a concern.
 

mcvf

Distinguished
Apr 22, 2009
126
0
18,690
Lets look at Crysis benchmarks, max resolution. Both system not overclocked give almost identical FPS! What the heck?! At that point (barely 40FPS) one can hardly assume that CPU would be such a bottleneck making 2x470 and 2x480 equal. Isn't there something fishy in 2x470 versus 2x480 or drivers? I guess best to resolve this would be 4 tests: different CPUs with both sets of cards.
 
Status
Not open for further replies.