System Builder Marathon: TH's $2000 Hand-Picked Build

[L0tus]

Are those efficiency figures correct?

Shouldn't the alternative $2000 system be a lot more efficient than the original one?

 

[hixbot]

It's sad that we have no PC games that can tax the graphics cards below 2560x1600. Personally I don't have a need for resolutions so high. I'd prefer games with tons more detail and lower resolutions. Such is the problem to being limited to console ports.
If we had a PC exclusive shooter that could make 480 SLI sweat at 1920x1080, it would be a damn pretty game that puts consoles to shame.

 

[digitalkryme]

screwed over in the UK again

 

[Crashman]

[citation][nom]digitalkryme[/nom]screwed over in the UK again[/citation]It's all legal and financial, really. Nobody wants to spend a bunch of money to give stuff away.

 

[youssef 2010]

Dump AVG for Bitdefender or symantec

 

[Cache]

I wonder if I sacrifice my aged P4 system to the PC gods, if I'll be rewarded for entering the contest this go around?

 

[Crashman]

[citation][nom]Cache[/nom]I wonder if I sacrifice my aged P4 system to the PC gods, if I'll be rewarded for entering the contest this go around?[/citation]Probably. It's worth a try. What have you got to lose?

 

[liquidsnake718]

good good

 

[triculious]

Aww not allowed for canadians

nor mexicans... nor anywhere-else-in-the-world-ians =S

that's the part that I don't like about tom's giveaways and newegg... I can only drool over stuff and prices (local prices double or triple their american counterparts... and I live 3 hours aways from the border!!)... alas, can't have everything *sigh*

I understand the legal and $$$ mumbo-jumbo... I just don't like it

 

[triculious]

...the new article focuses entirely on various Nvidia Surround resolutions.

which would make sense to me as I don't see anyone on their right mind purchasing several 480s for a 1280x1024 monitor

 

[tsnor]

Guess you've never tried an SSD drive.


citation][nom]hellwig[/nom]You SSD fanboys need to get a life. Sorry, but no amount of Tom's articles is going to make up for you wasting a couple hundred on an SSD drive instead of a better GPU/CPU.Add boot-up times to the benchmarks? That's dumb. How many times a day do you boot your computer, once? I'm running Win 7 on a Athlon 64 x2 off a RAID5 running 4 Maxtor 200GB drives. Guess what, it boots in less than a minute. Even if you could boot up in 5 seconds off an SSD, how is that worth anything to the gamers that would be considering a $2000 PC (sure, with a $4000 PC, you'd have to buy SSDs, you wouldn't have anything else to spend your money on).Face it, SSDs just are NOT practical for any setup involving a constrained budget. Gamers are not boot-time sensitive individuals. There's no real loss if Windows takes 30-more seconds to boot. There's no real loss if you have to wait an extra 30-seconds for your game to load. Gamer's are play-time sensitive. You need max FPS (frames per second) for max FPS (first person shooter) performance. Once your game data is loaded into memory, your harddrive doesn't even factor into the scenario.SSD fanboys are like those people who say you shouldn't use CFLs because the initial current draw when you turn a CFL on is too high. Yes, CFLs have a spike when you first power them on, but this is only a momentary thing. The reduced power consumption while they are on easily makes up for that initial power draw. Its the same with SSD vs. HDD. SSD fanatics want you to focus on the 30-60 seconds extra it takes windows to boot or a game to load. They ignore the hours upon hours of game play, internet browsing, movie watching, etc... that DO NOT BENEFIT from an SSD in any way (and, with reduced capacities, are often hindered by SSDs, and usually involve buying a second HDD for storage anyway).Buy all the SSDs you want, but Tom's Hardware already tested and abandoned SSDs in their system builder marathons because there just wasn't a performance increase to justify it. Yeah, they could add tests that are specifically designed to show an improvement with the SSD, but they could also add tests that show improvements with HDDS (i.e. I propose that Toms include a "writing 800GB of data to a single drive test". Every SSD out there would fail, does that make the test valid?)[/citation]

 

[deletemach_kernel]

dude why the ssd bashing ..... this not an article about ssd's...

but great to point out....to have our page file on a ssd and also create a readyboost partition with the ssd.....i had no idea we could readyboost with ssd's...let them get cheaper i will get myself one.... heard they can even run ide!!!

if i could sacrifice my age old p4 with onboard graphics...i wonder if they will start give aways in my country

 

[Crashman]

[citation][nom]triculious[/nom]which would make sense to me as I don't see anyone on their right mind purchasing several 480s for a 1280x1024 monitor[/citation]It's nothing to do with 1280x1024 being inadequate for pushing the cards, it's all about 1920x1080 being inadequate for pushing the cards.[citation][nom]tsnor[/nom]Guess you've never tried an SSD drive.[/citation]Perhaps not, but Tom's has for the SBM, and from the benchmarking perspective he's right.

 

[JamesSneed]

Quality-wise, the Three Hundred is probably the best case you can get for under $80, but there should have been one more fan in the system given the internally-vented graphics cards the system ended up with. On the other hand, a lot of builders would be more than happy to "settle for" 4GHz at 1.30V, and the CPU will certainly live longer at the lower voltage.

The only major dowside is the Antec case doesnt allow the PSU to draw air from outside the case which may have mattered since your GPU + CPU test is showing 872 watts. Using the graphics cards that don't exhaust out of the case would have increased the case temps which in turn would have increased the temp of the PSU reducing its max power output. Any idea if this could have limited the stability of the CPU OC?

 

[Crashman]

The only major dowside is the Antec case doesnt allow the PSU to draw air from outside the case which may have mattered since your GPU + CPU test is showing 872 watts. Using the graphics cards that don't exhaust out of the case would have increased the case temps which in turn would have increased the temp of the PSU reducing its max power output. Any idea if this could have limited the stability of the CPU OC?

Drawing air from inside the case means the power supply helps to reduce case temperatures. So in this instance, the design certainly didn't hurt anything.

A little experimentation might have shown that leaving both front fans in place and adding a side panel fan as exhaust may have helped to draw warm air away from the graphics cards. Heat rises, so it really would have to be tested experimentally.

 

[Avro Arrow]

I think that the value of the $2000 systems would have been significantly augmented by the use of ATi vidcards instead of nVidia. You could have gotten four 5850s in there for roughly the same price. I have a pretty strong feeling that having four HD 5850s in crossfire will destroy two GTX 480s in SLI by quite a margin. It's essentially the same as running two 5970s at stock speeds.

2 x GTX 480 ~ $1000
4 x HD 5850 ~ $1000

I'm sure that would have put it over the value line like you wanted.

 

[Guest]

I quite like this new setup. I think it is better rounded system. What with the nforce mobo in the other one, and running sli with and AMD cpu. Ugh.

The only thing I might change is dropping the cpu to the quadcore 950, and taking the $600 you save and putting it into a SSD, a blu ray drive, and upgrading the case to a Phantom.

 

[feeddagoat]

Appreciate the feedback on this, guys. This is something we'll consider for the next SBM in the planning stages now!

your welcome.

Just thinking about it more, the only way to evaluate an SSD is look at the main reason for buying one and that is boot/load times. How much weight you give this tho is hard since its time your saving and won't change much either since each system will either have an SSD or a mechanical HDD as a boot drive and this will probably be the same brand each month.
The only way I can think of showing an SSD's value is to add boot, application load times onto the test time. That means that adding a small ssd to the $1000 system could give positive benefit that is measurable but give too large a penalty or advantage. For photoshop you could set the disk as a scratch disk tho with the high end system builder, I can't see this being necessary due to it likely having plenty of ram.
Last possible test I could think of is loading a few large game levels like relic from crysis (crysis seemed to have epic loading times lol). Have a save point at the start of each level and total the time it takes to boot each level.

 

[hangfirew8]

[citation][nom]Crashman[/nom]Drawing air from inside the case means the power supply helps to reduce case temperatures. So in this instance, the design certainly didn't hurt anything.A little experimentation might have shown that leaving both front fans in place and adding a side panel fan as exhaust may have helped to draw warm air away from the graphics cards. Heat rises, so it really would have to be tested experimentally.[/citation]

Lowering case temperatures with the PS means raising the PS temps because it gets nothing but hot air- and with the bottom PS configuration and no fan on the side, the GPU's get a little convection from the side and the PS gets to vent all the hot air.

Why can't anyone at Tom's admit that not buying a couple of cheap 120mm fans was a big mistake? Its not a matter of experimentation, it is just a failure.

 

[Crashman]

[citation][nom]hangfirew8[/nom]Lowering case temperatures with the PS means raising the PS temps because it gets nothing but hot air- and with the bottom PS configuration and no fan on the side, the GPU's get a little convection from the side and the PS gets to vent all the hot air.Why can't anyone at Tom's admit that not buying a couple of cheap 120mm fans was a big mistake? Its not a matter of experimentation, it is just a failure.[/citation]Did you read the article? Everything was disclosed, including the fact (read the June build with the same case and fans) that externally-vented cards originally picked for this build would have reduced case temperature, followed by the fact that a side-panel fan was not ordered because the original order used externally vented cards. Further disclosure was provided in the article stating that, had we known the substitute cards would be internally vented, the fan would have been added to the order. That would be one additional fan, not three because the case is otherwise filled with fans. The only thing I mentioned for experimentation was using the added side fan as exhaust, rather than intake. With everything completely disclosed within the article, there's nothing left to admit.

 

[redgarl]

Not surprising seriously. I bet the 2 GTX460 with AMD 6 core cpu would have done better than the 2 GTX480.

Also, we need to admit that this cpu is... well... bad...

AMD really need to put something amazing on the table like they are doing for mainstream market where they are the best choice.

 

[hangfirew8]

[citation][nom]Crashman[/nom]Did you read the article? Everything was disclosed, including the fact (read the June build with the same case and fans) that externally-vented cards originally picked for this build would have reduced case temperature, followed by the fact that a side-panel fan was not ordered because the original order used externally vented cards. Further disclosure was provided in the article stating that, had we known the substitute cards would be internally vented, the fan would have been added to the order. That would be one additional fan, not three because the case is otherwise filled with fans. The only thing I mentioned for experimentation was using the added side fan as exhaust, rather than intake. With everything completely disclosed within the article, there's nothing left to admit.[/citation]

When I said "a couple" of fans I enumerate that to 2, not 3.

While there is plenty of disclosure (good for you) I maintain that the fail happened because original plan was not a good one. My experience with NVIDIA reference externally vented cards is that they still put a large amount of air and heat into the case. Just look at them, that have large gaps and/or slots in the shroud near the card edge.

While I love the 300 for budget builds (I've used several) a tight M/B dual card build with one is a poor plan unless special ducting is used to get fresh air into the sandwich space on the intake side, and hot air out of the sandwich on the exhaust side.

So to make it perfectly clear, I think the original plan was perilous to begin with. I would have ordered that extra fan along with the externally vented cards, plus another (custom internally mounted) fan for driving some serious internal ducting, plus some craft foam or something to make ducts with. If Tom's is not into that sort of thing, then the 300 was really the wrong choice. It is both a great budget case and a great modder's case, but if you're not planning to mod for a tight hot dual SLI setup, then you're planning to fail.

 

[Crashman]

[citation][nom]hangfirew8[/nom]When I said "a couple" of fans I enumerate that to 2, not 3. While there is plenty of disclosure (good for you) I maintain that the fail happened because original plan was not a good one. My experience with NVIDIA reference externally vented cards is that they still put a large amount of air and heat into the case. Just look at them, that have large gaps and/or slots in the shroud near the card edge.While I love the 300 for budget builds (I've used several) a tight M/B dual card build with one is a poor plan unless special ducting is used to get fresh air into the sandwich space on the intake side, and hot air out of the sandwich on the exhaust side.So to make it perfectly clear, I think the original plan was perilous to begin with. I would have ordered that extra fan along with the externally vented cards, plus another (custom internally mounted) fan for driving some serious internal ducting, plus some craft foam or something to make ducts with. If Tom's is not into that sort of thing, then the 300 was really the wrong choice. It is both a great budget case and a great modder's case, but if you're not planning to mod for a tight hot dual SLI setup, then you're planning to fail.[/citation]That's a fine opinion, but the June $2000 build didn't have the case heat problem when using the same case and motherboard with a pair of GTX 470 graphics cards. GTX 470 cards are hotter, but those were externally vented.

You'll probably get a nicer case next time anyway, since a cheaper CPU will allow for it. But I do have to say, yes I must say, that the case used in this article has better heat removal than most of the cases that readers recommended. Including the HAF's.

 

[Guest]

3.) The Antec Three Hundred is "proper bottom mount", please check the photos!

Nope. Actually, while PSU is bottom mounted, there are no holes on the bottom of the case, so (unless modded) PSU must be facing fan up. That eliminates the biggest advantage of bottom mount PSU: cooling it with ambient air, which obviously reduces temperatures and noise.

1.) The SSD is mentioned in the article. The problem is, it increases only synthetic test scores and the value analysis only uses "real world" benchmarks so it's basically a non-starter.

Indeed, and suite is the problem. Some multitasking benchmarks would highlight the difference.

 

[lamplighter]

I bought this rig on newegg. For the life of me I can't get the mushkin RAM to report all 6GB's with this Gigabyte mobo using standard BIOS settings (e.g. I have not tweaked or even attempted overclocking. When I did attempt using the settings in this article, the box refused to boot). I have RMA'd both the mobo and RAM, still the same problem. I went to Fry's bought some "Patriot" memory 3x2GB DDR3 PC3-10600 (1333MHz). Now 9 out of 10 times the mobo boots correctly seeing all 6GB's, the other 1/10th of a time the mobo reports 4GB at post, and right after "Verifying DMI pool", POST reports DRAM size mismatch and reboots immediately. This behavior occurs constantly with the mushkin RAM, and apparently the only way around it is to enter BIOS and quit without saving, then at least the box boots with 4GB detected.

BTW, once booted the system runs well, either with all 6GBs detected or 4GB. Win7 always sees 6.00GB installed(4.00 GB usable) with mushkin. I ran a stress test tool, name escapes me (probably found reference to it here on Tom's), system ran smooth at around 45C.

At this point I'm not even interested in overclocking it, I just would like to see the box boot with 6GB.

Am I asking for too much? Is this expected? I appreciate any advice.