System Builder Marathon, Q4 2012: $1,000 Enthusiast PC

[noob2222]

[citation][nom]silverblue[/nom]Well, per watt, PD is 15-20% more efficient in general. Remember you're getting higher clock speeds as well as about 10% more IPC.The 6100 had a base clock of 3.3GHz, the 6300 is 3.5GHz - a good 10 to 15% faster if you include IPC. Both turbo up 600MHz. The 6200 is 3.8GHz but the 6300, with a 300MHz LOWER clock speed, actually noticably outperforms it almost universally across the board in benchmarks as well as power usage, meaning that the hexacore Bulldozers really were lame ducks. Perhaps the 6200 constantly ran into thermal limits.[/citation]

I think you missed the part where the 6100 was overclocked to 4.45ghz, this build was slower oc at 4.3 ghz and blows the 6100 away in terms of performance. your clock speed arguement is voided.

 

[BreadWhistle]

[citation][nom]Onus[/nom]BreadWhistle, you should read the article, not just the parts list, before posting. A lot is explained. This build was designed to be comparative.[/citation]

I read all of the article thank you very much. I did not see anywhere where I saw "This build was designed to just throw random parts at each other and call it a $1000 dollar PC build. Maybe I'm missing something, but I doesn't make sense in the slightest why you would purposely try to make a build to be non-competitive. Instead of making an article for what ever this is, make one that tells everyone what the best is for $1000, not "k so we tuk sum partz and it is 1000 cash so k"

Bleh

 

[cleeve]

[citation][nom]BreadWhistle[/nom]Maybe I'm missing something, but I doesn't make sense in the slightest why you would purposely try to make a build to be non-competitive. [/citation]

I do not believe that you could choose a $1000 build around the FX-8350 that performs any differently, aside from cooling, which I talked about (to refresh, cooling would cost more money and power usage would soar).

Or are you able to magically get 200% faster RAM, SSDs, and hard disks for the same price? Perhaps you're suggesting another brand of motherboard would somehow provide a mystical performance increase?

Or are you saying the 8350 isn't competitive? I don't agree. In many multithreaded metrics it beats the Core i5. it depends what you're looking for, and the SBM builds have *always* been all rounders. That's why we test both apps and games...

 

[army_ant7]

The other was Skyrim at 2560x1600 O/C. The fact that the FX out-paced the i5 at the highest resolution but not at any lower resolutions - even when the FX CPU appears to be a bottleneck at lower resolutions - indicates that some other factor is enabling the current O/C build to outperform the August O/C build. Could this be GPU memory O/C, or could it be something like the PCIe BW and latency differences between the two architectures?

Aside from what Paul already mentioned, would it be possible that maybe Skyrim has more threads to be processed as certain effects or (higher) levels of graphics are set? Or maybe the lighter threads become heavier in a way that they start "craving" for other cores to process them or something?

Dude, the system pulled almost 500W under load.
I would not pair it with a 550 W PSU. Maybe I'm old school but there should be more of a margin than 50 W.

Well Don, It Gets Complicated when you're using a low-efficiency power supply. Say you have a system that pulls 600W peak power from the wall and has 70% efficiency, you need "at least" 420W and, even with padding (extra for unforseen circumstances), a quality 500W unit would do. In that situation, the choice between a high-efficiency 500W or low efficiency 600W unit should favor the higher-quality part. And then, maybe the draw at your socket will drop from 600W (at 70% efficiency) to 477W (at 88% efficiency)

Hehe... I look up to both of you guys, but as I saw your comment Don, I did get the idea that Sakkura and Thomas had, that efficiency comes into play when you measure at the socket. I'm guessing with age (in the case of Thomas, at least I assume he's older) comes wisdom?

Plus, if Sakkura's info is correct and that 550W-rated XFX PSU does have 4 more Amps on the 12V rail than that Corsair PSU, wouldn't that provide for better headroom for the CPU and GPU, especially when they're overclocked? I could be mistaken in thinking that both those components only or at least mainly draw from the 12V rail though. If ever you have some underlying reason as to why you'd still insist on the Corsair PSU you chose, I'd be interested in hearing it. I'm not questioning your choice as of now because I'm aware that this build was planned before the actual date of publishing and prices and stocks vary.

I don't mean to sound like I'm imposing. I'm just here to learn from you guys and share thoughts.

 

[BreadWhistle]

[citation][nom]Cleeve[/nom]I do not believe that you could choose a $1000 build around the FX-8350 that performs any differently, aside from cooling, which I talked about (to refresh, cooling would cost more money and power usage would soar).Or are you able to magically get 200% faster RAM, SSDs, and hard disks for the same price? Perhaps you're suggesting another brand of motherboard would somehow provide a mystical performance increase? Or are you saying the 8350 isn't competitive? I don't agree. In many multithreaded metrics it beats the Core i5. it depends what you're looking for, and the SBM builds have *always* been all rounders. That's why we test both apps and games...[/citation]

I didn't mean to come across as aggressive, my apologies. I've just always been argumentative and quick to point out any flaws in things. I did not insult your build, what I meant to suggest was that it was more of a subtle, okay build. Not really threatening to reign king for the most top-notch $1000 computer system IMHO.

I feel that if you replace the 8350 with the 8320 (-$40/Newegg), bump the cooler up to the Hyper 212 Evo (+$10/Newegg), Switch the Agility 3 for a 64GB Vertex 4 (+$15/Microcenter). Then, replace the Hatachi HDD with a Seagate Barracuda ST1000DM003 (-$20/Newegg). Lastly, trade the PSU for the Antec NEO ECO 620w (+$0/Newegg) For an extra 20 watts for these apparent "soars" in power usage.

So you get an almost equivalent CPU for $40 less that can be overclocked greater than or to the 8350 with the better CPU cooler. A faster and more reliable SSD with a mere 4GB more storage for... 4GB more things that need storing? A better HDD for less (No brainer here). Even a different power supply with 20 more watts (Though I believe is irrelevant) for $35 less than your proposed build. It'll do the exact same if not some things better than your build. I don't know though. Maybe I'm crazy?

 

[cleeve]

So you get an almost equivalent CPU for $40 less that can be overclocked greater than or to the 8350 with the better CPU cooler.

Read my conclusion. That's basically what I said I'd be doing for the next build... 8320 with more $ into the cooler.

A faster and more reliable SSD with a mere 4GB more storage for... 4GB more things that need storing? A better HDD for less (No brainer here). Even a different power supply with 20 more watts (Though I believe is irrelevant) for $35 less than your proposed build.

I'm not sure I'd agree any of those items are worth complaining about, since you'd see no difference in performance. As for $35, prices fluctuate on a daily basis on newegg, many of the prices that were in effect when I chose the components were different by the time we published.

I never said the build can't be improved, you basically agree with my own conclusion in the article. But I don't see the $35 difference in your component choice is the difference between competitive and non-competitive... seems more subjective to me.

 

[cleeve]

Hehe... I look up to both of you guys, but as I saw your comment Don, I did get the idea that Sakkura and Thomas had, that efficiency comes into play when you measure at the socket. I'm guessing with age (in the case of Thomas, at least I assume he's older) comes wisdom?

Nah, I'm older and DEFINITELY wiser than Thomas.
He's unquestionably better when it comes to PC technical knowledge though, no arguments.

They're right, I simply knee-jerked a response to Sakkura. As far as the PSU choice though, I'm comfortable with the CX600 because it's been very reliable for me.

 

[f-14]

[citation][nom]CaptainTom[/nom]So a 600w PSU for one 670? Get a 500w, get kingston RAM that is $20 cheaper, a $50-$70 liquid cooler for the FX, and BOOM! More performance for the same price. I get you wanted to test a similar system, but just make that a different article...[/citation]


i have to agree, considering windows 8 is being used and the test comparisons can't even be made to give this machine an even par match up regardless. i did how ever like seeing this chip looked at, despite the choices in parts made in effort to evenly compare something that can not be compared due to HUGE MAJOR differences in the test itself making this almost completely null and void as an on par match up when you mess with the benchmarks themselves it's not comparable any more. very poor scientific method.


clock for clock it's pretty much what i expected to see when the over clock was not matchable.

the on thing i did want to know more about is what is with the 1050 and 1080 droops? seems to me these resolutions are a bottle neck or just a fluke?

p.s. benchmark changes should not have been made until after the new year. this change right now skerwers so much and win8 is such a terrible OS i scream millenium every time i see it and vista. vista had the same problem, nothing was written for it even drivers so benching anything on them until those components and programs were made for it i don't even want to see it benched marked.

 

[Sakkura]

Stuff benchmarks right around identically under Win 7 and Win 8. Apart from browsers at least.

 

[drinvis]

In the photoshop cs6 tests,http://media.bestofmicro.com/W/H/362753/original/Photoshop.png
how is it that with OpenCL it takes much longer than the CPU.
Or are the tests for OpenCL was done with different sample?

 

[NucDsgr]

Seems odd that an engineering sample and retail FX-8350 exhibit temperature differences so. And such a large temperature delta over 200Mhz would lead me to suspect the sins of Sandy Bridge are at Fault here: that the thermal compound is inadequate.

Here is an experiment: remove the FX-8350 from the build, remove the cover plate on the FX-8350, clean off the compound off the silicon and cover plate, apply Artic Silver 5, then replace the cover plate. Try the same thing and see what the temperatures are.

 

[army_ant7]

Seems odd that an engineering sample and retail FX-8350 exhibit temperature differences so. And such a large temperature delta over 200Mhz would lead me to suspect the sins of Sandy Bridge are at Fault here: that the thermal compound is inadequate.

Here is an experiment: remove the FX-8350 from the build, remove the cover plate on the FX-8350, clean off the compound off the silicon and cover plate, apply Artic Silver 5, then replace the cover plate. Try the same thing and see what the temperatures are.

I'm too lazy to read back through the article but wasn't the issue with heat at least partly because of the CPU cooler used?

Did you mean Ivy Bridge? Because I haven't heard anything about problems with Sandy concerning what's under the IHS (integrated heat sink a.k.a. cover plate), if there was such a reported problem though, I probably must've missed it.
Also, I'm not sure if taking of the backplate of the FX-8350 might be such a good idea. I heard that what Intel replaced with ("crappy") thermal paste was solder, which I think might be solid. I'm just thinking if both companies used solder under their IHS's and AMD hasn't changed that since...

 

[cleeve]

Seems odd that an engineering sample and retail FX-8350 exhibit temperature differences so. And such a large temperature delta over 200Mhz would lead me to suspect the sins of Sandy Bridge are at Fault here: that the thermal compound is inadequate.

Remember that our launch sample was paired with a hybrid cooler for overclocking tests. I think Vishera is just really, really hot when overclocked.

 

[dkcomputer]

This build is hard to beat if you're going to keep the price range crippled with the ssd, however, drop the ssd, pick up $30 more for a z77, $30 more for a 7970, and same price for an i5k, it would blow this one out of the water.

Other option would be to get $35 ram instead of $50, and $55 psu instead of $70, and you can at least go to a z77+i5 and keep the ssd in play with a 670. (see newegg for what I'm talking about, its not trash)

If I was advising a build, I would suggest not getting a ssd for that price range to a customer. You're just on the edge of such a huge increase in performance for gaming, its not worth it for booting faster. It would be a good thing to add on later.

And If you're not gaming on a $1k computer, what are you spending that much money on it for anyways. There's simply not a large amount of people looking for advice on a whetstone benchmark computer, building one for such type applications on a website appealing to the masses would be like advertising sugercane drinks at a diabetes doctors office - if you keep throwing time/effort/money into it you're going to lose your funding.

I feel the job of the producers of these truly useful and awesome articles (I hardcore envy people who get to play with such things and think about what to build next, I do it all the time except - I don't get to build many machines over $550) is to appeal to gamers #1 with their suggested builds. Anyone seeking advice on a specific budget for ripping itunes 24/7 because for some reason thats all they do, can look at specific benchmarks on processors.

p.s. I suggest ssd's for $550 builds everyday for local small business machines. They're great for productivity and being able to shut down the work computers every day, just not for gaming.

 

[Guest]

[citation][nom]dkcomputer[/nom]Thats like... The worst possible $1k build. wow[/citation]
doesnt even have an i5, what is this sh*t

 

[army_ant7]

doesnt even have an i5, what is this sh*t

Read the article. Read some previous comments. Then you'll know what kind of "sh*t" it is. (I'd say the pretty good kind since it's a well-rounded PC competition and not a pure gaming one.)

 

[cmi86]

Your lack of cable management disgusts me.

 

[jonjonjon]

wow. i know you need to switch it up and cant just get a 3570k every time but who in their right mind would buy this? also dont you think a 128GB ssd for $90 and 750GB hdd for $60 would be much better then what you got? i this is as bad as it gets.

 

[jonjonjon]

do you not include MIR's? it seems like i can always throw something together thats much better then what you do.

Intel Core i5-3570K 3.4GHz Quad-Core Processor - $214.99
Xigmatek GAIA SD1283 56.3 CFM CPU Cooler - $19.99
ASRock Z75 Pro3 ATX LGA1155 Motherboard - $69.99
Patriot Intel Extreme Master, Limited Ed 8GB (2 x 4GB) DDR3-1600 Memory - $31.99
Samsung Spinpoint F3R 1TB 3.5" 7200RPM Internal Hard Drive - $60.60
Intel 330 Series 180GB 2.5" Solid State Disk - $129.68
XFX Radeon HD 7970 3GB Video Card - $349.99
Antec Three Hundred ATX Mid Tower Case - $49.99
Corsair Enthusiast 650W 80 PLUS Bronze Certified ATX12V / EPS12V Power Supply - $74.99
Lite-On IHAS324-98 DVD/CD Writer - $14.99
Total: $1,017.20

 

[army_ant7]

wow. i know you need to switch it up and cant just get a 3570k every time but who in their right mind would buy this? i this is as bad as it gets.

People who run productivity programs maybe? Though they are just your opinions, I'd have to say they're over-emphasized and not so well thought-out.

 

[SuperVeloce]

[citation][nom]jonjonjon[/nom]do you not include MIR's? it seems like i can always throw something together thats much better then what you do.Intel Core i5-3570K 3.4GHz Quad-Core Processor - $214.99Xigmatek GAIA SD1283 56.3 CFM CPU Cooler - $19.99 ASRock Z75 Pro3 ATX LGA1155 Motherboard - $69.99Patriot Intel Extreme Master, Limited Ed 8GB (2 x 4GB) DDR3-1600 Memory - $31.99 Samsung Spinpoint F3R 1TB 3.5" 7200RPM Internal Hard Drive - $60.60Intel 330 Series 180GB 2.5" Solid State Disk - $129.68 XFX Radeon HD 7970 3GB Video Card - $349.99Antec Three Hundred ATX Mid Tower Case - $49.99Corsair Enthusiast 650W 80 PLUS Bronze Certified ATX12V / EPS12V Power Supply - $74.99Lite-On IHAS324-98 DVD/CD Writer - $14.99Total: $1,017.20[/citation]
Unrealistic prices for most of non-us readers. I mean, Gaia for $20? Are you crazy?

 

[regina_49]

It's great to see a amd build being recommend for once.Thanks TH keep up the good work!

 

[energyman]

This system uses up to 191 watts (0.191 kilowatts) more than the August build. Average US power rates are 12 cents kw/hr and go up to 36 cents kw/hr. So depending on your usage, under load 24 hours a day, this system would cost $200-$600 more a year in energy cost. 0.191 kw * .12 (12 cents per kw/hr) *24 hours *365= $200.77 at average rates. At 8 hours a day under load costs would be about $67-$200 a year higher.

The money saved in energy costs by going Intel, far outweigh most of the other costs people are trying to save. Money better spent on other components, either now or down the line.

AMD would be well served to not only engineer a processor that can compete on raw stats and raw price, but one that is also far more competitive on overall costs.

 

[blazorthon]

[citation][nom]silverblue[/nom]It's not their fault that AMD makes space heaters. Conversely, nobody tried to hide away Ivy Bridge's poor thermal paste.Performance between top and bottom boards isn't much (unless we're talking Biostar, I guess) so is the money really worth it if you don't need the additional features?AMD needs much higher IPC along with the new decoder setup. It's coming, but will it actually mean anything when it appears?Hell, does AMD actually need to dump AM3+ - is it holding FX back?[/citation]

Differences between motherboards can matter, especially with high-power consumption CPUs that really benefit from more and better VRM.

Also, for everyone talking about IPC, please stop. You don't know what the words means. Piledriver and Bulldozer have identical IPC. Performance per Hz of the CPU frequency is not necessarily the same, but that is not what IPC means.

[citation][nom]energyman[/nom]This system uses up to 191 watts (0.191 kilowatts) more than the August build. Average US power rates are 12 cents kw/hr and go up to 36 cents kw/hr. So depending on your usage, under load 24 hours a day, this system would cost $200-$600 more a year in energy cost. 0.191 kw * .12 (12 cents per kw/hr) *24 hours *365= $200.77 at average rates. At 8 hours a day under load costs would be about $67-$200 a year higher.The money saved in energy costs by going Intel, far outweigh most of the other costs people are trying to save. Money better spent on other components, either now or down the line.AMD would be well served to not only engineer a processor that can compete on raw stats and raw price, but one that is also far more competitive on overall costs.[/citation]

If this was used as a gaming build like it was mostly intended to be, then it would almost never be at full load, even when gaming, so power consumption wouldn't be too much worse than the Sandy Bridge system that preceded it. A more efficient PSU could also shrink the differences greatly.

 

[rydercast]

I'm going to be building my first system. I want to build a gaming rig and I'm wondering if this is a good place to start.