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

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...

Could have used a 6 or 4-core FX and made more money for a better cpu cooler and case. You have already demonstrated that more than 4 cores aren't used in gaming and here you have an 8 core CPU...

Thats like... The worst possible $1k build. wow

@Serhat i agree with you but this would be a better all around build. I think he could have done better though

Swap mobo for ASRock Z68 PRO3 GEN3 LGA 1155 Intel Z68 and processor for a sandybridge i5-2500. No overclocking needed.

Dat hitachi HDD, you guys do know that Seagate Barracudas are around $70 for 64MB cache 1TB storage and SATA 6 right? and arguably are far more reliable

Why wouldn't you drop down a bit to the FX-8320, that's about $40 saved, that could save you enough money to get the 7970 which clock for clock is better than the GTX 670.

serhat359Could have used a 6 or 4-core FX and made more money for a better cpu cooler and case. You have already demonstrated that more than 4 cores aren't used in gaming and here you have an 8 core CPU...

Its not 8 core, its 4 core with dual modules per core. Shared resources. Its why you see an increase in performance between a 4300 and an 8320

mouse24Its not 8 core, its 4 core with dual modules per core. Shared resources. Its why you see an increase in performance between a 4300 and an 8320

No, its other way around. It is 4 module cpu. Each module contains two integer cores (thus 8 cores total) and one FPU. It is more like reduced 8 core than full 8 core. Neverthles, Intel still is better.

I am looking forward to 2013. An HTPC with a Haswell K-series processor and an HD 8850 and a Cable card. As long as Simcity 5 is playable at 1080p I'll be happy.

I have a system with a Phenom II that I'm really interested in upgrading to the FX-8350 and I want to see what it can do. For $1K that's not bad at all, but I would have purchased a Cooler Master HAF 912 over an HEC case.

Why not use the SSD and the HDD in RST to make a cache drive for the hard disk?

g-unit1111I have a system with a Phenom II that I'm really interested in upgrading to the FX-8350 and I want to see what it can do. For $1K that's not bad at all, but I would have purchased a Cooler Master HAF 912 over an HEC case.

I just got a Thermaltake Chaser MKI and it's not too shabby at all. It's lighter than my old TT Kandalf, and although it uses more plastic it's not a terrible thing. That said, neither one is good for hauling to LAN parties.

yyk71200No, its other way around. It is 4 module cpu. Each module contains two integer cores (thus 8 cores total) and one FPU. It is more like reduced 8 core than full 8 core. Neverthles, Intel still is better.

If power was more similar, I'd go with AMD still just because I like smaller companies and cheaper prices. That said, I just bought an Ivy Bridge i5 because I was loooong overdue for an upgrade.

It's great that you built a worse PC for $1000 than you did back in August...

But I'd rather see the benchmarks vs the $500 build so I know if that extra $500 is worth it or not.

Although I think MOST of us know that between $500-650 is about all one should really spend these days on a new gaming system.

yyk71200No, its other way around. It is 4 module cpu. Each module contains two integer cores (thus 8 cores total) and one FPU. It is more like reduced 8 core than full 8 core. Neverthles, Intel still is better.

Sorry, yeah your right. I don't know what I was thinking.

A...M....D...
CPU....
*FINISH HIM*
**FATALITY**

It's nice to finally see an AMD platform in the SBM although the performance and power consumption is really a sad one to look to be honest.

I get that this is an alternative to what is the usual setup for a $1000 PC but it's a poor one I guess. At least people get the idea of how hot the FX8350 runs and that regardless whether you're a fanboy or not, if you're spending $1000 for a gaming machine, buy an Intel cpu and a discrete AMD/Nvidia graphics card

Other than power consumption, for gaming, the FX-8350 here did about as well as the previous build's Intel CPU. The only times where it truly did worse, according to these benchmarks, was when it was already far above 60FPS anyway. As I've said before, measuring in frame latency would give a much more accurate show of things, but for gaming and going just by the gaming benchmarks from this article, the FX-8350 kept up excellently in performance, albeit suffering badly in power consumption (which is easily rectified by simply disabling one core per module since there really aren't any games where a 4M/4C FX-8350@4.3GHz (could go a little higher thanks to the extra thermal headroom while still dropping power consumption) is weak. For the FX-6300, slightly higher frequency could easily offset the (then) one fewer cores and the much lower price could put some budget elsewhere.

Thanks for giving Vishera a try, Toms, but you could have done better with it if you wanted to.

JonnyDoughIt's great that you built a worse PC for $1000 than you did back in August...But I'd rather see the benchmarks vs the $500 build so I know if that extra $500 is worth it or not. Although I think MOST of us know that between $500-650 is about all one should really spend these days on a new gaming system.

There will undoubtedly be such comparative benchmarks in the follow-up to the last SBM build this month.

Also, there's no bad reason for spending more than $500-650 on a gaming machine. It's not until far over $1000 where you start to run out of extra performance for extra money.

A few points ;

1. Most of the rendering software are FP heavy, disadvantaging the AMD from the start. Maybe add some other render software that are more integer heavy ?

2. The files you use in winzip17 OpenCL test are probably < 8MB in size. Hence winzip doesnt send them to GPU for compression. A better selection of files should be used in which some are >8mb in size and a few are 50MB or greater. Otherwise, the OpenCL test is meaningless.

A $70 Corsair CX600 is a waste of money here. If you pick that PSU, at least include the $20 mail-in rebate Newegg offers. If you don't, there are better options. Like an Antec Neo Eco 520C for $55 + $4 shipping:
http://www.newegg.com/Product/Product.aspx?Item=N82E16817371030

Or even better stuff from other vendors. An XFX Core Edition PRO 550W for $64 before a mail-in rebate, for example:
http://us.ncix.com/products/?sku=59615&vpn=P1550SXXB9&manufacture=XFX&promoid=1257

Note that the XFX 550W has 44A on the 12V rail while the Corsair CX600 only has 40A on the 12V rail.

A nicely written article illustrating a bad plan which was poorly executed. Don't get me wrong, I wanted to see an AMD comparison do itself some justice against an Intel platform, even though most people knew already in their heart of hearts that it wouldn't win. But what we have here is an attempt to play lipservice to AMD and nothing more. There definately could have been a much better choice of components however I would have to suggest that it would have still been futile. Slam the reviewer for that if you want, but when you're starting off knowing you're at a disadvantage it isn't always clear which components are going to be a better fit for your project. What this perhaps illustrates that a lot of people still haven't worked out the best way to exploit the poorly conceived AMD architecture.

Quote:

hat this perhaps illustrates that a lot of people still haven't worked out the best way to exploit the poorly conceived AMD architecture.

There are no obvious shortcomings in the AMD arch. There are some bottlenecks, which would have been cured by Steamroller.