System Builder Marathon, Q3 2013: $1300 Enthusiast PC

[geok1ng]

This build is a very instructive example of how to pick the wrong parts and waste someone's time and money. Pick up a processor that every single showed showed that can't overclock as well as a 3570k, combine it with an overpriced motherboard and VGA and you end up with this.
The old and trsty way is better: 3570k+ CF 7950. You may even save some cash to get a bigger SSD.

 

[slomo4sho]

Yet more proof if any were needed that a used 2700K is still far better value
when it comes to an oc'd setup. 5GHz no problem, no need for water
cooling, leaves the 4670K way behind. Ditto 2500K (I notice there are
loads on eBay US); why anyone would buy a 3570K or HW is beyond me.

I am unsure where this misconception arises from...

4670K average on air is 4627 MHz
http://hwbot.org/hardware/processor/core_i5_4670K/

2500K average on air is 5041 MHz
http://hwbot.org/hardware/processor/core_i5_2500k/

Clock for clock, the 4670K is roughly 10-15% faster than the 2500K so a 4627 MHz Haswell is equal to a 5089-5370 MHz Sandy...

 

[SvRommelvS]

The Haswell is drawing more power than the ivy bridge? I thought the whole point was a CPU designed around longer battery life.

 

[slomo4sho]

The Haswell is drawing more power than the ivy bridge? I thought the whole point was a CPU designed around longer battery life.

This is not a laptop...

 

[g-unit1111]

the vengence ram Worst chip in all the model and brands !
http://www.newegg.com/Product/Product.aspx?Item=N82E16820313399
Team Vulcan 8GB (2 x 4GB)DDR3 2400 best P/p memory with same price better choice.
and corsair H50 not is better whan aircooler.
the Phanteks PH-TC12DX or V6GT with same price range the beat performance.

for ssd i,m recomend Kingston HyperX 3K 120G with same price and beter speed.

Really, you think a terrible off brand with a high speed is better than Corsair Vengeance?

And that Kingston drive you recommend is based on the old, slow, and very problematic Sandforce 2.0 controller. The Samsung 840 and 840 Pro are far better choices.

 

[slomo4sho]

Really, you think a terrible off brand with a high speed is better than Corsair Vengeance?

Team is a terrible brand? That's news to me... You have any support for this claim?

 

[-Fran-]

So my 2700K at 4.6Ghz is still faster than this i5 I'll assume.

Hasfail all over it. I bet the $650 build with this cooling solution would be an even better buy.

Cheers!

 

[Onus]

I read this SBM article before 7:00a this morning, and have contemplated it (and these remarks) at some length before commenting. This place sounds like an echo chamber, and IMHO for good reason. I believe in "targeted overkill" to obtain future-resistance, but this build used an excessive motherboard and a CLC cooler to no clear purpose. Was the author relying on a 4.6+ OC that didn't materialize in order to "sell" this build? That might be some excuse, but illustrates the folly of relying on the silicon lottery. I believe that a $50-$60 cheaper mobo and a $20-$25 cheaper air cooler could have matched these results. The extra money could have bought better RAM which, in this case, seems to have made a difference, or a larger SSD. I'm not going to niggle at all on the graphics card; prices are different today from what they were when this was ordered, so comments on HD7970 prices are moot. This build also reinforced my belief that anyone running an i5-3570K has no performance-based reason to upgrade.
I prefer SBM builds that really squeeze the value out of every dollar spent, and I'm a little disappointed that this one failed to do so, despite otherwise good performance.

 

[egowhip69]

I think you have too much money for a build like this...

I cookie cuttered a very similar build over in newegg, and managed to shave a couple hundred dollars off.

You need to look to do the build with something like $600, 750, and $900.

With those kind of budgets, you could actually show what spending an additional $150 would get you. That will show what upgrades you can get fairly cheaply, and give people a realistic number to save a bit more money for.

I mean, really, who in the heck starts at a $650 budget, and then goes, "Fack it! I'll double it to $1300!"?

 

[iam2thecrowe]

would have liked to see less spend on CPU and motherboard, and certainly remove the crappy liquid cooling solution you used. Put more into the GPU(s). Crappy build. I disagree with all the comments about the 7970 instead of the 770, your saying to spend less on the gpu, and then complain they spent too much on the cpu/motherboard cooling......wouldn't you want to spend more on the gpu for the same price point......

 

[mapesdhs]

I am unsure where this misconception arises from...

It would be a misconception if the builder for the article had been able to get the
HW to that kind of level, but he couldn't, and most don't.

Simple fact is, it's much easier to get a SB to 5GHz than it is to get a HW to 4.6+.

Ian.

 

[slomo4sho]

It would be a misconception if the builder for the article had been able to get the
HW to that kind of level, but he couldn't, and most don't.

You do realize what an average is right... The greatest number of people will be within a standard deviation of the average.

 

[mapesdhs]

You do realize what an average is right... The greatest number of people will be within a standard deviation of the average.

I've just been trawling through a bunch of launch reviews; most could not get remotely
that high, or needed damn good cooling to do it, eg. H100i. Not necessary with SB.
hwbot is likely a biased sample anyway, those submitting more likely to do so because
they've achieved good results, been lucky with their samples. It's not a fair survey of
all those who've oc'd the chip, by definition it's a selective sample.

Go read the reviews, looks pretty clear cut to me.

Ian.

 

[rayden54]

I keep hearing people say to "save money by skipping Haswell." It's weird. When I bought mine, a comparable Ivy Bridge/Motherboard would've cost me almost exactly the same amount as Haswell. /shrug. Maybe it's because I went Micro ATX.

 

[slomo4sho]

I've just been trawling through a bunch of launch reviews; most could not get remotely
that high, or needed damn good cooling to do it, eg. H100i. Not necessary with SB.
hwbot is likely a biased sample anyway, those submitting more likely to do so because
they've achieved good results, been lucky with their samples. It's not a fair survey of
all those who've oc'd the chip, by definition it's a selective sample.

Quite a few reviews of the 4770K have overclocks of 4.5GHz or higher (yes, most use water cooling):

http://www.tomshardware.com/reviews/core-i7-4770k-haswell-review,3521-10.html
http://www.hardocp.com/article/2013/06/01/intel_haswell_i74770k_ipc_overclocking_review/7
http://www.techpowerup.com/reviews/Intel/Core_i5_4670K_and_i7_4770K_Comparison/3.html
http://www.anandtech.com/show/7003/the-haswell-review-intel-core-i74770k-i54560k-tested/3
http://www.overclockersclub.com/reviews/intel_core_i7_4770k/4.htm
http://www.guru3d.com/articles_pages/core_i7_4770k_review,12.html

It all comes down to the silicon lottery honestly. As long as your chip is capable of hitting 4.5 or 4.6 GHz with 1.25V or less, you can achieve those clocks with a good air cooler. The issue with Haswell is the how hot the chip gets once voltage is increased.

My current 4770K build is stable at 4.6GHz at 1.195V and I can install a ZALMAN CNPS9900MAX that I have sitting around to test temps on an air cooler this weekend to determine load temps on air instead of my CM Seidon 240M. Unfortunately, the caveat is that the chip is delidded with CLU applied so the temp readings would not reflect true factory temps.

 

[Max_x2]

chaospower i got my 4670k at 4.8GHz, stable, on air cooling (thermaltake frio). But hey, don't let facts blind you.

 

[Crashman]

chaospower i got my 4670k at 4.8GHz, stable, on air cooling (thermaltake frio). But hey, don't let facts blind you.

Try that with four threads of Prime95 with AVX ENABLED for a couple hours. It WILL thermal throttle. I'm not going to say you're wrong, I'm only going to say that your results are incredible.

A toast to magic samples!

 

[chaospower]

Yet more proof if any were needed that a used 2700K is still far better value
when it comes to an oc'd setup. 5GHz no problem, no need for water
cooling, leaves the 4670K way behind. Ditto 2500K (I notice there are
loads on eBay US); why anyone would buy a 3570K or HW is beyond me.

I am unsure where this misconception arises from...

4670K average on air is 4627 MHz
http://hwbot.org/hardware/processor/core_i5_4670K/

2500K average on air is 5041 MHz
http://hwbot.org/hardware/processor/core_i5_2500k/

Clock for clock, the 4670K is roughly 10-15% faster than the 2500K so a 4627 MHz Haswell is equal to a 5089-5370 MHz Sandy...

chaospower i got my 4670k at 4.8GHz, stable, on air cooling (thermaltake frio). But hey, don't let facts blind you.

Fact is, the person who made this article got it to 4.3ghz with water cooling.
Fact is, I was talking about the results he got from OCing it. Was I supposed to talk about your OC which I was not aware of?

 

[slomo4sho]

Fact is, the person who made this article got it to 4.3ghz with water cooling.
Fact is, I was talking about the results he got from OCing it. Was I supposed to talk about your OC which I was not aware of?

Not sure where you got the idea that I was talking to you or correcting your misconceptions but now that you have brought your illogical and ignorant reply to my attention...

Fact is that any Z87 board will provide roughly the same overclock capacity as the biggest variable in the overclock capacity is the CPU itself so the $200 motherboard provides almost zero benefit when it comes to overclocking headroom for this build.
Fact is that the H50 performs worst than most quality air cooler.
Fact is that Don lost the silicon lottery when when purchased the 4670K for this writeup.
Fact is that a sample size of one does not provide an accurate depiction of overclock capacity of the Haswell lineup.
Fact is that Ivy Bridge and Sandy Bridge counterparts also had a substantial percentage of the of chips that didn't overclock well (albeit the proportion may be larger in Haswell).

 

[Crashman]

Fact is...

Fact is...

Fact is it's not my quote 😛 Fix it please

 

[slomo4sho]

Fact is it's not my quote 😛 Fix it please

Any update on when we can expect these forum bugs to be fixed?

 

[ojas]

Hey nice build. Similar to what i'd build for myself given $1300, except the RAM and maybe the case.

Not sure whether the higher temperatures are Haswell's problem or the cooler's, but i'm surprised that a lower voltage at the same clocks is leading to such temperatures, but i guess it's the FIVR doing that.

I'd probably go with an Asus board too, but i dunno, really. Maybe the PSU was excessive?

p.s. Didn't see any efficiency charts! Hope we get them for the final comparison.

 

[rwinches]

Haswell OCs differently. Going too far with mem OC can hurt CPU OC and overall performance. Reading these threads will show you the right approach to get the most performance overall from the silicon you got. The current stock OCs better.
http://www.overclock.net/t/1411077/haswell-overclocking-thread-with-statistics
http://www.overclockers.com/3step-guide-to-overclock-intel-haswell

Running burnin apps for long periods is pointless (Prime95 et al) no real world use would ever tax a system that way. If it runs for 10-20 minutes you're good just watch the temp and when it levels off run a few more minutes that's it, going further proves nothing.

 

[Narcissistic_Martyr]

Personally I'd have changed some things around (bigger PSU, cheaper cooler) to get 2x HD7950s in there but they weren't priced at $200 a piece when they ordered the components so I can see why they went the way they did.

 

[Siddhartha gb]

The main problem I have with your system builds is that they are directed to the needs of the gamer. I am not a gamer and have no intentions of using my system to play games. It would help if you provided video card recommendations for other uses besides gaming.

I would not spend more than around $150 to $200 for a video card and use the savings in the budget on the other components.