System Builder Marathon, Sept. 2011: $2000 Performance PC

[Crashman]

[citation][nom]larkspur[/nom]I didn't understand you originally so I apologize. I didn't get your reference to the article. All other considerations aside, his PSU will probably do it. Don't forget that in this article they did not measure peak power, they measured power at full load. I'm not purporting to be some PSU expert, but I know that peak power can be a great deal more than load power (definitely in motors, probably not so much in electronics). And I usually don't suggest that others go against NVIDIA's own recommendations.[/citation]I continuously recommend people to go against Nvidia's recommendations. Unlike electric motors, the power use in these tests actually goes up over time as the system gets warmer.

 

[clonazepam]

I run the 2600k at 4.6ghz with 1.32v. The 2 580s are stock speeds. I get about the same power draw from the wall of 650-670. Probably the best thing about 2 580s is that you have no immediate need to overclock them. If they can't produce playable results for a particular game at a particular resolution at stock, then they probably won't when overclocked either.

I never thought to run prime95 (blend) and 3dMark 11 on extreme settings at the same time. It was nice to validate the fact that the overclock on the cpu was stable to do both and still churn out a score of X3207.

Since I have such a similar system to this build, I hope I don't win it, but whoever does, you'll be very pleased at its capabilities.

 

[clonazepam]

[citation][nom]Crashman[/nom]Easiest answer: If one does 16 FPS and the other does 4, neither are playable. Ergo, the performance difference therefore doesn't matter. Even though the performance difference between THOSE TWO scores doesn't matter, the lower score still hurts its machine when added to the "average performance" calculations.Hmm, what part of "no heat issues with this build" isn't clear to anyone?Let me use the BTX form factor as an example: The more closely the CPU area resembles a "duct", the better Tom's Hardware tests have found a tower-style CPU cooler to work. Narrow cases typically have the lowest temperatures in Tom's Hardware comparison, whether tall or short, likely because more of the exhaust air is forced to first go through the CPU cooler.The Antec Three Hundred is cheap and has good airflow. It's been used in many SBM builds because of the low CPU temperatures and price. Size only played a major role in one of those builds, because the graphics cards in that build were internally vented and all their heat had to go through the CPU cooler as well. So Tom's Hardware quit pairing internally vented graphics cards with the Antec Three Hundred, and the problem went away.It's really that simple. Look up the Hyper 212 Plus review to see how bad a cooler it is, then take a look at the temperatures in this article. Remember that you need to get into the high 90's before Sandy Bridge overclocking is significantly hampered.[/citation]

When pushing air through a given space, if its bigger, the air has a tendency to spread out and lose velocity, whereas in a smaller space, it won't spread out as much or lose as much velocity, possibly increasing its cooling potential. The same rules apply to a car's exhaust system. Bigger pipes sounds like a good idea until its put into practice and you realize the air is losing velocity due to having more room to spread out.

 

[Crashman]

[citation][nom]clonazepam[/nom]When pushing air through a given space, if its bigger, the air has a tendency to spread out and lose velocity, whereas in a smaller space, it won't spread out as much or lose as much velocity, possibly increasing its cooling potential. The same rules apply to a car's exhaust system. Bigger pipes sounds like a good idea until its put into practice and you realize the air is losing velocity due to having more room to spread out.[/citation]Velocity is good for heat sinks, but car exhaust? In order to create that pressure, you're making back-pressure and LOSING velocity at the exhuast valve. That's why Top Fuel cars use big short pipes.

Now if you want to get into header design you start talking about exhaust pulses and VACUUM...too complicated to discuss here.

 

[clonazepam]

More about Grand-Am racing... like BMW M3s... reducing the pipes' diamter slightly increased hp / torque by a measurable degree.... and....

It made me sound smart (or at least that was the goal)

 

[gmp23]

Huge thanks to the guys that commented, I really appreciate it. Normally I wouldn't even bother asking as I would just figure it out myself, but why not. People on this forum are pretty damn smart.

 

[Crashman]

[citation][nom]clonazepam[/nom]More about Grand-Am racing... like BMW M3s... reducing the pipes' diamter slightly increased hp / torque by a measurable degree.... and....It made me sound smart (or at least that was the goal)[/citation]The latest in race header design that I read about in the 1990's concerned tapered headers tuned to the appropriate length of one exhaust pulse, with the idea that low pressure at the back of that pulse would help to further evacuate the cylinder as the valve approached its seat (near the top of the exhaust stroke, where there isn't much pressure left to blow out remaining exhaust gasses). Past the header, low backpressure was key because it helped increase the velocity of the exhaust pulse at the header.

Computer cases are much easier

 

[Guest]

Is there enough room for 4 sticks of RAM? Really want 16GB of ram even if I have to go with a smaller (or no) heatsink on the RAM.

 

[Crashman]

[citation][nom]thefirstevil[/nom]Is there enough room for 4 sticks of RAM? Really want 16GB of ram even if I have to go with a smaller (or no) heatsink on the RAM.[/citation]Yes

 

[plasmastorm]

WTF you used a seasonic power supply, and you expect that to last?

Seasonic must be the worst power supplys I have had the misfortune to come accross.
Have tried them 3x in my shop and 3x they have either been faulty on arrival or brought back by customers within a month.
And that wasn't just 'a bad batch' issue either.

 

[Crashman]

[citation][nom]plasmastorm[/nom]WTF you used a seasonic power supply, and you expect that to last?Seasonic must be the worst power supplys I have had the misfortune to come accross.Have tried them 3x in my shop and 3x they have either been faulty on arrival or brought back by customers within a month.And that wasn't just 'a bad batch' issue either.[/citation]

So, what's you're saying is that you buy and sell counterfeit power supplies?

 

[clonazepam]

Power Supply Test Number 1: Can you throw it farther and faster than a softball? I threw one out based on that test alone, w/o even trying it out. So much for inexpensive parts from obscure websites touting a great deal

 

[flong]

[citation][nom]anonymous x[/nom]No, this case cools better than many (most?) cases more expensive than it. 2x 120mm intake is more than adequate for SLI cards with room for air to flow between them. Did you even look at the power consumption numbers? The system when overclocked only consumed 697w at load from the wall (actual consumption is less), while the PSU is rated at 850w.[/citation]

While the 850W PSU will power the system that requires 700W, it is hardly the best choice. It will not be running in its best efficiency range because it is running at 82% capacity. A much better choice would have been the Corsair AX1200 however, I do not think that this build's budget would allow it.

Really, the AX1200 is a much better choice for an SLI or CF system because it will run and nearly 91% efficiency, it will run cooler, much quieter and it has a superior warranty. It also is fully modular. Pushing 850W units to 82% capacity will push them into high fan, high heat, low efficiency situations. Even the best PSUs will heat up and make a lot of noise as they are pushed over the 80% mark.

Despite these obvious facts, many well-meaning people continue to recommend spec'ing PSUs into the 70%-80% usage range to save $50. Really this is short sighted and just sad.

The reason we see a rash of absolutely pathetic PSUs and a huge segment of the computer industry recommending questionable bronze units because they do not take the long term view. The excuse is, "we just can't afford to spend the money on a higher quality PSU, our budget will not allow it." Yet I will be willing to bet that the next day many of these people are in Starbucks spending $5.00 on a latte.

One poster wrote that this fascination with cheaper PSUs is like playing Russian roulette with your expensive components and I agree. It is just not worth saving $75 when your system costs over $2000 and EVERY component depends on the PSU for power and is affected by its stability.

 

[Crashman]

[citation][nom]flong[/nom]While the 850W PSU will power the system that requires 700W, it is hardly the best choice. It will not be running in its best efficiency range because it is running at 82% capacity.[/citation]No, because the power supply is only outputting 575-600W. Wall power x efficiency = power used by the computer. Peak power consumption of 75% capacity is will within the power supply's "optimal" range.

Anything you have to say about low-quality power supplies doesn't apply here as this is a high-end unit. The main reasons it costs less than the same brand's higher-priced units is that it's unpainted and has no removable cables. Soldered-in cables are also more reliable, so one could argue this power unit is out of YOUR league.

 

[flong]

[citation][nom]Crashman[/nom]No, because the power supply is only outputting 575-600W. Wall power x efficiency = power used by the computer. Peak power consumption of 75% capacity is will within the power supply's "optimal" range.Anything you have to say about low-quality power supplies doesn't apply here as this is a high-end unit. The main reasons it costs less than the same brand's higher-priced units is that it's unpainted and has no removable cables. Soldered-in cables are also more reliable, so one could argue this power unit is out of YOUR league.[/citation]

The article sites higher usage than what you quote and other reviews of two 580s in SlI have listed usage near 700W (I know because I hear your argument all the time in the forum and several professional reviews have been posted which show the actual usage of this system's typical power usage). Even using your figures I disagree with your statement that at 75% the unit will be in its most efficient range.

Before I bought my PSU I read hundreds of PSU reviews and I began to notice that nearly all of the PSUs start to drop in efficiency at 60%. This varies with PSU but it is very consistent. Thus if you spec your PSU to run at 50% capacity you get the best efficiency - and really this cannot be argued. However, you are correct that some PSUs maintain reasonably good efficiency at 75% usage but you are missing other factors.

Because you are pushing the PSU to its upper capacity at 75% (really closer to 85% with real world numbers), it will heat up and switch to high fan which increases noise and lowers efficiency. This in turn heats of the interior of your case.

Real world example, I had an I-7 920 PSU that came with a 520W PSU. I upgraded the GPU to the ATI 5850 and added a Corsair HX 750 (one of the highest rated 750W PSUs on the market). The PSU constantly went to high fan and ran VERY hot. While the HX 750 is a great unit, with high fan on it is distracting and noisy. I really hated it.

Fast forward to my current build. It is a 2600K with an ATI 6950. My current build has two more HDDs, a blu-ray burner, a sound card and the Noctua NH-D14 added to the PSU load that my I-7 920 did NOT have. With this build I got smart and I got the Corsair HX 850 PSU.

WHAT A DIFFERENCE!!!!!! I have never even hear the PSU once, I am not sure that the fan ever turns on. According to the reviews I have read it is running near 92% efficiency most of the time. It NEVER heats up. Now this is REAL-WORLD first-hand experience. Keep in mind also that my second build draws much more power than my first build and the HX 850 handles it with ease.

Thus spec'ing your PSU to run at 50% capacity is a much more intelligent approach and it really doesn't cost that much. I got my HX 850 for $144.

Really for a two 580 SLI build, I think that the AX 1200 is by far the best PSU for a high-end build. This is because it is much more efficient at low capacity usage than comparable 850W - 1000W units - so it will run more efficiently (by far) than an 850W unit. Also it can regularly be bought on sale for $250 and so you don't have to mortgage your house to afford it. Also, it comes with Corair's legendary 7-year warranty. Also keep mind that PSU technology is not changing much and so the AX1200 will not be outdated in 7 years.

Finally it is a myth that soldered cables are more reliable than modular units but even if it was true I would not buy a non-modular unit - they frankly are a tangled mess and are miserable to build with. Poor solder connections are as common as dirt, ask any experienced computer tech. It takes a lot of skill to make a good solder connection.Though you believe a myth, I would take the supposed risk for the ease of using a modular unit. Keep in mind that the best units on the market right now including most Gold-rated PSUs are modular. I think Antec's 1200W unit is the only unit I can think of that is not modular and it is not quite as good as the AX1200 which is fully modular (both are highly rated).

It really never ceases to amaze me how many people run to the cheapest PSU to save $50 even though it will run hotter, be much more noisy, will cost them more for power, will not be as efficient and will not be as stable so it is more likely to damage expensive components. But damn it, if I can save a whole $50-$75 it is worth playing Russian roulette with my $2000 worth of components.

 

[Crashman]

[citation][nom]flong[/nom]The article sites higher usage than what you quote[/citation] no it doesn't, do the math [citation][nom]flong[/nom]and other reviews of two 580s in SlI have listed usage near 700W (I know because I hear your argument all the time in the forum and several professional reviews have been posted which show the actual usage of this system's typical power usage). Even using your figures I disagree with your statement that at 75% the unit will be in its most efficient range.[/citation]Most efficient is a point Optimal range, well, take a look:
http://www.plugloadsolutions.com/psu_reports/SEASONIC_SS-850HT_ECOS%201092.1_850W_Report.pdf
Now forget everything you thought you knew about power supplies because it doesn't apply to modern high-efficiency units.

I used to agree with a lot of what you said. Then power supplies got better. The power supply in this review is a top-quality unit regardless of price, just as some other parts are junk regardless of price.

 

[flong]

[citation][nom]Crashman[/nom]no it doesn't, do the math Most efficient is a point Optimal range, well, take a look:http://www.plugloadsolutions.com/p [...] Report.pdfNow forget everything you thought you knew about power supplies because it doesn't apply to modern high-efficiency units.I used to agree with a lot of what you said. Then power supplies got better. The power supply in this review is a top-quality unit regardless of price, just as some other parts are junk regardless of price.[/citation]

I agree with you that this power supply is a good quality supply. It appears to be stable and it is reasonably efficient. That being said it is not nearly as efficient as the Corsair HX 850. Please do not think that I am criticizing Tom's Hardware for this choice - it is a reasonably good choice for a PSU.

However, judging from the 80+ certification that you posted for the unit at 75% power capacity it is roughly 86% efficient which is not that impressive in today's units and it is not at its optimum efficiency as I said.

There have been several power usage article posted for the SLI 580 combo builds (this is a popular build among enthusiasts) and they do push 700W (not at wall power). You have to remember that most builders overclock more aggressively than Tom's Hardware. I have had this discussion several times in the forums and I really don't believe that the 850W supply is the best choice for this setup - others are of course entitled to their opinion. It will be hovering around 84%-85% efficiency under about 85% load, it will heat up, it will go to high fan, it will be noisy and it will cost more to run.

This is true with modern PSUs, they are always most efficient at the 50% capacity range. This Seasonic unit is no exception.

The Corsair AX1200 costs about $75 more and it will run at nearly 91% efficiency ALL of the time (read Johnny Guru's review) over the entire power usage range because with two 580s, the computer will always be drawing over 200W. The AX1200 is a genius at efficiency at low power usage where most PSUs drop in efficiency precipitously (somewhere around 88%).

For the extra $75-$100, the AX1200 is by far the best choice for this build.

 

[Novuake]

for the motherboard I would have went with the asus p8p67 pro. Just my 2 cents

HUH!? P67? You are joking right?

 

[Crashman]

HUH!? P67? You are joking right?

T'isn't anything wrong with P67, his suggestion simply points to the weakness of the benchmark suite in that it doesn't include Quick Sync tests.

 

[mjmjpfaff]

"We’ve heard that Core i7-2600K also gets Intel’s top-binned parts, so that might give this one a little extra overclocking capability."
why would they pick a hyper 212 plus if overclocking was so important. it is not like you could get 5ghz out of it even with the top binned parts while on the 212 plus. get a 2500k and get a better cooler. these builds are starting to be ridiculous.

 

[Crashman]

[citation][nom]mjmjpfaff[/nom]"We’ve heard that Core i7-2600K also gets Intel’s top-binned parts, so that might give this one a little extra overclocking capability."why would they pick a hyper 212 plus if overclocking was so important. it is not like you could get 5ghz out of it even with the top binned parts while on the 212 plus. get a 2500k and get a better cooler. these builds are starting to be ridiculous.[/citation]You lose since CPU temperature was far below its target maximum. You should have said "2500K and a better motherboard" if you'd known anything about the system's overclocking limits, but I guess you didn't actually read the article.

 

[clonazepam]

If I understood the article correctly, this particular chip hit its overclocking maximum based on prime95 results, and to a lesser extent, the motherboard's ability to provide the higher cpu voltage. It hit this maximum before it hit its temperature maximum and would a better cpu cooler provide more cooling to the motherboard's VRM? idk. We see this on graphics cards all the time. I remember this same argument centered around a graphics card in an article. It hit one maximum before the other, so there was no reason to improve its cooling as it would have been for not. Past arguments around the processor choice in the $2k build is that its not entirely centered around gaming. It's supposed to be an overall performance leader so more cores = win for the people who would need it.

The only thing I might have changed is found a less expensive SSD for the OS, or go even less, for caching, and grabbed a different board, like ones suggested from previous Tom's articles. It is what it is, and its not like this system has some huge 5-10% performance deficit from previous builds... does it?

I won't question the psu choice. I would say though that if the eventual owner begins to look at pimping it out with water pump(s), external rads/fans, lights, etc than they can swap it out if they feel the need.

 

[Crashman]

If I understood the article correctly, this particular chip hit its overclocking maximum based on prime95 results, and to a lesser extent, the motherboard's ability to provide the higher cpu voltage. It hit this maximum before it hit its temperature maximum and would a better cpu cooler provide more cooling to the motherboard's VRM? idk. We see this on graphics cards all the time. I remember this same argument centered around a graphics card in an article. It hit one maximum before the other, so there was no reason to improve its cooling as it would have been for not. Past arguments around the processor choice in the $2k build is that its not entirely centered around gaming. It's supposed to be an overall performance leader so more cores = win for the people who would need it.

The only thing I might have changed is found a less expensive SSD for the OS, or go even less, for caching, and grabbed a different board, like ones suggested from previous Tom's articles. It is what it is, and its not like this system has some huge 5-10% performance deficit from previous builds... does it?

I won't question the psu choice. I would say though that if the eventual owner begins to look at pimping it out with water pump(s), external rads/fans, lights, etc than they can swap it out if they feel the need.

A downdraft CPU cooler would have helped the voltage regulator but would have likely increased CPU core temperature. The best solution would have been to pick a board with heat sinks on all PWM phases and slide the fan low enough on the sink that any air bypassing beneath it adds to PWM cooling.

The reason this particular board was chosen was that it worked so well in previous tests at 1.35V. It wasn't perceived that 1.38V would be too much for it until after it was tried.

 

[clonazepam]

The new owner can pick up a cheap Antec goose neck fan, find a place to mount it or drill a hole, and aim it straight there... hehe (not being 100% serious here)

P.S. any higher than low gets noisy

 

[Crashman]

The new owner can pick up a cheap Antec goose neck fan, find a place to mount it or drill a hole, and aim it straight there... hehe (not being 100% serious here)

P.S. any higher than low gets noisy

Actually that would have been my next suggestion. SBM has used Antec SpotCool in the past, and Newegg sells it.