System Builder Marathon, Q2 2013: The $400 Spirit Of Mini-ITX

[slomo4sho]

This build definitely deserved an APU...

Nah, not for gaming. It would have been a step back in most games.

I didn't realize that the "Spirit of Mini-ITX" was gaming, especially at the $400 price point.

 

[ceh4702]

^^^^----funny comments. Someone missed the point of the build. mITX isn't common but a segment of the market.

How come so many people are building computers with Mini-ITX? It seems very common to me. Just look at all the Mini-ITX cases. They wouldnt make them if they thought they would not sale.

 

[pauldh]

This build definitely deserved an APU...

Nah, not for gaming. It would have been a step back in most games.

I didn't realize that the "Spirit of Mini-ITX" was gaming, especially at the $400 price point.

True, it certainly would not need to be. Although if you read the article then you know (as stated numerous times), this was meant to be a small, cheap, gaming PC. Gaming was (by far) the number one performance goal. In fact, I'd say it was the only performance consideration going into this build.

 

[slomo4sho]

True, it certainly would not need to be. Although if you read the article then you know (as stated numerous times), this was meant to be a small, cheap, gaming PC. Gaming was (by far) the number one performance goal. In fact, I'd say it was the only performance consideration going into this build.

The article suggested the following:

I knew I couldn't replicate a Tiki, or even an X51. But enthusiasts willing to sacrifice graphics performance can still build a tiny machine to serve numerous computing needs (including the potential for decent gaming). Enter our budget-oriented "True Spirit of Mini-ITX PC."

Although the article goes on to discount the capabilities of the APU, the defined purpose of the "True Spirit of Mini-ITX" PC seems to fit well with an APU. In addition, the new Richland APUs(I am aware that they were not released at the time of this build) can run hybrid crossfire with the 7750 so one can always add a 7750 or , as an alternative, go with an Athlon II X4 760K with discrete graphics.

I look forward to reading the upcoming review of the 750K.

 

[ayushde]

Why not have a gtx 650 instead of the 7750....it can be found at as low as 90$ and performs much much better than the 7750

 

[pauldh]

Why not have a gtx 650 instead of the 7750....it can be found at as low as 90$ and performs much much better than the 7750

Priced exactly the same at $110, it was considered. But the only low-profile GTX 650 available to us was this Zotac below. It has a dual-slot cooler that probably would not fit in the chosen case, and requires two low profile slots for full connectivity. The Sapphire 7750 is single slot all the way and we knew it would fit.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814500290

I would not say it "performs much much better" though, at least not at stock clocks. The GTX 650 is a bit faster and wins more than it loses, but it's no clean sweep (Skyrim for instance).

 

[pauldh]

True, it certainly would not need to be. Although if you read the article then you know (as stated numerous times), this was meant to be a small, cheap, gaming PC. Gaming was (by far) the number one performance goal. In fact, I'd say it was the only performance consideration going into this build.

The article suggested the following:

I knew I couldn't replicate a Tiki, or even an X51. But enthusiasts willing to sacrifice graphics performance can still build a tiny machine to serve numerous computing needs (including the potential for decent gaming). Enter our budget-oriented "True Spirit of Mini-ITX PC."

Although the article goes on to discount the capabilities of the APU, the defined purpose of the "True Spirit of Mini-ITX" PC seems to fit well with an APU. In addition, the new Richland APUs(I am aware that they were not released at the time of this build) can run hybrid crossfire with the 7750 so one can always add a 7750 or , as an alternative, go with an Athlon II X4 760K with discrete graphics.

I look forward to reading the upcoming review of the 750K.

Don't misunderstand... I am not discounting an APU for a cheap mini-ITX build, not at all. Just saying it would have faired worse at this machine's purpose. Also, AMD loses some ground in mini-ITX, as mobo prices do throw a wrench into the "cheap" plans. There are just 5 AMD mobos to chose from starting at $85-90, and 38 Intel options starting at $50.

Numerous times, including the Title Page Heading, it is referred to as a "Gaming PC" or "Inexpensive Console-sized Gaming PC." That was my purpose for the build. In fact, when I shot the Newegg Cart to Chris, I originally called it a "$400 Slim-line Gaming PC." Maybe it didn't seem clear since every SBM we test/value overall performance, but as the idea guy, the builder, tester, and writer, I know what this machine's purpose was. hehe 😉 Honestly I didn't want to go below 7750 level graphics.

We still can't buy the X4 760K so it may never make it into the SBM. 750K will be considered moving forward. I'd have some concerns though with 750K and 7750 with this 150W PSU, but mainly such a build would be way over our budget. The cheapest FM2 mini-ITX mobo was $90, meaning (if 750K had been available) we'd have had $55-$60 left for graphics. Low profile 6670 costs way more than that ($85).

 

[yialanliu]

As an owner of this case, you must be crazy to try to put a 7750 with a stock intel. It is 1 watt away from max load. The PSU is 10amps on the 12 volt which is 120 watts, and you are running 113/119 over clocked. Honestly, I don't recommend anyone try to actually build it with this spec and expect it to last. Get an APU and then you'll be much safer.

I bet Toms added a DVD drive and then realized that it didn't boot the fans because that would have been enough to go over the top hence why they didn't include it. All in all, good idea, but clearly not the right build for this case/PSU.

Honestly, I agree with the people who say Intel paid toms off, this build is just utter stupidity.

 

[Onus]

Why not have a gtx 650 instead of the 7750....it can be found at as low as 90$ and performs much much better than the 7750

Priced exactly the same at $110, it was considered. But the only low-profile GTX 650 available to us was this Zotac below. It has a dual-slot cooler that probably would not fit in the chosen case, and requires two low profile slots for full connectivity. The Sapphire 7750 is single slot all the way and we knew it would fit.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814500290

I would not say it "performs much much better" though, at least not at stock clocks. The GTX 650 is a bit faster and wins more than it loses, but it's no clean sweep (Skyrim for instance).

Furthermore, if you take a look at the back of the board, you can see where it has a PCIe power connector. Unfortunately, it would not have been a viable choice in this build, even if it fit.

 

[pauldh]

Why not have a gtx 650 instead of the 7750....it can be found at as low as 90$ and performs much much better than the 7750

Priced exactly the same at $110, it was considered. But the only low-profile GTX 650 available to us was this Zotac below. It has a dual-slot cooler that probably would not fit in the chosen case, and requires two low profile slots for full connectivity. The Sapphire 7750 is single slot all the way and we knew it would fit.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814500290

I would not say it "performs much much better" though, at least not at stock clocks. The GTX 650 is a bit faster and wins more than it loses, but it's no clean sweep (Skyrim for instance).

Furthermore, if you take a look at the back of the board, you can see where it has a PCIe power connector. Unfortunately, it would not have been a viable choice in this build, even if it fit.

Yeah you'd think so, but oddly that low profile Zotac 650 lacks a connector. Take a look at the end, it's missing with two blank mounting holes. There goes the OC.

http://www.zotacusa.com/geforce-gtx-650-lp-zt-61008-10m.html

 

[pauldh]

As an owner of this case, you must be crazy to try to put a 7750 with a stock intel. It is 1 watt away from max load. The PSU is 10amps on the 12 volt which is 120 watts, and you are running 113/119 over clocked. Honestly, I don't recommend anyone try to actually build it with this spec and expect it to last. Get an APU and then you'll be much safer.

I bet Toms added a DVD drive and then realized that it didn't boot the fans because that would have been enough to go over the top hence why they didn't include it. All in all, good idea, but clearly not the right build for this case/PSU.

Honestly, I agree with the people who say Intel paid toms off, this build is just utter stupidity.

Ugh, a know it all crying we are paid off. Sorry, you are totally wrong!

Just a bit of advice: Before screaming foul play, and tossing out the "stupidity" boomerang, how about you read the article and learn something, like the difference between power draw from the AC source, and power supply output. 119W overclocked was pulled from the wall, this power supply is/was known for 70% efficiency, meaning likely less than 90W output at full load.

You could have saved some embarrassment by reading....

"I don't have efficiency specifications for Antec's 150 W power supply. However, older units with the same model number were notoriously bad in this area. Regardless, under full load, the overclocked configuration pulls less than 120 W from the wall. Assuming between 70% and 80% efficiency, the power supply was only outputting between 83 and 95 W under full load. That should satisfy any concern that a 150 W PSU is insufficient for today's build."

Besides all this, the 10A rating is the 12V rail, not all rails.

Oh, and BTW, what APU did you want us to use to be "much safer? A much slower dual-core, or a more power hungry quad? 5800K consumes more power under load. (Stinks these use 1000W and 1200W units, but look around the web, other sites would lead to the same conclusion).
http://www.tomshardware.com/reviews/a10-5800k-trinity-efficiency,3315-10.html
http://www.tomshardware.com/reviews/a10-6700-a10-6800k-richland-review,3528-11.html

 

[Tpoke1985]

I really hope allot of people don't build machines in a hurry based on this article. I created one of these for testing over 6 months ago and under 6 months ago in it's second stage. YOU CAN NOT USE A 65 WATT TDP CPU IN THIS SETUP, reason being the power supply ages very quickly being over 50% load all the time. By time i used the smarter choice a i3 3220t the power supply had already aged to much and the machine began to act faulty or hang on bf3 then weaker games at time went on. the total amperage and power output degrades in under a year this was with a additional 60mm fan pushing air from the bottom of the power supply. This was not only with one SSD or motherboard but two different variations h61s from foxxconn and intel 33d ssd 256gb and with 128gb 520 series, again both ssds tried with gigabyte z-wifi board same issues. THIS BUILD DOES NOT WORK LONG TERM. 6 months-1.5 years with 65 watt tdp cpu and 1-3 with 35 watt tdp cpu.

 

[Tpoke1985]

For those of you who speak 3dmark06 to estimate fps in games still:
http://www.3dmark.com/compare/3dm06/17084307/3dm06/16973524

two different cpu's details in one gives accurate wattage readings at wall.

 

[Tpoke1985]

THIS BUILD DOES NOT WORK IN LONG RUN

 

[Tpoke1985]

I have tried with build over 6 months ago with two different motherboards and ssd and got constant hangs from the power supply aging to quickly under gaming loads. The machine starts off barely withing amperage range to supply the machine from the get. 65 watt g530 and a 35watt tdp i3 3220t used could not save it in the long run. Plugging in a 400 watt i had laying around eliminated the system hangs. I tried it standing sitting sideways everything I even had a 60mm fan pushing air into power supply from other side of case from day one.

 

[pauldh]

I really hope allot of people don't build machines in a hurry based on this article. I created one of these for testing over 6 months ago and under 6 months ago in it's second stage. YOU CAN NOT USE A 65 WATT TDP CPU IN THIS SETUP, reason being the power supply ages very quickly being over 50% load all the time. By time i used the smarter choice a i3 3220t the power supply had already aged to much and the machine began to act faulty or hang on bf3 then weaker games at time went on. the total amperage and power output degrades in under a year this was with a additional 60mm fan pushing air from the bottom of the power supply. This was not only with one SSD or motherboard but two different variations h61s from foxxconn and intel 33d ssd 256gb and with 128gb 520 series, again both ssds tried with gigabyte z-wifi board same issues. THIS BUILD DOES NOT WORK LONG TERM. 6 months-1.5 years with 65 watt tdp cpu and 1-3 with 35 watt tdp cpu.

So you signed up for another account today to rehash this? Look, I'm sorry you had trouble with yours yialanliu, but I think you are hung up on TDP. Under full CPU load this PSU is under 33% load. What exactly did Antec have in mind when they released the case/PSU back in early 2010, Atom? Sandy Bridge is efficient and miserly compared to older desktop hardware.

 

[yialanliu]

I really hope allot of people don't build machines in a hurry based on this article. I created one of these for testing over 6 months ago and under 6 months ago in it's second stage. YOU CAN NOT USE A 65 WATT TDP CPU IN THIS SETUP, reason being the power supply ages very quickly being over 50% load all the time. By time i used the smarter choice a i3 3220t the power supply had already aged to much and the machine began to act faulty or hang on bf3 then weaker games at time went on. the total amperage and power output degrades in under a year this was with a additional 60mm fan pushing air from the bottom of the power supply. This was not only with one SSD or motherboard but two different variations h61s from foxxconn and intel 33d ssd 256gb and with 128gb 520 series, again both ssds tried with gigabyte z-wifi board same issues. THIS BUILD DOES NOT WORK LONG TERM. 6 months-1.5 years with 65 watt tdp cpu and 1-3 with 35 watt tdp cpu.

So you signed up for another account today to rehash this? Look, I'm sorry you had trouble with yours yialanliu, but I think you are hung up on TDP. Under full CPU load this PSU is under 33% load. What exactly did Antec have in mind when they released the case/PSU back in early 2010, Atom? Sandy Bridge is efficient and miserly compared to older desktop hardware.

Actually, I do believe atoms were the original intent of these cases. That's why they also sell the same case at 65w PSU attached to it. Probably the only type of processor that can have a full system under 65 watts. These aren't 250w ini itx cases, these are 65/150w cases and honestly, that's the type of build they are designed for.

And I was referring to the 5700 level of APU without dedicated gfx card. That's my build and something much safer in the long run.

And in all honesty, 55w for the gfx card + 65w for the cpu is at 120 watts and you are excluding power used by hard drives, and usb and a variety of other things though small, does matter with a 150w PSU. I question 90 as an accurate number for the system when in full load with those components.

 

[Crashman]

So you signed up for another account today to rehash this? Look, I'm sorry you had trouble with yours yialanliu, but I think you are hung up on TDP. Under full CPU load this PSU is under 33% load. What exactly did Antec have in mind when they released the case/PSU back in early 2010, Atom? Sandy Bridge is efficient and miserly compared to older desktop hardware.

You're 100% correct Paul. At 80% efficiency your PSU would be pushing less than 100W output at full GPU+CPU load. At 80% efficiency, CPU-only load is actually less than 60W according to your test. If this power supply doesn't hold up, it's defective.

 

[slicedtoad]

And in all honesty, 55w for the gfx card + 65w for the cpu is at 120 watts and you are excluding power used by hard drives, and usb and a variety of other things though small, does matter with a 150w PSU. I question 90 as an accurate number for the system when in full load with those components.

It's a 150W psu with a 10 amp (120W) 12v rail. Everything except cpu and gpu are not running on the 12v rail so they fit nicely into the remaining 30W (ie. ignore them).

TDPs are 55W and 65W as you said. Without going any further, we know the psu should be sufficient since the requirements are within spec.

Now, OCing. A quick google search will show you that the real peak draw of a 7750 is ~43W at stock. Without upgraded cooling, no oc is going to break +20% watts consumed. So peak power still under 55W and likely lower. The CPU is not oced and a google search (or the results here) will show you that the 65W tdp is something of an overstatement.

So, no matter what, peak power draw should remain under 120W on the 12v rail.

Next issue, you mentioned it won't last if it's being used at 100% all the time? Regardless whether that's true or not, why would it be running at 100% for more than small periods of time?

Games don't use peak power, only synthetics so unless you're mining bitcoins on the gpu and litecoins on the cpu, it should never be using peak power for any serious duration.

Now, let's ignore what I said and pretend google didn't exist and we couldn't look up research done by others. We still know that this build will run on that psu because they tested it! Power draw from the wall showed a perfectly safe system.

 

[pauldh]

You're 100% correct Paul. At 80% efficiency your PSU would be pushing less than 100W output at full GPU+CPU load. At 80% efficiency, CPU-only load is actually less than 60W according to your test. If this power supply doesn't hold up, it's defective.

Agreed. Defective and/or tortured with heat (stuck in a tight/unsafe operating environment). And, 80% is being generous. According to a Gabriel Torres review of this 150W PSU in Feb 2010, it ranges from 65-75% efficiency. BUT, the peak 75% efficiency was hit way above my loads here (at around 97% or 190W from the wall). Antec may have made changes/improvements over the last 3+ years, can't say, but if those numbers still ring true, that's 65% efficiency at max CPU load, and maybe 72-73% under max overclocked full system load. So (unless changes have been made) PSU output was likely about 87W peak at full system load. And Prime 95 (CPU load) likely less than 50W.

 

[pauldh]

And in all honesty, 55w for the gfx card + 65w for the cpu is at 120 watts and you are excluding power used by hard drives, and usb and a variety of other things though small, does matter with a 150w PSU. I question 90 as an accurate number for the system when in full load with those components.

Look, at this point I'm trying to help you here. Adding up TDP isn't the way to go.

Some folks get totally hung up on TDP, not realizing the 35W versions of these dual-core chips only drop 4-7W in power consumption under full load (less at idle), not the assumed 30W the ratings would imply. I need to go off-site for the best link to help you understand.

Here, look at 620/620T and 2100/2100T. Power draw using the iGPU is only 4-7W lower. Power draw with discrete graphics is even closer. Notice, there are 35W chips like the i5-2390T consuming WAY more power under load than 65W ones like our G860.
http://us.hardware.info/reviews/3169/21/intel-core-i3-3225-and-3220-review-entry-level-ivy-bridge-energy-consumption-idle--cinebench-115

I just don't have the time or space to grab any measurements myself right now, (AMD platforms on the bench), but here is another link in case you for some reason do not trust the other one alone.

35W G620T vs. 65W G620 -
http://www.anandtech.com/show/4524/the-sandy-bridge-pentium-review-pentium-g850-g840-g620-g620t-tested/3

So the thought a 35W chip is so much safer for this PSU is built upon false assumptions you'd see a 30W difference. That's not gonna happen, not even close, esp. with discrete graphics. Again, notice G860 consumes less power under load than i3-2100T, despite the higher TDP.

That's not knocking your CPU upgrade, T or no T, i3-3220 would be excellent for this type build. I don't really like paying more for underclocked parts. It's a shame if people do it thinking the difference is huge (like the TDP implies).

We logged power metrics with a Watts-Up Pro meter during all use of this system. Under full CPU+GPU load, it drew 119W from the wall when overclocked. The PSU was outputting about 90W, or likely less. Gaming, it drew even less power.

The main thing I'd recommend is keeping an eye on heat. This is an inefficient PSU and a small enclosure. I'd fear folks may tuck it into a cramped location with too little airflow. High heat (ambient/case temps) would more likely cause issues and PSU failure than simply going above 65% PSU load (which our $400 rig will never do).

 

[allanitomwesh]

Well, each builder was given one shot at this, and each had their own goals. So that is 3 enclosures out of quite a large number of options. (This $400 build wasn't planned, just IMO it needed to be done and Chris agreed).

Crash already stated why he didn't use the FT03. And really, although rather large, most folks loved his Prodigy build.
On the low end, the pictured SG05 (and V3+) are ones I bought myself, nothing to do with the SBM. Yes, personally that is exactly what I wanted to use for the $650 PC, along with a GTX 660 and 8GB if possible. Going in, that was my expectation for the Gaming PC. However, it was $115, robbing $25 more of this budget. I had to admit the $40 120 Elite with Tahiti LE grabbed far more sheer bang for the buck. More than ever this SBM I faced a conflict of interest between what would perform the best and what could best find a place in my own home.

This.
Didn't know the cases are personal.
SG05 lite is $40 too. It's the SG05-450 without the psu. I know the psu itself is about $60 but there are other,different brand psus in sfx format, or even a short ATX psu.
I have nothing against crash's build other than the choice of case. I've explained why.
I think you did a great job with the two budget builds. The idea after all is to save space,and that's why I think they're the best of this marathon.

 

[ojas]

Apparently you missed the AMD press event in Bangalore on the 13th of June. Consumer Division head let it slip that it's coming next year.

wut?! is there a link to the revelation? the rumors and leaks i've been coming accross so far pointed towards a q4 2013 launch (oct, likely). q4 used to be fx cpu launch like bd and pd. since this year there won't(likely) be any cpu launch, kaveri would take it's place. could it be that kaveri will launch in india in 2014?
2014 is good too, it's not like intel will (ever) throw in an hd5000 or iris pro sku under $180.

Now THAT is possible (2014 India launch). Though i dunno, Richland's international and India launches were barely a month apart, so December launch in the US and January launch in India?

No link, man, I'm your source! There were others at the event, not sure anyone caught on to it (because no one asked them anything about it) and I'm not sure if any of the other tech sites here decided to report it.

Their Steamroller based server chips are slated for Q1 2014 too, so maybe?

 

[de5_Roy]

Now THAT is possible (2014 India launch). Though i dunno, Richland's international and India launches were barely a month apart, so December launch in the US and January launch in India?

No link, man, I'm your source! There were others at the event, not sure anyone caught on to it (because no one asked them anything about it) and I'm not sure if any of the other tech sites here decided to report it.

Their Steamroller based server chips are slated for Q1 2014 too, so maybe?

ah, so that's what happened. i couldn't find any 3rd party link to the event let alone an official one from amd. glad you asked them

there might be multiple possibilities. for example: q4 2013(oct.) for mobile kaveri and dec 2013/jan 2014 for desktop, paper launch in q4 and retail one in 2014 - among others. kaveri launch isn't similar to richland launch. richland was made on 3 year mature node while kaveri will launch in glofo's new 28nm node. kaveri's launch is similar to zambezi's (fx8150).

 

[ojas]

Now THAT is possible (2014 India launch). Though i dunno, Richland's international and India launches were barely a month apart, so December launch in the US and January launch in India?

No link, man, I'm your source! There were others at the event, not sure anyone caught on to it (because no one asked them anything about it) and I'm not sure if any of the other tech sites here decided to report it.

Their Steamroller based server chips are slated for Q1 2014 too, so maybe?

ah, so that's what happened. i couldn't find any 3rd party link to the event let alone an official one from amd. glad you asked them

there might be multiple possibilities. for example: q4 2013(oct.) for mobile kaveri and dec 2013/jan 2014 for desktop, paper launch in q4 and retail one in 2014 - among others. kaveri launch isn't similar to richland launch. richland was made on 3 year mature node while kaveri will launch in glofo's new 28nm node. kaveri's launch is similar to zambezi's (fx8150).

Even then, not usually a big delay. Haswell, Ivy, Sandy all within the same month. Llano took a long time, iirc.

I didn't ask them, the Consumer Division head (forgive me, i don't remember his name!) for India (or Asia? I'm not sure) let it slip by accident when talking about hUMA. Just said it's coming next year.