What I have Learned about Computer Performance from SPCR

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Re: What I have Learned about Computer Performance from SPCR

Post by m0002a » Sun Apr 01, 2012 12:56 pm

Maybe I am wrong, but it appears to me that many of the quotes in this thread were not orginally posted in this thread. This compendium of knowledge will no doubt some day be considered to be one the great achievements of civilization, and a monument to CES rivaling that of the great pyramids.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Tue Apr 03, 2012 2:58 pm

Lan-gear Braveheart case
ces wrote:Have you taken a look at the lan-gear.eu Braveheart case?

http://www.lan-gear.eu/langear-microATX ... oduction-1
SoulWager wrote:Interesting, certainly LAN compatible, but it doesn't look too quiet, in particular the GPU looks starved. Also I don't think I can buy one, their US retailer site appears bugged.
1. I would submit, that you need to look to your components for quiet. If you get the right components, it will be quiet. If you do not, nothing you can do will ever truly make it quiet.
2. I am certain these cases are available in the US. Someone in the US has one of them. You can try emailing Matthieu Alirol at [email protected] for information about the availability of the Braveheart as well as their Mini-ITX version of it
Last edited by ces on Tue Apr 03, 2012 4:27 pm, edited 1 time in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Tue Apr 03, 2012 4:26 pm

Testing Variability in Reviews of CPU Coolers
highstream wrote: The other thing I've found is that the comparative reviews come up with very different test results, quite often substantially different between the same coolers. Methodologies and test rooms and equipment, I suspect.
There are a lot of variables... but the top coolers generally come out on top... just with slight differences in ranking. If you stand back and look at them all I think you will find a lot of consistency.

There are just limits to how precise this type of testing can be. See:
viewtopic.php?p=557298#p557298

Too much vs. Too Little Air Flow / Heatsink Sample Variance
Devonavar wrote:Congratulations SPCR! 10 years is significant — a recognition of a positive contribution to the internet. I was proud to be a regular part of SPCR for about three years. I wrote over 100 articles, and wrote the bulk of the content on the site while I was active...

I also have fond memories of the various fan round-ups that I did. #2 has over 1.5 million views. This task was years in the making, and we spent months working on a methodology that would test fairly. Even now, I don't think we ever figured out how to test for airflow properly, and we eventually dropped it from our testing. More importantly, we learned that airflow isn't really that relevant. What you need is enough airflow. Enough meaning this: Enough to carry the amount of heat coming off the heatsink. Less than this amount, and the system would cook. More than this, and you basically increase noise with minimal benefit to cooling. There's a specific amount of airflow that any given heatsink needs to operate efficiently, and trying to push more airflow than is optimal simply doesn't benefit cooling that much.

The last article that sticks out to me is memorable more in hindsight than for the article itself: Our original review of the Scythe Ninja. The Ninja was a stunningly good heatsink — at least our sample was — and it remained our benchmark heatsink for years. In fact, our praise of it probably helped launch Scythe to its dominent position today. But, in subsequent tests (of which there were many, since we used it to calibrate every time we switched test beds), we always had difficulty reproducing consistent results. One mounting would duplicate the original results, and the next would be 5° hotter. The problems got worse once we had multiple samples floating around the lab. In the end, we concluded that our sample was probably exceptional, and the average case was not quite as good as ours was. But, it took us over three years to find this out, by which time even its exceptional results had been surpassed by bigger, beefier competitors. In the end, it was a lesson about precision of results — the error tolerance for heatsink testing can't be better than two or three degrees either way, with variance coming from different samples, different mountings, and different ambient conditions. Ergonomics, fin-spacing, and weight all matter more than raw test results, but none of these can be quantified or ranked in quite the same way that thermal results can, so thermal results continue to get top billing even though the top performing heatsinks can all be considered equal within the margin of error. Devon Cooke
Last edited by ces on Sun Aug 05, 2012 11:15 am, edited 1 time in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Sat Apr 07, 2012 8:01 am

How Much Computer Can You Power With a Pico PSU?
Jim G wrote:I use a 150W brick with the 160W Picopsu and it powers an Asus ITX board, 2500K (a 95 watt TDP CPU, ces), SSD, DVD drive, 4GB RAM and one 140mm fan with no dramas at all. Prime stable for 24 hours.
Abula wrote:If you want higher than the 150W brick, maybe go with the picoPSU-160-XT + 192W Adapter Power Kit
ces wrote:
Vicotnik wrote:A picoPSU-90-XLP and a 80W brick powers my i5 3570K system btw
The 80 watt brick may end up delivering 70 watts. The 3570K in theory is capable of absorbing 77 watts all by itself. Isn't that cutting a a bit tight... in the case of an unexpected runaway process... say?
Last edited by ces on Sun Aug 05, 2012 6:25 pm, edited 1 time in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Sat Apr 07, 2012 1:35 pm

Intentionally deleted by the poster.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Sat Apr 07, 2012 1:38 pm

Intentionally deleted by the poster.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Sat Apr 07, 2012 1:40 pm

intentionally deleted by poster
Last edited by ces on Sun Aug 05, 2012 11:15 am, edited 1 time in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Wed Apr 11, 2012 11:20 am

Scythe "NINJA WIRE" to Prevent CPU Cooler Stress on Motherboards

"Ninja Wire is an optimal tool to attach or stabilize heavy objects and components inside and outside the PC chassis.
Bending VGA Card or tilting CPU cooler is now a thing of the past!"

Model #: SCYNW-1000

http://scythe-usa.com/product/acc/065/s ... etail.html

http://www.performance-pcs.com/catalog/ ... s_id=25672

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Wed Apr 11, 2012 11:34 am

array of custom mini itx cases

viewtopic.php?f=14&t=54582

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Tue Apr 24, 2012 10:34 pm

Noctua introduces low-profile NH-L12 with dual PWM fans
lodestar wrote:Fitted with both NF-B9 92mm PWM and NF-F12 'focused flow' 120mm PWM fans the NH-L12 CPU cooler is intended for use in small form factor/HTPC systems. Weight is 415g, dimensions are 93mm high, by 128mm width and 150mm depth. Without the top 120mm fan, the height is 66mm. The NH-L12 comes with the Noctua SecuFirm2 mounting kit. There is also a mini-ITX mounting kit that Noctua says is for "... some Intel based Mini-ITX mainboards (that) don’t allow for the installation of backplates. The NH-L12 thus comes with an extra set of Intel mounting bolts which make it possible to install the cooler without using the SecuFirm2 backplate...".

The NH-L12 is supplied with a 'Y' PWM splitter cable, and two PWM low-noise adapter cables. The PWM low-noise adapter cable changes the range of the fans from 300-1600 rpm to 300-1300 rpm.

For full details, see http://www.noctua.at/main.php?show=prod ... =46&lng=en

Image
lodestar wrote:This review http://www.nikktech.com/main/articles/p ... imitstart= compares the NH-L12 with Noctua's NH-D14 cooler (the LGA 2011 version which features two PWM fans). The NH-L12 stands up quite well in this comparison, running under test conditions about 3C hotter at idle and 4C hotter under load. In doing so noise levels were signifcantly less than the NH-D14 SE2011.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Fri May 18, 2012 9:11 am

Thermal Paste
Lsv wrote:
Mr Spocko wrote:Used to be a die hard AS-5 user I still use it a bit from time to time. Wondering what folks are using for thermal paste any suggestions?
I used to be a die hard AS5 user until a guy at a comp shop said try out MX4, so I did and temps went down 1.5c at idle, about 2-3 at load. Nothing fancy but it helps. The non-conductive property of the MX4 and the fast cure time are the real sellers.
XBITLABS

Thermal Interfaces Roundup: Retail Products
http://www.xbitlabs.com/articles/cooler ... dup-1.html
Thermal Interfaces Roundup, Part 2: Bundled Products
http://www.xbitlabs.com/articles/cooler ... dup-2.html

MX4 review
http://www.thinkcomputers.org/arctic-co ... -review/2/

Benchmark Reviews - 80 Thermal Pastes Tested
http://benchmarkreviews.com/index.php?o ... itstart=12
Note that Arctic Silver 5 Application Instructions recommend 200-hours recommended curing time (MX4 doesn't have a cure time requirement)

MX2 did well (MX4 was not then available)
http://benchmarkreviews.com/index.php?o ... itstart=12

Noctua NT-H1 Thermal Paste (does surprisingly well - but not as easy to apply)
http://www.overclockersclub.com/reviews ... t_h1/5.htm
"NT-H1 is a better performer than AS5 (only just)" but the reviewer still preferred AS5 for ease of use

MX-2 vs MX-3 vs MX-4
http://www.hardwareheaven.com/reviews/9 ... ction.html
http://www.hardwareheaven.com/reviews/9 ... usion.html
http://www.redlinepctech.com/product-re ... mal-paste/

bjorn3d Thermal Paste Roundup
http://www.bjorn3d.com/read.php?cID=1605&pageID=7138
the winners were Thermal Fusion and Noctua

DansData - Arctic Silver 3 vs toothpaste vs Vegemite
http://www.dansdata.com/goop.htm
http://www.dansdata.com/images/goop/fibgraph2.gif (fibgraph2.gif )
http://www.dansdata.com/images/goop/nofibgraph2.gif (nofibgraph2.gif)

I think summing it up, the top choices are
MX-4, AS5, and NT-H1 in that order... they all do well so the most important factor is a good installation. Therefore ease of installation among these three is my priority. One of the things these tests don't do well is assess how well these pastes perform over time. To me that would be the winning attribute... but unfortunately it is an unknown.

Pappnaas wrote: http://www.hardwaresecrets.com/article/ ... -2012/1490
Even toothpaste, butter and others were tested.
How to Apply Thermal Paste
Tephras wrote:
Mats wrote:Here's an old reminder of how to apply the paste, it makes a difference. http://www.madshrimps.be/articles/artic ... dup-2007/4
Here's more:
What is the Best Way to Apply Thermal Grease? - Part 1
What is the Best Way to Apply Thermal Grease? - Part 2


Thermal Paste Advance - EPIC T1000

A common cause for unexpectedly poor CPU and GPU cooling is a poor thermal connection between the CPU/GPU and its heatsink.

The thermal connection between the CPU/GPU and its heatsink has always been the Achilles heal of heatsink performance. Even when properly applied, test results in reviews show material differences in performance between different TIMS. Up until now they have all been based on silver or ceramics.

Maingear has recently introduced a new advance in TIM called 'EPIC T1000'.

"... this TIM is unique because it is actually a Phase Change Metal Alloy (PCMA) that provides extreme thermal conductivity (>20 W/mK) and the lowest thermal resistance of anything on the market today. This is not just a minor improvement in thermal conductivity it is two to six times better conductivity than most commonly used TIM's." legitreviews.com
http://www.legitreviews.com/article/1836/1/

That sounds pretty impressive. There are other attributes that this article doesn't address.
(a) How easy is it to apply. If it is difficult to get a good seating, that is an issue. It doesn't do you any good to have a good TIM, if you fail to apply it in a manner that results in a good seating.
(b) What kind of burn in time does it require? Some times reportedly need a very large number of on/off cycles before they are working up to spec. Some TIMS are designed to work with no burn in. Others, that may perform better, require a longer burn in.
(c) How long will the TIM continue to work after it is applied. Some TIMS dry out after a while and need to be reapplied in order to maintain good performance.

What TIM do you use, and why? If anyone sees any reviews on EPIC T1000, please post it here.
dukeymo wrote:At the bottom of the article, it says this:
One final note for those observant readers, yes this is an OEM of Indigo Xtreme.
Indigo Xtreme has many reviews out there, so there you go :)
Thermal Paste Advance - Cool Laboratory Liquid PRO
http://www.coollaboratory.com/en/
particleman wrote:From the results I've seen on overclock.net you can shave off roughly another 15 degrees switching from Arctic Silver 5 to Coollaboratory Liquid Pro.
http://www.overclock.net/t/1283797/ivy- ... ng-results
I've been thinking of delidding my 3770k as well, but it seems to be a pretty risky operation. There are 3 people in the thread below that damaged their CPUs.
http://www.overclock.net/t/1280584/proj ... -lapped/70
Reviews From FrozenCPU who sells this product
http://www.frozencpu.com/products/3784/ ... d=9bDkfkCj

"Coollaboratory liquid metal not only has a high thermal conductivity and works well for air cooled rigs but it does especially well for TEC and phase change applications as its melting point is 8 degrees C, so it solidifies into a block of metal giving a near perfect thermal interface. Great support from the company when asking them about sub-zero specs. A few application tricks: Use dielectric under and along sides of CPU topped off with a silicone sealant to prevent any from getting to CPU during application and afterwards."

"Without a doubt this is the best TIM I''ve ever used, but beware, if you''re one like me who rebuilds their system every few months, this is not for you. Once its hardened on your CPU and cooler, you will need to lap both surfaces to get them clean and smooth again."

"Applyed to a Opteron 170 @ 3ghz 1.475v with no IHS under water cooling and my temps are 35''c loaded dual prime, Compaired to over 45''c on Dynex silver compound. which is about the lowest of the low when it comes to silver compounds. this stuff is a bit tricky to apply but take your time and its well worth the extra couple of bucks for this stuff"

Tzupy wrote:I apologize to graysky for linking the (other) thread at Anandtech, but it's even better documented:
http://forums.anandtech.com/showthread. ... &t=2261855
Someone claims that by replacing the MX-4 with Liquid Ultra, he got another 8 degrees lower temps.
To me this is kind of risky, Liquid Ultra may be conductive, it also 'eats' some metals.
Changing the Stock Thermal Paste in the 3770K Integrated Heatsink

viewtopic.php?f=28&t=64782
Last edited by ces on Mon Aug 06, 2012 4:23 pm, edited 6 times in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Sun Jun 03, 2012 12:33 pm

Scythe Big Shuriken vs. Noctua NH-L12 Low Profile Cooler vs. Scythe Big Shuriken 2 Rev B

The Scythe Big Shuriken 2 Rev B looks like the original Big Shuriken with an extra heat tube. Apparently SPCR is about ready to publish a review on the Big Shuriken 2 Rev B... and apparently it represents a substantial improvement over the old one.
MikeC wrote:Here's a tip: We just finished testing the new Shuriken, and it is considerably better than the last one.
If you, as I do, like small cases, with big CPUs, this heat sinks along with the Prolimatech Samuel 17 are the only heatsinks that are really exciting to choose among.... with maybe also the addition of the Scythe Kozuti to the low end of this list.

1. Scythe Kozuti - only 40mm high with the under slung 10mm fan. And because it uses the underslung fan it needs liilttle breathing room on top to do its job
2. Noctua NH-L12 - Without the top 120mm fan, the height is 66mm otherwise it is 93mm high with the top fan. With just its underslung fan, just like the Scythe Kozuti it needs little breathing room. (one option that SPCR found was that it was able to fit a 120mm fan on the bottom, that means it can be run with a single 120mm fan on the bottom or two 120mm fans, one on top and one on the bottom)
3. Scythe Big Shuriken - 57mm high with the stock 12mm fan. Replacing the stock 12mm fan with a 25mm fan brings it to 70mm high
4. Prolimatech Samuel 17 - With a 25 mm thick fan, the heatsink is 71 mm tall. The bolt heads add an additional 2 mm.

The one thing I have against the original Big Shuriken is its 12mm fan. Unexpectedly at 12 volts, it actually cools better with the stock 12mm fan than with the 25mm SPCR reference fan also running at 12 volts... but it does it so so so much louder. See:
http://www.silentpcreview.com/article1018-page6.html

The Noctua NH-L12 with just the stock underslung 92mm fan alone, cools about the same as the Big Shuriken is its stock 12mm fan at 12 volts... See:
http://www.silentpcreview.com/article1261-page6.html
With both fans, it is 9C cooler at the same dB level. At lower dB levels the gap increases to 17C.

Keep in mind that that the Big Shuriken with the standard 12mm fan at 12 volts outcools the Big Shuriken with the 25mm reference running at 12 volts. And also that the Prolimatech Samuel 17, with the standard 25mm reference fan at 12 volts outcools the Big Shuriken with the standard 12mm fan at 12 volts:
http://www.silentpcreview.com/article1261-page6.html

This is all before you even bring in the issue of dBs. As soon as SPCR publishes the review on the new Scythe Big Shuriken 2 Rev B I will update this post with that info.

For the time being here is a recent review of the Scythe Big Shuriken 2 Rev B
http://www.hardwaresecrets.com/article/842
Surprisingly in this testing regime, the Scythe Kozuti outperformed the Scythe Big Shuriken 2 Rev B.
Frostytech found that even the old Big Shuriken clearly outperformed Kozuti:
http://www.frostytech.com/articleview.c ... 614&page=5
We'll have to see what SPCR finds to get some reliable results.

SPCR's Updated 2012 Small CPU Heatsink Test Platform

This includes a compilation of test results for a number of small CPU Heatsinks on the new test platform
http://www.silentpcreview.com/article1267-page1.html

The Scythe Samurai ZZ fails to impress given its 94 mm height. It's not a bad cooler by any stretch of the imagination, but it's certainly outclassed by the NH-L12. However, if you're planning on a LGA1155 mini-ITX build with a discrete graphics card, the Samurai might be your best bet. The CPU socket on mini-ITX Sandy Bridge boards are so close to the PCI Express slot that most heatsinks interfere with it. The Samurai doesn't.

The Winner Is

It looks like basically the Noctua NH-L12 Low Profile Cooler out performs the Scythe Big Shuriken 2 Rev B
http://www.silentpcreview.com/article1270-page7.html
Last edited by ces on Wed Jun 13, 2012 11:04 am, edited 5 times in total.

potsy
Posts: 136
Joined: Wed Aug 30, 2006 4:55 am
Location: Australia

Re: What I have Learned about Computer Performance from SPCR

Post by potsy » Mon Jun 04, 2012 2:01 am

If you're only going to use the fan underneath the Noctua, can you swap the clips around and put the 120mm fan at the bottom? That would be quieter... [edit]: oh, I see in the review you can put the 120mm fan below.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Tue Jun 05, 2012 4:40 am

Fan Roundup #6: Scythe, Noiseblocker, Antec, Nexus, Thermalright
CA_Steve wrote:Thanks for the great review. Here's the data in graphical form. It could be fine tuned to help show the overlapped data points, but this picture really helps to focus the results. Hopefully, there weren't any transcription errors :)
Image
Looks to me like the Scythe Gentle Typhoons are the obvious winners here, especially is you assume that they likely have the best static pressure performance. Though the Noiseblocker S1 certainly looks good as well.

[
Last edited by ces on Wed Jun 06, 2012 4:10 am, edited 2 times in total.

fjodor2000
Posts: 100
Joined: Sun Sep 10, 2006 8:58 am

Re: What I have Learned about Computer Performance from SPCR

Post by fjodor2000 » Tue Jun 05, 2012 12:43 pm

ces wrote:The one thing I have against the original Big Shuriken is its 12mm fan. Unexpectedly at 12 volts, it actually cools better with the stock 12mm fan than with the 25mm SPCR reference fan also running at 12 volts... but it does it so so so much louder. See:
http://www.silentpcreview.com/article1018-page6.html
Well, is the cooling performance per volt really of interest? I mean, you could have a crappy fan that runs at 5000 rpm @ 12V. It would probably cool pretty good, but it would be noisy as hell.
ces wrote: Keep in mind that that the Big Shuriken with the standard 12mm fan at 12 volts outcools the Big Shuriken with the 25mm reference running at 12 volts. And also that the Prolimatech Samuel 17, with the standard 25mm reference fan at 12 volts outcools the Big Shuriken with the standard 12mm fan at 12 volts:
http://www.silentpcreview.com/article1261-page6.html
Isn't all that matters is cooling efficiency per dB instead? So the 25 mm reference fan might still be better (despite not outcooling the stock Big Shuriken 2 Rev B fan @ 12 V), since it might be able to deliver more cooling per dB anyway?

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Why Fan Cooling Performance at Specific Voltages is Importan

Post by ces » Wed Jun 06, 2012 10:40 am

fjodor2000 wrote:Well, is the cooling performance per volt really of interest? I mean, you could have a crappy fan that runs at 5000 rpm @ 12V. It would probably cool pretty good, but it would be noisy as hell.
I run my fans at either 12, 7 or 5 volts. So how they run at those three voltages is important to me.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Mon Jul 16, 2012 3:20 am

Interesting Idea for the Antec NSK3480, which has limited room for a PSU
Ice Tea wrote:ImageImage Anyone tried an PSU extender bracket on the Antec NSK3480 with a Seasonic Fanless supply?


Lian Li PE-01 Power Supply Extension Bracket - Black (PE-01)
http://www.frozencpu.com/products/7798/ ... PE-01.html

It is 60mm deep:
http://www.lian-li.com/v2/en/product/pr ... dex=84&g=f

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Tue Jul 17, 2012 11:58 am

Intel 330 SSD vs the Intel 520 SSD
m1st wrote:http://www.anandtech.com/show/5751/inte ... -sandforce

The Intel 330 and 520 series use a Sandforce controller, while the 510 uses a Marvell controller. The 330 and 520 will give higher peak performance and probably higher real-world performance. Their only weakness is incompressible data, but they'd still probably outperform the 510. The 510 was one of Intel's first 6Gb/s-capable drives, so that's probably why it's so overpriced.

Between the 520 and the 330, the only difference is the NAND used. Both use top-quality Intel NAND, but the 520 uses NAND rated for at least 5k writes. The 330 uses NAND rated for at least 3k writes. Even 3k is more than most will ever use, but it's something to consider. the 520 comes with a 5-year warranty, and the 330 comes with a 3-year warranty.

Lastly, the 330 has slightly lower IOPS ratings due to the NAND configuration. Nothing drastic, but it's something to consider.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Thu Jul 19, 2012 8:52 am

Intel To Release Haswell Processors & 8 Series Chipsets in April 2013
http://www.legitreviews.com/news/13682/

It will use an LGA 1150 with the same 75 mm between each screw hole as the 1155 and 1156
http://en.wikipedia.org/wiki/LGA_1150
The integrated VR and shared cache, among other features compels a change in pin count. But exiting cpu coolers should work OK with the new LGA 1150.

"The next generation scheduled for mid-2013 codenamed Haswell is a second-generation 22nm part with the Lynx chipset that has 25 percent lower TDP and 50 percent lower average power consumption than the current Panther Point chipset." The reason this is so important is that the chip sets are now the greatest wattage sink in ordinary day to day usage.
http://www.fudzilla.com/home/item/27992 ... -lower-tdp

The CPU and Graphics will have shared last level cache and will have fully integrated voltage regulation
http://forums.anandtech.com/showpost.ph ... ostcount=5

Haswell is going to be a "graphics monster"

The Intel Sandy Bridge HD2000 has 6 shader units. The HD3000 has 12. The Ivy Bridge HD4000 has 16 (each of which is materially more powerful than the SB shaders).
The High End Haswell will have 40 shaders.
http://semiaccurate.com/2012/02/08/hasw ... s-monster/

Each of these Haswell shader units may be qualitatively superior, by not a little, to the SB units because of the large chunk of shared last level cache with the CPU. The shared cache will give them an unfair architectural advantage over video cards. This could turn out to be a real game changer.
http://semiaccurate.com/2012/04/02/hasw ... ystalwell/

A Changing Thermal Target

According to Anandtech, while the Haswell may have a similar range of TDP to Ivy Bridge, the line will focus on a lower target TDP:
http://www.anandtech.com/show/4371/inte ... notebook/3
Last edited by ces on Sun Jul 22, 2012 6:25 am, edited 4 times in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Thu Jul 19, 2012 1:29 pm

A medley of Fan Tests

SPCR

80x25mm Fan Round-Up #1
http://www.silentpcreview.com/article689-page1.html

SPCR's Fan Round-Up #2: 120mm Fans
http://www.silentpcreview.com/article695-page1.html

SPCR's Fan Round-Up #3: 92mm Fans
http://www.silentpcreview.com/article720-page1.html

SPCR's Fan Roundup #4: 120mm Fans II
http://www.silentpcreview.com/article739-page1.html

Fan Roundup #5: Attack of the 120 Scythes
http://www.silentpcreview.com/article832-page1.html

Fan Roundup #6: Scythe, Noiseblocker, Antec, Nexus, Thermalright
http://www.silentpcreview.com/Fan_Round ... ermalright

XBITlabs

Roundup: 11 Fans
http://www.xbitlabs.com/articles/cooler ... undup.html

Ultimate 120/140 mm Fans Roundup
http://www.xbitlabs.com/articles/cooler ... undup.html

140 mm Fan Roundup
http://www.xbitlabs.com/articles/cooler ... undup.html

120 mm Fan Roundup, Part I: 1350 RPM or Lower Rotation Speed
http://www.xbitlabs.com/articles/cooler ... dup-1.html

120 mm Fan Roundup, Part 2: 1350 RPM or Higher Rotation Speed
http://www.xbitlabs.com/articles/cooler ... dup-2.html

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Mon Jul 23, 2012 9:08 am

FSP Group FSP300-60GHS-R 300W SFX12V 80 PLUS Certified Power Supply

This is a very under appreciated power supply. It is an SFX powersupply that comes with a plate that permits it to mount as an ATX power supply. It is perfect for small ITX cases such as the Lian Li Q07. It leaves extra room in these tight cases. In the Lian Li Q07 it provides an extra critical inch of headroom for a CPU heatsink.
ces wrote:
merlin wrote:I'm looking for a good Mini-itx case to house a sandy bridge Core I3-2100 with mini-itx htpc myself, the Antec ISK-300/310-150 would be perfect except the rather poor psu and I'd prefer not to go picopsu if possible. Plus who wants to pay for a second psu? :) I wonder if antec would come out with a quieter high efficiency psu model...
Consider a Lian Li q-07 or q-11 with a small FSP Group FSP300-60GHS-R 300W Micro ATX / SFX 80 PLUS Certified Power Supply mounted as an ATX to make room for a larger cpu cooler (the ATX mount comes with the PSU)
http://www.newegg.com/Product/Product.a ... 6817104075
mikewu wrote:I finally built a truly silent system without going through a lot of trouble or using really special components. Even with really good/expensive silent components I've had issues with electronic noise (my DH61AG motherboard squeals some when doing I/O, my Seasonic X power supply emits a high tone when in standby).

- Intel DH77DF Mini-ITX motherboard
- Intel 3570K CPU (2 cores disabled in BIOS to reduce peak power, make the codes stay at 3.8GHz turbo)
- Rosewill RCX-Z90-CP heatsink (copper core w/aluminum fins)
- Raidmax ITX-0907-BP enclosure
- Fortron FSP300-60GHS-R 80 plus micro-atx power supply (with cover and fan removed)
- Sandisk Extreme 240Gb SSD
- 2 sticks of Samsung 2Gb DDR3 1600 memory (CPU-Z says that they are running at 1.35V)

The two components that matter the most are the DH77DF motherboard and the FSP300-60GHS-R power supply, the rest of the components were just personal preferences. In my case, the FSP300-60GHS-R sits at the bottom and I remove its top cover and fan completely so heat rises up into the enclosure. Similarly I remove the fan from the RCX-Z90-CP (it rubs a lot so it's not quiet even at slow RPMs). Then I mount an Artic cooling 120mm case fan (Arctic F12) running at 5V above the CPU heatsink where the DVD drive would go. It provides airflow to the CPU, motherboard and FSP300-60GHS-R components while gently forcing heat out of the enclosure.

There is no electronic buzzing in any situation. Standby power is 1W, idle at the Windows 7 desktop driving a 1080P display and connected to gigabit ENET is 21W. Load w/2 cores at around 3.7/3.8GHz is a little over 50W. 4 cores w/graphics comes to around 75-80W. Audio output is perfectly clean as seems to be the case with Intel motherboards in general.

The FSP300-60GHS-R is a good power supply. I've had a pair of them running 24/7 with the cover and fan removed as I did here for about a year. At the low idle power of Intel's Ivy/Sandy Bridge CPUs I actually get about the same measured power as my Seasonic X PSUs (21W is only 5% of my Seasonic X's rated load, to low to be in its ~90% sweet spot).

Anyway, this system is silent with any acoustic dampening. And it is fast, I didn't have to make performance tradeoffs to get it to be quiet.

Mats
Posts: 3044
Joined: Fri Dec 26, 2003 6:54 am
Location: Sweden

Re: What I have Learned about Computer Performance from SPCR

Post by Mats » Mon Jul 23, 2012 12:11 pm

Silverstone has rebranded FSP SFX PSU's, including the 450 W version.

Then we have the upcoming 80Plus Gold version with modular cables, but it's not made by FSP.
viewtopic.php?f=6&t=64249

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Thu Aug 02, 2012 4:15 am

Interesting new approach to noise generated by the tips of fan blades

Noiseblocker
IanM wrote:New Noiseblocker fans just appeared on Caseking today! I don't know anything about them but they look very interesting, especially the B12-PS that is specified as 400rpm-1500rpm. Noiseblocker web page for these is in German only at the moment: http://www.noiseblocker.de/eLoop120.php
  • B12-1 : 800rpm, 7.83 dBA, 57.3 m³/h
    B12-2 : 1300 rpm, 16.7 dBA, 86.9 m³/h
    B12-3 : 1900 rpm, 26.5 dBA, 121.2 m³/h
    B12-4 : 2400 rpm, 34.3 dBA, 150.3 m³/h
    B12-PS : 400-1500 rpm, <21.2 dBA, 98.7 m³/h
    B12-P : 800-2000 rpm, <28.4 dBA, 132.4 m³/h
grab the pdf (in German) to read the claimed static pressure figures and other specs: http://www.noiseblocker.de/Datenblaette ... 082012.pdf

Image Image
Pappnaas wrote:http://www.evologics.de/en/products/pro ... index.html
Seems like the connected blades are part of the innovation.
Antec TrueQuiet Pro 120

http://www.silentpcreview.com/article1266-page6.html
http://www.silentpcreview.com/article1266-page7.html

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Thu Aug 02, 2012 4:50 am

Quiet Water Cooler
josephclemente wrote:Case: SilverStone SG05
Fan: Scythe Gentle Typhoon AP-13 1150 RPM 120mm, used as front exhaust
PSU: Seasonic X-460 Fanless
Motherboard: ASRock Z77E-ITX
CPU: Intel Xeon E3-1230 V2
CPU cooler: Antec Kuhler H2O 620
RAM: Samsung 8GB 30nm 1.35V DDR3 1600
GPU: PowerColor Go! Green AX7750
SSD: Samsung 830 Series 256GB

I replaced the original restrictive front hex grill with a much more open hex grill (ModRight ModMesh).

Image

I modified this case to accommodate a full size ATX modular power supply.

A sheet of aluminum was cut to hold the SSD drive at the side of the case.

I measured AC power consumption with a Kill A Watt.

Power Consumption:

IDLE: 39W
Prime95: 98W
Prime95 + Furmark: 138W

Prime95 was set to Torture Test In-place large FFTs.
josephclemente wrote:The pump and fan noise is very low. Audible, but low enough where I often mistake the computer for being off.

The fan and pump speed is controlled by the motherboard in BIOS. Too slow and the pump makes a light clicking noise. So I have the speed up a few notches from the lowest setting.
Intel Xeon E3-1230 V2 Ivy Bridge 3.3GHz (3.7GHz Turbo) 4 x 256KB L2 Cache 8MB L3 Cache LGA 1155 69W Quad-Core Server Processor BX80637E31230V2
http://www.newegg.com/Product/Product.a ... 6819117286

If you know you are going to have a serious heat problem, and price is no object.
ces wrote: If you know you are going to have a serious heat problem, and price is no object.

1. Start by feeding fresh cold external air directly to your CPU cooler. That is good for 10 to 15C right there.
2. You can do this with a fan duct. Get a red or blue (one is longer than the other) Thermalright 120mm fan duct. Get both and experiment for which will fit your setup best.
http://www.sidewindercomputers.com/th12fandubl.html
3. Get a good downdraft CPU heatsink. The Noctua NH-C14 will work. Use two Nexus 120mm fans or two 120mm Scythe GT or two 120mm PWM fans.
4. Use the Thermalright fan duct to duct cold air to the top fan. The bottom fan will make sure to cool your motherboard VRMs

Now, is that isn't good enough, you will have to try a water cooler. Just remember though:
1. If one of your two air cooler fans breaks down, it is unlikely to do any damage at all.
2. If your pump breaks down you are going to over heat something. You hope that your CPU goes to 110C and then shuts down. If you are just idling or using it to browse, you could cook your motherboard, and its tender capacitors, with 100C for days without knowing it. Same thing if your water cooler fan breaks. And you also have serious problems if you spring a leak.

viewtopic.php?p=563366#p563366
Last edited by ces on Fri Aug 03, 2012 12:59 pm, edited 1 time in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Fri Aug 03, 2012 11:43 am

Static Pressure - What you need to know

Here is a good Youtube video by NCIX that, with the accompanying downloadable charts, does a good job of explaining static pressure and why a good case fan can fail miserably as a fan for use on a dense, high static pressure, CPU heat sink.

http://www.youtube.com/watch?v=9uUXt7mE ... tu.be&t=9m

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Fri Aug 03, 2012 8:45 pm

Use A Paper Collar To Get A Few More Degrees Of Cooling
MikeC wrote:
ces wrote:
MikeC wrote:But you can feel a lot of airflow coming through from the sides of the stack (ie, perpendicular to the fan), as well as around the periphery of the fan where it meets the fin stack.
What do you think of the idea of placing paper around the sides of a CPU cooler?
I trust you mean so that the air cannot flow out the sides of the fin stack? It would force more of the airflow through the center, but this depends a lot on the details. If the fin stack was especially dense w/ narrow spacing, that might just create more back pressure.
Seems like it would be very very interesting to
(a) block the sides and top of a Scythe Mugen (a tight 1.89 mm fin spacing, but with multiple channels to help airflow)
(b) block the sides and top of a Prolimatech Megahalens (2.00 mm fin spacing, but with a channel to help airflow)
(c) block the sides and top of a HR-02 Macho (a generous 3.12 mm fin spacing and deep enough to benefit from the help)
(d) block the sides and top of a Scythe Ninja 3 (a still healthy 2.64 mm spacing and deep enough to benefit from the help)

Though it is unclear whether doubling up on fans helps deep coolers more than skinny coolers. Maybe:
http://www.silentpcreview.com/article1279-page6.html

It seems like you can use the collar to pick up a few degrees of cooling, or to remove the second fan and lose a few dBs of sound.
ces wrote:
ces wrote:
suchageek wrote:Thanks for the tip. I guess it doesn't matter that it mounts horizontally? How does one fasten the collar?
how about aluminum foil?
suchageek wrote:Nope. Not going to use aluminum foil.
Yeah, I think you are right. What about some craft paper or some cardboard (from say a cereal box) and some tape. Certainly good enough to experiment with.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: What I have Learned about Computer Performance from SPCR

Post by ces » Sun Aug 05, 2012 11:08 am

Fan Vent Covers
JJ wrote:I'm building a file server in an older Cooler Master RC-590 case. The case has two top fan openings, two door fan openings, and even a fan opening in the door behind the motherboard tray. I want to cover these fan vents. I can use duct tape or something similar, but would like to find something neater. Does anyone make vent covers that are mounted using screws?
Image
Das_Saunamies wrote:
Thanks, now I feel embarassed; why did I search for "fan cover" when it's obviously a "fan mount cover". First hit is the Puget. :D

QUIET!
Posts: 238
Joined: Sat May 11, 2013 8:33 am

Re: What I have Learned about Computer Performance from SPCR

Post by QUIET! » Sat May 11, 2013 4:10 pm

ces wrote:Intel To Release Haswell Processors & 8 Series Chipsets in April 2013
http://www.legitreviews.com/news/13682/

It will use an LGA 1150 with the same 75 mm between each screw hole as the 1155 and 1156
http://en.wikipedia.org/wiki/LGA_1150
The integrated VR and shared cache, among other features compels a change in pin count. But exiting cpu coolers should work OK with the new LGA 1150.

"The next generation scheduled for mid-2013 codenamed Haswell is a second-generation 22nm part with the Lynx chipset that has 25 percent lower TDP and 50 percent lower average power consumption than the current Panther Point chipset." The reason this is so important is that the chip sets are now the greatest wattage sink in ordinary day to day usage.
http://www.fudzilla.com/home/item/27992 ... -lower-tdp

The CPU and Graphics will have shared last level cache and will have fully integrated voltage regulation
http://forums.anandtech.com/showpost.ph ... ostcount=5

Haswell is going to be a "graphics monster"

The Intel Sandy Bridge HD2000 has 6 shader units. The HD3000 has 12. The Ivy Bridge HD4000 has 16 (each of which is materially more powerful than the SB shaders).
The High End Haswell will have 40 shaders.
http://semiaccurate.com/2012/02/08/hasw ... s-monster/

Each of these Haswell shader units may be qualitatively superior, by not a little, to the SB units because of the large chunk of shared last level cache with the CPU. The shared cache will give them an unfair architectural advantage over video cards. This could turn out to be a real game changer.
http://semiaccurate.com/2012/04/02/hasw ... ystalwell/

A Changing Thermal Target

According to Anandtech, while the Haswell may have a similar range of TDP to Ivy Bridge, the line will focus on a lower target TDP:
http://www.anandtech.com/show/4371/inte ... notebook/3
Since April 2013 has come and gone but Haswell has not come, I think tempered expectations are in order.

The Intel road maps showing Haswell parts do not include desktop parts with the high end graphics, the special embedded dram graphics version is mobile only where the CPU is soldered to the motherboard, great for laptops but SPC, not really.

The other dust up was a stock analyst that downgraded Intel stock when the leaked Haswell specs showed higher tdp than the ivy bridge being replaced. This isn't much of an issue because the vrm is moved on die (I think) but it suggests that power wise Has well is not a tremendous improvement.

My prediction is that buying a PSU with good light load efficiency will provide much more bang for the buck than paying premium prices to be a Haswell early adopter.

And speaking of PSUs, it turns out most PSUs are not compatible with Haswell's low power states so add that to the shopping list or look for Haswell compatibility if you are in the PSU market if you want something a little bit future proof.

Any way, the Haswell value proposition at the low to mid end might not be there in the beginning while low end ivy bridge will have an inexpensive upgrade path as high end ivy bridge parts get to EOL.

Post Reply