Buildegg Articles Computer Hardware Analysis

20Sep/092

Striped Array: Mobo Hardware Differences

Striped arrays are accessible to almost everybody ( we're focusing on PCs with Windows). Windows offers a software solution, and many motherboards support RAID arrays. Some boards, like the Gigabyte Ep45-ED3P ( a mobo based on Intel's P45 chipset ), have multiple options for hard drive setups due to the inclusion of their own controller ( Gigabyte SATA2 Chip ). If one is running the aforementioned motherboard, that provides one with three options for RAID 0, and that is not including buying a stand-alone RAID card. Naturally, one would ask one's self, "Which will offer the best performance?".

This comparison is not exhaustive whatsoever. It is a quick, simple test of different ways the average PC enthusiast can setup a striped array. Tested setups:

  • Windows 7 Striped Array by Dynamic Volume (pure software, cannot be booted from / will not have an OS)
  • Intel ICH10 South Bridge (apart of the P45 Chipset)
  • Gigabyte SATA2 Chip (Gigabytes own controller)
  • A single drive as a 'baseline'
Two WD5001AALS used in the tests

Two WD5001AALS used in the tests

The arrays were synthetically tested with HD Tune Pro 3.5. The first test is the HD Tune 'Benchmark' for transfer rates, access times, and bursts rates. NOTE: The Windows striped dynamic volume is not available for the HD Tune tests.

HDTune Benchmark Single Drive (WD Black 500)

HDTune Benchmark Single Drive (WD Black 500)

HDTune Benchmark for Intel ICH10

HDTune Benchmark for Intel ICH10

HDTune Benchmark for Gigabyte SATA2 Chip

HDTune Benchmark for Gigabyte SATA2 Chip

The RAID 0 array on Intel ICH10 had an average transfer rate almost double that of a single drive. The Gigabyte SATA2 Chip benched slower, at 83% of the Intel chipset.

The next test was HD Tune's 'File Benchmark'. This is a more practical test, as it simulates reading and writing files.

HDTune File Benchmark Single Drive (WD Black 500)

HDTune File Benchmark Single Drive (WD Black 500)

HDTune_File_Benchmark_Intel___Raid_0_Volume

HDTune File Benchmark Gigabyte SATA2 Chip

HDTune File Benchmark Gigabyte SATA2 Chip

The Intel ICH10 was about 80% faster on write speeds than the single drive, and the Gigabyte was 25% faster. The read rates were even more revealing, with the Intel array at about 55% faster, and the Gigabyte array coming in slower at around -5%. This test was performed a few times along with the 'benchmark' test on write mode, and each time the Gigabyte SATA2 Chip performed extremely poorly -- in the above case, the single drive was faster than the Gigabyte striped array!

Finally, the arrays were subjected to a simple copy/paste of about 8 gigs of data from and to themselves. There are probably a ton of parameters at play in this situation, but the only variable is the hard drive controller. The Windows 7 striped dynamic volume was included in this test.

Running the RAID 0 array purely through Windows provided the same performance than the hardware controlled Intel ICH10 array. The Gigabyte SATA2 Chip based striped array performed poorly in comparison.

There are two take home messages from these tests:

  • Do not use Gigabyte's supplied hard drive controllers for your RAID setups! Its performance was ghastly.
  • If you're not booting off of your planned striped array, considering letting Windows take care of it for you.

A hardware approach ( like the ICH10 RAID controller ) may be more efficient ( but based off of the above results, that is just speculation ), but the big benefit of a Windows based array is that it does not depend on specific hardware. One cannot move a RAID 0 or RAID 1 array from one hardware controller to another, but if the array is managed by Windows, hardware in the computer can be upgraded without killing the array.

Filed under: Uncategorized 2 Comments
14Sep/092

PC Explosion Project: Part 2

We return to our test bed with lessons learned from 'Part 1'. Now that the weakest link, the hard drive, is mounted outside of the case, can the computer be blown up and sustain an operating environment?

All of the original components were functional after two firework artillery shell explosions from 'Part 1', except for the hard drives. We discovered that hard disk drives were prone to failure when submitted to extreme temperatures and shock (wow, really??). Armed with another old IDE drive and long-ass cables, we mounted our storage outside of the case, underneath the blast zone.

HDD Mounted Outside

Windows XP was installed again, along with 3dMark2001SE, and this time: SpeedFan 4.38 for monitoring system temperatures. As long as we can keep the computer running, we can see what kind of temperature abuse the motherboard and CPU are experiencing.

The first artillery shell of the day was placed in the case, at the bottom, pointing up. The wick was ran out the back of the case through a vent hole. After the benchmark software and SpeedFan were loaded, we blew it up.

The Explosion

Unlike the results in 'Part 1', the Lian Li case remained, somewhat, intact. We believe that the earlier explosions warped the side panels enough that most of the blast pressure was expelled through loose cracks (and through the open 3.5" drive bay). As a result, the majority of the firework burned on top of the computer innards.

It Did Not Survive...

The operating environment did not survive; the computer was immediately shut off during the blast. Inspection of the debris revealed melted capacitors, resistors, and solder points. The BIOS chip completely removed, and those things are in there tight. Unfortunately, simply sticking the BIOS chip back onto the motherboard did not bring the test bed back to life. The A7Pro, GeForce 3, and Sound Blaster were dead.

Slightly bummed, we did a quick replacement of the motherboard and video card. The new components were from the some time period, but cheaper. We think the mobo was Asus, but couldn't find a model number (and it was seriously lacking in features). The video card was another GeForce 3, but card manufacturer was some unknown brand. The hard drive survived, but Windows XP had to be reinstalled due to the change in hardware ( didn't have to do that with Windows 7 ).

To make a night-long story short, we blew up the computer three more times and were never able to keep the operating environment intake (still running). Our best shot was from 'Part 1', the second explosion. In that instance, the computer froze, but did not shut off... it may have continued to run if the hard drive did not die. Maybe we'll revisit this project with a modern computer, but for now, we leave you with an explosion montage from 'Part 2'.

Filed under: Uncategorized 2 Comments
14Sep/092

PC Explosion Project: Part 1

With July 4th right around the corner (Independance Day and overstocks of fireworks), we set out to answer the question on everyone's mind: Can a computer survive an explosion from a 1.5 inch artillery shell? We built a fully functional PC and exploded an artillery shell firework from within its case. The PC was be subjected to shock, pressure, heat, and amputation of components; survival seemed most improbable.

A modern computer would have been awesome to use for this project, but that would be like setting money on fire, right? We looked around the ol' closet for some old parts and found an eight year old gaming rig.

* AMD Athlon Thunderbird 800 MHz  *Asus A7pro  *NVIDIA GeForce 3  *Create Sound Blaster Live!  *Lian Li Case

System Components

We were going to try this project using Windows 7 RC 1.0 (atm, Buildegg = total Win7 fanboi), and thought we would be successful, but we could not get through the installation. So we reverted back to Windows XP, and chose to use 3dMark2001SE as the software to be running during the explosion.

-= Disclaimer: This is probably a stupid thing to do. Don't do it. =-

The artillery shell was placed inside the case, at the bottom, in the center, and in an orientation so that the first propulsion charge would shoot the shell up. We ran the wick to the rear of the case, and out through a vent hole.

Shell Placed Inside Case

The case was sealed up in a way that:

  • All bay covers in place (except for one 3.5 inch "floppy" cover... that was lost)
  • All expansion card slot covers were screwed into place
  • The case sides were slid / clipped into place, but rear screws were not installed

Running 3DMark

The size of the explosion was unknown. It might have been as violent as mullet guy's, or quite dull. In any case, the cameras were positioned at a safe distance, and zoomed in. We booted to Windows, began the 3dMark2001SE "demo", and lit the wick...

... not as loud of an "bang" as we were expecting, but that may be because the artillery shell packaging did not label which shells had what kind of effect. It turned out we used one of the "explode then crackle" shells. Nevertheless, the result was ultimate carnage.

We found the Lian Li case completely dismantled. The front panel, and every 5.25 bay cover, were strewn upon the ground. Both side panels were blown out; now mind you, even though we didn't use rear screws, there were relatively massive clips and hooks holding those panels in place, all of which were bent by the explosion. The video card and sound card were half way out of their AGP/PCI slots. The case, and all of the components, were very hot (almost too hot to touch). There were char marks and smoldering fragments throughout. The outlook for this old gaming legend was grim.

Inside After First Explosion

Did it survive? After the blast, the LCD panel went dark from no data. The power LED on the case was blinking, instead of being constantly on. We pulled the plug to the PC, brushed out any fragments, and reinserted the expansion cards. Power was restored, and we turned on the computer. As the power button was depressed, we were prepared for sparks and/or total failure, but instead we were greeted with spinning fans, POST information to the LCD, and even a "all is well" beep. It was really un-freakin-believable.

There was a casualty however; other than the sound of fans, the first thing we noticed during that first boot was the loud, consistent clicking of a destroyed hard disk drive. Western Digital 15 gigger... rest in peace.

We were not exactly prepared for the system to be POSTing, as it caught us off guard. We grabbed the last IDE drive laying around, hooked it up, and we were installing XP on a just recently blown up computer.

Reinstalling XP on a new drive

Can a computer survive an explosion?

Yes

New question: Can a computer survive an explosion, and continue running?

We had time for another round.

Why did the computer shut off during the explosion? An obvious answer would be the destruction of the hard drive, but there were other factors that could have caused the crash, like the heat and the movement of the expansion cards.

In an effort to protect the drive, we mounted the next artillery shell up in the 5.25 bay area. Also, this "new"  drive was a WD 60 GB; last known to be in good shape (that 15 gig drive was really old). The expansion cards screws were torqued down tight.

We ran the wick out the front of the case, sealed the case up, started 3dMark, and blew the crap out of the PC again. This time the artillery shell firework was not a crackler, it was a one that would have lit up a neighborhood. The blast was so bright, the cameras were overexposed for most of the explosion.

The Second Explosion

The frame that the 3dMark demo was on during the blast remained on the LCD panel, definitely a triumph over the first blast, even though that frame remained frozen on the screen. Other than that, the aftermath was similar to before... utter devastation.

Miraculously, the PC started to boot again after this second explosion. As the system tried to boot from the hard disk, we unfortunetly learned that we had lost another drive. Everything still seemed fine, but we did not have another IDE drive to continue testing.

PC Survived, HDD Did Not

* * *

Before we started this project, we thought the hard drive would be the last thing to be affected by the explosion. Of course, we were thinking about it all wrong; even though the drive is enclosed in a seemingly indestructible shell, its mechanical nature makes it susceptible to shock. Be it shock, pressure, or heat, so far the hard drive does not survive, and is the weakest link.

The stakes are higher now. How awesome would it be to see a computer explode, and continue to run a benchmark. The new goals:

  • Have a PC continue to run a game demo / benchmark as a 1.5 inch artillery shell (firework) is exploded within the case.
  • Mount the hard drive externally / remotely, as it is known to fail in this situation
  • Monitor and log internal temperatures (motherboard temp, CPU temp)

Spare IDE drives are on the way. The project will be continued under Part 2.

Filed under: Uncategorized 2 Comments