Cyrious

Idle power consumption and the effect it will ultimately have on the house cooling system also needs to be taken into account, as cards with high idle wattage will ultimately cost the user more in the long run even if they can be bought cheaper immediately.

No, but MSI could have had the damn decency to add VRM protection like every other manufacturer out there. Hell, my cheapo Asus AM2+ board has VRM protection, and I've managed to trip it on at least one occasion before I almost melted the CPU power plug through OCing the Phenom II too hard. If a $50 motherboard can have VRM protection, MSI's $200 AMD boards can have it too.

Intel's Sata blunder didn't stem from cost cutting, but from a tiny mistake that snuck its way past all of the verification processes and made it to production. It's happened quite a few times from just about every major silicon manufacturer, and they're referred to as errata. In some cases, like with the TSX bug in all Haswell based chips, it doesnt detract from overall chip performance or isnt that severe. In other cases, like the Pentium FDIV bug and AMD's TLB bug, it can be quite damaging to the company's reputation.

The gist of it is that there was some corporate espionage, and one person stole the designs for more efficient electrolytic capacitors from Rubycon after leaving them and sold them to basically every chinese electronics manufacturer who was willing to pay, which was quite a lot. However, he failed to get the name of a critical additive that would prevent the electrolysis of the water in the capacitor, which in turn would prevent hydrogen generation and oxidation of the aluminium components inside. Once the pressure inside built up too much, the capacitor would explode. This ended up plaguing every electronic device using electrolytic capacitors, but was most notable on computer motherboards. Solid polymer capacitors came around, and not only did they not have the flaw of being explodey, they had superior performance compared to a equivalent electrolytic, and didnt wear out over the years. So, everyone's high end stuff switched to solid capacitors.

http://www.overclock.net/a/database-of-motherboard-vrm-failure-incidents Note just how many are MSI boards. Yes, they were overclocked, but the fact of the matter is that MSI does not have any sort of VRM protection, and if they do its tuned so high that its useless. Any other board would have auto shut-down long before the burn.

Oh god, i had the misfortune of having to deal with a 1U server while in high school. 30mm fans, 9000RPM, and they kicked up at the drop of a hat. My Phenom II box is only configured as a network storage as I needed a place to bounce things back and forth between my desktop and laptop as well as more bulk storage. I dont do anything fancy with it, even though I could theoretically install ECC Unbuffered Dram in it and get the benefits, as the board and CPU apparently support ECC. Uh, you touch my Xeon I'll run you over with my 400 ton super mining crawlers. If you need moar coars for rendering, hook yourself up with a 4P G34 rig and load it up with 12C Thuban based or 16C Piledriver based Opterons. The same friend who gave me the Xeon had 4x12C 3.8GHZ Opty Rig that got over 40 in Cinebench 11.5 and for about a year held the world record Cinebench Score on HWbot. It also drew >1000W at the wall, and burned itself up after a week because supermicro boards werent meant to handle 300W a socket.

MSI's boards, especially their AMD boards, have a well earned reputation for blowing the VRMs and taking the CPU with them. As I do distributed computing alot, it greatly increases the odds of the board going up in flames and destroying the CPU. This in turn would have me sitting back on my Q9400, and I really dont want to go back to Core 2 Quad.

Yeah I was thinking about grabbing that for my Xeon, but it doesnt officially support it, and the only other X79 boards to support Xeons are MSI's offerings, and I'll be dead before I stick a $2000 Xeon on an MSI board. What makes selecting the new board even harder is that it must have some sort of supplemental PCI-E power plug, as I overclock the crap out of my graphics cards (including overvolting), and the GT 440 144 doesnt have a PCI-E 6-pin to feed it extra juice, which means all that it gets comes from the slot.

Heh. The Xeon doesnt require much to keep it cool; the Xigmatek Gaia w/ 2x Coolermaster Blademaster 120mms on it does a wonderful job even when its sucking in heated air from the GTX 460. I had to orient the heatsink so it was venting up though, as the VRM heatsink needed some loving and there's a pair of 120mm fans for the upper exhaust. Meanwhile my own "fileserver" is sitting on a heavily underclocked and undervolted Phenom II 940. I want to use my Pentium M box for that purpose, but the CPU was a huge bottleneck during testing for that purpose.

Please note my Q9400 results much further down.

Yep, thats why I intend on getting the Asus P9X79 WS motherboard for my Xeon used instead of new. New, its over $370 USD on newegg, yet used I can get one in excellent condition off ebay for $250.

The Xeon? I got it as a freebie from a friend who works in the industry and who saved it from the scrap bin. I use it as my daily driver, as its replacing my Q9400 rig. The Q9400 in turn was suffering a bit on single or lightly threaded programs even OCed to 3.6ghz. The Xeon, even though the all core turbo clock gives it a 300mhz disadvantage against the Q9400, beats the crap out of it in single and multi-threaded programs. If you want I'll post a screenshot of Cinebench R15 so you can get an idea of just how much of a beast this thing is.

Huh, didn't know that about digital meters. My Electronics and Aviation Tech classes never really got into the really deep nitty gritty of circuit design, just the basics and how to fix em. It's also been about 5 years since I took those classes, and haven't really put them to much use. Even then, while in class unless it was specifically required for the lesson I always used a digital meter as I had issues figuring out how to read an analog meter.

"Replace it or butcher something else to fit. Case fans and cable ties work surprisingly well." I have a Tesla M2050 that I had to lash a trio of 80mm fans to it so it wouldnt overheat. If i could bolt on a third party heatsink I certainly would, but the PCB, heatplate, and backplate designs kinda forbid it, and i dont have an AIO cooler or waterblock to hack onto it. As for the multi-meter: ye olde multi-meter much? Mine is purely digital.

I got mine from my high school magnet program. They had a crapton of multi-meters that they were tossing (mainly student damage, the bulk of the students I went to that magnet program with could best be described as mongoloids), and I asked if I could tear them down for good parts and assemble a working multi-meter out of the results, I got the go-ahead and about 30 minutes later I had a multi-meter to call my own. But seriously though, not that many people other than us geeks have their own multi-meter or the training to properly use one. Its a fine thing to be trying to diagnose a live circuit of a fault, make a mistake because you didn't know what you were doing, and short something that really shouldnt have been shorted (12v DC directly to the GPU for example.).

Secondhand and generic on fleabay, with other brands starting at $70. I almost never buy new hardware, as I can generally get what I want/need used and for a fraction of the price of new. When my GTX 460 hit the market it was over $250. I bought it used for under $60, and considering the games that I play aren't too demanding, it works just fine.

Indeed, its why I can get away with an 8c/16t Xeon (E5-2690), an overclocked GTX 460 1GB, an overclocked GT 440 144SP, 3 HDDs, 1 SSD, and a bunch of fans on a Seasonic G550.

I merely suggested it as its by far faster than the posted 8400GS for approximately the same price. Yeah, I agree, its an old card, but if you need something better than integrated on a budget its a decent choice. A newer choice would be maybe a 650TI, as its easily twice as fast as the GTS 450, and can be had for about $60.

I didn't suggest he use a multi-meter for testing the voltages at the fan header simply because its not often that someone owns a multimeter or knows how to use it properly. And as for a quality power supply, Seasonic or anything derived from Seasonic for sub 1KW is usually the best choice, if a bit expensive. Past the 1KW barrier theres a few other options, as its not easy to cheap out a >1KW PSU in terms of build quality.

OpenHardwareMonitor when it works is pretty solid. CPUID's HardwareMonitor is guaranteed to work on anything but doesnt come with the snazzy graphing feature of OHM.

Heh, thats why I always have a chunk of screen real estate taken up by monitoring graphs and stuff. Its how I know that with firefox and KSP running at the same time I have ~600MB free out of 8 gigs of ram.

I was asking because you may have no choice but to get a third party cooler. We've already determined that the card itself is functional, but that something involving the fans isnt working. With any card manufacturer, generally removing the stock heatsink is enough to void the warranty. If the problem is with the fans, then the issue will be resolved with the install of a new cooler. If the problem is with the card though, it CAN be fixed with a new cooler, but the new cooler cannot be a PWM cooler as the card cannot feed power/a PWM signal out through the fan header.

http://www.zotac.com/products/graphics-cards/geforce-700-series/gtx-760/product/gtx-760/detail/geforce-gtx-760-amp-edition/sort/starttime/order/DESC/amount/10/section/gallery.html This? If so, I'm not seeing any consumer feedback regarding failed fans, so its not a common problem. Is the card still under warranty? If no, how comfortable are you with pulling the heatsink/fan assembly off and putting a replacement on?

So, saw this mod while looking for a jump drive mod, and this one is certainly what I'm looking for. Problem: I'm on .90, and the latest available copy of the mod is for 1.0.2. Lets see what happens when I mix the two together. EDIT: it works! Just did a jump test to Jool. Now to set up a Beacon network.

With very rare exceptions, every launch done these days is a fairly public affair. When the X37B went back up, ground observers found it and its orbit within weeks. A manned capsule would be even easier to spot due to the solar panels being deployed