Kingwin Revolution RVT-12025D Heatsink: A quick review

Aah, Kingwin..... or more like AAH!!! KINGWIN!!!! Or so I thought, until just today.

You see, I bought this cooler on Newegg about 8 months ago for $18, when they still carried it. Pretty cheap for a heatsink. It got mixed reviews. Some say that the heat pipes aren't wicked, so you have to attach it with the pipes facing upward. Some say that it's just a plain out piece of junk. Others say that it's great. For the past many months since I bought it, I was one of the people who thought it was a piece of junk. Today, I slightly changed my mind

1. Physical Description

The RVT-12025D isn't a small heatsink. It's definitely necessary to pull the power supply out of your case when you install it. It's got 4 heatpipes that run directly up to its cooling fins. This doesn't translate into the usual six heatpipe-paths of a standard tower heatsink with 3 heatpipes. This heatsink contains 4 pipes, with exactly 4 paths for the heat to go (remember this for later). Installing the heatsink is a bit of a pain, since it has the stupid LGA775 pushpins, but you'll eventually get it in there stable.

2. Cooling Performance

I don't have any exact numbers, but I can say this:

With my Pentium 4 520 @ ~3.7GHz, 1.55v, this thing was awful. A bit better than the Intel heatsink at idle, but about the same at load. Not worth $18.

BUT, with my Celeron D 356 ES @ 4.5GHz, 1.5v, it's quite nice. It kept the CPU in the mid 60s Celsius under full load. I only just tested it with this CPU today, having used a different heatsink (that doesn't work as well) for many months, since I thought that the 12025D was awful.

Here's the deal: there are two differences between these two CPUs that would make a major cooling difference:

1. The Pentium 4 520 was running .5v higher than the Celeron.

2. The Celeron has a D0 stepping, so at it's stock speed, it only requires 65w. This is compared to the 84w power consumption of the Pentium 4 520.

Why does this make such a huge difference? Because of the heatpipe transfer rate. As I said before: a standard tower heatsink supplies 6 paths for heat to travel through the heatpipes. This heatsink only supplies 4 paths. So, although the cooling fins can dissipate heat perfectly fine, the heat has trouble getting to those pipes. I calculated out the overclock wattage for both my CPUs, and it looks like the Pentium 4 consumes 135w and the Celeron consumes 115w. Just a 20w difference pushed the heatsink over the edge.

3. Conclusion

So, in conclusion, this heatsink will work great with most CPUs. Stay away from CPUs that have a TDP of more than 120w, but for most CPUs, even with a little overclocking, this is a great heatsink. I highly recommend it. (with caution)  :-)

Overclocking the Geforce 8400 GS: Decent Graphics on a VERY Low Budget

Welcome to Spynoodle Tech Stuff! On this blog I will pretty much cover my latest ventures and discoveries in the world of computing. In short: I'll write about my experiments (usually on the "not new tech" side) and the latest tech news that I think is interesting. This is my first post, so I guess I'll start out with my first real experiment: overclocking the Geforce 8400 GS.

Part 1: The Card

When it comes to PCI-e video cards, it pretty much doesn't get any cheaper than this. The Geforce 8400 I bought was from a seller on ebay who sold these cards in bulk by the hundreds for $15 each. He called the card a "Geforce 8400 HP" and I believe he said it was manufactured by HP, although I didn't find any hint of it being manufactured by them at all. In fact, I didn't even know they manufactured cards until I looked it up and it turned out they did manufacture a PCI-e x1 version of the Geforce 8400, although this was PCI-e x16. The card had 128mb of ddr2 memory running at a base clock of 500MHz. The fact that it only had 128mb of memory surprised me, since it's the newer, faster G98 version of the card vs. the old G86 version. Some programs actually thought that it had 512mb of memory, although I seriously doubt they were right. The GPU by default runs at 540MHz, which is also a bit slow for this model card. Most run at 567MHz. The GPU/Shader ratio remained the same when overclocking.

Part 2: The Testbed

Like the card, my testbed is also El cheapo. In fact, the testbed was actually given to me for free. It was a Dell Optiplex GX280. For my most recent tests I use a custom computer that I built with some of the parts from the GX280, but I ran this experiment earlier this year. Here are the specs:
-Pentium 4 520, LGA 775, E0, 2.8GHz, 1m cache 
-Crap Dell Intel 915 motherboard. No support for anything but Pentium 4s and early Netburst Celerons. >:0
-512mb (2x 256mb) Micron ddr2 @ 400MHz + 512mb (2x 256mb) Samsung ddr2 that would run at 533mhz but instead ran at 400MHz. One of the 256mb sticks was standard in the Dell. The other three were salvaged from other systems. The sticks ran in dual-channel.
-Geforce 8400 GS + Radeon 7000 for a dual-monitor setup.
-Crap Dell 250w PSU
-DVD-RW drive (missing cover)
-200gb Maxtor Maxline something-or-other. Never wrote down the specs. XP Pro was installed on this drive.
-40gb Western Digital SATA 1.5gb/s WD400 with Crunchbang Linux installed on it.
-Samsung Syncmaster 172T 17" DVI LCD monitor as main monitor.
-Old Viewsonic CRT monitor (15"?) as secondary monitor.


Software Used:
-Atitool 0.26, to overclock the card. I've actually stopped using it on my new build in favor of MSI Afterburner.
-Rivatuner, to adjust fan speed. I didn't realize when running this experiment that the card wouldn't at all get hot if I didn't raise the voltage (which I didn't).
-3dmark 03, to benchmark the card.

Part 3: The Experimentation
  
Then came the fun part: seeing how high the card could go. Sadly, when running these tests on that crappy GX280 I couldn't raise the PCI-e voltage. Now that I have my new setup I might update the OC with volt-upped results. Still, here are the results of the OC:

^ My max speed I achieved was 591MHz on the GPU (meh) and 601MHz on the memory (okay). Overall, it gave me a pretty nice 3dmark score-up (from 5729), probably mostly from the decent increase I got on the memory. At this speed I did have a few artifacts in 3dmark, so I decided that I would keep it at 585MHz on the GPU and 590MHz on the memory. At those speeds I got a 6333 in 3dmark.


Part 4: The Conclusion

I guess I would evaluate this OC as a good ups for the 8400, but overall still a pretty low score. Although 6401 was a nice score for the card itself, it's still probably not comparable to something like the Geforce 8500 GT. Either way, if you're looking for a card that's exceptionally cheap but still gives decent performance, I highly recommend it. I'm not a gamer, but still DEFINITElY want better graphics than the GMA 900 I was dealing with. (Ahhh!!!)