| || |
| Quality and Setup of the eVGA e-GeForce4 Ti4600 w/ACS² |
| Looking Good... |
As a whole, the eVGA e-GeForce4 Ti4600 is a standard reference design, very similar to other cards we've already reviewed. At the heart of the eVGA e-GeForce4 Ti4600 is a GeForce4 GPU running at 300MHz. backed by 128MB of Samsung 325MHz. RAM (650MHz. DDR). The DDR RAM uses the newer semiconductor packaging called BGA or "Ball Grid Array." This is a more advanced process of mounting the chip which allows for a smaller form factor. In turn, the newer RAM can perform more efficiently while running cooler than other "leaded" chips, maintaining better over all signal integrity. The card provides both VGA and DVI outputs as well as a S-Video output for TV-Out purposes. What truly sets the eVGA e-GeForce4 Ti4600 apart from other Ti4600 cards is the custom tailored cooling package they've applied to the card. Dubbed the Asymmetric Cooling System² (ACS²), the cooling unit is designed to be a superior solution for efficiently removing heat from the card. We've seen a number of these "unique" cooling designs from other manufacturers in the past, but this one really caught our eye, making us want to delve a bit deeper into how this unit functions.
Click To Enlarge
The ACS² is a 2 piece cooling unit that utilizes a liquid filled copper heat-pipe to channel heat away from the RAM as well as an oversized heat-sink for keeping the GPU cool. Actually, the theory is pretty standard, warm water will normally move toward cooler water as will warmer air to cooler air. So with the liquid filled heat-pipe, the coolest portion of the liquid should be at the fins located at the end of the pipe, at the fan's exhaust. Through natural attraction, the warm liquid over the RAM should be drawn to the fins, effectively removing the heat from the system. On the reverse side of the card, the system is less efficient, utilizing finned RAM sinks that rely more on ambient airflow from the CPU and chassis fans rather than the GPU fan. Thermal pads are mounted over the RAM chips to help conduct their heat to the ACS², however, we think that maybe the point of contact could have been slightly concave to allow the use of a thermal paste instead. Over all, the system looked like the ACS² had a lot of thought behind it, that is until we started to take it apart.
After releasing the two spring-loaded pins that hold the assembly in place, we were quite surprised by what we found. Although the ACS² system appeared to by a superior design on the outside, once we exposed the core, we saw that something wasn't right. It turns out that the heat-sink doesn't completely cover the GPU. Our first thought was that this can't be good. With that said, we decided to run a few tests to see how well the ACS² performed at dissipating heat.
After seeing how the ACS² didn't cover the GPU fully and taking into account that it is an extra feature that adds $50 to the cost of the card, we decided to run a few quick tests to see how efficient the ACS² actually was. To start things off, we cleaned off the ACS² and applied a fresh dab of Arctic Silver II to the GPU and then remounted the ACS² to the card. Before reapplying the back-plate of the cooling package, we adhered a temperature probe from a Thermaltake HardCano II to the back of the GPU. We then reinserted the card into our test system and applied a barrier between the back of the card and the CPU of the test system so there was no interference in the accuracy of the test from its exhaust. Once Windows was loaded, we let the system run for a few minutes to let the temperature of the card equalize, determining a baseline idle temperature. After determining the idle temperature, we applied load to the card with 3 runs of Comanche4 at 1600x1200x32 with 4X AA enabled in the video drivers. At the end of the 3rd test we averaged the peak temperature recorded during each test, establishing an average maximum temperature under load. After the test was completed, we installed a reference cooler from a Gainward Ti4600 video card and ran the same tests under the same conditions.
Instead of being surprised by how well the ACS² system cooled the GPU of the Ti4600, we were doubly surprised at how poorly it performed over the nVidia reference cooling design. The ACS² ran an idle temperature that exceeded the reference cooler under load! Perhaps if the ACS² covered the entire surface of the GPU instead of leaving a good portion of it exposed, the scores may have been better. It is unfortunate to think that many users may pay an extra $50 for something that will, in fact, make there new video card run hotter and potentially shorten the life of their very expensive video card. Of course we've only scratched the surface of this issue, however, if you are interested in a more complete examination of the ACS²'s performance, head on over to HardOCP where Steve Lynch did a terrific job at testing the ability of the ACS² versus a standard nVidia reference cooler.
OK, before we move on to actually benchmarking the eVGA e-GeForce4 Ti4600, we thought we'd throw a little eye-candy your way and show the visual capability of the card.
A Few Pics, Overclocking and 3DMark2001SE