Sorry to keep you guys waiting for quite some time, but I ran into some probems today.
I am very happy with the board so far. It is a very good overclocker and in comparisson to 680i, it's lightyears ahead when it comes to booting and oc behaviour in general. I noticed when the CPU is overclocked to the edge of stability, the board hangs at bootup (only sometimes and very are) and needs one or two resets. Memory overclocks very good too, a little better than what I was able to run on P35.
When I wanted to test 3D, I ran into some problems though. I am already in contact with EVGA, so the problem is known and being worked on. Seems like I chose the only video card, that is troublesome with the board: a overclocked 9600 GT.
If you are not aware of what Nvidia implemented into the 9600 GT, I recommend this article by W1zzard: http://www.techpowerup.com/reviews/N...Shady_9600_GT/. In short: The GPU of the 9600 GT can be overclocked in two ways: using a software tool like Rivatuner or by increasing the PCI-E frequency. "Now where is the problem?", you might think. The problem is: you cannot adjust the PCI-e frequency in the BIOS on a 750i chipset yet. Nor is there any tool out that allows you to make changes to the PCI-e frequency. Note, that this is my assumption and I might be totally off: I think the 750i chipsets adjusts the PCI-e frequency depending on the load of the video card.
What does that mean in my case? That means the 9600 GT is being overclocked dynamically depending on the load because the PCI-e frequency most likely changes on the fly. That results in a black screen i.e. a crashing video card once it enters 3D mode, because I am using a factory overclocked version from Zotac (amp! edition), which is clocked to 725MHz on the GPU by default at a PCI-e freqiency of 100 MHz. Now if the PCI-e frequency is around 120 MHz, the real GPU speed could be closer to 800 MHz which makes the card crash once it enters 3D mode. Again: the 9600 GT is the only video card I know, that is being overclocked by increasing the PCI-e frequency, so any other video card will work just fine. A test with my old 7300 GT proved that. No issues whatsoever.
Luckily, doing hardware reviews is my job and I tested the exact same video cards on a Asus Striker 2 780i and a EVGA 780i reference board a while ago using a QX9650 @ 400 * 9, 3.6GHz and DDR2-800 4-4-4-8. I used the 174.12, but the performance is the same with 174.74. I tested booth drivers with several video cards and results were within a small margin. With that config, I got 24,130 3DMarks in 05 with cards running 745/1798/1120 MHz in SLI. Using a E8500 at the same speed on the 750i, I got 24,249 3DMarks with cards running at 630/1503/1120 MHz with 2 cores less and half the L2-cache. 3DMark05 does not make use of more than 2 cores, but might benefit from the bigger cache, so the E8500 has a slight disadvantage in this test. Ok, now how can I compare those 2 results without knowing what the video cards real clockspeed was? I tested the cards for max clocks without artifacts. I didn't see any on the 750i, so I assume the real clockspeed of the cards was around or less than what it was on the 780i. That means 750i is as fast as 780i.
I know this expalnation and assumption is very adventurous to say the least, but as I won't have other video cards until monday, that's all I can do for now.
Reply With Quote
Bookmarks