AMD Phenom II Review Thread

[Miss Banana]

Miss Banana...hypothetically let's say that you had Phenom II with 8GB of RAM and 2x 4870x2, and you compared it to i7 with 8GB of RAM and 2x 4870x2 in a Far Cry 2 benchmark. Let's say the i7 whupped the Phenom's ass. I am imagining you quoting my benchmark and saying... "Well the i7 had the x58 chipset and the Phenom had the AMD 790 chipset so you can't compare! Try again!" Certain variables (like chipsets) are impossible to eliminate. Other variables (like 2 vs 4GB of RAM) sometimes just don't make a damn bit of difference in the end.

You know what, even though AMD and Intel are stuck on different chipsets I did the benchie anyway just for s and giggles. I did a Far Cry 2 benchmark of a Phenom II system with 8GB of ram and 2x 4870x2 and it totally smoked an equivalently configured i7 clock for clock...but I'm not going to post it because the stars aren't in the right alignment anymore so it's pretty much invalid.

Nah, just kidding.

One last thing about the HardOCP review. ( http://www.hardocp.com/article.html?...50aHVzaWFzdA== )
The problem that arises when you desperately WANT to test the phenom II with tri sli, is that you have to use an nvidia chipset for the AMD system.
So what chipset did kyle use?

In his test setup specs he claims he used the MSI DKA790GX Platinum, which is strange because besides only having two PCI e slots, this motherboard also does not support sli. This means he used a different chipset and screwed up by giving the wrong information in his test setup section, no real surprises there.

Later on he claims he used a MSI K9N2 Diamond, based on the NVIDIA’s 780a chipset. Note that this chipset only supports tri sli in a 16* 8* 8* configuration, while the MB for the intel system (ASUS P6T6) supports a *16 *16 *16 configuration.

If it's not obvious yet, this is where the problems begin. Surely it is not possible to test different platforms with the exact same chipset, but wouldn't it make sense to build configurations that at least have slightly similar specs? Not to kyle.

Back to the chipset used for the AMD system, kyle himself once reviewed the motherboard he is using, so you would think he is aware of the shortcomings right?

Look at this page
http://enthusiast.hardocp.com/articl...50aHVzaWFzdA== and note the winrar performance.



Something weird is going on with this nvidia tri sli chipset for AMD cpu's and kyle has the following to say:

Here the performance takes another nose dive. I'm not sure what the deal is with this particular test, but the results are less than stellar. Hopefully this type of problem will be corrected as the drivers for the platform mature. This test is solidly bandwidth limited in our small spectrum of CPUs, but this is just terrible and it was continually repeatable.

Now let's look at gaming, again the nvidia motherboard is acting weird by underperforming in every gaming benchmark and then suddenly performing better in a crysis benchmark. Kyle is very surprised:

Again, the behavior of this board in the benchmarks is a bit odd. I am not sure if these odd results are a design issue with the MSI board specifically or a factor of the drivers which will hopefully change over time.

In the end of the review kyle has the following to say:

Dan and I had much difference experiences when it comes to the MSI K9N2 Diamond. There is just no way I can suggest using this motherboard after my personal experiences with it.

Really kyle? Quite strange you decided to use this board for the phenom review then. I guess you had to since it was the only board that does tri sli on a phenom system? Or was the whole trisli thing an excuse to use this board for the amd platform? Surely not.

Obviously this is only the beginning. The nvidia board is mysteriously not working correctly, as kyle discovered, and then (surprise) does not work with 1066 mhz memory. Kyle then decides to give the AMD system half of the memory of the intel system, and uses the GPU's that have substantial bandwidth requirements.

So we have slower memory, less memory, a buggy chipset, and less PCIE lanes for the AMD system. You are right, some variables do not make a difference in the end, but I don't think that is the case here.