Originally Posted by neojack
do you know where i could find a review of the XSPC RX480 blade waterblock ?
I plan to build a watercooled multi GPU mining/gaming rig with 4 GPU's, and onother rig with 6 GPU's
XSPC waterblock costs 100$CAD
Ek's costs 150$
I would like to find a review especially about the VRM temps and pressure drop.
Also it seems that those blocks can only be used in serial since there is only two ports. 6 of them in serial would be too much pressure drop maybe ?
I can't elaborate on pressure drop, but the VRM cooling is definitely not as good as the EK.
With all stock settings on my reference 8GB I am seeing 67C after 20 minutes of heaven. Running 1400/2250 with +50% power and max voltage (stock BIOS) peaked at 78C. This is with a single XSPC EX240 radiator with (2) absolutely silent fans pulling through the radiator that doesn't have a good source of fresh airflow. Temperatures could likely be better with a more optimal cooling setup...but this is in my HTPC so size/noise were priority as temps are well within operating range.
The question is though, what kind of settings were you planning on running all the cards at?
If you're looking to flash BIOSes and max all the cards out, then the EK blocks are going to be what you want.
If you plan on running stock voltage/power restrictions, or even with an undervolt, then the temperatures for VRMs should be acceptable with the XSPC block. If you were to add backplates and use some thermal pads to make them passive heatsinks, the temps would definitely get better. I did this with a R9 290 (which I recently sold), and utilizing the backplate as a heatsink for VRM1 brought it's temperature down to just under the core and on par with VRM2...which is generally unheard of on Hawaii cards.
I upgraded thermal pads for the VRMs, to Fujipoly Ultra Extreme, this would likely be a mandatory upgrade to ensure they run as cool as possible since you'll be loading the cards for extremely long periods of time.
Originally Posted by Synister
Hi folks, I've got an MSI RX 480 Gaming X. While running the valley benchmark, my GPU at times dips from 100% usage. Why is this? You can see below, a short run of Valley to show an example.
What settings are you running the card at? Is there any overclock applied? Did you increase power limit or voltage?
Please give as much information as possible so we can help you determine what is going on.
Here's what my card looked like after running Valley looped for 15 minutes with my waterblock installed. I believe there to be dips when scenes change, but I could be wrong.
I also found another 15 min Valley loop screenshot and this one has much fluctuation in GPU usage. I can't remember what the settings were exactly, but GPU usage in this scenario is much rockier than the one above. The biggest difference is that this was when i was using the Kraken bracket and an AIO to cool the core, with a homebrew VRM heatsink.
In both scenarios I'm fairly certain the card was set to 1400/2250 with max voltage for core and 50% power target, but I am not 100% certain of this. I may have been experimenting with RAM voltage for this run. I would have to run the bench again to be certain.
Originally Posted by looncraz
I ran some tests - it's looking to me like you're experiencing a CPU bottleneck. I get some dips even at 4.5GHz with an i7-2600k in Valley (Heaven is even worse), so don't expect to be nailed at 100% in any scenario...
What matters is how well the games play, and you have enough performance that most games should play very well.
I don't think that the CPU is the problem. How have people run bigger GPUs than the 480 with FX CPUs then? I feel there is something else at play here, unless Valley is extremely CPU dependent for good performance.
There can be scenarios where an Intel CPU would perform better from the gaming stand point, but that would most likely be a game that is more reliant on single core performance where Intel obviously mops the floor with an FX CPU.
I was looking, and am currently unable to locate, a great CPU comparison article for gaming purposes. In most 'modern' (last few years) gaming titles, even an i3 or FX-4xxx CPU can give extremely good performance due to how games are coded to leverage multiple CPU cores/threads. The bottom line from said comparison was that a majority of the performance in gaming can be had with 4 cores/threads; and after that it is diminishing returns. I'm not saying there isn't extra performance to be had from bigger/better CPUs, but you would be surprised how well a HyperThreaded i3 or 4 'core' FX chip will play games.
Edited by Roboyto - 9/16/16 at 8:48am