Discussion in 'Videocards - AMD Radeon Drivers Section' started by d2kx, Apr 5, 2016.
2k is a bit much, i have a vapor-x which should be very close..
Strange. Even his PCS+ is overclocked 2k difference in graphics indicates something isnt right. Check on that.
I still wonder why I have so much better scores than you.
I tested again with the newest version of Firestrike. As I said about my previous posted result, it was with memory tweaks and 16.3.2, but the one run I did without saving without the memory tweaks and 16.4.1 was very close to it. Well, it officially is. These are my results with the Vengeance memory at stock settings 1600 and 16.4.1. They are on the margin of error on the previous ones.
The condition of my card after all these years and all of the juicing by the drivers:
Someone remember this?
Hey Undying nice logo.
LOL haha. Just don't post it in the Async Shaders or the 970/390 thread of the NVIDIA subforums.
That 13224 graphics score sounds about right for the r9 290 pcs+ at 1125/1500. Was getting the same score as well. The Mac, your score i used to get if i didn't use msi afterburner to force constant power/core clock. Try it, and put it at 1125/1500 and you should see the same number.
This is with factory oc.
Well my score is from a while ago. Factory overclock used to get me 12300 points.
Ran another test and the result is similar.
Did you do some bios tweaking?
I can't do any testing with an r9 290 anymore, i no longer own one.
For GPU no, and I don't even dare to oc it since you burned yours.
Update: so, I installed 16.3.2 and with this driver I get ~12300
It seems 16.4.1 is nice for my gpu. :banana:
Alright but would you just try it for once. I promise you won't burn your card.
Put the offset on 50mv with 50 powerlimit, and core at 1125 and memory at 1500, and run firestrike once and afterwards just put the numbers to normal. I want to see how much difference it makes then.
Nope, it's a glitch ROFL. So, I installed 16.4.1 again, and got even lower result than 16.3.2, after that I tried to figure out what was the different previously. Things to know: everytime I install a driver I turn on VSR, GPU scaling and set global tesselation to x16. Ofc firestrike told me about the tesselation change, so I turned it back to default. It seems after I set it back to def and run a new test, firestrike boosts my graphics score. :bigsmile:
Now, I'm back at 13200 again.
i dont have core fluctuation problems, it stays constant when it loads.
im at 1030/1400, that factory OC.
270X CFX @ 1150/1500
4790K @ 4700
HELLO 290X single card :grin:
i get ~20FPS without hyperthreading in physics only test
and ~40FPS with hyperthreading
10.5k on 4770k @ 4.5 and 290 @ 1000/1300
Driver seems to work well enough
It isn't core fluctuation, rather a power issue that i also started to have with the first crimson. Just try it and force even the stock clocks on msi afterburner, and you will definitely pass the 12000 points on the graphics score. Don't even touch the powerlimit, voltage offset or anything. Just put the same core and memory clocks as the factory oc and apply it. You won't lose anything. It is the same issue i had and already pointed it out on an earlier thread of a crimson driver. As for why he is going passt the 13000 on the factory clock, maybe his tessellation is somehow connected.
So, I ran two tests today, and it's still 13200. Before the second test I changed AMD optimized tesselation to Use application settings for both Global and 3DMark, but it's the same. The only thing I can think of is firestrike applied x16 AFTER I set it back to AMD optimized and not WHEN I set it to x16, and now it became permanent. It seems I cannot even restore it, unless I'm reinstalling the driver. That is some weird bug if you ask me, also this is the Steam version.
Edit: But feel free to mess around with tesselation, maybe you can trigger the same glitch as well. :nerd:
16.4.1, same settings as my above result.
i guess its just variance but i can rerun if i have the patience some time.
the interesting thing which i just got reminded of, because i didnt have 3dmark in use & installed since several months... is... how it cripples crossfire performance with its combined score. it seems to be an issue for quite some time according to some forum posts at futuremark forums.
windows 10 64, same clocks, 16.4.1
edit... yo my image hoster ... he switched some base url stuffs... now all imgs are unlinked...
so i just leave this here :nerd: 11836