What I like about the picture above is that the backplate isn’t solid, which mirros Sapphire’s outstanding Fury design. Maybe I’ll pick it up, I’m not sure though.
Can anyone tell me what the difference between the reference manufacturer’s fans are? The Gigabyte one looks beefier than the MSI one. Is it just because the label on the GB one makes it look bigger?
The openings are exactly the same. This makes me wonder if they made a trade off between air flow and noise or something?
On a side note: I think I need a reference card because my case thermals aren’t great (brand new Phanteks Enthoo Evolv matx TG). My r9 390 (MSI) gets to 91 degrees in benchmarks before I chicken out and shut it off. If there are any other ideas to remedy this i am open to suggestions. I would rather not have to buy a new GPU, tbh. Heh, I’ll move this to another thread if this is not the place for it.
They look exactly the same to me. Look at the length and spacing of the actual fan blades. The lighting on the MSI photo makes it harder to see that part that the blades connect to but it’s certainly there.
3 Likes
All reference cards are AMD built. The only involvement from the board partners was sticker and packaging
So I guess they do that so they dont have to deal with warranties, customer service and distribution. I see the logic in that.
Interesting video from one of my favorite YT channels.
5 Likes
When the heck are these non-reference cards coming out?!? Not a team red fanboy, but want my GPU to be future proof being virtualized with ESX.
SPOILER > 3x 8 pin 
4 Likes
Two for power, one for “LOOK HOW OVERCLOCKABLE THIS THING IS!!!”
4 Likes
True. Even with a 3 fans, aftermarket cooler. Pulling 3 x 8 pin power will set fire to your PC 
The XFX card looks neater imho…
1 Like
I think it’s pretty cool that so far the AIB versions we’ve seen are so different. More choices for different purposes methinks.
2 Likes
Indeed, the article also noted that the card might not be finished yet. So it might still change…
If that’s the case, then it needed a fourth one for delivering a furnace.
“Winter’s Never Coming”
I like how wccf doesn’t mention why the chart was taken down and just throw it in there.
more cut copy and paste journo, Vidcards seems to be the better site.
WCCF atleast does link to the original.
Hi guys,
I have gathered some questions that you might be able to help me with regarding Vegas performance in some edge cases.
How does the HBCC impact performance in workloads that use a lot of VRAM compared to small VRAM usage and CPU rendering due to no HBCC capabilitys like on NVIDIA cards.
The question came up while talking with some dude who had 3Titan Xes for Blender Rendering.
He additionally needed some crazy dual Xeon setup to render stuff that didn’t fit in VRAM.
The question basically is:
Does VEGA with HBCC expanded VRAM, perform good enough to be able to combine and rival such a setup?
I have a Vega64 and 32GB RAM myself so i will start testing this on my own in the far future once i find the time.
But it is the “far future” due to me not having any experience with blender or any other things i currently do that fall into the same realm.
I’m just very curious about that since i couldn’t find anything closely related to that online.
Additionally, i couldn’t find information on the “Crossfire, ehem mGPU” and dual Vega HBCC situation and what it even does in those cases.
Does it combine it’s VRAM over HBCC to 16GB ?
Do they become rainbow unicorns and fly you off to AMD Headquarters ?
I’d like to try that, but a second Vega64 is still far away for me : (
Thanks in advance
Sincerely RageBone
You could check out this thread as well: Blender Benchmarking (2x Rx580)
I don’t have any time at the moment to do testing, but I’m not going to get much further than you as I only have one card as well.
Wtf how did I miss this. This is more important than the hardware!! Lol.
Vega and AMD gfx in generaly is a minority vs Nvidia on games. There is software to be dialed on on drivers (amd) and games (developers) that can leverage Vega and even the FP16 performance.
But if humans dont make the code to use it. It will be silicon burning power for nothing.
Games are the shittiest code on the planet. Often rushed and pushed out early.
1 Like
you are totally right.
But that wasn’t the question
Lets assume that Vga64 is close enough in performance to rival the titan XP or XPp (what a bullshit name) under some circumstances like price to performance etc.
If Your Task is VRAM constraint and likely to breach the amount of available VRAM, you are basically fucked for days or money.
If Vegas performance doesn’t drop bad enough with HBCC expanded VRAM to fit the task, then Vega could become the next Go To GPU for such Tasks.
And it shouldn’t be that hard for vega to be better then CPU rendering.
The question was: Is that assumption right?
Sincerely RageBone