• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Possible Radeon 390X / 390 and 380X Spec / Benchmark (do not hotlink images!!!!!!)

Status
Not open for further replies.
Don't think we can make any true assessment of a 4gb HBM card in 4k until we actually see the reviews.

Too many people on this forum see "4GB" and remember threads with people complaining it's not enough for 4k but that is current existing tech in the form of gddr5.

For all we know this could be totally irrelevant under HBM architechture, however there are a lot of armchair scientists and Internet warriors in this forum who would tell you adamantly that this new tech which we know sod all about really, is not up to the task.

It is common to compare new tech with old and believe they both have the same limitations, however I say wait and see, not long now to wait either.

Let's wait and see the fury x and its 4GB HBM perform in 4k before we get the pitchforks and torches.

One thing I will say, AMD are professional makers of gpu's (albeit often rubbish ones) and I don't see many forum members here who can claim that same a accreditation, I'm fairly sure they know what they are doing with HBM, if they release their rebranded 390x cards with 8GB ddr5, then push out the fury x with 4GB HBM I'm fairly sure it's because they have tested their halo product in its intended enviroment and found it capable.

While it may be hard for some people here to fathom or comprehend it may just be possible that "New" tech is significantly better than "old" tech, to a degree that makes newer tech seem like some for of magic or wizardry is happening.
 
2 x 8 Pin for Fury X ?

Hell yeah, this is getting serious now. If the HBM is low-power draw this means the GPU is gonna be able to draw some serious juice. This is great news for the Overclockers.
 
Last edited:
Nope, the noose is already in situ. If it can beat the Titan X convincingly at 1080/1440p i'll cut it some slack.

I commend to you a tweet by Dan baker of oxide games. He tweeted that this is the first single 4k card in the world. I asked about capacity vs bandwidth. He compared having more capacity to more seats on a bus if above the working set. I.e doesn't make the bus faster or performance better. Why would he say that it is the first single 4k card despite 4gb if this were not the case?

EDIT: This is the tweet: https://twitter.com/dankbaker/status/609445259597213696
 
Last edited:
Don't think we can make any true assessment of a 4gb HBM card in 4k until we actually see the reviews.

^^^
This

This thread didn't get to 8202 posts with reason and logic.

:D

I commend to you a tweet by Dan baker of oxide games. He tweeted that this is the first single 4k card in the world.
I asked about capacity vs bandwidth. He compared having more capacity to more seats on a bus if above the working set.
I.e doesn't make the bus faster or performance better.
Why would he say that it is the first single 4k card despite 4gb if this were not the case?

Only time will tell, but I have my doubts - hoping to be proved wrong though.
 
I commend to you a tweet by Dan baker of oxide games. He tweeted that this is the first single 4k card in the world. I asked about capacity vs bandwidth. He compared having more capacity to more seats on a bus if above the working set. I.e doesn't make the bus faster or performance better. Why would he say that it is the first single 4k card despite 4gb if this were not the case?

I suppose if it can move the data in and out of the HBM at a fast enough rate under DX11 then the user won't notice and there will be a limited perf. degradation. Obviously doing this using normal VRAM cripples perf once you run into the vram ceiling as its not fast enough.

Under DX12 I think this is even less of a problem as you can stream even more efficiently in and out of the HBM.

I wonder if the AMD Demo will be DX11 or something else. For this Christmas I imagine the AAA titles will have some sort of DX12 capability and so the demo may be Windows 10/DX12 based.
 
Seems AMD are invested with DX12 and Microsoft, I expect there to be a suprise at this event in the form of a VR demo on probably the occulus rift and potentially playing a VR version of BF4 or something similar.

Don't be surprised if you see something along these lines, it's just a hunch I have, mainly due to AMD sudden ditching of mantle and then praising dx12, followed by their talking up of VR and their VR sdk stuff, and that recent bf4 screenshot makes me think all this will tie together at the event
 
Best card ever made is about to be presented soon.

9jq649.jpg
 
While we dont know if 4GB HMB will be ok compared to 6gb or more of DDR5, i'd guess if AMD is planing to release a 8GB version later its still has the limitation of not being able to hold enough for some uses.

Happy to be awed and suprised thou :)
 
So far the devs who have tweeted have praised these cards as 4k capable. Haven't seen anyone say anything bad yet which is very promising.
 
I commend to you a tweet by Dan baker of oxide games. He tweeted that this is the first single 4k card in the world. I asked about capacity vs bandwidth. He compared having more capacity to more seats on a bus if above the working set. I.e doesn't make the bus faster or performance better. Why would he say that it is the first single 4k card despite 4gb if this were not the case?

Because his game isn't memory intensive and he's invested with AMD and DX12. All he's saying is that capacity isn't everything or relative to performance. He's more qualified to comment on DirectX 12's ability to cope with buffer management with features like tiled resources than anyone in here.

But 4GB is still 4GB, so the real question is if you have a large scene with high resolution textures, and you're working to render that scene and have it perform as excellently as you can - Is it not a case that you're still not able to use as high a resolution as you would if you were working on the premise of a 6 to 8gb frame buffer, and is there any benefit in performance between a GDDR5 card and one sporting HBM.

Basically it means very little till next year lol. It's very difficult to see a clear argument that less VRAM is a good thing. A very convenient one maybe.
 
It seems AMD want to make as big of an impact with this launch. What gives a launch a bigger impact? Increasing the silence beforehand.

Edit: Someone at a US tech shop (am I allowed to say BB?) ***** up
http://imgur.com/6aGGKLm

On the AMD subreddit somone bought two from the shop, and benched one on Firestrike extreme. Although no official drivers for the cards as of yet so...
 
Last edited:
mantle was more of a kick in the a** to microsoft showing them what an API can do and to show them they needed to step up their game a bit.

the big thing i will miss about mantle though its native support in Linux... I dont see microsoft openning DX to linux
 
Status
Not open for further replies.
Back
Top Bottom