>if only they made Radeon 7 without HBM2 it would be so much cheaper
How can you know nothing of technology despite being on a technology board? It's literally a nerfed enterprise compute GPU. They didn't sit down and plan to do this Vega GPU for the consumers. It's a compute card first, just repurposed so the consumer can pick it up. Them making a whole new Vega with different memory controller and the same specs costs a fuckload of money, hell getting that memory bandwidth is close to impossible with GDDR6. Recent history has showed that even when AMD is competitive, the consumer still flocks to Nvidia. It is so bad that the fucking 1080 outsold the midrange Polaris, you think they're going to risk making a big GPU so everything gets outsold? Lurk before posting you pieces of shit.
Vega/Radeon VII/7
damn the resistors on the right look so kawaii OwO
>Put high-tech expensive memory on your GPU to BTFO traditional GDDR
>It can barely compete with a 2 y-o GPU
So this is the power of HBM
NPC's can't compute the information you gave them, OP. how sad
>Being /v/ tard
Go hang yourself
Lurk before posting you piece of shit.
>Ayydvanced Mirco Devices releases inferior products 95% of the time for the last 20 years
>"How dare people flock to Nvidia?!?"
>I started to follow the GPU space after Maxwell
Lurk more you piece of shit.
Hurr durr, AyyMD released cards better than Nvidias. It used twice the power and lasted 2 months before Nvidia countered i
Just admit the Radeon VII is an underwhelming piece of garbage, a disappointment. 7nm hot garbage that barely keep up with a 2 y-o 14nm card.
Freesync still doesn't work after all these years to the point Nvidia has to interfere again and put actual certifications on monitors because AyyMD is slapping the Freesync sticker on anything and everything.
AMD will be crushed once again nd you shills will go back to "just wait" and praying for your "magic drivers" while you keep "supporting the underdog because it's good for competition"
People are also suggesting 8GB vram for the VII
while there are no 2GB stacks, so it would be impossible to get 1TB/s of bandwidth which is the only real difference from the Vega10 chip.
It's a repurposed enterprise compute part, it's a miracle it even does remotely well in gaymes.
Be glad AMD even bothered to shit out anything that tries to compete in the top end.
The fanboyism shit isn't what this thread is supposed to be about, so please kindly fuck off. All the whining isn't going to stop nvidia from fucking gaymers (you) in the ass.
> Be glad
Are they losing money though. I heard VII costs $750 to make.
When has anyone said it was a good card ITT? It is underwhelming because it's just not only Vega, but bottom of the barrel trash tier MI50s. Again with the TDP, holy shit do you guys love that. Unless heat really fucks you up, it's a non issue.
I will admit AMD is dumb on the freesync part. At this point Nvidia will make freesync their own by putting g-sync certification on freesync monitors. They should have made a standard so people who don't research don't get fucked by different freesync ranges.
Does Nvidia even have a product stack this year apart from the rumored GTX series? I've heard nothing of them after RTX bullshit. I mean, they just released the 2060, having most of the RTX line filled.
Nice generalization by the way. There are plenty of reasons to buy from both AMD and Nvidia, I am one who buys the former.
>I heard VII costs $750 to make.
Even if that was the case, these cards are MI50s that didn't make the cut. Better to get some $$$ back from a failed card than to get nothing at all.
Who knows? Unless the yield is so incredibly bad I don't believe they're selling at a loss, even though 16GB of HBM2 will be expensive.
Remember, these are the trashed MI50 chips anyway, they had no purpose as they could not be made into MI50s, they only had the garbage bin waiting.
Wow literally lyrical bullets
Fuck off zoomer
They HAVE to use HBM.
GCN is horribly bottlenecked by Vram bandwidth.
HBM is the only way to remove that issue.
On top of that, using Gddr6 would add an extra 50W to the TDP, they also cannot do that.
AMD has engineered themselves into a corner.
If they gave it 12GB instead of 16GB of memory and then priced it at 599 USD they would've been in a great position.
they would need to cut down the chip even further, including cutting down the bus to 3072-bit and thus reducing bandwidth (what sets it apart from Vega10).
This is a low volume, enterprise reject project.
it would be an unnecessary cutting down of the chip's bus, essentially only cutting the cost of 1 HBM stack.
Not possible.
Reducing bandwidth Murders GCN.
It has 16gb of HBM for a reason.
>user can't even read
>I started to follow the GPU space after Maxwell
You're confusing based Nvidia with lousy AyyMD
It has an insane amount of compute performance for the price
>It is so bad that the fucking 1080 outsold the midrange Polaris
>ever buying poolaris
>amdrone's arguments
That's not the point you blabbering mouth breather. When the $700 card outsells the $200 card, you know there's a problem.
It would be retarded from a business standpoint to release a Vega 7 card without HBM, it would be too costly.
That 2060 looks very sexy.