As mods neither give a shit about ecelebs nor brand shilling anymore, let's start another one on the AMD Radeon 7 or Vega something.
Buildzoid made some pretty good points in his video, on what the card is and why it's priced like it is.
youtube.com
Gotta say I agree to a whole lot of that, which leads to the conclusion that AMD has basically given up for an high end gaming GPU, as why should they bother what can't be won by now.
Memes aside. it'll mean NVIDIA hold their monopoly even stronger and prices(MSRP not some mining craze fallout) will likely even more increase with time and even though this board was ever so obsessed with technology that it hated all kinds of consoles, that's exactly what we're going back for.
The huge majority of people won't be paying for a GPU that's double the cost of the rest of the system and we'll go back to some controller experience. But hey instead of criticizing the companies we got some spicy poorfag memes, so good job Jow Forums.
As mods neither give a shit about ecelebs nor brand shilling anymore...
Other urls found in this thread:
youtube.com
wccftech.com
amd.com
nasdaq.com
nasdaq.com
nasdaq.com
nasdaq.com
techgage.com
youtube.com
twitter.com
Average AMD user
i don't care about this loser's uninformed opinion about a card that isn't even in public circulation yet
It doesn't need to be, they already showed off FPS in verious games and benchmark in various benchmarks. If anything it only could only get worse in performance in what's already been showed on stage.
Navi isn't ready yet so AMD is trying to bring in money from people who need the compute power. If Navi can't compete with Nvidia at the 2070 level then PC gaming will die a very slow death because Nvidia will continue raising prices. We're going to see a $2500 3080 next year if Navi doesn't deliver.
Does it though? A 2060 beating would be more than enough for a global scale, remember neither are most gamers as affluent as americans, nor are high ends cards necessary at all.
i like radeon 7
at first I was kind of dissapointed they didn't release any new gpus or cpus that are cheap, but readeon vega 7 is actually comparable to the performance of 2080 ti or titan v for $700.
Same performance for 500-2300 dollars less sounds like a good deal to me(even though it's too expensive for me)
I like Noctua fans and my build is Intel/Nvidia.
My monitor is 1680x1050 though, I love 16:10 ratios.
It's supposedly comparable to a 2080, almost. What it will be comparable to after the reviews come in... We'll see.
Vega VII are just failed MI50s, the fuck are people expecting?
Also, is it true they're leaving SR-IOV on? I'd buy one day 1.
Wait a minute. It's going to be pcie 4?
yes
What does that do?
Same, but that's still to be revealed.
ye
Will it have hdmi 2.1?
WRONG I only have a single noctura fan D15 and 2560x1080p monitor
Try again seething nvidia intel
People wouldn't be as mad if they announced the cut down card that's actually for gamers at like $399 for 90-95% of a 2080's performnace
Yet he's clearly holding a Titan.
The display should really be a CRT.
>trash tier but expensive MI50s
>lmao let's sell it for $400
Actually, didn't the die also get shrunk down when moving to 7nm? Vega 64 was pretty big, but I think this one is much smaller. So going for $500 might have been possible to undercut the living shit out of novideo.
What's the matter, can't afford dell's latest 16:10 monitor you poorfag?
Because I can't.
Who cares retards are gonna buy it I refuse to spend $500usd or more on a midrange gaming gpu anyway it's just gonna be outdated in a year or two and age like milk regardless.
Maybe Vega 7 will be a steal used who knows
>I refuse to spend $500usd or more on a midrange gaming gpu anyway
user, how can you call these midrange? Have you seen the fat Nvidia dies? They're huge.
I assume the cut down card would be like 75% of the die with 8GBs of HBM2
$400 is probably too low, but $500 is likely doable, could step down the VRM a bit on the cheaper cards too.
Seems like a leftover Radeon Instinct that couldn't make the cut for businesses. I am interested in seenig how low it can be undervolted and how it can be overclocked with said undervoltage.
>AMD has basically given up for an high end gaming GPU
Then why does it cost the same as a 2080? Why did AMD not make a card that performs like a 1080ti for $100 less?
>I love 16:10 ratios.
I think 16:10 + 4:3 is just right.
Would this card be good for VR?
>stock fan configuration
>THREE HUGE FANS
im sure im not putting that fire hazard on any system
I'm pretty tempted to find out, but not $700 tempted. Maybe I could write it off as business expense or something for tax purposes.
they want to sell to datacenters and for that type of use. Watch the video. It's a tacked on product just to have something. No dev time was spent on it.
proprietary garbage
it wouldn't even be as fast as a Vega 64 then, especially as you're back to half the bandwidth
again, thanks to 2x4GB HBM2 stacks. those 16GB aren't for fun, they're for 1TB/s of memory bandwidth.
Yeah, too expensive.
Next time I upgrade I'll get a 2560×1600 monitor.
It would be maybe 10% slower than Vega VII at stock, and could likely make up for much of it with an overclock
just like Vega 56 can catch up or surpass Vega 64 with overclocking.
The memory bandwidth really isn't too big of an issue for most things since it becomes overkill for gaming anyways, and you can just use 4 stacks still.
Look at Vega FE Vs Vega 64 dies.
As far as I know there are no HBM2 2GB stacks for 4x2gb configs at 1tb/s.
>le overclock makes i as fast
whatever man, that is just retarded
The extra cores on Vega 64 don't matter to most games, memory bandwidth is important to a point, but just some clock speed instead on 56 catches it up or passes 64
of course power gets exponential and all of that.
Nvidia is charging 400usd for low end 2060s so 500ish is the new midrange now
>but overclocking
yes, just run the product way out of spec with massive unknown failure rates
>Then why does it cost the same as a 2080?
Like the Titan X, it's a gimped work card. Work cards were always expensive.
>Why did AMD not make a card that performs like a 1080ti for $100 less?
Because they can't, nor bother anymore. That's the whole point of the video.
wccftech.com
it cost $750 to make this
Article only says he suggested a so named Radeon 7, which could have been engineered to something different we're seeing today, or it's just components like HBM2 getting cheaper.
Anyway AMD is desperate to show something off out of their GPU segment on 7nm yes, but it should be a zero game at least. Throwing a product with loss on the market would be suicidal in fron of investors, as Radeon 7 is definitely not made to gain share.
>as Radeon 7 is definitely not made to gain share.
unless miraculously AIB cards will be cheaper like with nvidia. for $650 it's worth considering.
he doesn't understand basic things about how GCN works, 1tb/s memory is what makes it possible to compete with 2080
he also doesn't understand that Fp64 is integral part of the architecture, AMD can't afford to design a card without Fp64 capabilities at that performance level
>has basicly given up
we know navi is coming and we know its gonna be at 2H of 2019
the reason 590 and radeon 7 exists is just to fill the gap
few gamers will buy this card but many workstation class prof will buy it also considering that kind of fp 64 perf exists only at 4000+ bucks cards..
Please don't associate based Noctua with freetards and AMPOO garbage, thanks.
>Vega VII are just failed MI50s
Buildzoid seems to have missed that the R7 is a re-branded MI50. Looks like they cut the infinitifabric connectors to tie a large number of these together and that seems to be it.
>like a leftover Radeon Instinct
amd.com
That's exactly what it is. AMDs already making this card anyway. Wrapping it in a slightly different piece of plastic and selling it to a broader market makes total sense.
They could make a cut-down version and I wouldn't be all that shocked if they do that down the line. The Radeon 7 name alone is why I see it as possible that there could be a Radeon 5 and a Radeon 3. But that's just guesswork. An argument against it is that the R&B required to make a scaled-down version isn't cheap and that's probably why they just rebranded the MI50 in the first place.
>is that the R&B required to make a scaled-down version isn't cheap
you need additional fab capacity, you need TIME=>money
they simply can't do that, that's what I said, that's why nvidia couldn't make turing 7nm
fabs aren't infinite.
>awww he thinks the competition is real when people will wake up
nasdaq.com
nasdaq.com
nasdaq.com
nasdaq.com
Normies buy from whichever company has the shiniest halo product
>Gotta say I agree to a whole lot of that, which leads to the conclusion that AMD has basically given up for an high end gaming GPU, as why should they bother what can't be won by now.
They're not giving up on anything, their mgmt is a bit too stubborn for that.
That Vega20 rebage along with 590 are just some product decisions they did before pulling Navi in.
2080ti is still a good 20-30% ahead. It's only as fast as a 1080ti
>then PC gaming will die a very slow death because Nvidia will continue raising prices.
This isn't bad.
>We're going to see a $2500 3080 next year if Navi doesn't deliver.
That won't happen either.
Nvidia is walking a thin line between prices and volumes.
techgage.com
AYYMD IS FINISHED & BANKRUPT
AYYMDPOORFAGS CONFIRMED ON SUICIDE WATCH
No one ever expected a $700 card to have 1:2 DPFP.
They'll probably make a Radeon Pro SKU that will have it.
i hope Navi focus on FP32 like Maxwell
yes, for efficiency
>buying high end gpus
Honestly these people are dumb and deserve to pay the goyvidia tax
I'm not fanboy enough to buy something that loses specs every day.
It focuses on FF hardware.
fixed function???
Yes, that thing they always had problems with.
Their FF setup is as old as R600, they just doubled it twice.
what's all that FP64 good for? As in specific applications
Scientific computing and that's about it.
techgage.com
To bad it won't matter if you buy one. And here I thought this card might be interesting in some way. RIP
I see a lot of "creative" work mentioned. What creative application actually need so much Ram on the GPU side?
Say it with me:
SR-IOV enabled.
Source: my dad is David Wang.
Don't toy with my emotions user
no one's going to answer that.
because there's none.
There's plenty of scientific/compute workloads that would benefit.
"Creative" though, not sure on that one.
PS4 emulation.
All it has going for it is the HBM2.
>Can't be used for FP64 shit
>Gets mauled by a 2080 for FP16 with tensor meme cores
>Probably mediocre FP32
see It's completely trustworthy. Surely, no one would lie in these places.
Skip Vega 2 guys I saw the benchmark from amd VS a 64 is barely any faster maybe 5-15% on average and no more then 20-30%
Still front end bottlenecked and the hbm 2 1ghz 1gbit does bugger all for gaming either 16gb is not only overkill but as usual the shit slow gcn arch can't make use of it.
390x all over again fury x all over again
At least the 56 Is a decent underclocker but apart from that amds high end gpus are trash.
Youve been warned and yes I owned a 56 it's underwhelming and gets shit on by a 1080 oc
Maybe I missed what he said but, why can't AMD make an 8GB version?
nah, they just wanted the bragging rights to the world's first 7nm gaming gpu, even though it's more suited for content creation. They even name it 7.
HBM bandwidth is determined by the amount of stacks. Vega needs the bandwidth. This is a salvaged faulty MI60 Vega Instinct GPU clocked up because consumers don't care about power as much and a couple of broken CUs disabled. This card only exist because Nvidia is pricing there cards so high AMD can sell these and still be competitive. This is NOT a replacement for Navi. Only a stopgap.
Does any commercial applications uses fp64 or is it reserved to speciallized scientific phython scripts?
why the FUCK does AMD insist on going with fucking HBM2 modules? For fuck's sake just go with GDDR5X or even GDDR6. The modules are way cheaper. This new card is $700 but $300 of that comes from the fucking HBM. AMD would be a shit load more competitive in the GPU market if they stopped this shit.
Rebranded workstation gpu, retard
Don't be so dumb while posting my hobo wife, you piece of shit.
It is a salvaged MI60 Vega instinct die. This GPU like the original Vega was not designed for gamers. They are simply using this as a way to salvage dies that didn't make the cut to be server chips and to provide a stopgap until Navi.
The GPU literal does not have a GDDR6 controller and the card is already pulling 300 watts with HBM2.
>This is a salvaged faulty MI60
More like bottom of the barrel MI50.
rude
fuck your hobo wife. She more than likely has aids from all the homeless cock she sucks
I knew it was a salvaged work station card, I however didn't know there was no controller for GDDR based memory. And you're right. I didn't even stop to think that using memory other than HBM would result in huge power requirements.
Man fuck. I just want AMD to stop half assing shit and make a decent high end card without so many caveats.
>drumpf is a RACYSS
amarite?
user just stop being a dumb piece of shit.
youtube.com
It's the future.
>I just want AMD to stop half assing shit and make a decent high end card without so many caveats.
They need a new µarch and they know it. A new architecture from scratch can not be shat out overnight. Navi is there last GCN based card and even that will have significant architectual changes.
4k video editing
Watch Linus's video.
I think he's actually telling NVidia their pricing is insane.
My problem with the card definitely isn't the HBM2 memory. Maybe the amount, but not the fact that it's HBM2.
It's sad, but they couldn't the bandwidth without the amount.
8GB VII would be basically a Vega64, with somewhat better clocks.
699$
LOOOOOOOOOOOOOOOOL
8gb (4x 2gb stacks) would not be significantly cheaper than 16gb (4x 4gb stacks) as a significant portion of the cost of HBM is the packaging.
Vega NEEDS the 4 stacks or it will be bandwidth starved.
I retract the part about maybe taking issue with the amount then. Still like the HBM2 part.
I don't know, it looks pricey as fuck for not much performance.
It is almost double the performance you can get out of GDDR6. Vega has major issues with its front end and lacks the same level of memory compression Nvidia has. HBM itself is brilliant. There is a reason Nvidia uses it on there top of the line server SKUs.
Nvidia using HBM on their gaming cards fucking when?
When they need to.
Why is it bad?
>PCIe 4
>Have to buy a new MB so you can use your subpar GPU
AMDrones will defend this