Secret AMD Vega tech day is happening right now and AMDGPU-PRO

>>57901004
Poodrones forgot that vega is expected 2019

>>57901074
how much do you get paid to invent shit

>>57901004
AMD handpicked who they told about this event, they were super secretive about it. Vega might be pretty big.

File: trashman4.jpg (167KB, 650x429px) Image search: [Google]

167KB, 650x429px

>AYYYYYYMD
>moronix

Nobody cares about your pooware Rajeed

>>57901141
No, handpicking means that it's another disappoinment and they don't want people to figure it out right away.
They did the same shit with their APUs- they get 'approved' journalists to report on nothing but the positives, so that's people's first impressions, and then wait for everyone else to trickle out the negatives too.

not really secret since every blogger and every big jurnalist of the big sites already said they are going to canada to a secret trip...

i highly doubt they all go to see the nagiara falls..

>>57901174
>Look at me make shit up: The Post
Its amazing how dedicated you autistic shitposting children are.

>calling everyone who doesn't buy into your AYYYYMD bullshit an autist

That'll show them hehe

>>57901004
Why it's not overhyped like hbm or polaris?

>>57901562
because its 2-3 months till its been launched...

>showing off shiny new tech
>in absolute secrecy

Failure confirmed.

>>57901562
This "Secret AMD Vega tech day" literally exists to generate hype and "leaks".

File: truth_chart1.png (47KB, 700x700px)

47KB, 700x700px

>>57901562
They're trying to crack down on clickbait rumormill media perpetuating false expectations which end up hurting them.
>echochamber sites take certain lines out of context and push a salacious narrative
>clickbait titles say X is going to be the next best thing ever
>fanboys believe it
>people talk about the rumors as if they're fact
>months go by with people only talking about the rumors, completely forgetting where they came from
>product gets released
>its not what the clickbait said
>fanboys on the internet whine about product not living up to expectations
>shitposting autists like the ones ITT do nothing but spam memes
>rinse
>repeat

Its a terrible cycle. Companies like intel and Apple closely control who has access to their insider information, who reports on it. AMD is trying to do the same which they have every right to. Anything that takes money away from talentless cunts from currytech and fudzilla is a good thing.

>>57901128
3.5 rupees

>>57901644
fudzilla videocardz wccft are examples of what a jurnalist shouldnt do

>So it makes perfect sense that both events are connected. Did you notice that CES takes place in Las VEGAs? ;)
This is conspiracy theory shit

>amds newest driver update added freesync support for windowed mode
the last benefit of gsync.. gone like that.

>>57901644
>>57901731
Videocardz isn't that bad, also sometime they blow the NDA

File: 1450123517338.png (25KB, 609x749px)

25KB, 609x749px

>>57901004
I hate Radeons for only one (1) reason: no Shadowplay.

>>57901944
are you kidding videocardz is worse than wccft..
they posted so much bullshit
so much clickbait shit
so much rumors from 480 being a 4096 card to 480 to the lowest one from 1060 to be a monster from 1060 to be another 960(well this one was true)
they literally posted SO MUCH shit that they eventually were spot on

>>57901973
they have relive like literally launched today....

>>57901973
:^)

>>57902013
I know but it's too late.

>>57901973
>*rips open a bag of gamerfuel*
>opens up Overwatch powered by Geforce experience on 1080p screen with 1080 GPU

>>57902054
too late...
an argument that is so stupid that there is no sane comeback to it

>>57901004
O FUCK

BRB RETURNING MY 1070

>>57901220
I'm just going by what AMD has done in the past, dipshit. Not my fault you don't have a memory that lasts more than 2 weeks.

>>57902082
>nvidia take almost ALL the twitch marker because amd can't fix gvr
>not an argument

But I bought my 970 and not the 290/390 because of that. They lose customers.

>>57901973
Facebook integration bontnet kys

>>57902162

>>57902160

>>57902121
AMD didn't heavily vet and invite selected handfuls of journalists to the press events for Llano, Trinity, Kaveri, or Carrizo. They had big public events for all of them, including ISSCC keynotes where they detailed the chips for literally anyone who showed up.

Autistic little kid just making shit up.

>>57902160
lol
https://www.youtube.com/results?search_query=relive+amd
relive is faster and with less impact generally...because you know they actually have a dedicated chip for video encoding and decoding instead to just throw it on the gpu..

>>57902254
But like I said, it's too late.

>fall for the ReLive meme
>this is what it does to audio

puuDOTsh/sIGRJ/0b535fc511DOTmp4

I still don't get it. Is Polaris just a stopgap, which Vega will replace entirely? Or is Vega just the higher end cousin of Polaris, and both "architectures" will continue to be sold in the same series?

And if so, why on earth have AMD made two architectures in such close proximity? Why not Polaris for the whole lineup?

>>57902419
Cause Poolaris is shit and not efficient

>>57902419
Polaris has 4(6 if you add the special cards that we wont see outside of china)that is from low to mid
vega will have 4 that will be from mid to high and if the rumors are correct we might see the first dual gpu from amd that will be the test bed for navi (2 cores inside the interposer essentially killing the need of a plx chip and thus any shit scaling from devs)

>>57901973
ReLive launched literally today with a better featureset than Shadowplay for the exact same performance hit.

>>57902419
Polaris was more power efficient low and mid range cards.
Vega is high end.

These aren't separate architectures, they're both 4th generation GCN. The names Polaris and Vega are just code names for the dies they designed.

>>57902461
what a insightful comment pajeet

little vega is approx 1070 perf with 12gb of gddr5x and a ~500mm2 die, launching in the first week of q2 '17.

big vega is between 1080 and titan xp perf on a 600mm2 die with 16gb of hbm2 and launches 3-4 months after little vega.

screencap this kids, AMD is done for

>>57902295

Pick your poison - data mining drivers or features turning up really late (but aren't datamining you).

File: 1424715856582.jpg (32KB, 312x342px)

32KB, 312x342px

>>57901973

http://www.gamersnexus.net/guides/2710-amd-relive-vs-nvidia-shadowplay-benchmarks

>>57902556
>draw like a GTX 1080
>the biggest chip match with the GTX 1060

>>57902612

GTR 480 says no u.

>>57901174
>>57901220
>>57902121
>>57902189
>AMD did this.
>No they didn't
>You're a shill
>No you
The thread

>>57902602
>I can't read a thread
BUT IT'S TOO LATE

>>57901074
>>57901160
>>57901243
>>57901607
ITT: Isaacs on full damage control

>>57902497
>>57902602
Reliva is better than Shadowplay but I will not sell my 970, too late.

>>57902681

> 970

You've already made one mistake so I suppose continuing the trend doesn't hurt.

>>57902681
I have 3.5 reasons to all you a faggot.

>>57902254
Would you stop with the fucking ellipses already? It makes you look like a retarded child.

>>57902602
>not using OBS

you all fail

File: αρχείο λήψης.jpg (12KB, 199x253px) Image search: [Google]

12KB, 199x253px

amada always brings the awesome gifts last

>>57902702
>>57902717
B-b-but stop bully me. In 2014, the 970 was the best option. The 980 was overpriced and the 290/290x had GVR...no thanks

>>57902681
It's okay. No one wants to watch your GTA 5 videos you post on Youtube anyway. The only good thing Shadowplay has saved us from is Unregistered Hypercam 2 videos.

>>57902783
>No one wants to watch your GTA 5 videos you post on Youtube anyway
I pay my car, apartment, college with Twitch...

File: GCN.png (139KB, 924x576px) Image search: [Google]

139KB, 924x576px

>>57902419
AMD mixes generations all the time, 3** series cards had different architecture levels. 370 was labeled GCN 1st gen, 380 GCN 3rd gen, 390 GCN 2nd gen. Polaris is a smaller chip, Vega will be bigger and much more powerful

>>57902535
>These aren't separate architectures, they're both 4th generation GCN
It's not that clear cut. GCN is also kind of a marketing thing, it consist of different modules that can vary even in the same "GCN generation". For example Fiji and Tonga was both in the GCN 3rd gen but had some slight arch differences and Polaris isn't radically different than them even though it is called GCN 4th gen. We know Vega will have Graphics IP V9.0, up from IP v8.0 of fiji and polaris

>>57902854
the basic design is the same since 2011 only thing that has changed now is that they added the HWS the hardware raster chip and the practicly made all the cores to be flexible enough (flip/switch mid cycle) other than this nothing really has changed

>>57902783
>The only good thing Shadowplay has saved us from is Unregistered Hypercam 2 videos.
And what about performance issue?

>>57902726
>Would you stop with the fucking ellipses already? It makes you look like a retarded child.

...why?

>>57902745
>OBS
It's complete shit for AMD cards unless you use the ancient OBS with work arounds.

>>57902461
it is efficient, however the silicon lottery is real with this process. same fucking chip can draw 90 watts or 160 watts for stock performance.

>>57903085
obs studio has built in compatibility for amds encoding.

>>57903085
isn't there an OBS fork for AMD cards?

>>57901174
This. They pulled the same shit with the first generation Bulldozer too, and it turns out that it was poo.

>>57903385
No, no they didn't.
Bulldozer had a big press launch.

>>57903385
I don't think Vega is going to be a "disappointment" because the only people expecting it to be unreal are stupid fanboys or false flagging Nvidiots. It'll be good. Just good. Nothing more, nothing less.

Are these new drivers going to add support to Ubuntu 16.10?

>>57903475
>I don't think Vega is going to be a "disappointment"
That's real funny
How much are you being paid to say that, Poojeet?

there is a very real possibility that the card we keep seeing on aots nowdays isnt a vega but a full fledged polaris with 4096(the actual polaris 12) and this makes sense actually because that means that vega will be the true successor of the hbm monsters

>>57903475

saying anything AMD launches will be 'good' is a pretty unrealistic expectation, most people would consider behind 2 generations worth of perf/w behind nvidia to be 'good'.

>>57903606
Keep dreaming, Pajeet. Polaris 12 is still a rumor at this point and Vega isn't coming out for another 3 quarters. It's just another shitty dual GPU Polaris 10 piece of shit that no one will buy.

>still falling for the NVIDIA perf/watt meme

>>57903637
lisa said q1 random idiot on 4chan says end of the next year

yeah no one bought polaris 10 cards lol i guess amd gained 12% so far on selling potatoes

>>57903616
The initial Polaris chip was a bit of a disappointment in terms of performance per watt but GloFlo have now got the production process under control and we are now seeing 14nm chips coming out that equal the GTX 1060's ppw target in the XFX RX 480 GTR Black. Blame GloFlo for the shaky start if you must.

>>57903670
Only idiots bought Poolaris 10 and 11. OEMs avoided them like the cancer they are because they're in the business of making money, not losing it from expensive RMAs.

>>57902419
>Is Polaris just a stopgap, which Vega will replace entirely?

Yes, Polaris is stopgap as fuck.
It's barely more than a shrink of older GCN and lacks rumored major features of Vega like double-speed fp16 and completely restructured SIMD scheduling (and even variable width?).

Polaris 11 will probably continue to be sold under one name or another for years, but supposedly even Polaris 10 is being superseded by a smaller (1x HBM2) Vega chip.

>>57903687
because surely there were so many OEMs out there with a 1060..
oh wait....OEMs dont care about mid gpus at all

>>57903756
Last I checked, 75% of pre-built gaming PCs on Newegg and Amazon had what midrange card?
Oh right, the GTX 1060.

>>57903783
75% literally pulling a number out of your ass lol fact is every retailer that sells worldwide has both cards as an option only few like shillied maingear and dino shit pc that sells nvidia infested pc with a propiertary motherboard with an i7 and 16gb along with a ref 1060..

>>57903783
90% of prebuilt gaming PCs have a low-end card like 950.

>>57902583
Good lord at least try to make plausible speculation.
A Polaris 10 expanded by 50% would be ~340 mm^2 and would outperform the 1070 pretty handily.
A 600 mm^2 chip would crush the TXP but would have a huge TDP.

Something more reasonable:
Little Vega: 1x8GB HBM2 (256 GB/s), 2560 shaders @ ~1.5 GHz = 8 TFLOPS, ~160W, ~300 mm^2
Big Vega: 2x8GB HBM2 (512 GB/s), 4096 shaders @ ~1.45 GHz = 12 TFLOPS, ~230W, ~500 mm^2
* plus binned variants of both

>>57903895
a 600mm2 would be like a fury x x2 lol i would love to see this shit with a 300 watt tdp

>>57902588
Enjoy your old as fuck technology

i live in toronto and i went to markham to see what's going on, i was in the parking lot and i seen some of the youtube reviewers chilling outside

i can confirm some shit did go down

>>57904048
nobody cares about a shitty 970 owners opinion, fuck off.

>>57903917

Given the pro duo effectively has a full cover block for two dies a proper premium AIO on the reference model (basically the pro duo's cooler as that is better than the fury x's) would easily contain such a beast if the supplied fan is good. Especially if AMD decides to do some crazy binning ala nano and only release the best voltage curve cards as full fat vega (they won't, but it would be insane if they did).

>2017
>falling for amds shiesty tactics

https://www.youtube.com/watch?v=04ITA1_XoqM

>>57901004
I bought a 6850K for 200 and plan to resell it, so I really hope Zen crashes and burns so I can that max profit.

>>57903616
>Perf/watt guys
Does anyone give a shit about spending one dollar a year more on their video card that cost half of nvidia's?

It's an absurd argument to make.

>>57904485

Nobody cared until Nvidia said perf/watt was important. When fermi was a nuclear reactor vs VLIW it was justified because "muh performance".

File: AMD-Polaris-10-XT2-Polaris12-1.jpg (769KB, 1514x839px) Image search: [Google]

769KB, 1514x839px

>>57903637
>Polaris 12 is still a rumor
no it isn't

>>57902419
AMD doesn't do 1 arch per series.

Probably, 500 series will be a mix of Polaris on the low end and Vega on the high end.

>>57904531
it's odd how every 400 series runs so absurdly hot
gtx 400 = housefire
rx 400= housefire
legit 80-85 ingame bcus the stock fancurve is retarded

>>57903130
Is that because lack of quality control during manufacture?

>>57904719
It's because GloFo can't get their shit together and they overpromise and under-deliver.

>>57904719
It's because the best polaris chips are used for other things then GPU's. Gpu's are actually the last thing amd uses its chips for, medical shit goed first so we gamers get the worst chips.

>>57904749
>>57904785
Will those things ever change?

>>57904877
when the process is more mature they will get better results and release an RX 485 that clocks higher with less power consumption.

>>57904719
First of all the statement is entirely incorrect.
The Ellesmere die used in the RX 480 pulls 110w at stock settings. This is the same for every die they ship out. This does not vary unless an OEM alters stock voltages and clocks.

In the foundry world some chips always end up a little better than others, some will be able to reach higher clocks with less drive voltage. That means generally you can hit a target clock with less power draw than average. You only would see the effect from this is you manually undervolted the GPU yourself though. OEMs almost never bother releasing undervolted/underclocked cards unless they're specifically low power or passively cooled variants.

>>57904749
GloFo is just offering Samsung's 14nm process. Any perceived issues(which don't exist in reality) have nothing to do with Global Foundries' past fuckups.

>>57904785
AMD didn't do any specific clock/voltage binning for consumer Polaris cards. They set margins wide enough that millions of dies would be viable, and they if they passed QC they were shipped.
Parts sold to Apple were the pick of the litter.


>>57904877
There is nothing to change. They didn't do anything wrong.

>>57904908
Off topic but I saw a comment on a previous Zen thread about the quality of a Zen based Chipset, ppl speculate it will be bad in quality because of past, is that true?

>>57905013
AMD purchases their external chipsets from a 3rd party manufacturer ASMedia. They've generally been pretty spotty. Random SATA issues, things like that. Never major issues, but annoying small things pretty consistently. For the record intel has had their own fair share of issues as well over the years, but no one ever mentions that because affected users in both cases are low single digits.
Some shitposter keeps spreading FUD about a USB3.1 nonexistent USB3.1 issue, which has been the only rumor casting doubt on them. Its just obnoxious fanboys being fanboys.

>>57905013
Do you mean that AsMedia USB3 or SATA signal quality FUD?
Nothing is wrong with that. Every USB3 controller needs a signal booster (Intel's chipsets do, too), but for some reason people started screaming about how every Zen AM4 motherboard is fucked.

>>57905013
It will most prob have a 400W draw judging by their last shit.

>>57905109
(You)

>>57903526
What kernel are you on?

>>57905013
Do you mean some external vendor chipset, Zen itself, the southbridge, USB3 false flag bullshit, or what.

Either way, no.

Like >>57904908 hinted at the issue isn't the process itself it's the tolerance margins for everything, which is why some are like this >>57903130

We won't have the opportunity to see it in use until the 13th and even then I'm sure it's going to be heavily sanitized.

>>57905117
>fx83xx 125W TDP
>fx93xx 220W TDP

Ok so maybe only 330W TDP

>>57904639
You know nothing about housefires. GTX 480 easily went over 90 degrees.

>>57903142
>>57903364
The old application with the shit UI has the AMD encoding the new 'modern' one does not, "yet"

File: amd1.jpg (71KB, 552x661px)

71KB, 552x661px

>>57905181
DELET THIS

I hope we get to see vega this year hyped, anons.

>>57905890
Gonna molest your best sheep to celebrate?

>>57901004
>unironically posting wcftech, videocardz.com or any other Pajeet tech website as source.

You know better than that Singh.

>>57905069
Would there be anything, and I mean anything specific to not buy Zen(including a mobo obviously) for playing games/movies if you were not to take into consideration the CPU's IPC and clock speed? I really want to hear all aspects of its design, the chipsest, mobos etc. I am getting a PC after almost 10 years and I want to have as litle buyers remorse as possible for the years to come.

>>57905943
Far too early to tell.

>>57905914
Calling yourself sheep only helps that much, goy. I'll be gently :^}

>>57906134
>another /g/ retard with grade 1 education

>>57905148
Yeah, if anything they'll show a full DotA 2 match with Zen/Vega running without any game benchamrks againts the competition, just building hype.

File: 1479090146735.jpg (123KB, 720x540px)

123KB, 720x540px

>>57905890
I hope it launches alognside Zen tbqh familia.

Is there any way to make the ReLive recording not pick up your equalizer?

File: 1479089045623.jpg (10KB, 164x164px) Image search: [Google]

10KB, 164x164px

>>57905924
What's a good rumour site then? If you're not into rumours don't reply.

>>57906258
LMAO, enjoy your AYYYYYYMD

AMD is in such a weird spot when it comes to GPUs. They have nothing in high end/enthusiastic 1440p/4K/144Mhz market. Their 480RX is arguably better than 1060 GTX but because DX12 is still non-existent and many many dx11 games favor Nvidia by using workshop etc. they still look weak. Fact that their cards were hard to obtain similar to 1070/1080 didn't make it better - However 480RX could never compete with 1070/1080 and by the time you could get something good like Nitro, 1060 gtx came in full force and made AMD lose once again cuz of brand recognition. By the time DX12 will be actually viable (which won't happen at least till 2018-2019, especially since patching up dx11 games to dx12 makes them run like shit across the board and no dev will sacrifice big chunk of pc crowd by making dx12 only games where AMD could really show how good async is) - Nvidia will come up with some other, go to normie card that is insane for its price like 970 (meme all you want, 3,5 was never an issue, this card is still very much more than enough for your 1080p/60Mhz crowd which is majority of people). It's like they constantly come up with innovative and good tech but it's always too late or too early. It pisses me off.

>>57905181
There's literally nothing wrong with high TDP :^^) (if the performance is good and it can scale well)

>>57901004
the only thing I cared was 4k hevc decoding,and it's still doesn't work.
At least not on windows 7 and chrome.

>>57906422
It turns out that nobody gives a shit about the .01% market share of $800+ video cards and the real money is in products people actually buy.

The 970 is a meme and at the cost it was a double meme.
>3.5gb was never an issue
Whew lad, how's that 1080p and medium texture quality on your 350 dollar card doing?

>>57906422
Honestly, I got a 1060 anticipating I would buy a new graphics card in 2 years. The lack of dx12 support kept me from getting a 1070 from nvidia, and the lack of dx12 games(and other features) kept me from getting a 480. This is just a really awkward time to be in market for graphics cards, with New APIs, slowly rising popularity of 4k monitors and VR.

That said, its still an interesting time for AMD, they've planted a lot of seeds to retake the market. Both DX12 and Vulcan are based off of their Mantle API. Console's use AMD hardware which means games will be optimized for AMD. They are putting a big focus on Mainstream gaming as oppose to high end in order to take market share. And they have been making a big show of improved software support ever since they switch to the QT back end in their control panel. These are all savvy business moves.

Whether they will pay off against nvidia brute performance lead and push into the AI market is still up for debate though.

>>57906252

I really think this is what they are trying to do. Can you imagine having CPU+GPU bundles available at launch? Fuck, Zen and Vega could be mediocre, as long as its better than FX line and RX they would be laughing to the bank. I really want them to release something even if its just OK. Doesn't have to be good, just something for the right price.

> I just want my AMD stock to go up
> I haven't made a 20% profit already since October

>>57904639
RX 400 series aren't even close to housefires, like at all
The R9 290X was called a housefire because it had one of the shittiest coolers ever in a reference design
The GTX 480 was a housefire because it was close to 100 degrees C, had a brand new cooler and even then they had to disable parts of the chip to have a viable die
Power consumption wise, the GTX 480 is still among the worst ever made
>>57902602
>Encoders for ReLive can be switched between AVC and HEVC, while nVidia’s uses its own solution for x264 encoding
>x264 encoding
Why are GPU/tech """"""journalists"""""" so retarded?
They can't even get the most basic facts right

>>57908143

The funny thing is, AMD has almost had the better hardware, while Nvidia had the drivers. This is why AMD ages well, because there is more headroom to have performance improvements with driver optimization. Nvidia releases cards that are pretty much as good as they are going to get, but damn, they are good off the bat.

Nividia also has actual money to spend on teams that go around helping developers dealing with their hardware and drivers. AMD doesn't have anything like that right now. If AMD can get some cash flowing and a non retard marketing team, I think they will make a YUGE comeback. About time IMO.

>>57908398
Yep, price is all what will come down to.

>>57908143
>Console's use AMD hardware which means games will be optimized for AMD
this is already starting to show, dx12 is having some major improvements for framerates on both phenom II and FX cpus.

>>57908143
>push into the AI market
i thought they were going for mobile (nintendo switch, and then a successor to their shield)

>>57908402
>RX 400 series aren't even close to housefires, like at all
a gpu which target is 85C and throttles itself rather than turning up the fan speed is awful and housefire material.
fermi was just bad all around, they could've and should've released the 5xx variant first but were already far behind AMDs 5870 which was top dog for a loong time

>>57908402
>tfw it turns out nvidia actually embed libx264 on their cards to do their h.264 encoding and that journalist knew it all along

>>57909211
>it turns out nvidia actually embed libx264
Ignoring the massive licensing problems this would entail, Nvidia's quality is too shit to be comparable to x264

>>57901004
Nice Fake News

>>57908485
>Nvidia releases cards that are pretty much as good as they are going to get, but damn, they are good off the bat.
That's a big problem for AMD. There's still the perception that you have to wait for AMD to get good. Most people buy whats good today. They really need to shake that imagine they built up


>>57902602
>>Encoders for ReLive can be switched between AVC and HEVC, while nVidia’s uses its own solution for x264 encoding.

Both Nvidia and AMD support HEVC, Nvidia blocks it for Shadowplay due to the fact that Youtube WILL NOT transcode from it. Which is annoying because it can cut bandwidth in half for the same quality if they supported transcoding from it.

Good to hear AMD is finally catching up on the streaming side, if not even overtaking. Its been way to long to blissfully ignore it.


>>57909147
That's true as well, but Google, Amazon and Microsoft have all been working deals with Nvidia expressly for AI.

>https://lists.freedesktop.org/archives/dri-devel/2016-December/126516.html

Linux cucked! DC aka DAL not making it into Linux. Not sure if both freesync and eyefinity were packaged in it but this means it's gonna be a while until we see it crop up. It's about 93000 lines that's going to be rewritten at the speed the 970 approaches when it hits the 3.5GB limit. fug.

In other news, some ledditor found that async is disabled for Tahiti (and probably all GCN1.0) cards on drivers 16.9.2 and up. A lot of people are accusing AMD of intentionally shittying up their GIMPCN 1.0 GPUs, though some maintain it's a bug.

Also modded bios for Polaris cards are out the window with the new drivers, BIOS verification is strict. However, you can run a modified driver prior to verification then disable verification to workaround the issue.

>>57907487
See posts like this is why I hate to admit that I'm an AMD owner, drones like you just make me cringe.

>>57908143
I really want to believe that HBM2 is going to be the shit but somehow I feel it's going to be Fury all over again.

>>57909139
>>57908143
Haha... Are you guys for real? If anything consoles bring pc down harder than ever before. Look at shit like Dishonored 2, Mafia 3, Forza Horizon 3. Ye I guess it's nice that people with $100 cpus and $150 gpus will have a performance as terrible as guys with $300 cpus and $250 gpus.

>>57910937
I (>>57908143) never said it was a good thing for PC. I said it was savvy of AMD.

As game developers now have to re-target their games to be optimized for AMD this will, with little doubt, negatively impact PC games who are mostly composed of Intel and Nvidia based systems. So yeah this will be a large set back for PC games that already suffer from bad ports.

Even AMD users will be on the short end of the stick for a while as their cards still fall short with DX11 games and DX12 is still a number of years off.

>>57902956
it's faster than shadowplay

>>57911071
amd drivers just got +20% boost in synthetic benchmarks on dx11 threads

>>57911612
You got references other than >>57902602, because they only tested 2 games.

>>57911641
>+20% boost in synthetic benchmarks
>boost in synthetic benchmarks
>synthetic benchmarks

No one has mentioned that ReLive can stream 1440p60

>>57909171
GTX7xx/GTX9xx series are aimed at 85C

>>57911757
scroll down for other tests, basically in general it hits less than shadowplay for about 1%
that's in nvidia titles like rotr and farcry primal
in dirt rally it's 2% difference
avc and hvec is just better than whatever nvidia got, they jewed on paying licencing fee my guess
shadowplay might've been faster with those codecs who knows

>>57901973
>I hate Radeons for only one (1) reason: no Shadowplay.

They had play.tv before today. That worked fairly well. And today they launched relive.

>>57911992
Same way no one has mentioned that ReLive causes constant stuttering.

>>57912760
Consider how small the difference is, it's likely just the memory bandwidth difference. Nvidia calls their hardware encoder NVENC and in the 10 series cards it support AVC and HEVC. You can actually use them both with ffmpeg or xmedia right now.

>>57909675
Man I enjoy reading Linux mailiing lists so much.
There is just something comfy about them.

>>57904121
well if vega isnt going on any apple shit we might actually see very good binned shit around
>>57904611
indeed this will be the last of the currect gcn design
>>57904616
actually no vega is the v9 its a totally different design than polaris and they do have 4+4 cards (4 mid/low) (4 mid/high)vega as a series is going to replace everything even polaris understandable tho since the whole series will have hbm 2

>>57913723
>they do have 4+4 cards (4 mid/low) (4 mid/high)vega
Sauce? So far only Vega 10 and 11 are rumored.

>the whole series will have hbm 2
HBM2 seems unlikely to be in mainstream cards soon. Nvidia only uses it in their $5000 server cards, not even the $1200 Titan.

>>57909495
considering that google just made a deal with amd for AI then no..
afterall amd already had a hip tool that can translate cuda code to c++ and given how much better amd is when it comes to paraller workloads the future will be quite interesting to see..
gotta say getting rid of the former ceo of amd and bringing back raja koduri was actually 2 of the best moves amd made since 20 years now..

>>57913843
those are the sources lol vega 11 is the mid low and vega 10 will be the high even wccft got this right(which desu the leaked design specs of vega pretty much sealed it)

well nvidia doesnt have licencing deal for hbm2 neither does have priority since amd was the one that developed it.. and unlike the mistake amd made with gddr memory they wont let nvidia take advantage of this anytime soon..

File: butthurt.jpg (226KB, 1280x720px) Image search: [Google]

226KB, 1280x720px

>>57901220

>>57911071
nobody gave a fuck when gimpworks negatively impacted any game running on AMD systems and some older nvidia cards so really, I don't give a fuck

>>57913852
>considering that google just made a deal with amd for AI then no..
Nvidia has been all over the news recently in regards to AI.

Nvidia's surge into AI:
http://www.forbes.com/sites/aarontilley/2016/11/30/nvidia-deep-learning-ai-intel/#45766d0339cc

>>Tech giants like Google, Microsoft, Facebook and Amazon are buying ever larger quantities of Nvidia's chips for their data centers. Institutions like Massachusetts General Hospital are using Nvidia chips to spot anomalies in medical images like CT scans. Tesla recently announced it would be installing Nvidia GPUs in all of its cars to enable autonomous driving. Nvidia chips provide the horsepower underlying virtual reality headsets, like those being brought to market by Facebook and HTC.


AMD and Intel playing catch up::
http://marketrealist.com/2016/11/growth-potential-artificial-intelligence-attracts-intel-amd/

>>The increasing adoption of AI could see other companies such as Intel (INTC) and Advanced Micro Devices (AMD) create their own versions of accelerators. However, they would have to do a lot of work to develop a large ecosystem to equal the one that Nvidia has built over the years.

>>Returning to the AI space, AMD has launched its Radeon Open Compute Platform, which eliminates the need for developers to port their code to OpenCL to run on AMD GPUs. Although AMD has the required software in place for data center applications, its hardware still lags behind Nvidia’s hardware.

File: amd.png (42KB, 653x726px)

42KB, 653x726px

>>57914655
>its hardware still lags behind Nvidia’s hardware.
BUT MUH ASYNC AND BUTTCOINS

>>57901996
>from 1060 to be another 960(well this one was true)
what? no it isn't.

the 1060 is roughly equivalent to the 980.

the 960 was even weaker than a 780 lmao.

>>57902160
>has 970
>complains about anything AMD
kys.

>>57906422
they need to hit the high end market in full force with vega. if they release a 1080 contenter they have lost.
AMD needs something than a 1080. Their 590 should be at least 7% better than 1080 at the same price. Then nvidia will answer with 1080ti , and amd must have a competitor there to one them up.

480 won the mid range vs the 1060, but they gave up the high end. A 490 would be a failure, they need to skip a generation with vega and release stronger products than nvidias current offerings.

Stocks falling again.

>>57915308
Probably just profit taking and stops triggering. Nvidia dropping just as hard.

Why do the mods even let these threads stay up. Nvidia shills shit up every graphics card related discussion with shitty memes and constant racism.

Delete these threads and ban the shitposters fucking cancerous faggots.

>>57905287
there was an old branch version that did, and then there is the obs studio that does. why lie?

>>57904908
there is no real way for us to measure just the chip though, so we have to settle for entire card or entire system and do math from that.

>>57905943
>>57905987

from what we can tell, amd shaved off things like a large fpu to make the chips cheaper.

what this does is in a large fpu workload will be intel's clear advantage.

What this means for normal use is completely up in the air, especially after the blender demo as rendering 3d on a cpu is one of the more demanding things you can do and it effectively tied the 8 core at equal clocks.

The best I can think is possibly video rendering will be effected, possibly some render engines will shit the bed, and as amd put it 20% of server workloads intel will have a clear advantage.

This cpu is going to be very interesting because they trimmed the fat for a cheaper chip, and the question is will we miss it

>>57915251
>480 won the mid range vs the 1060
I assume you mean in performance cause they are lagging behind hard in sales. By more that a factor of 4.

>> NVIDIA GeForce GTX 1060 ---- 1.47% - +0.48%
>> AMD Radeon RX 480 --- 0.35% --- +0.08%

Source: http://store.steampowered.com/hwsurvey

They didn't even see as much growth as the 1060. I think a big part of it is the fact they didn't launch relive WITH the 400 series. I personally would have considered them more if it had the software features it has now when I purchased my card.

This how bad AMD has flubbed marketing and software development with their card. It is better performing than a 1060 and is cheaper than a 1060. But their lack of software support and poor marketing has kept it behind nvidia.

>>57906209
personally, and i would find this fucking hilarious, if they showed in every single shot, something that pushes a cpu to its limit.

In the background you see dishonored 2 playing around 140 fps but in the foreground, you see 'so what can you tell me about this dota match'

the whole god damn time, something fascinating in the background, interview with a player 'so you switched from lol to dota'

It goes on this way for a while till you see in the background, out of focus but just in focus enough to see, half life 3

no one questions it, no one even acknowledges any of the things in the background exist. every site that went there doesn't talk about the background stuff at all, and the internet collectively shits itself over it.

Im guessing this is why im not in charge of marketing.

>>57915893
>constant racism
poojeet got his feelings hurt

>>57916113
1060 STRONK tho
>>57915630

>>57913843
because hbm2 isnt/wasnt close to ready for main stream use, nvidia used it in their server parts at a massive fucking premium because they had no other choice.

>>57916020
100% incorrect. The ASIC monitors and reports its own power draw.
Upon boot the die does a self check and monitors its own leakage and power draw characteristics based on its set voltages in all its pstates.

>>57914774
1060 in dx11 is on par with a 980 possibly because of gimping
1060 on dx12 is way behind a 980..

>>57910937
dishonored 2 is on id tech 5, an engine that performs like shit in every aspect possible. the only time it was decent was in wolfenstein if you had nvidia because they had to implement cuda to help with the textures.

mafia 3 is a true open world game, has there every been one that preforms well?

and to top it off, a windows store exclusive game that is made by a team with 0 pc experience... call me shocked when that plays like shit.

what im saying is the consoles didnt fuck these games, retarded devs did.

>>57916518
>mafia 3 is a true open world game, has there every been one that preforms well?
GTA V has excellent performance and is very scalable too. Witcher 3 also runs much better than Mafia 3. Mafia 3 is exceptionally shit. I'm pretty sure even Dishonored 2 runs better.

>>57915251
>Their 590 should be at least 7% better than 1080 at the same price
By the time Vega is in stores with decent availability the 1080 will probably be ~1 year old, AMD need to do much better than an extra 7% over last year's card for the same money. Everyone who wanted 1080 performance at 1080 price already bought one when AMD said they weren't even going to compete.

>>57916227
I'm white but holy shit does it get boring after a while. You must have a really sad life to fill it with garbage like that.

Personally I don't give a shit about AMD winning anything. I go where the smart buy is. That means similar or better performance for less money plus better longevity and that's the 480. Why should I care what babby Nvidia no brainers buy? Why should they care what I buy? I buy a product based upon informed reasoning not marketing hype.

>>57916518
Any reason why they didn't go with idTech6?
The renderer is ridiculously robust in D44M.

>>57903045

Different anon here but ellipses are used by chat kiddies.
They also make it seem like you have not finished your thoughts before typing them.

>>57904639

Are you kidding? My RX 480 runs at 33° C in desktop use (that is lower than my CPU, i7-2600K) without under clocking or anything.

Will they finally fix their multi monitor power consumption?

>>57915251
>480 won the mid range vs the 1060

Only in that broken head of yours. AMD did fuck up literally everything they could when it comes to launching 480RX. I'm not really gonna write everything down cuz as always I will just be greeted with some sort of an Indian rupees meme that AMD shills stole from Nvidia shills.

>>57920357
you already gave up your powerbill running two monitors, why care?

>>57916493
https://www.youtube.com/watch?v=qoGSpWItULY

Overall it seems like the newer counterparts run slower than the last series. 980 Ti beats 1070, 980 beats 1060. Marginally but still. I don't try to say that 980/Ti are better cards because they're not even on paper. I assume shitty drivers but that video is now almost 2 months old

>>57920420
Wait for new drivers, the new cards will go up by like 0.1% and the old cards will drop by 20%

File: dex_1920.jpg (158KB, 548x629px) Image search: [Google]

158KB, 548x629px

>>57920471

Its the Nvidia way.

>>57920665
>980 Ti 2 fps behind 1070
>1060 not even on list

I think you're projecting.

>>57920779

I don't see a fury either, guess there must be some sort of hidden agenda! Or maybe Ivan and friends didn't have a card to test at the time that benchmark was done.

>>57921004
>I don't see a fury either, guess there must be some sort of hidden agenda!
It's the second and sixth from the top

>>57921457

Fury x and nano =/= fury. I chose my words carefully.

>>57916082
Thank you.

>>57916666
witcher 3 is 'open world'
realistically you get a single big playground, its not a real open world.

gta5 runs like shit and doesn't scale down well to bad cpus, granted I played it at launch when optimizing the game was still a mystery, I may be able to get it to play ok now.

and as for amd needing to do much better, why? no one buys their cards anyway, what amd always needs to do is hit that sweet spot, design a card that is cheap for how good it preforms. chasing nvidia is fucking stupid for them till they can mcm low end gpus into one big non crossfire using socket.

>>57918577
because bethesda cant do fucking anything right.

most likely though, they were making the game in rage very shortly after the first dishonored, possibly wolfenstein, and sense mega textures are a fucking hellish thing to work with they couldn't port.

>>57919998
...ok...

>>57920665
>releasing a game which the latest gpus can't play at decent framerates without having more than one
what an unoptimized piece of shit. what kind of fuckstick thinks it's okay to release trash such as this? reminds me of crysis

>>57904639
>we wuz kangz
t. nigger

>>57920384
my three monitors consume 60 watts altogether
AYYYYMD GPUs do the same, while my 1070 only draws 20 in idle

AMD better delivers an efficient card that is a competition for the GTX1070.

Anything else is a failure.

>>57925068
go back to the subreddit fag

File: 1471362590570.jpg (171KB, 1062x1062px)

171KB, 1062x1062px

>>57916518
>true open world game, has there every been one that preforms well?
skyrim

File: kek.jpg (53KB, 1167x165px)

53KB, 1167x165px

Most of /g/ is a special kind of retarded. You can seriously all go fuck yourselves.

If I book profits, your opinions don't matter. It's as simple as that.

>>57927183
Yeah, I bought £100s worth back in feb, when they were down at about $1.60.

Yeterday, bought an extra £200s worth, to ride the hype train of Vega and Zen.

>>57908143
>mfw i got a fury nitro for $270 on black friday
>mfw i bought it to play games more smoothly and at higher settings on 1080p while i wait for vega/zen to do a new build
>mfw i didnt even research it, just pulled the trigger
>mfw turns out its better than 480, 1060, and super clocked 980 imd dx11 across the board
>mfw its better than 1070 in dx12
>mfw with a modest oc its better than fury x
>mfw its horribly bottlenecked by my 8320 @4.2

I might even have to keep this thing even after zen is out. Its a lot more beastly than i expected. I feel bad for all you cucks fighting over 480 vs 1060, this thing was so worth the $260.

File: self intoxicated, twisted detonator.png (314KB, 1131x1006px) Image search: [Google]

314KB, 1131x1006px

>>57928846

>mfw its horribly bottlenecked by my 8320 @4.2

You really should have that chip at 4.7ghz or higher. Just don't do what i'm doing - running my motherboard way beyond spec just because.

>>57928975
I would need a better cooler to bump it any further. At full load the fan sarts to ramp up a lot and its unbearable. And frankly its not worth it. I can play most games at 60fps 1080 with all the completely over kill retarded graphical settings turned up pretty comfortably.
Lately ive finally been playing witcher 2 since my 7870ghz ran that game like shit. I can run it at 4k with everything set to ultra and uber sampling turned off.
I wont be making any more purchases until im building a whole new system. And with the fury i can probably wait to buy a new gpu untill vega's first price

>buying AMD
>2016-17

>>57928975
>Jan 20 2016, $1.80
>Dec 9, $10.34

Well meme'd my friend

File: gta-v-cpu-4k-vh.jpg (52KB, 806x532px)

52KB, 806x532px

>>57929041

> I can run it at 4k with everything set to ultra and uber sampling turned off.

That is because cpu load doesn't change with resolution and at 4k its the gpu that crumbles first, not the cpu.

>>57929071
Oh i know friend. I was praising the gpu and just saying the cpu is ok for the time being. But its sadly pretty much only able to do 60fps. Be it at 4k/ultra/no sao/aa or 1080 with everything set to low, 60 is the most it will get.
I know ive seen a handful of benchmarks showing a pretty big improvement on the 8 core amd cpus in dx12 mode. Ill be curious to see if that will allow this system to last some time longer than expected. Either way come tax season ill either be getting zen or kaby lake.

>>57920420
have you heard the story about 780ti and dx12? and why nvidia actively tried to tell to any developer not to run any bench with maxwell cards and dx12 when tthey use async?
did you know that because 780ti is a superscalar design and it has a hardware sc it was faster than a 980 till 1440p on aots?due to the fact that it was simply faster on paraller workloads? if 780ti had 8gb of ram back then this card would have smoked the shit out of any maxwell till now..

>>57929286
Maxwell being cut down Kepler with more tessellation/ROP throughput is the biggest tech scam of our time.

>>57929286

>when tthey use async?

Async kills performance on Nvidia's architecture full stop - pascal still chokes under such workloads unless Nvidia tune the pre-emption. In practical terms you can't kill a GCN 1.1 or newer (you can kill 1.0 though) with async workloads - by the time you overwhelm the amount of queues available (assuming the workload is broken up to fit correctly) the rest of the chip would've choked long ago.

Its why fury x can see such enormous gains as async lets it use its enormous compute power.

>>57929544
Fiji benefits most since it's the most unbalanced GPU in terms of compute vs. pure graphics fixed function.
> 8.5 TFLOPS with only 4 GTRI/s and 64 GP/s

>>57929891

Imagine fiji with 128rops - it would be a monster in every aspect. Shame 20nm fell through.

>>57926493
the game that limits you to 60fps or the game shits itself and still took years to max everything out without dropping below 60, oh and the best part modding was broken for quite a while because of how the game handled ram in a game that is sold on the premise of mods? I mean sure, if you want to call a game that requires a load screen for every single thing you do open world then mario galaxy is also open world.

File: 9YNJkEB.jpg (63KB, 960x640px)

63KB, 960x640px

AMD NAPLES
M
D

N
A
P
L
E
S

File: 6MZzqWo.jpg (51KB, 960x640px) Image search: [Google]

51KB, 960x640px

>>57930326

>>57930057
they didn't even use sse instructions properly when it came out, so needed mods to speed it up.

File: 1480858901324.jpg (74KB, 627x627px) Image search: [Google]

74KB, 627x627px

>>57930326
>>57930339

>naples

More like my nipples are hard.

File: 1467429571517.jpg (344KB, 1752x1703px) Image search: [Google]

344KB, 1752x1703px

>>57930577

>AMD, do we matter? Lisa Su? AMD has a hugely influential and substantial fanbase waiting to wave your flag again. We all still have that Blue Core Thunderbird and 9700 Pro love in our hearts. We are older now and have lots of money to spend on tech and its toys. We are established, influential, and well informed, and all our family members and all their friends ask our advice on computer purchasing and then it trickles down. That is the HardOCP reader profile. Wouldn't you love to have us once again direct all those purchasing dollars with a comment like, "Just look for the AMD Zen (and beyond) badge and you will be getting a quality product."

>If we have learned anything this year, it is that grass roots support is monumental. God knows us enthusiasts can market better for AMD than Roy Taylor can any day. AMD, the door is open, give us a reason, please.

http://www.hardocp.com/article/2016/12/09/intel_kaby_lake_core_i77700k_ipc_review/6#.WEwTBuYrJQI

>>57930632

Hardocp will never reccommend AMD again no matter how good their products are.

>>57930339
inb4 pins again
Guess I'm staying Red Team again

When's amdgpu-pro 16.50 hitting the AUR? they're still on 16.30

>>57930326
>>57930339
a few questions:
why show the empty socket?
will 16c Naples have the same socket as 32c Naples, or will the latter have more pins to expose more DDR4 channels (8?) and PCIe lanes (128?)?

>>57901004
>1990+26
>South Bridge

>>57930326
>>57930339
God damn, I never like the shape of the thing.
Completely useless I know but it's ascetic as fuck

>>57931970
I even* sorry
(Wait 20 seconds so nobo)

>>57931862
The socket isn't empty.

>>57932164
fuck you're right, didn't even notice that was an IHS filling up the entire fucking space.

is this really the 2-die model?

>>57932239
Difficult to tell with the lid on.
Someone in another thread said that snowy owl (16c zen) is BGA and that this isn't it since it looks like LGA.

>>57930632
do people take this guy seriously after his tantrum?

>>57930684
what practical difference does it make?

>>57932341
Which tantrum? There's been a few

>>57932239

Naple has 2 dies on 16core version.

>>57929286
>hardware sc
Didn't they remove it after the Fermi mess?

So, /g/, can anyone give a simple answer to the following question: Have AMD's GPU drivers for linux (either open source or their proprietary blob) reached parity with their Windows drivers? I will be in the market for a new GPU in the next two months, but I've largely kept to nvidia cards, simply because the last time I bought a new GPU (2+ years ago), AMD's drivers for linux were nowhere near parity with their Windows drivers, whereas nvidia's were. Has the situation improved?

>>57929544
async kills the nvidia cards that have a pipeline uarch desing
780ti is a superscalar it was a massive paraller chip instead of a wonky pipelined one that was built to work only with gameworks..
take a read on it
https://www.ukessays.com/essays/information-technology/pipelining-and-superscalar-architecture-information-technology-essay.php

>>57934065
till yesterday amdgpu open source driver was actually pretty good
but the kernel maint said they wont allow amd to bring a hybridshell so that they can share linux and windows driver aspect and thus they killed the driver
we dont know what will happen i assume they will continue somehow developing it since vega will need a new driver for linux also..

>>57934080
Fermi and newer and all GCN arches are SIMT, not normal "superscala".

>>57934152
there hasnt been a normal superscalar cpu or gpu for like aions i said superscalar because it gives a general idea

>>57934065
Not really the same as Windows, but it has improved a lot. It's no longer a fucking mess like in the fglrx times, the new amdgpu driver is pretty good.

>>57934142
It's not the first time they reject AMD's patches, and they didn't kill the driver, you can still use the proprietary driver that includes DC and has FreeSync support.
https://www.phoronix.com/scan.php?page=article&item=amdgpu-1650-relive

>>57934214
but compared to amdgpu this is a total shit also having a closed down driver isnt really the way to go..look at nvidia and how bad it is on linux compared to amdgpu..

>>57934142
>>57934214
Thanks anons. Glad to hear things are better than the fglrx days at least.