/w4v/ - Daily Wait for Vega Thread

>>61566654
Oops, dropped the /w4n/ link like a retard: >>61514222

>>61566654
>that cooler

We shall name it Vega THROTTLEMAX

By the time this thing has proper drivers and partner versions with coolers that don't throttle the fuck out of it, Volta will be less than a year away. RIP.

>>61566692
Navi is taping out Q4 this year, we might see it Q4 next year, so Vega will be rather short lived.

Would be nice if it goes toe to toe with the 1080 at a slightly cheaper price. Next gen cards are a while away. I'm regretting buying a 2160 monitor before there were cards capable of driving it without costing a small fortune.

>>61566901
Slightly? It better be significantly.

>>61566654
Those stock coolers weren't enough to handle Polaris let alone fucking Vega. Why even bother making any retail consumer cards with a stock cooler like that?

File: psu_quarter.jpg (210KB, 1024x768px) Image search: [Google]

210KB, 1024x768px

>>61566654
pic related of what you need for run this housefire

File: lol.png (1MB, 1000x1000px) Image search: [Google]

1MB, 1000x1000px

>>61566654
Jewish slide thread, nice try brian

>>61567303
could you assign monetary value to slightly and significantly?

for me slightly is $20 which would be irrelevant and eaten in different regions making 1080 a better buy
and significantly everything $50-70 or more

>>61566712
So I should Wait™ for Navi?

>>61567653
For making me wait for a good 8 months using my shitty iGPU cause I fell for the wait meme, it better be under 400 dollars.

>>61567742
why did you wait? 480s were dirt cheap end of last year
I can wait,I have 390
For what reason you wait with iGPU, you have only yourself to blame.

>>61567782
>why did you wait?
Cause people told me Vega was on the horizon and it was going to BTFO out of whatever Nvidia had. No point in buying a mid range card that I am going to use only for 4-6 months cause Q2 release.

File: 1422463258579.jpg (25KB, 400x387px) Image search: [Google]

25KB, 400x387px

>>61566654
So the "gaymin" drivers are out and tested internally by aymd
>scores the same on 3d mark as FE

hold me /g/

>>61567742

Keep dreaming. The RX 580 is selling for nearly that price. We'll be lucky if low tier RX Vega cards are $600.

>>61567827
I fucking hate all of you and I fucking hate everything if it isn't below 400.

>>61567844
>if it isn't below 400
it will probably be below 400W power consumption if you're lucky

>>61567816
Don't worry buddy I got some excuses for you too keep waiting
>Secret features are turned off
>Anyone who doesn't believe the secret features will make it 1080ti performance don't know about GPUs
>something something full gpu 102
>something something go back to /v/
>something something backhand tactics
You're good bud

>>61567732
vega is literally gcn 5.0 the last of that kind navi is the 100% card that raja worked on and since it is rumored to be an mcm with IF we are going to see another ryzen

I may just stick with my 480 if this turns out to be shit. I can run most games at 1440 with some tweaks. Heck I can let it drop as low as 40 fps on Freesync.

File: eb0ffdef4c.gif (4MB, 460x344px) Image search: [Google]

4MB, 460x344px

So a Vega FE tier GPU for half the price?
Count me in

>>61568110
it would be pretty shocking if a 60% core clock bump and a total rework of the pipelines and back end resulted in 0% gaming performance increase. They would have needed to do literally nothing else but throw transistors into compute and associated functions for the scores to be so similar with P6000 but so vast between the 1080 Ti and RX Vega.

just a reminder on what to expect from the proper drivers..

nfinity is agnostic on topologies and will be implemented like a mesh on Vega, said Maurice Steinman, an AMD fellow for client SoC architectures and modeling. It can provide the full bandwidth of any attached DRAM.

“That was something that we could not do in the past,” Steinman said.

“We had multiple on-die protocols trying to do the same thing that gave us some inefficiencies,” Steinman said. He added that creating a new interconnect “was a huge investment, but we are seeing the ability to do variants and architectures we could not do that we are now embracing.”

In the past, an on-chip network change early in an SoC design cycle might have taken six months, “but we can do it now in a few hours,” he said. In addition, AMD is able to offer more interconnect variants to its ASIC customers, such as the videogame console makers.

The Infinity link is the conduit for a new suite of uses. They range from test and debug functions to new algorithms to check hundreds of on-chip sensors and dynamically adjust power and frequency as the chip has thermal headroom.

>>61568955
more overpromising like XFR.

>>61568955
what does this mean though? I'm not an microchip engineer.

>>61568968
you are saying that infinity fabric doesnt work on ryzen?
are you mad?
go back to /v/
>>61568969
means that they can use infinity fabric for pretty much everything that is on chip and outside of the chip and is directly tied to the speed of the ram(in the case of ryzen) and hbm2(in the case of the gpus)

>>61569009
you seem to think its an actual fabric. I don't think you even know what he is saying.

>>61569032
yeah obviously
meanwhile on the real world
http://www.freepatentsonline.com/y2017/0200672.html
>The described embodiments include an interposer with signal routes located therein. The interposer includes a set of sites arranged in a pattern, each site including a set of connection points. Each connection point in each site is coupled to a corresponding one of the signal routes. Integrated circuit chiplets may be mounted on the sites and signal connectors for mounted integrated circuit chiplets may coupled to some or all of the connection points for corresponding sites, thereby coupling the chiplets to corresponding signal routes. The chiplets may then send and receive signals via the connection points and signal routes. In some embodiments, the set of connection points in each of the sites is the same, i.e., has a same physical layout. In other embodiments, the set of connection points for each site is arranged in one of two or more physical layouts.

on size fits them all by amd

>>61569064
it should be called interchangeable puzzle pieces tech instead

>>61569064
So what is the actual, real world performance advantage with this? What sort of percentage real world performance increase are we going to see?

Like XFR, it's all for nothing on the user end.

>>61569099
and thus coming back to your previous statement

xfr wasnt the real OC braker on ryzen IF was
essentially the faster the ram the faster the cpu intercommunication therefore a faster cpu on the same clock

same with gpus tho i assume they wont throw IF on everything since L2 and ROP's have a better interconnections..

File: msrp.jpg (236KB, 700x476px) Image search: [Google]

236KB, 700x476px

is there ANY chance to buy a Vega at its launch price? will the miner faggots drive it up instantly or will it wait? will the retailers anticipate the miner faggots and drive it up anyhow?

we already know that Vega FE has around 35 MH/s, with 300 Watt TDP. I assume RX Vega will have a bit less of both.
if they launch at around 400$ shouldn't the miners still prefer RX 580? although overpriced, is is sold for much less than 400$.

>>61570919
Vega is shit at Etherum but is good at other currencies like Monero where the 1080ti was the best choice, Vega gets about double of a 1080ti

>>61570933
but monero has a low price and it wouldn't be worth to invest a 400$ GPU that needs 300 watts. also as soon as miners flock to it, monero will tank and will become hard to mine. ethereum is the only annoying shit that many normies mine. ethereum and only ethereum is at fault for the GPU prices.

>>61570997
Monero bubble when?

If the thing is even remotely cheaper than a 1080 and can be underclocked/undervolted, miners will still grab it.

>>61570997
There are people that mine Monero with Xeons, some have dedicated 20k sockets since it's good with both GPUs and CPUs(especially Zen arch ones)

If anything it's possible the next fad after Ethereum.

>>61568951
>Creo
>Maya
>Solidworks
>compute
?

>people waited 15 months for this
>15 months
>for a 1080: power hungry edition

>>61571051
No new currency will have significant value.
Bitcion works for buying drugs and paying bribes.
Why would the users want anything else?
What purpose do ther coins have?

>>61571578
Ask the ones mining them.

>>61568110
Threadly reminder that anyone who claims to "know" how RX Vega is going to perform is a clueless shitposter since all of Vega's new uarch features remain disabled in Vega FE's gaming drivers.

These include: Advanced Clock Gating and Adaptive Voltage and Frequency Scaling, it should also have Primitive Shaders, Tile Based Rasterization and a High Bandwidth Cache Controller

Proof:
http://www.gamersnexus.net/guides/2990-vega-frontier-edition-undervolt-benchmarks-improve-performance

GN's article proves that Vega's Advanced Clock Gating isn't working at all, and that Adaptive Voltage & Frequency Scaling are in a failsafe mode where 1.2v is being used regardless of clocks. This is why GN was able to increase the performance of Vega FE by undervolting it to 1.09v and ~280w.

http://digiworthy.com/2017/06/30/amd-vega-frontier-tile-based-rasterizer/

Here's proof that Tile Based Rasterization is not working on the current Vega gaming drivers.

http://www.overclock.net/t/1633406/so-what-went-wrong-with-vega/120

HBCC and Primitive Shader are also disabled currently in gaming drivers.

Just to maybe get to to at least TRY and educate yourself a little bit, here is a nice article about what the implementation of TBR meant for Nvidia in going from Kepler to Maxwell:

http://www.anandtech.com/show/10536/nvidia-maxwell-tile-rasterization-analysis

What does this mean? It means that until we see IF and how well RTG's driver team manage to implement these Vega uarch features by the launch at Siggraph it's actually impossible to know how RX Vega performs compared to Vega FE in gaming workloads.

Will RTG's driver team fail to implement ANY of these by Siggraph and it flops at launch? Who knows? How well will any of these actually be implemented by RTG? Who knows? There is way too large a gap between the floor and the ceiling here to make confident predictions.

>>61567500
>Those stock coolers weren't enough to handle Polaris let alone fucking Vega. Why even bother making any retail consumer cards with a stock cooler like that?

Polaris used a 10 cent aluminium block as a "heatsink", if they put a proper, heatpiped block in there, it could handle Vega.

But then, they have a terrible record for stock coolers so I wouldn't get my hopes up.

And third party versions will only ship in late august / early september.

>>61571552
So i can get a high end gpu without having to pay to the company that brought us
>3.5
>vendor locked features
>goyim works
>vendor locked monitors
>gimped performance of products sold in favor of next generation

File: 7e2d7b40aac2ff3ac54afc9a7689ff56---stages-of-grief-grief-counseling[1].jpg (110KB, 634x500px) Image search: [Google]

110KB, 634x500px

>>61571588
So we're at bargaining now, right?

>drivers are not ready at all
Of course, they only had 2 years to make them ready for launch of their 1500$ prosumer flagship, but nothing works. They just need the one extra month to destroy their brand image and add some finishing touches.

>i-it's not a gaming card!
Yet it uses the same core, higher model even has a super beefy CLC. But somehow it isn't comparable to cheaper gaymur versions.

>b-but they're sandbagging!
And that's why they are comparing it to 1080 themselves, right? In official events.

>>61571585
Empty hopes for another bitcoin?
Why would there be an another bitcoin?
The old bitcoin is alive and well.

The whole deal is that while having benefits for miners (immune to ASICs) Ethetium and other crap have ZERO benefits to the end user (drug dealers)
Why would they switch and bother with accepting other coins?

File: amada kokoro.jpg (98KB, 781x1177px) Image search: [Google]

98KB, 781x1177px

post amada kokoro
I NEED MORE AMADA KOKORO

>>61571610
>3.5
True, it was a great card though.

>vendor locked features
Same with AMD

>vendor locked monitors
Same with AMD

>gimped performance of products sold in favor of next generation
Tinfoil hat engaged

Btw I thought the 1080 wasn't even high end according to AMDfriends?

>>61571643
I'm not a miner so I'm not privy to their mentality.
Only thing I know is that people are using a lot of power and hardware for these things nowadays.

>>61571658
>Same with AMD
Name it.
Nvidia can use Opencl
Nvidia could use Freesync and Mantle if the wanted to.
>Hairworks gimps AMD
>AMD tress FX doesnt gimp Nvidia

Also AMD never had disasters like gtx480, drivers killing hardware, drivers crashing the system.

>>61571637
You can feel free to engage with the evidence that those uarch features remain disabled in Vega FE's gaming drivers, or with whether or how successfully any of those features will be implemented by RTG in drivers by Siggraph, or even with what those feature would be worth in terms of performance (I'll note GN got a ~7% performance gain JUST from undervolting Vega to 1.09v and 280 watts) if you are capable of doing so, but I have no interest in entertaining your mindless shitposting.

File: amd_fury-x_28.jpg (116KB, 640x425px) Image search: [Google]

116KB, 640x425px

>>61571653

>>61571713
It's GPU war thread.
It's all mindless shitposting.
Enjoy it!

>>61571637
>well reasoned post with sources
>HURR DURR UR IN DENIAL

Yep, it's /g/

>>61571713
>evidence
Forum conjecture isn't evidence.

File: amada kokoro1.jpg (110KB, 1024x1024px) Image search: [Google]

110KB, 1024x1024px

>>61571719
I always wondered what is the intel-tan. all I have found is deviant art.
>inb4 jew

>>61571793
>testing with B3D suite is not an evidence
?

>>61571710
>Nvidia can use Opencl
AMD didn't invent OpenCL.

>Nvidia could use Freesync and Mantle if the wanted to.
Freesync probably.
Mantle? Sure, they could """use""" Mantle if they started making GCN cards.

>Hairworks gimps AMD
It's not their fault they have a worse tesselation performance
>AMD tress FX doesnt gimp Nvidia
Ditto

>disasters like gtx480
What is Vega?

>drivers killing hardware
Source aside from wccftech?

>drivers crashing the system
You're right, that has never once happened with AMD in recorded history.

>>61571832
>Mantle?
AMD literally proposed jacketman to adopt Mantle.
He refused.
>What is Vega?
GP102 competitor?
>Source aside from wccftech?
Newfag?

>>61571846
>AMD literally proposed jacketman to adopt their proprietary "standard". He refused
Gee, I can't imagine why. He really must be evil.

>GP102 competitor?
That's a weird way to spell GP104.

>>61571869
(You) should try harder next time.

move this thread to /bant/

>>61571658
>Same with AMD
Freesync isn't an actual technology on the monitors. It's a certification. The actual technology is the adaptive sync specification originally found in edp for mobile devices. AMD pushed for the specification to be modified for dp and pushed for vendors to adopt it in monitors. Freesync is a GPU side tech that takes advantage of that and also a brand/label, which is why there have been some non-freesync labeled monitors that were compatible with it. Nvidia could come up with their own solution to take advantage of the same specification without even touching freesync but they won't

>>61571846
>>61571882
it's slower than most 1080s and you somehow think it will be as fast as a 1080ti?
this is some hardcore denial right there

>>61571973
This bait is getting stale.
Make another thread and try to fish for (You)s.

https://www.nordichardware.se/nyheter/radeon-rx-vega-prislapp-sverige.html
This doesn't look good. 730 euros without tax.

>>61572068
But that includes a 30% Khalifate upkeep tax already.

>>61571999
i'm not even that other guy
you must be completely blind to not see that vega is pretty bad, they should fire that indian guy and get someone competent for the job
they had so many years to improve, they fixed their CPUs but for some reason they still suck at making competetive GPUs

>>61572121
This bait is getting really-really stale.

>>61572130
you forgot to add the words nvidia shill to your post

>>61572136
(You)

give me a quick rundown

so far ive got compared to a 1080;

-double the price
-worse performance
-higher TDP

how is this redeemable?

>>61572503
>how is this redeemable?
You can shitpost about it.
You are shitposting about it.
Kill yourself.

>>61571643
bitcoin takes ages to transfer. ethereum doesn't. also the transfer itself has lower fees. another thing about ethereum is the scam hype around it with the empty promises of apps being built on ethereum's gas. most people who buy it have no idea what it means, they just saw some TED talks about blockchain's future and dumped money on ethereum/GPUs
monero's only good value is in its privacy. you can't track it like other crypto. however, it can never be the next fad since it will never have such demand. only pedophiles, murderers, and money launderers use it. simple druggies don't need that shit.

>>61572503
>I believe literally every stupid rumor and leaked whatever I read on the internet: The post

>>61572592
enlighten me on the truth oh wise one

>>61572632
It's gonna beat 1080Ti sli ez.

>>61572687
cool!

>>61572745
Very

>>61572687

*in price only.

>>61571588
>Advanced Clock Gating and Adaptive Voltage and Frequency Scaling,

Fancy marketing names for something chips have been doing for 10+ years now.

>it should also have Primitive Shaders, Tile Based Rasterization

We have no white papers on what those are, only marketing slides. They might be names for 2 bugfixes or a layout change in the rasterizer.

>and a High Bandwidth Cache Controller

That's the HBM2 memory controllers shiny name.

>>61573112
>new shader stage
>2 bugfixes
?

File: radeon vega.jpg (364KB, 917x1469px) Image search: [Google]

364KB, 917x1469px

AMDrones confirmed on suicide watch.

>>61573112
>>Fancy marketing names for something chips have been doing for 10+ years now.
How is that relevant? The point was that the Gamers Nexus undervolting testing clearly and unequivocally shows that neither AVFS nor ACG are currently working in the Vega gaming drivers.

>We have no white papers on what those are, only marketing slides. They might be names for 2 bugfixes or a layout change in the rasterizer.
Now this is olympic level shitposting, betting on the ignorance of the /v/iddiots reading this to not know or look up that TBR has an established meaning and was the sine qua non of the kepler -> maxwell jump for Nvidia.

>That's the HBM2 memory controllers shiny name
The feature that Raja already demonstrated in a previous demo and yet isn't working in gaming on Vega gaming drivers yet, you mean?

http://digiworthy.com/2017/02/28/amd-radeon-rx-vega-hbcc/

>>61573255
top kek tb honestly

>>61566654
>Had cool looking stock cooler for vega
>Replace it with Polaris cooler Team Rocket edition
Gay.

>>61574076
That's an engineering sample, it's possible that the retail cooler looks different.

>>61572136
Don't worry
He does this every thread

When he runs out of arguments he just starts (You) posting to hide the fact that it won't come close to competing with the 1080ti. It's not worth your time

>>61573255
Gonna steal this if it's cool with you bud

>>61572068
If it's higher priced than a 1080ti it's also going to be faster. Basic logic.

File: tenor.gif (4MB, 400x253px) Image search: [Google]

4MB, 400x253px

>>61575182