Will AMD release any high performance GPU's in Q2 or Q3?

[Bigvizz]

I think you don't realize that much of that is because Nvidia simply has shitton more supply.

AMD simply has no resources to put up the same stock as Nvidia, after all it's no secret that Polaris is sold out.

Sounds like a AMD problem huh? Nvidia not going to stop producing cards just to let AMD catch up and all the while Nvidia gobbles up more of the gaming market pushing AMD closer to what I said earlier. At this point it's not IF, but WHEN AMD loses the PC gaming market entirely. /Tick Tock

[Gaidax]

I think you guys do not understand what I say and I am saying all that not because I'm some AMD fanboy, but simply because it's reality.

AMD simply is not big enough to put out any sort of decent stock. Their midrange is actually pretty decent, they only really fall off at 1070+ range.


I, personally, use 1080Ti so I could not care less about the 1060/580 range of equipment, but the fact I have Nvidia in my rig does not suddenly make me blind to reality there.

[Denpepe]

consoles / mobile

vs tiny pc market.

I do not consider this tiny

https://arstechnica.com/gaming/2017/...than-expected/
https://newzoo.com/insights/articles...ile-taking-42/

[Evildeffy]

Actually if talking in pure production capacity AMD can easily outdo nVidia.

Whilst AMD no longer owns/have a majority stake in GlobalFoundries they still command most of it's production as well as a large chunk of TSMC's production.
In sheer numbers AMD does have the advantage.

What they lack are products.

Also when it comes to GPUs vs. PS4/XBX ... I wouldn't jump the gun quite so quickly on that nVidia has sold more GPUs than all AMD-enabled consoles together in the last 3 years.
Having said that AMD is too big a player in the market, they will not and cannot withdraw.

If anything AMD's CPU division was on it's deathbed before Ryzen and look how well they are doing.
AMD's Radeon Technology Group actually had a lot of leeway and funds still so they can go on for a long time without issue.

In general you people think of doom, gloom and disaster scenarios far too quickly.

[Life-Binder]

large chunk of TSMC ? I doubt that very much



anyway, GPU IPs are too important for AMD and they will keep making console SoCs with GPUs, as well as mid-range GPUs and mobile GPUs (whether those will be power efficient enough to put into laptops is another matter)

they may or may not keep trying to compete at the $400-500+ PC gaming GPU market .. but they definitely want to keep making powerful GPUs for HPC and AI markets at least .. and logically if they already have high-end compute market chips - they can try to repurpose them into high-end gaming cards .. although that may not end so well like with Vega, since compute/pro and gaming graphics want different things from a chip and making one jack-of-all-trades seems too ambituous

[Evildeffy]

large chunk of TSMC ? I doubt that very much

Everything prior to Ryzen in the CPU department was produced between TSMC and GlobalFoundries with TSMC having the bigger share.

Everything prior to Polaris/Vega in the GPU department was produced solely by TSMC.

For example:
AMD produces a great crapton of lower power CPUs (like Intel's ATOM) that you are unaware of and goes in business devices.
These are still produced by TSMC alone and in VERY large quantities.

At some point AMD delivered 20% of the ENTIRE CPU market with JUST the AMD Fusion APU, the low power E-350 to E-450 models.
AMD's production capabilities easily outclass that of nVidia if they so desire but like I said...

They don't have the products to do so right now. (what I mean by this is across the entire GPU spectrum, as Ryzen is covering the CPU spectrum pretty damn well)

[Thunderball]

Their "High-end rebrand" which you so fault AMD for as an example, you want to blame one but not the other?
Funny how you described the G92 as low-end since it only became that with the last jump to GT250 and not before.

I value transparency. I dont want chips RENAMED to hide that it's the same chip. It sucks that Nvidia didnt have anything better to offer at that time, but you at least knew what you get. G92 was a top chip for one generation: 9000 series (8000 launched with a G80 chip originally), G92b is a different chip.

Wrong again, GM107 and 108 were which were identified as the GTX 750 and GTX 750Ti.
There was never a commercial GM100 chip, Maxwell was prototyped as GM107/108 only.

I know what thay launched with it first. I sincerely hope you understand that it always starts with top silicon, that gets cut down to make cheaper cards later.

Sure the logic is there, nVidia wanted more money if you were using DP, there was no technical reason to.
Also no-one, at any point in time, suggested crippled Vega FE, what was suggested was rushed launch, broken drivers, firmware immaturity etc.
Not all that uncommon, neither of which is calling out to wait and see what they have in under 2 weeks instead of assuming it's 100% shit regardless of how much of an educated guess you may throw at it.

Fixing all of those cannot attribute to the deficit Vega faces right now, and you know that perfectly. Noone is saying that it's 100% shit, people are saying that given the power consumption differences and the fact and it's competing with a card that's out for over a year now doesnt bode well for Radeon Technologies. Vega doesnt have the potential to change the market situation in the highend segment, which is all I care about. This is honestly Fury X all over again, Fury X was a good card, but as irrelevant on a market scale as it could possibly be.

Granted I did assume because of the fact of them naming a major graphics card company (there's only 2, technically 3 counting Intel but they don't use GDDR) and out of those 2 who's more likely to use GDDR6?

https://www.skhynix.com/eng/pr/press...ew.do?seq=2086

Go, find Nvidia in there. They didnt name anyone, it's the press who automatically assumed that Nvidia will use that memory (which is not even guaranteed to be the first GDDR6 chip to be commercially available) due to Hynix doing a presentation at GTC (which is an Nvidia conference). Again, all 3 major memory suppliers announced that they will be making GDDR6, all 3 estimating late 2017 - early 2018 availability. Sure, Nvidia stated that they will use GDDR6 for Volta but that doesnt mean that it's a subject to change: earlier Volta was expected to use HBM.

it would be highly expensive

Again, why? Using GDDR5 and GDDR5X controllers not too expensive, but GDDR5X and GDDR6 is too expensive?

Ok so... leaked slides that should never have reached public are AMD's creation of the hype where actually showing it at their actual events which are meant for the public is not and by no means an indicator of their performance?
Including by releasing the same "test" you so abhor into the wild for people to do their own math and confirm it.
It's funny how dismissive this has all of a sudden become with your arguments being refuted.

Dont leak them yourself if you dont want them to be seen. Also, I never argumented on Ryzen performance, you're twisting arguments here again. You wanted to see where the hype comes from, I'm showing you. Other results are irrelevant because they showed actual test numbers without "this processor is better than that processor" (most of which ended up being bullshit anyway due to AMD's testing setups, check Tomshardware review for example). Their presentation and performance comparison on both press event and launch event was based on Cinebench numbers, which are honestly completely irrelevant, that's hype.

Really? Intel seems to love hyping their new "Mesh" core communication design .. or the fact that they have finally built a monolithic die for their 28C behemoth... but I guess those aren't technical details...
However on the name... I do agree.

Noone is talking about it except you. "Better for VR" argument would've been a lot more valid.

Actually yes there was quite a bit of hype and like you could see a lot of controversy and failure for the X299 launch, not to mention it being drowned out entirely by the showing of the X399 boards for ThreadRipper.
Also X299 cannot be compared to EPYC (Naples) as Intel's counterpart to(the Xeon chips) use LGA3647 and not LGA2066 like X299 does.

The hype Intel tried to create was solely and utterly destroyed and drowned out by what AMD is set to release, which in a way worked out for them better as if there'd be more attention to the failed power delivery design of X299 ,or the insanely stupid amount of power draw from their CPUs where Intel has actually classed their own CPUs as lower TDP than they are factually putting out, would be way more devastating than it already is.

Which hype? Created by Intel? Are you serious? Also, noone is comparing Naples to Skylake-EX/EP, both are not out, there is nothing to compare. Honestly AMD's EPYC claims look even more ridiculous than Ryzen, but those actually have a possibility to be true, so I wouldnt dismiss them as Intel seems to be doing. But server market is quite different, there a lot more going into it than performance, potential buyers in that market just dont have the same priorities are PC enthusiasts do.

- - - Updated - - -

Actually if talking in pure production capacity AMD can easily outdo nVidia.

Whilst AMD no longer owns/have a majority stake in GlobalFoundries they still command most of it's production as well as a large chunk of TSMC's production.
In sheer numbers AMD does have the advantage.

AMD doesnt have any stake at GlobalFoundries. They also dont command any production of TSMC (their TSMC production is commanded by people who ordered the development and production of those chips), so theoretically if GlobalFoundries kick them out there is no way AMD can make any chips.

[Evildeffy]

I value transparency. I dont want chips RENAMED to hide that it's the same chip. It sucks that Nvidia didnt have anything better to offer at that time, but you at least knew what you get. G92 was a top chip for one generation: 9000 series (8000 launched with a G80 chip originally), G92b is a different chip.

G92 was top of the line for the 8800 models, namely the 8800 GTS.
It also offered the 8800 GT and 8800 GS which were all G92 chip.
As far as "knowing" what you got... bollocks... both companies don't tell you upfront what you're getting, it's thanks to savvy tech users we figure it out for the most part until some spec whitepaper is released by nVidia which in itself is a rarity.

However whether you or I like or dislike rebrands they make perfect sense from a business point of view.
The cost of developing an entirely new uArch each generation with 1 year apart is extremely costly and was actually 1 of the reasons ATi (before AMD) was bleeding money.

I know what thay launched with it first. I sincerely hope you understand that it always starts with top silicon, that gets cut down to make cheaper cards later.

Wrong again, the Maxwell architecture in the GM107/GM108 series was a prototype to see how it'd run, the GM100 designs were there but they remained designs only, there was never a prototype, they simply used the GM107/GM108 as a stepping stone, testing out the uArch and designing the next generation around it.
Also starting off with the top silicon is also a myth, it's something done because AMD/ATi started that line and nVidia followed because it was a marketing advantage, start with top of the line and grab headlines, deal with smaller later.
nVidia originally started with the "smaller" dies.

Fixing all of those cannot attribute to the deficit Vega faces right now, and you know that perfectly. Noone is saying that it's 100% shit, people are saying that given the power consumption differences and the fact and it's competing with a card that's out for over a year now doesnt bode well for Radeon Technologies. Vega doesnt have the potential to change the market situation in the highend segment, which is all I care about. This is honestly Fury X all over again, Fury X was a good card, but as irrelevant on a market scale as it could possibly be.

And my point, which you still don't seem to understand, is that you still don't know all the details.
You don't know what RX Vega is going to cost you.
You don't know exactly how it'll perform.
You don't know (for a fact) yet what it's pitted against.

Let me put it in a different light for you: Say the RX Vega is 10% faster than a GTX 1080, has a 285W TDP and costs 400 USD (current rumour, use it in a hypothetical situation) ... would it still not have an impact?

Answer:
Of course it would, even at increased power consumption it would still take a chunk away from possible 1080 owners or new builders.
Will this change the highest end card demographic? Not even in the slightest.

Does that mean just because RX Vega is late to the party it has no place and should be abandoned?
Hell no... it's an entry into the high-end, experience in that uArch with memory subsystem and simple presence in the market.

This is why I keep telling you to wait until SIGGRAPH because you're already declaring it dead before knowing anything in detail and that is a dangerous and stupid move to do.

https://www.skhynix.com/eng/pr/press...ew.do?seq=2086

Go, find Nvidia in there. They didnt name anyone, it's the press who automatically assumed that Nvidia will use that memory (which is not even guaranteed to be the first GDDR6 chip to be commercially available) due to Hynix doing a presentation at GTC (which is an Nvidia conference). Again, all 3 major memory suppliers announced that they will be making GDDR6, all 3 estimating late 2017 - early 2018 availability. Sure, Nvidia stated that they will use GDDR6 for Volta but that doesnt mean that it's a subject to change: earlier Volta was expected to use HBM.

Doesn't change my statement... has AMD announced they would use GDDR6 directly on availability or do they want to continue with HBM2?
Considering there are no other players using GDDR memory it is a correct assumption to think nVidia is that very client.
Yes if you look back on nVidia's roadmaps you'll see HBM there but that was several years ago (2 IIRC) and they've announced later (but still well before launch) to be using 5X and 6 instead so stating that Volta originally was planned with HBM2 is rather irrelevant.
And you've just reinforced my earlier point regarding GDDR6 and GDDR5X particularly.. GDDR5X is Micron only, GDDR6 ... look at that ... competition.

So I'll ask you again... who's more likely to use GDDR6 at this point? nVidia or AMD?

Again, why? Using GDDR5 and GDDR5X controllers not too expensive, but GDDR5X and GDDR6 is too expensive?

Since you and I cannot see eye-to-eye on this particular matter I'll simply say "refer to our earlier discussions on this point" and leave it at that.
uArch design is expensive, especially if it needs to be developed, tested and validated for every single variation.

Dont leak them yourself if you dont want them to be seen. Also, I never argumented on Ryzen performance, you're twisting arguments here again. You wanted to see where the hype comes from, I'm showing you. Other results are irrelevant because they showed actual test numbers without "this processor is better than that processor" (most of which ended up being bullshit anyway due to AMD's testing setups, check Tomshardware review for example). Their presentation and performance comparison on both press event and launch event was based on Cinebench numbers, which are honestly completely irrelevant, that's hype.

Cool ... so we're blaming press leaks on AMD now, nice.
Also I'm twisting arguments again? Funny ... my prior statement stands to this point still.
Also it's hilariously obvious you've not seen, interpreted nor given a crap about the presentations .. that much is obvious.
But I would GLADLY read your Tom's Hardware Review that you're referring to about them being bogus, please give me a link to see where Tom's Hardware states that.

Noone is talking about it except you. "Better for VR" argument would've been a lot more valid.

Funny how I'm the only one talking about it ... I'm sure it never appeared in Intel's (EXTREME MEGATASKING) marketing slides.
Gamers Nexus has quite a few videos up on this, have a look.
Or the whole "Up To" memes out for Intel's Tech Specs... but no, no-one but me talks about it... sure.

Which hype? Created by Intel? Are you serious? Also, noone is comparing Naples to Skylake-EX/EP, both are not out, there is nothing to compare. Honestly AMD's EPYC claims look even more ridiculous than Ryzen, but those actually have a possibility to be true, so I wouldnt dismiss them as Intel seems to be doing. But server market is quite different, there a lot more going into it than performance, potential buyers in that market just dont have the same priorities are PC enthusiasts do.

The fact you're unaware of this makes you either completely oblivious to marketing or you simply ignore it, I don't know which and I'm not going to guess.
That said both AMD EPYC (Naples) and Skylake-EP/SP are actually out and can be bought by companies and AnandTech even has a review out, the claims are not so ridiculous at all.

PC enthusiasts in general want a Bugatti Veyron or McLaren F1 where server markets can vary from a small lorry to a "road train", this still doesn't change the fact that Intel has, for the first time ever, violated their own TDP rules and have a chip on their hands that simply was never meant for this market in the first place but still try to cater to it, near 400W CPU power draw for a 10C overclock is not OK and considering there's still an 18C to come... it's just going to get worse.

AMD doesnt have any stake at GlobalFoundries. They also dont command any production of TSMC (their TSMC production is commanded by people who ordered the development and production of those chips), so theoretically if GlobalFoundries kick them out there is no way AMD can make any chips.

Wordplay... funny.

TSMC is still doing a lot of business with AMD and is still producing their technology on a pretty large scale so yes "AMD commands" still a large chunk of TSMC's output capability and as far as GlobalFoundries is concerned... if they "kick" AMD out they would instantly go bankrupt with all the clauses and lawsuits they'd face.
Since x86 technology is still licensed to AMD and not GlobalFoundries, as are IBM's orders from GlobalFoundries as well as the graphics technologies.
GlobalFoundries also uses the 14nm FinFET process from Samsung which is licensed yet again to AMD and IBM does the same with their 7nm process to AMD.
So them "kicking out" AMD is pretty much impossible at this point.

So yes AMD actually still has a larger production capacity than nVidia if they so desire... the problem is they do not have the products to ramp it up to full.
Though yes "Theoretically" if GlobalFoundries wishes to go bankrupt and take the tech industry with them, they have a shot at doing so and heavily reduce AMD's output capacity... this still doesn't change that AMD still has a large output going with TSMC.
Hell they even have a clause in their contracts with GlobalFoundries that if they should ever so require it that TSMC takes over production partially or entirely if GlobalFoundries cannot deliver promised products in time.

AMD got burned by this a while back, they aren't taking another chance like that again.

I could continue with naming things you specifically skipped mentioning after being told otherwise but it'd be pointless.. you'd accuse me of twisting arguments again whilst trying to cherry pick responding statements... Pot meet Kettle.
So I'll do myself, and others on this forum, a favour and end the argument between you and me with this response.
Therefore you will no longer be seeing any response regarding the topic you and I are currently discussing from me, whilst others will.
That means this is your chance, you could write the most ridiculous stuff there is and I'd ignore it making you seemingly to have the last word even though factually it is not... don't waste it now... I don't give many people this opportunity.

[Bigvizz]

Here we go again.....

[moremana]

Here we go again.....

Do you read their posts?

Its kind of like watching Siskel and Ebert. It makes for good reading and they both are basically doing all the research for you. I dont see it as bickering, I see it as constructive points from 2 different minds.

[Denpepe]

Do you read their posts?

Its kind of like watching Siskel and Ebert. It makes for good reading and they both are basically doing all the research for you. I dont see it as bickering, I see it as constructive points from 2 different minds.

Too much wall of text for me, then again my eyesight is crap so that does not help. Kinda sad they go on like this, there might be good info in there, just cba to keep following it.

[Life-Binder]

http://www.guru3d.com/news-story/amd...in-3dmark.html
https://videocardz.com/71090/amd-rad...ke-performance

RX Vega Firestrike (driver technically unknown, but one would assume its the RX Vega driver, which only has 1-2 weeks left to get finished)

WhyCry VideoCardz.com • 3 hours ago

Something I didn't verify yet, so take it with a grain of salt. Rumor has it it will launch on 8th August. No custom cards at launch.

[Bigvizz]

Do you read their posts?

Its kind of like watching Siskel and Ebert. It makes for good reading and they both are basically doing all the research for you. I dont see it as bickering, I see it as constructive points from 2 different minds.

No I rather complain, watching Evildeffy argue with anyone is an exercise in patience.

User was infracted

[Deleted]

Well hopefully this doesn't get buried, for radeon owners, keep an eye out for drivers tommorow;

https://twitter.com/CatalystMaker/st...46633575780352

AMD have been delivering on the driver front pretty well recently so its worth a watch.

[Gaidax]

Having recently switched from AMD to Nvidia, my only complaint is their absolutely ancient NVIDIA Control Panel, I swear that shit looks absolutely the same as it looked back when I had frikkin' 8800GTS and GTX280.

Everything else is sweet and Geforce Experience with Shadowplay are brilliant, IMO.

At least AMD did a good work on their old shit control panel with crimson update, too bad they don't have any GPU worth my time for a couple of years now.

[Pickynerd]

Tiny PC market? In hardware terms? I think Nvidia has shipped more GPU's then PS4's sold in the past 3.5 years since the consoles launch.

- - - Updated - - -

Yeah, just ignore mobile and jump on the ps4's sales alone.

[Life-Binder]

what about mobile ?

[Evildeffy]

No I rather complain, watching Evildeffy argue with anyone is an exercise in patience.

So you would rather complain than be informed properly with information or be misinformed with faulty information than consider what I have to say constructive even though, as post history so far has continually indicated, I am almost always correct in my predictions and acknowledge when I'm lacking?

And specifically mentioning me because I wrote detailed well thought out and constructive posts as a hassle... hey guess what .. don't like it? Don't read it!
And most certainly don't respond to it when you're going to be as childish as this.

Well done, that's a good way to go through life without actually having to think for yourself a single second.

[Gray_Matter]

Having recently switched from AMD to Nvidia, my only complaint is their absolutely ancient NVIDIA Control Panel, I swear that shit looks absolutely the same as it looked back when I had frikkin' 8800GTS and GTX280.

Everything else is sweet and Geforce Experience with Shadowplay are brilliant, IMO.

At least AMD did a good work on their old shit control panel with crimson update, too bad they don't have any GPU worth my time for a couple of years now.

I have both at home and my pet peeve with Nvidia is that you can't get any of the features with GForce Experience unless you register and log in. That and there are some serious issues using their controller with Windows.

[Bigvizz]

So you would rather complain than be informed properly with information or be misinformed with faulty information than consider what I have to say constructive even though, as post history so far has continually indicated, I am almost always correct in my predictions and acknowledge when I'm lacking?

And specifically mentioning me because I wrote detailed well thought out and constructive posts as a hassle... hey guess what .. don't like it? Don't read it!
And most certainly don't respond to it when you're going to be as childish as this.

Well done, that's a good way to go through life without actually having to think for yourself a single second.

Just like hassling you cause it's fun and you take yourself so seriously. Jeez, did I hit button.

Oh, btw your posts tldr....