• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Capsaicin live stream

CAT-THE-FIFTH said:
There is a THIRD Polaris chip??
Click to expand...

They didn't state how many there are, but they said that 10 & 11 are just the first two, and when asked if we had to wait until Vega for a new high end chip, he said no and that they'd show faster stuff later. Like I say, I don't see how a bigger chip than P10 could work without HBM1, given that HBM2 is too expensive / low supply until '17 Q1 and Vega / NV Titan product ... it'd be so bandwidth starved with GDDR5 unless it had something stupid like 16GB .. but then the PCB would be gigantic and cost / power / heat advantages would be lost entirely.
 
Last edited:
pmc25 said:
They didn't state how many there are, but they said that 10 & 11 are just the first two, and when asked if we had to wait until Vega for a new high end chip, he said no and that they'd show faster stuff later. Like I say, I don't see how a bigger chip than P10 could work without HBM1, given that HBM2 is too expensive / low supply until '17 Q1 and Vega / NV Titan product ... it'd be so bandwidth starved with GDDR5 unless it had something stupid like 16GB .. but then the PCB would be gigantic and cost / power / heat advantages would be lost entirely.
Click to expand...

Wow! So AMD is going on full steam - so they will have a £100 to £500 stack and then 9 months later a new pair of GPUs??

Hoping this means RTG is far more proactive in competing.
 
CAT-THE-FIFTH said:
Wow! So AMD is going on full steam - so they will have a £100 to £500 stack and then 9 months later a new pair of GPUs??

Hoping this means RTG is far more proactive in competing.
Click to expand...

Right now they have the perfect window - new node - new memory. Nvidia's memory controllers have never been as good as AMD's when it comes to new memory - they always have to catch up/brute force it. Second they always have issues on doing die shrinks.

AMD comes out full guns blazing - full stack; then bring another full stage 9 to 12 months later - they will taken all the pressure off themselves and put it back on Nvidia. We've seen how Nvidia reacts to pressure. Along with the fact they are really starting to feel it in professional realm. This is Nvidia's market share there has ever been - they've never been pressured in the professional market like they are now. Between AMD and Intel - Nvidia'a bread and butter; aka milking the market there isn't as easy as it was 5+ years ago.

Polaris isn't make or break though for AMD - I do think it will be smash; they will also need to hit it out of the park with Zen - if they are both as good as they possibly could be....late 2016/2017 will be awesome years for AMD
 
andybird123 said:
why? Micron are said to be mass producing it already, why would they be making it in live production if it had no customers to go to?
Click to expand...

They aren't mass producing it for one thing, they said they plan to mass produce it starting at some point in summer, this puts the most realistic time frame for the first gpus with gddr5x as maybe Q4 or next year.
 
Orangey said:
...meanwhile, back in reality...
Click to expand...

This.

With adequate async allowing the furyx to use all of its shaders and up to its theoretical performance, it beats down overclocked 980ti's. All while Fiji is running stock.

HBM is needed as gpus gain more shaders. HBM allows parrallel access by different shaders to different memory blocks simultaneously. With gddr the reads and writes are queued to each gddr5 die.

Polaris should be able to make use of its theoretical performance, so even having fewer shaders than Fiji it should perform better in the majority of cases. So gddr5 should not be as limiting in this regard.

The 980ti has around 3000 cuda cores while Fiji has 4000 odd.
 
Mauller said:
This.

With adequate async allowing the furyx to use all of its shaders and up to its theoretical performance, it beats down overclocked 980ti's. All while Fiji is running stock.

HBM is needed as gpus gain more shaders. HBM allows parrallel access by different shaders to different memory blocks simultaneously. With gddr the reads and writes are queued to each gddr5 die.

Polaris should be able to make use of its theoretical performance, so even having fewer shaders than Fiji it should perform better in the majority of cases. So gddr5 should not be as limiting in this regard.

The 980ti has around 3000 cuda cores while Fiji has 4000 odd.
Click to expand...

I don't think Polaris would suffer one bit from using the 390/x's 512 bit Vram, They'd probably of been better of using it instead of HBM1 on Fiji too considering how little using HBM has helped it. It certainly doesn't pull away from Grenada much and not in any way you can put down to the memory (ie: Grenada gains as much ground as Fiji at higher res's overall).
 
Someone didn't get an invite?

AMD fishes for hollow headlines with Capsaicin
Spinning masterclass with nothing of substance
Mar 16, 2016 by Charlie Demerjian
2015 AMD LogoAMD is fishing for headlines with hollow teasers void of real substance. This time around it is a dual Fury GPU, a VR headset, and a VR-ready PC.

All of these things have two common themes, they are not actually available to anyone, and are only vague specs meant to grab headlines from sites that aren’t bothered by hollow BS. What is the problem with the announcements? First is that they were done with no notice to the press while being spread across a few hours of a painfully worthless ‘event’ called Capsaicin. More problematic is the content of these ‘news’ releases, or lack thereof.

AMD of late is all about grabbing social media hits as a stand-in for actual news. Social media ‘events’ are manufactured, shiny, devoid of anything but a splashy headline, and give serious outlets absolutely no chance to ask questions. This isn’t by chance. If the press actually asks questions of AMD, they will very quickly see how hollow those headlines are. AMD’s way of spinning the fact that they are shoveling crap is to do it via channels that will eat it up and smile knowing the real sites are obliged to follow or appear left out.

Take for example the dual Fury card, something that you had to have the misfortune to sit through a few hours of the AMD Capsaicin ‘event’ to hear about. What was said? The ‘release’ of this card was accompanied by massive technical detail, 2x Fiji GPUs, 8GB of HBM memory, 3x DP and 1x HDMI ports, and 16TFLOPS, or things that were all known. That’s it, no detail, no ability to ask questions, and all facts that were known.

Price? Who knows. Availability? When SemiAccurate showed you this card last June we had all those details, and probably more were said in confidence. What we can say was that card was supposed to be out in the fall of 2015. That became late 2015, then CES, then the end of January, then nothing until it was sprung on us with hours of notice. We couldn’t ask simply questions like, “Why was this dog woefully late?”, and “Will it be priced not to sell like your other dual GPUs?”. From there a few good follow-ups would be, “But seriously, if it actually becomes available, can we get a real date?”, and “Should we read anything into the fact that you didn’t sample the press at what is ostensibly the ‘launch’, like maybe that it is a dog?”.

Don’t think we are intoning that this card is effectively both vapor and DOA because we are not, we are stating that outright. Three quarters ago a dual Fiji would have made sense, now it doesn’t, and at the absurd prices SemiAccurate is hearing, it definitely is not going to fly, if it ever actually is released that is. Like the launch vehicle, this card is meant to grab shallow headlines and will not move the marketshare needle for AMD, two of a market failure does not a winner make.

Then there is the VR headset, it may be a cool device but since it is backed by AMD, it is doomed to failure. The company simply does not have the market traction to make inroads with their own VR technology even if it is showstoppingly great, Oculus marketing execs are not going to lose sleep over this one, trust me on this ‘prediction’. Backed by the awesome power of an AMD APU, at extremely low wattages, this this has all the brute power to almost pull the skin off a pudding, with a lot of help and a horse or two. About the only decent thing you can say about it is the drivers are less broken than Intel’s, but that is not high praise. Another headline grabbing attempt that no one can ask questions about, no technical details were released, and no press will ever get the chance to independently evaluate one before it fades away. But the coveted headlines will be gotten.

Last up is the AMD VR-ready PC which is effectively a Maingear system with a bunch of high end components. Nothing against Maingear, they make a fine system, but does AMD really think anyone is going to buy a VR-spec PC because they blessed it? AMD’s stamp of approval vs Oculus’ or Steam’s, kind of hard to pick a winner there, or even two, isn’t it? They are all so close in gravitas that there is almost nothing between them. Snicker. But it did grab social media headlines which were aided by an inability to ask questions and a complete lack of accompanying data. “Echo” says the echo chamber.

In summary, AMD is carrying on their traditions of worthless ‘events’ designed to hide the truth and keep people from asking questions. Shiny things are tossed around, the tame dance, and metrics about retweets and likes are somehow tallied into successes. Unfortunately the truth and real information are victims here. Did anyone write about how late the duo was, how pointless an AMD VR stamp of approval is, and how much traction an AMD blessed headset will get? Nope. Our apologies to those who wasted a few hours of their lives with another pointless headline grab ‘event’ even if it had a catchy name like Capsaicin. At least we didn’t have to sit through another turkey this time, the last one was scarring enough.S|A
Click to expand...

https://semiaccurate.com/2016/03/16/amd-fishes-hollow-headlines-capsaicin/
 
nashathedog said:
I don't think Polaris would suffer one bit from using the 390/x's 512 bit Vram, They'd probably of been better of using it instead of HBM1 on Fiji too considering how little using HBM has helped it. It certainly doesn't pull away from Grenada much and not in any way you can put down to the memory (ie: Grenada gains as much ground as Fiji at higher res's overall).
Click to expand...

I think fiji would have burst into flames with all those GDDR5 chips around it ;)
 
tommybhoy said:
Someone didn't get an invite?



https://semiaccurate.com/2016/03/16/amd-fishes-hollow-headlines-capsaicin/
Click to expand...

He is kinda right though, isn't he :) It was cool to see more info on Polaris and new announcements of new games and collabs, but who here was so excited to hear something new about VR or Radeon Pro Duo?

I for one was watching stream for Polaris demo and game announcements. The rest of the fluff in the event was just that. I completely understand Charlies disappointment in the event, since AMD was going for attention of game devs. And as I said before event was run by engineers :D
 
These events suck, fans think these events are for them so they are becoming less about information and more about tech people trying to be cool because random tech geeks think they should present more like Apple. Actual tech talks and presentations held to industry people 3+ years ago were great, then they were streamed, great, then idiots on forums started criticising them for not being flashy and instead of talking in depth about the hardware to industry people they started doing these things for end users and they are pointless.

Nvidia's tech thing at CES was barely watchable, "look at this video... of a car going down a motorway, woooooo".

Even before this event same stuff, people expecting benchmarks, real information despite not being a Polaris launch. People have no self control, no common sense and no perspective. They expect more so AMD and Nvidia are making these events longer.... for absolutely no reason except people started watching industry presentations and expecting information for the end users and complaining that it wasn't flashy, cool or everything they randomly chose to believe it was going to be.
 
muziqaz said:
He is kinda right though, isn't he :)
Click to expand...

Oh yes he was right about there being no notice to the press, (the fact that certain tech sites, were in attendance and they knew there were going to be in attendance last week, seeing as Ryan Shrout told everybody where he would be in their weekly podcast.)

He was right about, no ability to ask questions, ( please ignore the earlier video of Ryan Shrout interviewing Raja Koduri and that is just the one interview that I know about)

He was right about not knowing the price and availibility of the Radeon pro duo..er ok maybe not.

Oh look
The AMD Radeon Pro Duo is planned for availability in early Q2 2016 at an anticipated SEP of $1499 USD.
Click to expand...
Source: http://www.amd.com/en-us/press-releases/Pages/amd-radeon-pro-2016mar14.aspx

The only thing he is right about is the VR headset probably is doomed to failure, as in my opinion it isn't powerful enough to play the Occulus/Vive targeted games and its a windows platform so no good for all the phone style games.
 
You must log in or register to reply here.
Back
Top Bottom