390X coming soon few weeks

Zarathustra[H];1041643465 said:
I obviously don't have my launch dates perfectly.memorized, but was the 290x even readily available at retailers by the time the 780ti was launched? They got at most a month, probably less out of that one before they were leapfrogged again, so IMHO I don't feel it really counts.

Well I counts though I think I was 2 weeks.
 
yes, but the damn miners had snapped them all up and the retailers were price gouging like no ones business.

took me about a month after release to finally get one.

IIRK, AMD finally put their foot down and told any retailer who charged over MSRP would loose their shipments.
 
Zarathustra[H];1041643418 said:
So its been a while since we have seen a performance crown on anything AMD. I'm hoping that happens with Fiji XT :)
No, it hasn't been awhile. The 5870, 7970 and 290X were all the fastest single GPU cards available at the time they were launched.
 
Well I counts though I think I was 2 weeks.

Well, to me it doesn't count if all AMD did was push a product across the finish line slightly before Nvidia only to get beaten again within a few weeks to a couple of months of the release.

To confidently say "performance crown" I'd like to see it be generation long, or at least 6 months.

I didn't realize that the 7970 OC models were reliably beating the 680 though, so i guess the 7970 was the last one then, because - as mentioned - I just can't bring myself to call having a back ordered product very few could actually get, be faster than the competition for a couple of weeks, like the 290x, a real performance crown.
 
But what about AMD developed features like HDMI FreeSync? Wouldn't that be reason enough to at least consider an AMD solution even if it was 2 frames less than the competition? FreeSync in your living room could be a really big thing especially with the Steam Machines launch and the possibility of consoles getting it.

The Displayport FreeSync was backwards compatible with the GCN architecture. I hope this one will be also.
 
But what about AMD developed features like HDMI FreeSync? Wouldn't that be reason enough to at least consider an AMD solution even if it was 2 frames less than the competition? FreeSync in your living room could be a really big thing especially with the Steam Machines launch and the possibility of consoles getting it.

The Displayport FreeSync was backwards compatible with the GCN architecture. I hope this one will be also.

While FreeSync on TVs for consoles and Steam machines would be cool, I do not think that will happen in the lifetime of your 390X/Fury. Look how long it has taken to get it on monitors and we are still waiting for a good implementation of it. Monitor manufacturers are learning they need quality scalars. Maybe they can directly pass that over to TVs, but maybe not. That said, I doubt GSync will ever be on a TV due to licensing and the way NV scrutinizes the implementation as well as added cost of their extra hardware. Though I would never say never.
 
Well I counts though I think I was 2 weeks.


Less than that when you factor in the first 3+ months the supply was drained and prices were artificially higher actually helping sell the 780/780Ti. The 290/290X became almost a laughing stock as quickly as it became hyped. Not just because of performance against Nvidia, but due to heat, power consumption, and supply/pricing issues for months before it could be resolved and aftermarket coolers came in to alleviate the bad press.

AMD had 2 critical release errors in the past 4 years. They waited 2 years to do what was supposed to be a "Sea Islands" refresh claiming the "Never Settle Bundle" was making them boat loads of money that they didn't need to do a refresh. Remember that controversial leaked roadmap without a new generation GPU in it?

Second is what they're doing right now. Waiting almost another two years before making a significant splash giving Nvidia plenty of time to strategize a way to maximize their products even on the off chance it gets tanked by an AMD product. This is why Nvidia has boat loads of money and AMD doesn't. If you include these Titan like products that ultimately become a semi-refresh, we're basically seeing a generation and refresh from Nvidia every 1 1/2 years now.
 
Zarathustra[H];1041643333 said:
The major selling point for these products are fivefold, and four of them are engineering specifications:

1.) Performance

2.) Stability

3.) Features

4.) Power use; and

5.) Price


Nothing else matters.

New tech matters.
I rather buy a HBM card than a GDDr5 card especially at high end.
Looking for a card for me and nothing nvida is doing appeals for me but AMD Fury has me excited like nothing else the last few years.

Its why you see so many posts on all forums about the Fury not the 980ti or TX.
 
While FreeSync on TVs for consoles and Steam machines would be cool, I do not think that will happen in the lifetime of your 390X/Fury. Look how long it has taken to get it on monitors and we are still waiting for a good implementation of it. Monitor manufacturers are learning they need quality scalars. Maybe they can directly pass that over to TVs, but maybe not. That said, I doubt GSync will ever be on a TV due to licensing and the way NV scrutinizes the implementation as well as added cost of their extra hardware. Though I would never say never.

I thought FreeSync took less than 2 years to implement and the monitors that we are seeing on the market are generation one. I would assume that improvements will happen with generation 2. Also FreeSync was backwards compatible with multiple generations of GCN cards. Well at least it is supposed to be with my 7950's and R9 290. I don't have a FreeSync monitor to test with. :)

I don't consider a couple of years to get new tech out a long time; then again I'm getting old. How long has AMD been developing HBM? Even Nvidia announced support for HBM on their 2016 cards. Can't be but so bad if your rival uses your tech on their future products. :)

I'm looking forward to HDMI standardized FreeSync. It would synergize nicely with the future consoles and TVs.
 
New tech matters.
I rather buy a HBM card than a GDDr5 card especially at high end.
If that new tech has no practical benefit - performance, cost, energy efficiency, etc - then why would you or anyone else care?

I'm excited for HBM but at the end of the day we need to see what the benefits are in real world usage.
 
i for one dont whant another "exchanging blows with nvidia" card like 290x was to Titan, not with these specs

Fury will be or will be not the fastest card on the market (against an OCed 980ti); if not im goin green
 
i for one dont whant another "exchanging blows with nvidia" card like 290x was to Titan, not with these specs

Fury will be or will be not the fastest card on the market (against an OCed 980ti); if not im goin green
If it trades blows with the TX for $499 will you change your mind? :D
Price matters.
 
New tech matters.
I rather buy a HBM card than a GDDr5 card especially at high end.
Looking for a card for me and nothing nvida is doing appeals for me but AMD Fury has me excited like nothing else the last few years.

Its why you see so many posts on all forums about the Fury not the 980ti or TX.

IMHO, new tech is important, but only in as much as it affects the other items on the list.

If new tech doesn't provide either a performance improvement, new features, lower power use, or a cheaper price, then what is it good for? :p

HBM - for instance - may be helping AMD prepare for their next gen chips, but in this generation it will make little to no difference in the areas that really matter.

For the consumer a GPU is a black box that provides output. If you can satisfy the needs of the computer, it doesn't matter if it's a box of gremlins riding exercise bikes, inside, or if its high end silicon. The results are what matter.
 
Well HBM does save power, but it makes sense for AMD to put that to use elsewhere.
I think AMD realized they needed the appeal of new tech, plus the power-savings to remain competitive with Nvidia in the slightest. Hope I'm wrong though.
 
Now we are arguing who had the "performance crown" and for how long? In the context of the 390 who ****ing cares?
 
I thought FreeSync took less than 2 years to implement and the monitors that we are seeing on the market are generation one. I would assume that improvements will happen with generation 2. Also FreeSync was backwards compatible with multiple generations of GCN cards. Well at least it is supposed to be with my 7950's and R9 290. I don't have a FreeSync monitor to test with. :)

I don't consider a couple of years to get new tech out a long time; then again I'm getting old. How long has AMD been developing HBM? Even Nvidia announced support for HBM on their 2016 cards. Can't be but so bad if your rival uses your tech on their future products. :)

I'm looking forward to HDMI standardized FreeSync. It would synergize nicely with the future consoles and TVs.

I like the idea of freesync and gsync, but I haven't been shopping based on them at all, and they won't influence my next purchase.

The reason?

I already have my monitors. I bought them in 2010 and will likely use them until they die, and none of them have free/g sync capability.

They will with all likelihood outlive the next 2-3 GPU's I'll buy. Next time I shop for a monitor I may look at free/g sync, but until then the tech will have 0 influence on my GPU buying decisions.
 
Also FreeSync was backwards compatible with multiple generations of GCN cards. Well at least it is supposed to be with my 7950's and R9 290. I don't have a FreeSync monitor to test with. :)

Nope sorry you got manipulated by AMD PR hype machine- only gpus that can use VRR are Hawaii, Tonga and Bonair.
7950 can only use it for power savings.
 
Zarathustra[H];1041643712 said:
I like the idea of freesync and gsync, but I haven't been shopping based on them at all, and they won't influence my next purchase.

The reason?

I already have my monitors. I bought them in 2010 and will likely use them until they die, and none of them have free/g sync capability.

They will with all likelihood outlive the next 2-3 GPU's I'll buy. Next time I shop for a monitor I may look at free/g sync, but until then the tech will have 0 influence on my GPU buying decisions.

I agree with you about the monitors, but I already have a old PC in the living room that I've been wanting to upgrade for the past year. If HDMI FreeSync is adopted and comes to a TV, that's what I'll base my next HTPC around. It just makes sense to as it won't need another upgrade for 5 years or more I bet.
 
Neat comparison of GDDR5 vs HBM usage:
https://jsfiddle.net/xhhqthrc/

Source

Fail as hell... too much variables that aren't taken into consideration (the more important, Direct Bus and vRAM clock), it read as HBM its 32x faster than GDDR5 without even think in Memory speed. which isn't true.. the direct comparison in the test would be with a 384bit bus vs 1024bit which i edited for you.. and for the sake of a more comparable test, without take into consideration clock speed. :D

https://jsfiddle.net/xhhqthrc/40/

EDIT: another heavier one.. the "app" requesting 5mb each 1ms, adding to the pool 10mb.. with 1.000.000 elements to render. instead of 100.000 requesting 1mb each 1ms adding 1mb...

https://jsfiddle.net/xhhqthrc/45/
 
Last edited:
What are they trying to help people visualize? The amount of vram actually used and needed to play a certain game assuming both cards were the same otherwise
don't laugh but for some reason the point went over my head

Yeah, I'm not quite sure what they are trying to demonstrate here either.

Are they trying to suggest that HBM uses the RAM more efficiently, and thus it fills up slower?

I'm not sure I buy that for a second.


Are they trying to show how much faster HBM is than GDDR5? I DO buy that, but I very much doubt it will amount to a hill of beans worht of performance in the first generation, as the bottleneck will with all likelihood still be the GPU, not the RAM.


In a typical render pipeline data is sucked from the VRAM to the GPU, which renders it, and stores the in process frame back into VRAM for post processing, after which point it is sent out to the monitor.

If we assume a steady state 60fps, the render and post processing is done in 1/60 of a second.

Provided the VRAM can provide the GPU with data fast enough, that the GPU is never sitting there waiting for data, the VRAM is fast enough to support the GPU. You are GPU limited, and no amount of extra VRAM bandwidth is going to be helpful.
 
I think we should stop doubting that HBM is the way to go, since NV is also following suite next soon.... if both of them are saying it's the way, why the hell are we doubting it :p?
 
I disagree. You can say that about stuff like mobile phones, but high end GPUs... No casuals besides 15 year olds with well off parents are rolling into best buy and buying a 500-1000 dollar GPU. Majority of the people buying the high end GPUs are on the forums. Whether that be [H] or Overclock or Guru3d or AnAnd, that's where their market is located. AMD doesn't know it's market right now. They don't realize everyone's been F5ing all day waiting for leaks and news. Regardless if they set a date of June 16th 10 years ago and decided to stick to the plan you have to play to your market and adjust based on competition and they're doing neither.

Exactly, AMD is dropping the ball. People are eating up the 980 ti while AMD is "sticking to the plan", and losing sales.

I'd just like to point out nVidia said absolutely NOTHING and I mean NOT A DAMN THING about the 980 Ti. We didn't have no dates, didn't know if it was cut down, and performance benchmarks literally leaked hours before the official launch. While AMD who at least threw out an official date gets blasted. Just sayin'.

As for people eating up the 980 Ti, how much you wannat bet 90% of them were gonna buy the 980 Ti regardless of what AMD had to offer?
 
I'd just like to point out nVidia said absolutely NOTHING and I mean NOT A DAMN THING about the 980 Ti. We didn't have no dates, didn't know if it was cut down, and performance benchmarks literally leaked hours before the official launch. While AMD who at least threw out an official date gets blasted. Just sayin'.

As for people eating up the 980 Ti, how much you wannat bet 90% of them were gonna buy the 980 Ti regardless of what AMD had to offer?

AMD didn't have an official date until Monday. Everyone thought AMD would present some info at Computex and they did not. AMD also did nothing to try and clarify nothing would be said about Fiji at Computex, I'll assume to try and damper Nvidia sales for a few days.
 
I think we should stop doubting that HBM is the way to go, since NV is also following suite next soon.... if both of them are saying it's the way, why the hell are we doubting it :p?

No one is doubting whether or not HBM is the way to go long term.

What is being questioned is:

1.) Will HBM provide any real performance benefits in this generation, or are we still going to be GPU limited.

2.) Is HBM VRAM more capacity efficient than GDDR5? (in other words, can you get by with less of it?)

HBM IS THE FUTURE, it's just that there are some doubts as to what difference it will make in the short term, and if Nvidia made the right decision to hold off for a generation.
 
AMD didn't have an official date until Monday. Everyone thought AMD would present some info at Computex and they did not. AMD also did nothing to try and clarify nothing would be said about Fiji at Computex, I'll assume to try and damper Nvidia sales for a few days.

And nVidia never had an official date. Not sure what point you're trying to make.

And if you want to go by assumptions, well it was known as early as late January AMD was gonna present at E3, so there's also that.
 
I'd just like to point out nVidia said absolutely NOTHING and I mean NOT A DAMN THING about the 980 Ti. We didn't have no dates, didn't know if it was cut down, and performance benchmarks literally leaked hours before the official launch. While AMD who at least threw out an official date gets blasted. Just sayin'.

As for people eating up the 980 Ti, how much you wannat bet 90% of them were gonna buy the 980 Ti regardless of what AMD had to offer?


Context is important. People have been waiting for this ultra high end potential Titan killer for half a year or more. Titan X came out, what what? The middle of March? No one expect them to release a card that's near Titan X performance for that price point, so shortly after it's release.

So 980ti becomes a pleasant surprise, where as the 390X/Fury becomes an overdue underwhelming irritation.
 
AMD Fury only tech that people are interested in as old tech passed by without much noise.
its why so many are impatient (demanding) for news :)
 
And nVidia never had an official date. Not sure what point you're trying to make.

And if you want to go by assumptions, well it was known as early as late January AMD was gonna present at E3, so there's also that.

nVidia didn't need an official date. Titan X was already king of the hill. Everyone assumed a Ti would show up with marginally less performance than the Titan X shortly after, just like it has previously. It did.
 
previously it was 6 months

this time is was 3.

i think nvidia was concered.
 
previously it was 6 months

this time is was 3.

i think nvidia was concered.

Concerned? Sure, why not. Titan X was expensive. AMD cards are usually not. Makes sense to accelerate the cheaper Ti as a response and undercut AMD prior to their release.
 
Zarathustra[H];1041644105 said:
HBM IS THE FUTURE, it's just that there are some doubts as to what difference it will make in the short term, and if Nvidia made the right decision to hold off for a generation.

Not sure nV had the option to do anything sooner. AMD developed the tech, so I think they were a lock to be first to market.

I agree that the benefits are nebulous, but this is certainly something where nV is playing catch-up for the next gen. Lucky for them AMD didn't just lock competition out of the tech...
 
I'm starting to grow disinterested. It's like HL3, eventually I just don't care.
 
Not sure nV had the option to do anything sooner. AMD developed the tech, so I think they were a lock to be first to market.

I agree that the benefits are nebulous, but this is certainly something where nV is playing catch-up for the next gen. Lucky for them AMD didn't just lock competition out of the tech...

yes, Hynix gave AMD one year lock advantage over nvidia for their partnership in the development of HBM..
 
Back
Top