Radeon Fury to be slower than GTX 980 Ti

[erocker]

TPU seems to disagree, as noted above.

Going off of what? A graphic on a box?

 

[irani]

Intel is the biggest rebrander of all time. From SB to skylake lol

big like

 

[RejZoR]

You can't say it's a rebrand if it's using a different core, jesus christ. So GTX 970 is just a rebrand of GTX 750Ti you'll say? It isn't. It's a different core. Stop spreading nonsense and learn what is a rebrand and what is a refresh. R9-390X is a refresh!

 

[the54thvoid]

TPU seems to disagree, as noted above.

He knows best. Everyone else is wrong. They may be. Only RejZoR knows.

 

[erocker]

He knows best. Everyone else is wrong. They may be. Only RejZoR knows.

Welcome to the internet. Where everyone see's a box and thinks the sky is falling.

 

[RejZoR]

He knows best. Everyone else is wrong. They may be. Only RejZoR knows.

Because I AM correct. It can't possibly be a rebrand by any definition if the chip is nothing like the old one. HD7970 transition to R9-280X is a rebrand. It's exactly the same core, they tweaked clocks a bit and slammed a new BIOS on it. This is what we consider rebrand. But when you take a foundation, replace all shaders with new ones, tweak the clocks, potentially slam additional instructions for new version of DX, based on what fucked up logic you can call this "rebrand"?

It's also not a rebrand of Tonga core despite sharing same GCN version. R9-285 has less shaders, less compute units, less ROP and narrow memory bus. Ergo, not the same again.

With your logic, Fiji is just a rebrand of ATi Rage from the freaking 90's. Um oh, they both have GPU on it, so they are the same.

 

[arbiter]

You can't say it's a rebrand if it's using a different core, jesus christ. So GTX 970 is just a rebrand of GTX 750Ti you'll say? It isn't. It's a different core. Stop spreading nonsense and learn what is a rebrand and what is a refresh. R9-390X is a refresh!

Kinda hard to call something a rebrand when one has 640 cuda cores and other has 1664 cores. SO your logic is completely retarded. One requirements for rebrand is would need to be THE SAME.

 

[RejZoR]

You just proved my fucking point. 2816 cores found in R9-290X are not the same as 2816 cores found in R9-390X. Assuming they will have the same number, there are no official numbers yet mind you. But one thing is known for sure, the 390X will NOT use GCN 1.1. It will have GCN 1.2 That is without a shadow of a doubt a solid fact. Which by itself makes my point correct.

By your logic, 400 HP Mitsubishi Lancer Evo is just a rebrand of 60 HP Mitsubishi Colt. They both have four cylinder engines, meaning they are EXACTLY the same aka it's a rebrand. My sides, my sides are gone...

 

[R-T-B]

Because I AM correct. It can't possibly be a rebrand by any definition if the chip is nothing like the old one.

From TPU today, they say they have NO evidence the silicon is modified in the least for at least the R9 390x. So at least that one is possibly a rebrand.


No, I don't like that idea any more than you, but it's what's being indicated...

You just proved my fucking point. 2816 cores found in R9-290X are not the same as 2816 cores found in R9-390X. Assuming they will have the same number, there are no official numbers yet mind you. But one thing is known for sure, the 390X will NOT use GCN 1.1. It will have GCN 1.2 That is without a shadow of a doubt a solid fact. Which by itself makes my point correct

Again, this is NOT certain.

@erocker
I was supposed to be quoting RejZor, my tablet thought otherwise aparently. It would have made more sense in context.

 

[RejZoR]

Sorry, but that would be galactic scale fail. If anyone thinks that way, he's not thinking with any logic.

Why would AMD develop a Tonga core between Tahiti and Hawaii around GCN 1.2 and then rebrand an OLDER GCN 1.1 core into a newer generation card. That's like NVIDIA doing work on Maxwell 1 and then making GTX 970 based on Kepler instead of Maxwell 2 which is what they did. It would be that kind of level of stupid.

 

[HumanSmoke]

Sorry, but that would be galactic scale fail. If anyone thinks that way, he's not thinking with any logic.

Why would AMD develop a Tonga core between Tahiti and Hawaii around GCN 1.2 and then rebrand an OLDER GCN 1.1 core into a newer generation card. That's like NVIDIA doing work on Maxwell 1 and then making GTX 970 based on Kepler instead of Maxwell 2 which is what they did. It would be that kind of level of stupid.

Bookmarked this baby!

Knowing something as fact (GCN 1.2 isn't confirmed for Grenada), and assuming that it so because you cant possibly entertain the fact that it could be otherwise aren't really the same thing at all. Case in point is the first verified leak of AMD's (non-OEM) 300 series. The R7 370 gets a fancy new GPU name (Trinidad) , yet still retains a Crossfire finger - so no XDMA, and likely is at best a metal layer respin of Pitcairn/Curacao.....which means that AMD is releasing a high-volume part that doesn't have support for a number of features AMD itself are pushing to sell the card series - notably TrueAudio, FreeSync, and VSR.

 

[R-T-B]

Sorry, but that would be galactic scale fail. If anyone thinks that way, he's not thinking with any logic.

Why would AMD develop a Tonga core between Tahiti and Hawaii around GCN 1.2 and then rebrand an OLDER GCN 1.1 core into a newer generation card. That's like NVIDIA doing work on Maxwell 1 and then making GTX 970 based on Kepler instead of Maxwell 2 which is what they did. It would be that kind of level of stupid.

I know and there's only one very sad explanation for it in my mind:

AMD is very, very strapped for cash.

 

[HumanSmoke]

AMD Radeon Fury X 3DMark performance

Well, I certainly hope that isn't indicative of the Fury X's actual performance. 2% faster than the 980 Ti with 45% more cores, 52% more bandwidth, 33% more raster ops, 45% more texture address units, and nominal 20% higher board power limit doesn't make for great reading. Still, if it's slugging it out with a $650 card, maybe the pricing will be better than originally mooted.

 

[GreiverBlade]

aaaaaaaahhh we don't care (well at last I don't ...) about rebrand ... if there is a rebrand even with some optimization or refinement, it only mean they need only 1 new sku : the rest still has enough potential to hold the line.

if the price is right and the performance are here ... then : everything's OK
will still wait till a Fiji XT or Pro hit the 2nd hand market, depending the price, before deciding on a upgrade ... or keep my 290 BUT first! i wait for REVIEW, enough rumors BS and other joyfull things!
nvidia AMD : BOTH HAVE INTERESTING OFFER! (performances and power consumption for the first and bang for bucks for the later )

 

[MrGenius]

Refresh is NOT rebranding. It even goes against the dictionary definition of it.

New architecture = entirely new chip that has very little in common with predecessor
Refresh = better process node, optimizations
Rebrand = Same everything, maybe different clocks and different BIOS with (obviously) different name.

HD7970 transition to R9-280X is a rebrand.

Because GCN 1.0 is the same as GCN 1.1...Open GL 4.2 is the same as Open GL 4.3...and DX 11.1 is the same as DX 11.2...right?

 

[Xzibit]

Source: https://www.reddit.com/r/Amd/comments/399775/new_picture_of_fury_nano_w_backplate_tubing_and/

That doesn't look like a 25/27mm radiator.

 

[HumanSmoke]

if the price is right and the performance are here ... then : everything's OK

No, everything isn't OK.
AMD have had price and performance on their side for years, and look where it has got the company.
The whole reason AMD are only launching a single new GPU is because they are R&D constrained because their account books look like a dumpster fire being dowsed with raw sewage. Those great prices, and their potential customer base queuing up to buy second hand cards, pretty much mean that the company cant invest in R&D today, which translates to reduced amount of product on a slower timetable 2, 3, 4, 5 years in the future.

will still wait till a Fiji XT or Pro hit the 2nd hand market, depending the price, before deciding on a upgrade ... or keep my 290

I'm sure AMD's bean counter is VERY happy to hear that.

 

[the54thvoid]

BTW @HumanSmoke - I dig your CD tagline. Was that not from when Intel was working on Larabee/Knights Corner and he was spouting stuff about it finishing Nvidia etc or was it on die gfx talk? Either way I do recall reading the ramble he was on that day. I also recall the time he said a viable GF100 core was impossible (with all cores enabled) and then about a month later Nvidia dropped the GTX 580. T'was funny.
I liked S|A at the start but it got ugly and the mods were dicks. If people think TPU has a bias, hell - need to look around... Anyway, enough off topic stuff.

That doesn't look like a 25/27mm radiator.

I'd prefer if they release AIO's with optional components. I'd rather switch out most of these parts, in other words do it custom.

 

[Xzibit]

I'd prefer if they release AIO's with optional components. I'd rather switch out most of these parts, in other words do it custom.

Just curious... If they were going to use something smaller then the 295x2 38mm since its a single gpu.

CoolIt is rumor to be making it so it might end up something similar to the Corsair H80i underneath the shroud

 

[HumanSmoke]

BTW @HumanSmoke - I dig your CD tagline. Was that not from when Intel was working on Larabee/Knights Corner and he was spouting stuff about it finishing Nvidia etc or was it on die gfx talk? Either way I do recall reading the ramble he was on that day. I also recall the time he said a viable GF100 core was impossible (with all cores enabled) and then about a month later Nvidia dropped the GTX 580. T'was funny.
I liked S|A at the start but it got ugly and the mods were dicks. If people think TPU has a bias, hell - need to look around... Anyway, enough off topic stuff.

Back when Fermi was undergoing its troubled gestation-the thread title is the top right hand corner. Charlie and noted AMD shill Spigzone

Spoiler

I and five other friends who learned computer tech as school back in the 70's shared a user account at S|A when it first started up. Used to drive some people nuts as the posting viewpoints came from all over the show. One of the guys was in Taiwan doing chip layout, one in Hong Kong working for OEM supply chain, and the others spread far and wide - and all rather critical of the industry as a whole

The mods there don't suffer anything that steps over the party line. The major "no no's" are questioning Charlie's facts, criticizing Intel content written by Charlie (who is extremely Intel friendly), and pointing out deficiencies with AMD's management/strategy. Anyhow, back on topic...

I'd prefer if they release AIO's with optional components. I'd rather switch out most of these parts, in other words do it custom.

Seconded. Even adding some quick disconnects and an options list would give the products wider appeal, although I guess the whole reason behind AIO's is that they cater to people who don't want to be bothered with mix-and-match - or are too intimidated by bespoke watercooling.
On a related note, that fan blade profile looks somewhat like a Scythe Gentle Typhoon.

 

[GreiverBlade]

No, everything isn't OK.
AMD have had price and performance on their side for years, and look where it has got the company.
The whole reason AMD are only launching a single new GPU is because they are R&D constrained because their account books look like a dumpster fire being dowsed with raw sewage. Those great prices, and their potential customer base queuing up to buy second hand cards, pretty much mean that the company cant invest in R&D today, which translates to reduced amount of product on a slower timetable 2, 3, 4, 5 years in the future.

I'm sure AMD's bean counter is waiting happy to hear that.

well you got me ... i will buy a day one Fury X

altho i can't think about anything else than : AMD is doing the job of intel and nvidia together so, no wonder they are more limited and only "middle of the road perf" than the 2 other who focus on only 1 sector ... but... ah whatever ...

as long as the reviews are not up, this thread as no more meaning for me

cheers

 

[R-T-B]

well you got me ... i will buy a day one Fury X

altho i can't think about anything else than : AMD is doing the job of intel and nvidia together so, no wonder they are more limited and only "middle of the road perf" than the 2 other who focus on only 1 sector ... but... ah whatever ...

as long as the reviews are not up, this thread as no more meaning for me

cheers

Yeah, and AMD does it with like a small slice of the capital of either of those groups. Amazing they are relevant at all, really...

 

[TheGuruStud]

Yeah, and AMD does it with like a small slice of the capital of either of those groups. Amazing they are relevant at all, really...

It's amazing what can happen when you care about competing and aren't committing crimes just b/c you're afraid to compete *looks at inhell*.

 

[HumanSmoke]

It's amazing what can happen when you care about competing and aren't committing crimes just b/c you're afraid to compete *looks at inhell*.

Just before you petition to have AMD's BoD canonized, you might want to check AMD's record. AMD aren't adverse to bending (false claims) and breaking the law (IP theft, price fixing), its just that their execution is less than stellar and generally isn't worth the effort.
Sure, they aren't a patch on Intel, but neither are they spotless.

 

[arbiter]

Bookmarked this baby!

Knowing something as fact (GCN 1.2 isn't confirmed for Grenada), and assuming that it so because you cant possibly entertain the fact that it could be otherwise aren't really the same thing at all. Case in point is the first verified leak of AMD's (non-OEM) 300 series. The R7 370 gets a fancy new GPU name (Trinidad) , yet still retains a Crossfire finger - so no XDMA, and likely is at best a metal layer respin of Pitcairn/Curacao.....which means that AMD is releasing a high-volume part that doesn't have support for a number of features AMD itself are pushing to sell the card series - notably TrueAudio, FreeSync, and VSR.

Easy to say their gpu if faster then a gpu that is what? almost year and half old already? 1.5 years, only manage 20%. Figure most 750ti's can overclock a good 40+%.

Well, I certainly hope that isn't indicative of the Fury X's actual performance. 2% faster than the 980 Ti with 45% more cores, 52% more bandwidth, 33% more raster ops, 45% more texture address units, and nominal 20% higher board power limit doesn't make for great reading. Still, if it's slugging it out with a $650 card, maybe the pricing will be better than originally mooted.

Well if you look at 980 vs 290x. 290x had 40% more shaders, 50% more memory bandwidth and look which of the 2 cards were faster in most testing? So really you think about it, its well possible that is only that.

fiji has almost 2x the memory bandwidth of the 980ti. You kinda wonder where that limit is of where it doesn't help the gpu. For an intel cpu for example, 1333 to 1600mhz you get an ok jump in performance but 1600 to 2400 there is very slim jump in most uses cept in applications where it needs it.

Nvidia and even Intel both prove its not how many cores you have its how you you use them.