[VC]AMD Fiji XT spotted at Zauba

[boxleitnerb]

According to your own link,

GTX-780Ti at default 928MHz boost is only 37% faster at 1080p and 40% at 1440p than GTX-770.
GTX-780Ti has 87,5% more shaders, 87,5% more Texture Units, 50% more ROPs and 50% more memory Bandwidth than GTX 770.

I believe we may see almost the same with GM200 vs GTX-980.
So if 390X is 20% faster than 980 it will be fine at $499-$550.

You also cannot read. Read again, maybe you will understand. Hint: Clocks.
If you clock both GPUs the same the performance delta will increase. The exact increase depending on the extent of bandwidth limitation of course.

GTX 770@928: 66.8 index
GTX 780 Ti@928: 109 index
109/66.8 = +63.2% >> 37-40%

 

[Silverforce11]

I still don't think you guys are paying attention or acknowledging history.

All of what you speak of won't matter if AMD is not competitive on the high end. Where NV got the halo branding is not due to being 2nd best and offering great perf/$ for years. They got it by being THE BEST from the top.

Pushing GE vs Gameworks is a game of throwing $ at developers via manpower & resources, always has been. Fact is, NV has more money to throw around.

Definitely AMD needs to keep pushing with GE & Mantle, especially easier for them now due to Radeons in major consoles. But they also need to push for raw performance WITH a focus on efficiency for their new architecture. They need efficiency to compete on notebooks, without it, they are giving that market away to NV.

The 7970 situation was a small window where they led. Everyone knew NV was going to respond soon, and they are willing to wait a few months for it. After that, it was too close to call, then NV dropped the bomb with Titan (overpriced but still THE BEST), then 780 hammered home the win on that generation. R290/X was far too late, and while it enjoyed a good run, 970/980 is now again the better choice forcing major price drops on AMD's lineup.

See, to someone loyal with NV, they see this and they know NV delivers. They may be slightly late here and there, but they will be on top. The common man where I am think NV is the best & they are willing to pay extra for it.

As soon as AMD is able to lead a generation utterly, then follow up with a 2nd generation in the lead, is when heads will turn.

They have not done that ever since the 8800 series from NV. AMD has been 2nd best and bleeding marketshare.

 

[raghu78]

You clearly didn't read my post. I said "up to the GTX 780". Everything above scales worse due to bandwidth beginning to be a limiting factor. All other reviews besides HT4U are useless since they don't benchmark with fixed clocks, thus the influence of clock speed is uncertain and can shift results depending on cooling/boost.

The ht4u results show GTX 780 is 31% faster than GTX 680 at 1440p for 50% more shaders/TMU/ROPs/bandwidth (GPC increase is only 25% from 4 to 5). Now don't try some funky straight linear interpolation of GTX 780 performance with clocks because thats never the way things work. btw GTX 780 is a salvaged GK110. the GTX 780 Ti is the full GK110.

And why do you ignore my example of 7870 vs 7970 GHz? Afaik Tahiti has 2 raster engines, just like Pitcairn. Who is to say that those don't have an impact on scaling with the rest of the function blocks? The results are inconclusive here.

thats the point. there is rarely a case where every resource is scaled perfectly between 2 chips- shaders, TMUs, ROPs, shader engines, geometry engines, raster engines , bandwidth. Even in the rare case of R9 290 vs HD 7870 the perf scaling is not perfectly linear even though resources are exactly scaled 2x.

Why will it not happen, how could you possibly know? Did you design GM200? It is absolutely possible that GM200 increases ROPs, TMUs, SP and bandwidth by x%? Why would it be unlikely that performance would increase by those x% too - be it 30, 40 or 50%?
Only if for example GM200 were to double the shader count (while retaining similar GPU clocks) but increase bandwidth by only 50%, it would be quite certain that scaling would be non-linear.

Fact is: Linear scaling can and does happen as evidenced by my examples. Another example would be GTX 680 vs GTX 650 Ti.

Sorry but perfect performance scaling of 1:1 never happens with added resources even when all resources are increased in the same proportion. Thats just the way things work. btw the GTX 650 Ti was a salvaged GK106 chip. The full GK106 chip was GTX 660. I will take a bet that GM200 will increase 50% on shaders but the rest of the specs won't scale the same 50% especially GPC count.

 

[AtenRa]

You also cannot read. Read again, maybe you will understand. Hint: Clocks.
If you clock both GPUs the same the performance delta will increase. The exact increase depending on the extent of bandwidth limitation of course.

GTX 770@928: 66.8 index
GTX 780 Ti@928: 109 index
109/66.8 = +63.2% >> 37-40%

Hint: GM200 Graphics Cards will have lower clocks than 980, like 780Ti has lower clocks than 770. (assuming 28nm)

 

[therealnickdanger]

Hint: GM200 Graphics Cards will have lower clocks than 980, like 780Ti has lower clocks than 770. (assuming 28nm)

Reference or aftermarket? Given how much headroom there is between reference Maxwell and the on-air limits, I could see GM200 launching with higher clocks.

 

[lavaheadache]

^ I am in the camp that thinks GM200 780Ti successor (not 780 successor) will be 40-50%+ faster than 980, not 20-30%.

me too

 

[Keysplayr]

Sounds about right.

 

[AtenRa]

Reference or aftermarket? Given how much headroom there is between reference Maxwell and the on-air limits, I could see GM200 launching with higher clocks.

We are always talking default. Otherwise there are OCed GTX770, GTX780Ti and GTX980 as well.

 

[exar333]

me too

If NV follows the previous 7xx/Titan path with the new Titan II and xTi part, 50% definitely seems about right. Hopefully the new 980Ti (for the lack of a better name) is ~ par with the Titan II in gaming, with potentially more OC headroom. If the price is right, this could be a great part!

Back on topic.....

I am excited for Fiji XT myself. The wild-card to me is how well AMD can tweak perf/watt. That will directly impact how many SP they can get under the power envelope. The scaling for AMD has been INCREDIBLE lately and I really like the advances on bridge-less XF as well.

2015 could be the most exciting GPU year in a LONG time.

 

[monstercameron]

I will never be able to understand this massive pivot to perf/w for enthusiasts gpus. I would think spending $300+ an a gpu, the buyer would want as much perf as possible.

 

[f1sherman]

You also cannot read. Read again, maybe you will understand. Hint: Clocks.

wth you guys. I saw this

In the chart you mention the GTX 780 Ti (966 Mhz base) is 44% faster than GTX 770 (1084 Mhz base) at 2560 x 1440. GTX 780 Ti has 87.5% more shaders than GTX 770. But the actual perf increase is 45 - 50% .

http://ht4u.net/reviews/2013/nvidia_geforce_gtx_780_ti_gk110_complete_review/index42.php

http://www.computerbase.de/2013-11/nvidia-geforce-gtx-780-ti-vs-gtx-titan-test/4/

http://www.hardware.fr/articles/912-22/recapitulatif-performances.html

Even the classified GTX 780 Ti which boosts to 1100 - 1150 Mhz at stock and can be considered on a clock for clock comparison with ref GTX 770 which boosts to 1150 Mhz is just 64% faster (61 * 1.64 = 100.04) than GTX 770 at 1600p and just 51.5% faster than GTX 770 (66 * 1.515 = 99.99) at 1080p. So the claims of perfect scaling are just wrong. Linear scaling maybe at a factor of 0.75x is what we see. ( 87.5 x 0.75 = 65)

and started typing frantically. But then I saw that everything has already been explained by boxleitnerb

What do you mean "even"? Like paying attention on flops (instead of just cuda cores) and comparing on same clocks is somehow unfair to 770.

As Boxleitnerb has already pointed out: Clocks + bandwidth.

And yes its tough comparing due to boost mechanism.
There has already been so much confusion due to this. Ppl still keep forgetting that real clock(s) (which rarely gets mentioned in reviews) is neither base, nor boost

 

[f1sherman]

I will never be able to understand this massive pivot to perf/w for enthusiasts gpus. I would think spending $300+ an a gpu, the buyer would want as much perf as possible.

And I am equally puzzled by how many times this ^^ gets repeated around here.
Just ask AMD if he's equally unimpressed by all this perf/W move on NV part.

Not every discussion is a Buyers Guide.

 

[Enigmoid]

I will never be able to understand this massive pivot to perf/w for enthusiasts gpus. I would think spending $300+ an a gpu, the buyer would want as much perf as possible.

I agree with you. However its really simple. Given the market conditions and the gradually diminishing of the GPU due to the igp GPU makers are under more and more pressure remain profitable.

AMD and Nvidia compete is 4 major markets.

1. Desktop GPU
2. Mobile GPU
3. SOC/APU (kaveri, beema/mullins, tegra)
4. Professional

In two of these markets perf/W reigns supreme. In the professional market efficiency is often a huge factor though other factors may dwarf it. Simply put any design that doesn't focus on efficiency cannot compete well in mobile or SOC/APU and is handicapped in the professional market. Desktop is really the only market where efficiency (sometimes) comes second fiddle to absolute performance and compared to the amount of money that the other three markets comprise is a small portion. Not to mention that a design that wins in performance but looses in efficiency will lose design wins in mobile resulting in a poor ROI and thus handicap the next generation. In the desktop space the trend is toward smaller and more power efficient designs.

Efficiency is the reason why Nvidia owns the mobile GPU market, especially the mid range and up (AMD pretty much sells 95% Oland chips in mobile). On the desktop given similar performance and price efficiency is a good reason to go with GPU A over B. Efficiency also enables the upward scaling of performance as there are practical limits to how much power a GPU can use.

Price as always is the deciding factor to most consumers. However, high efficiency does not imply high price.

 

[3DVagabond]

This is only going to work for brand agnostic price/performance buyers who bought 4870/5850/6950/7950/290. Price/performance strategy failed AMD many times on the desktop. Now what if you could CF a 370X with an APU. Someone on a budget who can only spend $130 on a CPU but wants 380 level of performance could get close by pairing an 512-768 GCN 2.0 APU with a 370X. This is a strategy neither Intel nor NV can compete with. It's a totally unique value add for both AMD CPUs and GPUs. Think about how many people purchase $120-190 i3/i5s and $200-250 NV GPU. Now take the additive performance of GCN in the APU and a $200-250 AMD GPU (which happens to be faster than NV due to price/performance anyway).

This combination will allow the 2 AMD components in tandem to punch WAY above their power class in GPU limited games. NV and Intel will be scrambling to respond to that aggregate GPU power. Unfortunately, AMD has never been able to get Hybrid CF with a low-end and mid-range/flagship GPU to work properly. All of a sudden the 15-20% GPU advantage of GM200 will be wiped out by the GPU power in the APU. As memory technology improves, AMD needs to create a Unified shared memory pool and allow any APU and GCN to CF. AMD would price the APU cheap but make up the lost APU profits with added sales of AMD motherboards and GPU sales.

If this is too complex, AMD needs to allow CF for any GCN GPU combination (390 with 370X). This would also increase sales.

I don't think perf/$ is a failed plan. They need to win over public mind share as well. getting AMD graphics into the OEM market is needed. Many people's first PC is an off the shelf unit. They form their attachments to brands right from then and it's a hard to break bond.

 

[monstercameron]

I agree with you. However its really simple. Given the market conditions and the gradually diminishing of the GPU due to the igp GPU makers are under more and more pressure remain profitable.

AMD and Nvidia compete is 4 major markets.

1. Desktop GPU
2. Mobile GPU
3. SOC/APU (kaveri, beema/mullins, tegra)
4. Professional

In two of these markets perf/W reigns supreme. In the professional market efficiency is often a huge factor though other factors may dwarf it. Simply put any design that doesn't focus on efficiency cannot compete well in mobile or SOC/APU and is handicapped in the professional market. Desktop is really the only market where efficiency (sometimes) comes second fiddle to absolute performance and compared to the amount of money that the other three markets comprise is a small portion. Not to mention that a design that wins in performance but looses in efficiency will lose design wins in mobile resulting in a poor ROI and thus handicap the next generation. In the desktop space the trend is toward smaller and more power efficient designs.

Efficiency is the reason why Nvidia owns the mobile GPU market, especially the mid range and up (AMD pretty much sells 95% Oland chips in mobile). On the desktop given similar performance and price efficiency is a good reason to go with GPU A over B. Efficiency also enables the upward scaling of performance as there are practical limits to how much power a GPU can use.

Price as always is the deciding factor to most consumers. However, high efficiency does not imply high price.

I don't disagree that is why I used the qualifier enthusiast gpus. Efficiency is very important in many markets but high cost, high performance gaming gpus? seems like a new smear campaign.

 

[ShintaiDK]

I don't disagree that is why I used the qualifier enthusiast gpus. Efficiency is very important in many markets but high cost, high performance gaming gpus? seems like a new smear campaign.

Does it? It seems the smear is to hide the inefficiencies instead to accept subpair products.

Like the wast majority I dont have AC in the house I can turn up just because I bought a hummer of a GPU. And I dont intend to buy one just so I can endure the summer while sitting infront of a socalled proclaimed "enthusiast" product. Nor do I want some kind of turbofan setup that will make me half deaf, unless I go all the way for water cooling. Not do I want some kind of 1990s tower PC just to house it in.

 

[monstercameron]

Does it? It seems the smear is to hide the inefficiencies instead to accept subpair products.

Like the wast majority I dont have AC in the house I can turn up just because I bought a hummer of a GPU. And I dont intend to buy one just so I can endure the summer while sitting infront of a socalled proclaimed "enthusiast" product. Nor do I want some kind of turbofan setup that will make me half deaf, unless I go all the way for water cooling. Not do I want some kind of 1990s tower PC just to house it in.

then dont buy an enthusiast gpu? I really dont understand what any of your post means. Don't want it don't but it. get a low end or mainstream gpu.

 

[ShintaiDK]

then dont buy an enthusiast gpu? I really dont understand what any of your post means. Don't want it don't but it. get a low end or mainstream gpu.

So a GTX980 is not enthusiast? Just the fastest and most efficient GPU there is?

 

[AtenRa]

To be fair, when AMD had the better perf/watt people were asking the same questions.
I do care about perf/watt and i want AMD to get better in that, but it is not the first thing im watching when im in the market for a high-end GPU. Performance/$ is the first one im looking for and i believe its the one the vast majority of buyers care as well.
Also, people buying $500+ GPUs are 99% concern about performance alone and nothing else.

 

[monstercameron]

So a GTX980 is not enthusiast? Just the fastest and most efficient GPU there is?

Dont know if it is the fastest gfx card, when you can have a 295x2. I also don't know if it is the most efficient gpu, that all depends on the work load.

 

[exar333]

I don't think perf/$ is a failed plan. They need to win over public mind share as well. getting AMD graphics into the OEM market is needed. Many people's first PC is an off the shelf unit. They form their attachments to brands right from then and it's a hard to break bond.

Perf/$ has worked-out REALLY well for for AMD on the CPU front. Oh wait...

Love or leave it, AMD has marketed AMD GPUs as a 'budget' option pretty much since they purchased ATI in 2006. In my opinion, that has cheapened their brand and is a main reason they are forced to sell the same (or sometimes better) products vs. NV for less $. I can't fault AMD for pretty much trying everything from a marketing standpoint recently, but it doesn't seem to sway a lot of diehard-NV fans.

My personal fear is that when a company goes the 'budget' route that they decrease investing in R&D (which we see from AMD now). That means you sometimes get into a death-spiral situation. You don't need to look further than AMD's CPUs...

 

[monstercameron]

Perf/$ has worked-out REALLY well for for AMD on the CPU front. Oh wait...

Love or leave it, AMD has marketed AMD GPUs as a 'budget' option pretty much since they purchased ATI in 2006. In my opinion, that has cheapened their brand and is a main reason they are forced to sell the same (or sometimes better) products vs. NV for less $. I can't fault AMD for pretty much trying everything from a marketing standpoint recently, but it doesn't seem to sway a lot of diehard-NV fans.

My personal fear is that when a company goes the 'budget' route that they decrease investing in R&D (which we see from AMD now). That means you sometimes get into a death-spiral situation. You don't need to look further than AMD's CPUs...

death spiral? seems too strong a sentiment for a company that still holds at minimum 20% of a multibillion dollar graphics industry.

 

[geoxile]

I don't think perf/$ is a failed plan. They need to win over public mind share as well. getting AMD graphics into the OEM market is needed. Many people's first PC is an off the shelf unit. They form their attachments to brands right from then and it's a hard to break bond.

AMD is automatically an inferior good, not only because of quality but also because of the pricing and image relative to Nvidia. Considering many of the PC gamers buying high-end parts are also enthusiasts with a lot of disposable income, you can see why this value-oriented image might not matter

I've also noticed (on here, OCN, gfaqs, even /g/) that people seldom consider the value of AMD's game bundle when making recommendations.

 

[AtenRa]

AMD doesnt need to be number one, but they DO NEED to get more mobile market share. And that means they need to innovate more and be first with technology.

But from my experience, the thing they need to fix the most is their ability to sell their products and increase sales. You can have the best product in the world, but if you cant sell it you gain nothing.

 

[el etro]

Project Denver approach(the CPU cores on the GPU to reduce CPU dependancy), Driver overhead elimination techniques such as Mantle or DX12 and HBM(reducing PCB complexity by adding Vram on the GPU die) is the technologies that will drive the GPU industry foward.