This is just a post about a thought that I had today. People are looking at fiji and looking down on it because it did not beat the 980ti, and due to all the hype that is expected. Now I am just curious if AMD's zen ends up preforming in the same respect to say the intel i7 part when it drops, I would say that is an achievement, but would everyone else think that, or will they be impressed?
EDIT This topic is for open discussion about AMD. Now that is not to say that any questions that you have relating to AMD should be posted here. If you have more complex/should I buy AMD chip or Intel chip, you are still better off making your own topic.
Impressed? No. FuryX hasn't been tested and circulated enough for benchmarksbto be trustworthy + drivers... On CPU stage and can only hope to improve cache and single core performance- I don't see the second to happen.
I think it's important for AMD to close the gap before it starts worrying about passing up it's competitors ... Them releasing things that are on par, and even in some cases better than what the other offer is a huge jump in the right direction. Granted they've kind of always done this in the GPU market, it would be nice to see them release a CPU that could compete with current gen Intel chips.
I'm not at all upset about the recent Fiji release, I do think that it could have been handled and delivered better both internally and externally, however I don't think that it takes away from the fact that AMD is offering competitive performance in a very wide price range with their new line-up.
That being said I don't think that there's any AMD user out there that isn't hoping Zen will finally let AMD reclaim some of its former glory in the CPU market, but only time will tell. The important thing to remember is that the market needs the competition, because if nobody puts the pressure on to innovate, nobody will.
I have my preference for AMD. i don't like Nvidia's Business practices and the closed platform they try to promote with their Graphics cards. But AMD thrives on False Hope.
I'll name a few examples
AMD mentioned having a 12-Core processor coming... it turned out to be a Quad-Core APU with 8 graphical cores. they market it as a twelve core processor and it's not, and perfomance wise didn't really beat anything besides an i3 Intel Processor.
AMD mentioned the 9590 being revolutionary in the sense that it was going to be the worlds 5Ghz Processor and it's going to destroy benchmarks. it didn't. in fact it didn't really show more performance than an overclocked 8350.
Fiji XT is meant to compete with Titan X/980ti. it doesn't beat it. mind you it's in the same price bracket. then again you can argue that it's drivers. but we won't never know till time goes on. Nvidia drivers are mature. the Fury X drivers are new and are prone to performance issues.
HSA - AMD keeps mentioning it's revolutionary and it's going to change the way we get our work done. and you know what.. they are right. but where's the progress? no one is investing in APUs for it.
Now for AMD's Zen. they mentioned that there will be a 40% increase as compared to the current FX line-up. lets be realistic. that is not happening. maybe 20 or 30 percent. but not 40%. it's just going to appear huge cause it's been almost 5 years since an architectural change. Fury X makes no sense in it's price bracket in my honest opinion. it should of been at MOST $549. the Fury non-X should be around $479. the price gap between the R9-390x and the Fury X is too much.
overall time will tell. AMD knows they aren't doing well. they need to stop with the non-sense and get their act together.
Your going to have fan boyz on each side regardless of the products or benchmarks, I don't want to piss anyone off but I've never been a person to buy into bleeding edge hardware so that pretty well keeps me away from the latest and greatest cpu's and video cards till they are at least a year old and the price has stabilized.
To be honest I'm not impressed with the Fury, but at the same time I dislike nVidia just about as much as I dislike Apple or Microsoft, but I might buy a fiji based GPU if it will work in the environment I'd like to use it in which would be Linux running a KVM which I know the 980Ti won't work in because nVidia gimped the drivers to keep us from using any nVidia card in a KVM which I think is a pretty shitty way to do business.
I do hope the Zen architecture does level the playing field a little, we as consumers need to have competition in both the graphics and cpu markets not only to drive innovation but to keep competitive pricing alive, what we don't need is more walled gardens put in place by a company that feels they own the market for their product and the consumer has no choice but to buy from them regardless of price or performance.
AMD is also stuck in a rough situation when it comes to project management and R&D if you ask me ... They're the only in the industry that are attempting to stay competitive in both the CPU and GPU market at the same time. Twice as many problems, still just one company.
Technology grows at such a fast rate, and the demands for advancements in the field are so high in order to stay on the competitive edge, that it makes it hard for a company to compete in any single market let alone multiple markets.
I honestly think that the Fury-X will end up doing better than the titan-x and 980ti due to the HBM. Also with slight overclock it could probably perform better. I agree with the arguement on how nvidias business practices are anti competitive. The Fury-X does seem slightly over priced though. Given driver maturity it will probably beat out the rest of the pack though.
One of the things I think will help AMD is HBM, it's revolutionary, different, and maybe a edge in the graphics market if it proves to be a successor to GDDR5, being stack-able will make graphic cards more compact like the Fury-X while providing more performance using less power.
If it all works as they say, and can be scaled as they say it can.....time will tell.
The "problem" with AMD is that the technologies they invest in are more of a long-term thing than a short-term performance boost. HSA and APUs are phenomenal tech. The problem is that for both of those to be worth anything to you, you have to use applications that are written using (at least partly) OpenCL - which most big application suites (for example by Adobe and Autodesk) still do not. If you want performance metrics for why APUs could be fucking amazing - look up how many floating-point operations per second a general purpose x86_64 CPU can do versus a graphics card (NVidia cards are a lot weaker than AMD cards in terms of raw power). So yes, even a very weak AMD APU beats any Intel CPU... at some specific given tasks. I don't approve of AMD labeling GPU CUs as "cores" in regards to processors, either... but to me, that's marketing and I don't really care about it since I expose myself to it as little as possible. The Fury X is really a quite powerful card. Only it's gimped in regards to double precision (FP64, it's slower at that than a single HD7950, heck it's slower than my M6600's FirePro M8900 mobile GPU - though it still beats a Titan X by a factor of 3 in this discipline...), and it has - in true AMD tradition - not yet that great drivers. Not saying it's worth the money right now, but it's not as bad a graphics card as some people are making it out to be. I have to agree, though, that the pricing makes no real sense. Yes, they can ask for some more money because it's quite quiet and comes watercooled out of the box, but I can't see anyone buying it at a price point of SEVEN HUNDRED FUCKING EURO. That's just insane.
When it comes to Zen and future CPU plans by AMD, there have been some leaked and reasonably authentic looking slides. According to these, a single Zen core will have four times as big an FPU as a single Vishera core, as well as a 1.5 times as many pipelines. This means that without any improvements to the components of the design, Zen will probably be four times as powerful as Vishera in games and at the very least on par with Haswell (probably better than Haswell though). If some similar slides are to be believed, there will be a high performance APU model with 16 cores, as well as a high performance Opteron or possibly even enthusiast chip with 32 cores and no iGPU. I cannot verify that either of those are true, however, and so we're probably looking at a good improvement over Vishera, but per core still not too close to Haswell.
As CyklonDX said, the Fury X really hasn't been properly tested yet. We can blame this both on reviewers, and on a poor launch.
Anyway, that said, the Fury X is still roughly on par with the 980 TI, which is still stupid-good. Secondly, the Fury X would have been another "Titan killer" if Nvidia had not undercut itself. Nvidia pretty much needed to, because imagine how bonkers people would have went if Nvidia had not released the 980 TI? Since Nvidia gave us, essentially, a Titan X-performer in a more affordable (but still exceptionally expensive) price bracket, you don't hear people bitching about it. Oh, I suppose people who invested in a Titan X will have 6gb of VRAM more... because it was worth the extra $350.
GPUs aside, the biggest thing that AMD needs to do for its CPUs is increase its power efficiency - preferably by a lot. If they can increase performance, even by 20-30%, that would be great, especially for people who are more productivity-orientated. However, with DirectX12 coming out, CPU performance for gaming will not be of dire importance for a certain span of time. As long as games are decently coded and optimized, like Tomb Raider was, then it will be the GPU that will be the only realistic limiting factor. That is not to say that AMD should not concentrate on creating a higher performance product, I am just saying that the most important thing for them to do is to increase power efficiency by a lot.
Well people are disappointed in what AMD has been releasing lately, and rightfully so. AMD enthusiasts have wanted something that is actually superior to what Nvidia has been pumping out, and AMD just hasn't delivered. Fury X is 980 ti money, and not beating the 980 ti in games. It just doesn't live up to the hype, plain and simple. AMD hasn't released a performance cpu in years. What is there to be excited about? Zen is a scramble move from AMD since their apu's haven't done anything impressive in taking market share. AM3+ doesn't even have native support for more then what, like 2 or 4 Usb 3.0? Its still on pcie 2.0. It is this old people, no matter what numbers it still can put out. AMD needs a 40% improvement in IPC, and they really needed a 2015 launch that was worth any meaning. Apu refreshes are cool, I guess. But I'm interested in something that can blow the socks off my Xeon 1231-v3, not something that is promised to be somewhat competitive if people actually optimized for it. Who cares what the potential is of a product if that potential is based off a failed "if" statement...
Like others above I really want Zen to kick ass, but I don't think it will be the Intel killer people want.
The one thing AMD have on their side for this is DX12, because there are large similarities between their Mantel (now largely Vulkan) and DX12 I expect them to know what to shoot for with a new CPU. IT would be business insanity for them not to aim toward that with a new enthusiast CPU and their big talk about their new GPUs and how they are build for it.
I would love to see what happens and whatever the outcome It will be the CPU of choice for me. Regardless of performance, it may sound like fanboyism and that is a part of it but it is also just the want for competition and supporting the underdog. I have never liked being that person who wore the same shirt as everyone else but I try not to shout about it.
On that Vulkan note though. I mentioned this before there has been no news about it at all recently. This worries me a great deal. Mantle was all talk and it had the go to back it up but it was ultimately canned (they can call it what they like) and then they said their effort with Mantle went into Vulkan. I don't doubt this but at all the events and releases surrounding the 300/Fury all the talk from AMD was heavily emphasising DX12 and almost nothing was said about Vulkan. OpenGL has floundered for years, it needs Vulkan to come good just like AMD need their CPU to be good too but here we are with AMD hyping and Vulkan nowhere to be seen even slightly while once again DX steals a march and secures a user/developer base locking them into windows for yet another generation.
Add to this the just abysmal showing from Valve with Steam OS and the future of open source gaming is looking as bleak as it always has. It was a once a big thing with AMD open source, but now look, where is it once again.
Some things came up that I've looked at a decent deal in recent time: 1. Power efficiency: this might sound surprising, but AMD APUs in laptops have a far lower TDP than most Intel laptop CPUs (A6 APUs for laptops - quad core + halfway decent GPU - have a TDP of 19W, whereas Intel dual core laptop CPUs still seem to stay at around 25W TDP largely, with the notable exception of ULV parts). It's of course to be noted that a new manufacturing process will make processors more power efficient as well and that Bulldozer (FX) CPUs are being manufactured in the 32nm process (which Intel ditched after Sandy Bridge).
Vulkan, as opposed to Mantle, is not an AMD-driven thing. Vulkan is being worked on by the Khronos group, a consortium that also works on technologies such as OpenGL and OpenCL. This means it is going to, just like OpenCL, be implemented by more companies and is not going to be limited to Microsoft or AMD platforms. This means a likely Linux implementation (at the very least for Intel platforms, probably for AMD as well), and we have Valve apparently looking into using Vulkan for their game engine, which is probably going to force a more widespread adoption of Vulkan.
Now, here's the deal with Vulkan and why I literally couldn't give a toss about DirectX 12: it is designed to unite in itself a graphics API and OpenCL. Now, most people don't know this, but OpenCL allows you to use not only a GPU, but also a CPU and any coprocessor in the system that supports OpenCL, without having to necessarily change your application to accomodate for several different makes of devices, or even types of devices. This could allow developers to tremendously improve how their games perform, and knowing Valve, they definitely will make use of this part of the Vulkan featureset, meaning that Valve games, when they use Vulkan will simply run phenomenally well when compared against Valve games not utilizing Vulkan.
I'm happy where AMD is in terms of hardware. It's just they don't have the software to take advantage of what's there. We have a GPU with HBM, so if they can release an APU with HBM that can be used along with HSA then I think they'll take the performance crown. I hope Zen has integrated HBM alongside an iGPU as with HSA they could prove a powerhouse and even outdo Intel. Linux as of kernel 4.0 has support for HSA baked in so utilization on that side of things should be phenomenal especially since AMD is now Linux first.
As to where they are now, I have an HD 7870 and am more than happy with its performance at 1080p, and even elected to get a 5820k over a 4790k because in terms of GPU performance I have enough. AMD seems to be more future proof in that respect. But then again I've probably only used it for around 5 days in total. It's practically brand new. What's more is the technologies AMD endorses are for the most part open source so Nvidia or Intel for that matter can utilize them.
You have people who are pissed off about the rebranded 300 parts, but I can nearly guarantee you Grenada will remain useful for at least another two years. Look at the 290s performance in Star Citizen, the best looking game I've ever seen. It's fantastic.
Where I think all companies need to improve is the performance gains they offer with each generation. I mean come on Intel. The 990X from 6 years ago nearly performs as well as a 4930k in multithreaded tasks. The 5-15% gains each generation that's been going on for nearly a decade needs to end. Nvidia, the same goes for you too and AMD isn't excluded however I can kind of make an exception as they focus on two segments... Unless you count Nvidias Tegra and Intels Iris Pro... Which I wouldn't considering they're embedded.
just to ease people, as they are expecting too soon Vulcan and DX12 to roll out in games. That won't happen right away. Most likely we'll start getting games on those api's by the fall of this year or Q1 2016; and there going to be like 1 or 2 titles in all. Don't expect wonders from both of those, and don't expect a lot of titles to actually support them.
it was only taken so fast because AMD forked out a lot of cash at EA DICE. DX12 is the other way around, if you want it game devs have to fork out cash for licensing. Vulcan looks open, but maybe valve has something in store to profit from it, its still not here - why? -
If you are referring to the Carrizo chips when you say '12-cores' I think you need to rethink it. I watched the AMD press conference and Lisa tsu was talking about Zen when she said 8 and 12 core chips. Which I think is more likely since AMD was the first to Dual core, Triple core, Quad core, Hexa core, and Octa core. Plus AMD was the first in Server processors to hit 16 cores. They even had a 24 core that didnt come out. It sounds like Zen is where they will come out with 12 core chips. Especially if they switch to SMT to match Intel's IPC. AMD has almost always upped the core count with each major platform release. Phenom marked the hexa-core release. Bulldozer marked the Octa core release. I think Zen with it's DDR4, will mark the Dodeca- (greek) or Duodec- (Latin) Core release.
If the fury had even one more gig of ram I'd probably be more likely to take the fury over the 980ti, but with the two having basically the same performance I'm going to go with the one that will handle more. Which is hi likely to be the 980ti because of ram. Sure, most things don't need more ram, but my skyrim sure does. Not enough grass being drawn.
I am excitedly awaiting more news on AMDs next line of CPUs. I love that both the A or whatever series APUs and the next FX series both use the same socket. Love it. So long as they can match intel's current line at that time and then beat out core count I'm game.
I'm referring to the last two generation of APU's not Carrizo. they are marketed as 12 Compute cores. anyone could mis-read that and think it will has 12 actual processing cores. AMD told us prior "we have a twelve core processor coming", and it happened to be 4 processing cores and 8 graphics cores. that's what i meant by AMD giving us false hope. you tell us something amazing is coming and it really isn't.
Not for the Enthusiast market. gamers or PC builders the highest end are strictly 8 core parts. the 16 cores and above are for servers.