Of course i have, we know scalability is coming with navi,but to make 2 gpus work as one will take more than an interposer and infinity fabric i would think, there have been many ultra deep discussions about this over at beyond 3d forum from developers and such.Not sure if you understand. It is a single video card... surely you have heard of this. Perhaps you have not thought about how AMD might use their technology across their platforms..? You must be aware of AMD's APU and how the gpu works, now look beyond that, to a Vega + Vega on single chip.
dERP!
That will be amazing when it happens. Given the massively parallel nature of GPUs, such an arrangement is the most intuitive fit. AFR as well as all the other kinds of Crossfire/SLI arrangements always seemed like a scabby band-aid and they of course didn't work a lot of the time. Making multiple GPUs transparent to the API will be revolutionary.Infinity Fabric (found in Vega and Ryzen) should, in theory, allow for memory coherency between the GPUs. With the proper drivers, two chips should be able to act as one big GPU and stay out of the AFR mess. Whether this will actually be ready for Vega or have to wait for Navi is the big question. It's definitely the future.
I think he's just angry at the launch scenario: not quite ready. Or not, I could be wrong. FWIW, I understand that some people are legit underwhelmed or even angry that Ryzen launched with (entirely predictable) bugs. People just want things to work. Maybe he's new to the whole bleeding edge hardware thing and hasn't experienced every single brand new architecture launch up to this point, like most have. So, I can excuse that type of ignorance. Ryzen is obviously in a rather stellar performance state right now...with obvious MoBo supply issues and obnoxious Windows and BIOS bugs to be ironed out, but it's a solid performer in its advertised use scenarios.
At the same time, Vega is going to be a much harder sale if it releases with similar, or even worse non-optimized issues. 1080ti will be out and (Assuming no supply issues...thought that does seem to be the case?) saturating the market. nVidia has not been stagnant all these years, unlike Intel (yes, I understand there is a difference in physicals with actual hard limits when comparing modern CPU design vs GPU design), so the competition on the GPU end will be a wholly different beast with AMD.
I fully expect Vega to be a monster on paper--but with even more of the "Well, if only devs utilized the cool tricks and designs we have in our silicon, it would be that much better!" issues that you often have with AMD. I'm definitely going to be on board with Vega 11 or maybe even 10, but with the understanding that it will mostly likely underperform the 1080ti on launch, with a pile of performance left on the table.
Ask and you shall receive
Ryzen Tested as "Quad-Core" by Forced Disabling
Not bad IPC compared to Kabylake (Skylake D0)
Some CompuBench results
VideoCardz - AMD Vega with 64 Compute Units spotted
Not sure about frequencies... final ones or will it change for a production cards?
BTW, CompuBench detected 4GB VRAM only
VideoCardz said:With a single precision compute of 12 TeraFLOPs per second on a GPU with 4096 cores, and considering TeraFLOPs is a function of Clock Speed * 2 Instructions Per Clock * Cores, you are looking at a Vega 10 graphics card that is clocked at roughly 1500 MHz assuming it has the same amount of NCUs. Considering the fact that the MI25 is passively cooled and also if the consumer version has less cores, it will be clocked significantly higher than the 1500 MHz mark!
Of course i have, we know scalability is coming with navi,but to make 2 gpus work as one will take more than an interposer and infinity fabric i would think, there have been many ultra deep discussions about this over at beyond 3d forum from developers and such.
Im not sure how they could make it work, but im pretty sure its not as easy as you think.
There is multi adaptor in dx12 which would solve the issue in software, problem is game support.
Why would I wait for Vega to Shine? I could get a 1080Ti, then get the 2080Ti by the time Vega starts to "shine"?My expectations as well. AMD has trouble with new architectures. GCN didn't find some legs until Never Settle drivers 9 or 10 months after launch. And since then there have been years of gradual relative performance increases, like watching the relationship of 290X to 780 Ti and 980 change. It may take a while for Vega to shine.
Of course i have, we know scalability is coming with navi,but to make 2 gpus work as one will take more than an interposer and infinity fabric i would think, there have been many ultra deep discussions about this over at beyond 3d forum from developers and such.
Im not sure how they could make it work, but im pretty sure its not as easy as you think.
There is multi adaptor in dx12 which would solve the issue in software, problem is game support.
Also, how are we getting 1500 mhz from 1200....
Not only are you wrong, your condescending tone isn't appreciated.
Maybe I'm new to the whole bleeding edge hardware thing? Lol, please take your condescending remarks elsewhere.
I won't stoop to that level in pointing out the flaws in your post in the same condescending manner.
Not sure if you understand. It is a single video card... surely you have heard of this. Perhaps you have not thought about how AMD might use their technology across their platforms..? You must be aware of AMD's APU and how the gpu works, now look beyond that, to a Vega + Vega on single chip.
dERP!
Ahh of course, everything is integrated at the silicon level not the substrate level as i was thinking. Jesus. I wonder how many they can put together. Then Navi architecture with scalability might be on 7 nm with increased bandwidth and 'core' count. So 2 Vegas just like there is 2 CCXs in Ryzen?
At LiquidSky, we are constantly upgrading the cutting-edge tech we make available to you. While some of our upgrades are just minor tweaks, others are huge hardware changes with massively improved performance specs. Now, it’s time for our biggest hardware upgrade yet. With the new LiquidSky, we’ve upgraded both our hardware and pricing plans, giving every gamer more flexibility than ever before.
AMD Vega GPUs to Power LiquidSky Game Streaming Service
Finally, while AMD isn’t releasing any extensive new details about their forthcoming Vega GPUs at the show (sorry gang), they are announcing that they’ve already landed a deal with a commercial buyer to use these forthcoming GPUs. LiquidSky, a game service provider who is currently building and beta-testing an Internet-based game streaming service, is teaming up with AMD to use their Vega GPUs with their service.
Not sure if...
AMD's current push is away from the single "monolithic" die idea. All current APUs are on a single die, but this has nothing to do with Vega.
I'd say the price point is nothing new except this time they have no competition that we can see and Nvidia are about as greedy as it's possible to be.So you're saying that the third year time in a row of high end ti cards would surprise amd?
Sorry but your theory is just not based in reality.
$650-700 price point for the ti card is nothing new.
780ti
980ti
1080ti now surprises you?
Please tell me how this price is a surprise...
I still think the best solution for both AMD and consumers is multi adapter dx12, sniper elite 4 gets 100% efficiency out of 2 Rx 480s, perhaps interposer/infinity fabric/HBCC will make it much easier for AMD to get dual gpu support working in games where the devs have not bothered with multi adapter, probably easier to write drivers for and better more consistent performance, certainly over crossfire.No one said it was easy, but with Infinity Fabric it should at least be doable. The biggest barrier up until now was maintaining cache coherency, and Infinity Fabric allows that. It will definitely be a lot of work to get the drivers working, but if done right, it should be possible to make it transparent to the system and appear as one big GPU instead of two (or more) smaller ones. I have no doubt that this was one of the specific use cases AMD had in mind when they developed the Infinity Fabric IP.
Developers will never do the necessary work to get multi-GPU working. This is a chicken-and-egg problem: very few gamers have multi-GPU rigs, therefore developers don't think it is worth the time to code for, therefore few gamers see the need to buy multi-GPU setups... It's impossible to break this deadlock unless the vendor can make multiple GPUs fully transparent to the system. Add to that the fact that on smaller processes it is becoming increasingly harder to make big monolithic dies, and it's clear this is really the only way forward. Raja Koduri has talked about the need to "move past CrossFire" so this is definitely going to happen; the only question is how soon.
AMD has freesync 2, at some point I expect normal TVs will use the technology.The way w3rd talked you'd have thought infinity fabric was Navi....
It's not.
If Vega has scalabilitiy, they'd have talked about it. Like you've so kindly noted, this is coming in Navi apparently, and how they intend to make that work we don't know.
Also, how are we getting 1500 mhz from 1200....
Why would I wait for Vega to Shine? I could get a 1080Ti, then get the 2080Ti by the time Vega starts to "shine"?
It's just not the same value proposition at the high end. I'm not interested in waiting for a Vega GPU to shine. I'll get Vega, because I'm screwed into it with Freesync. But I feel I'm getting poor value, then I'm done with AMD for high end gaming and will eat the Nvidia tax. It's not doing AMD any favors that Nvidia gets the best monitor tech released for them first.
AMD's saving grace is the monitor choices for Freesync. I can't get a monitor like this from Nvidia Gsync. AMD needs to get their partners to talk about new upcoming 4K High refresh rate monitors like Nvidia has. Otherwise, it actually seems like GTX 1080Tis would be better since High refresh rate 4K monitors haven't been announced on AMDs side.
Not only are you wrong, your condescending tone isn't appreciated.
Maybe I'm new to the whole bleeding edge hardware thing? Lol, please take your condescending remarks elsewhere.
I won't stoop to that level in pointing out the flaws in your post in the same condescending manner.
I can finally say with confidence that reviewers and users are stupid. Just utterly stupid. I've been saying for awhile now that Ryzen 7 is a TERRIBLE pick for gamers looking for value.
This just confirms it. You can get amazing performance out of the cheaper Ryzen CPUs in comparison to the 7700k. There is NO reason to compare Ryzen 7 with a 7700k when the true comparisons in Ryzen 5/3 will show far better price/performance.
I need to see the 2 core/4 threads for my own personal usage since 3 cores is out of the question due to the makeup of Ryzen. That's a huge shame but whatever.
Once SMT+Windows works well together in ways that do not cause regression, I expect Ryzen to be a top notch performer. This again comes down to AMD not being ready/prepared. They should have known about SMT causing a regression in gaming applications and worked with Microsoft especially regarding Windows to ensure that Windows didn't do idiocy with SMT just like it did with HT.
I'm in no rush for Ryzen 7, so I've got time to wait for the bugs to be ironed out like I said earlier.
Either way, your benches prove my main point about Ryzen 7 vs the 7700k.
I can run 2 4770k like experiences with Ryzen 7.... That's something a 7700k can't do PERIOD. So no point in comparing the processors. 7700k should only be compared with low end Ryzen 3/5. It's not an HEDT platform CPU and should NEVER be treated or talked about as such.
It's a straight up crime that reviewers did not test Ryzen extensively in a 6 core configuration. It would shut down all this 7700k talk instantly.
Ryzen 5 > 7700k for gaming in $/Perf. Easily.
If we assume Vega should be running atleast @ 1500mhz core, then we can extrapolate what the scores would be once clocks are finalized
1500/1200 = should net a 25% performance increase with 100% clockspeed scaling
Face detection: 1080TI is ~60% faster then Vega
"simulated 1500mhz Vega" = 194.58
1080TI = 313.26
TV optical flow: Vega is ~4% faster then 1080TI
"simulated" 1500mhz Vega = 72.13
1080TI = 69.22
Ocean surface: Vega is ~20% faster then 1080TI
"simulated" 1500mhz Vega = 4760.57
1080TI = 3944.8
Particle simulation: Vega is ~16% faster then 1080TI
"simulated" 1500mhz Vega = 2123.35
1080TI = 1816.88
T-Rex: 1080TI is ~7% faster then Vega
"simulated" 1500mhz Vega = 18.48
1080TI = 19.773
Video composition: Vega is ~5% faster then 1080TI
"simulated" 1500mhz Vega = 202.2
1080TI = 192.586
Bitcoin mining: Vega is ~7% faster then 1080TI
"simulated" 1500mhz Vega = 1509.07
1080TI = 1408.061
Now the big question is, how much above 1500mhz will the consumer Vega10 be running at ?
Afterall we have gotten that 1.5ghz number from a professional workstation-card leak, and as we know, consumer version tends to be clocked higher..
Not bad considering the 1500mhz Vega10 should be running at a 225w terminal envolope.
https://videocardz.com/65521/amd-vega-10-and-vega-20-slides-revealed
Not to be annoying, but you're arguing against yourself. In one sentence you say Nvidia is leading due to their brand, in the other due to a history of making good products.
What does this have to do with anything I said?Your metrics and weights are far inferior to what an actual
"gamer" is looking for.
Most PC gamers don't do so in a vacuum. They have friends... they are not people perusing the internet and click on an add and install a "game". Most PC gamers are Platform gamers. Only since Steam, which capitalized on young kids, who were out of the home, has PC "gamer" become more casual and at ease. In which, anything is a form of entertainment.
There is a big difference between a PC gamer and even a Mobil (laptop) gamer (massive difference). The broad strokes in which you are trying to paint Your picture, of a typical PC builder is laughable. Arnold Swartz commercials for his Mobile Strike on are TV all the time & those are the dumb, proto-typical being you are trying to bring to life in your argument, as a gamer?
Enthusiasts today, who are building rigs today... are not looking to stifle their experience, they are looking out towards the future, & what will enhance their abilities for future engines, etc. Not hard to understand, how & what a human needs/thinks about. They are looking for efficiency of values (things important to the want/need). I can could persuade you (as a 37 year PC gamer), that Ryzen is the easy choice for a base platform, for any build you wish to endeavour. The case is simple: AM4
Gamers look at the whole SYSTEM when building. AMD is providing more legroom with their platform. Now, and looking forward, to me AMD's platform has what the People need & want. As a consumer, it is plain and simple. Intel is not even in the picture because it's platform, isn't in the picture.
The 7700 would be a stop gap, or wasted choice in just a few years. Because the 7700 today's price/value, doesn't exceed Ryzen's future value... at any tier.
I run three rigs, one is always top percentile (ie: 1080ti incoming), and the other two rigs just get trickle downs. As a gamer, One does not even need a benchmark to see Zen's value. I have three i7 (3770k, 4770k & 4790k) that need upgrading. I am patient, the two builds I am doing, are over the next 2-4 months. I agree that AMD's birthing pains could've been better, but I am sure in 2 months time when Vega drops, full AMD will be the gamer's choice..!
I am looking to go mATX on AM4. Why not..?
AMD has freesync 2, at some point I expect normal TVs will use the technology.
Besides if Vega is only 10% slower out of the box yet consumes less power and is 25% cheaper? Why would you think Vega is a bad deal?
Considering it has a more advanced uarch and will almost certainly get faster over time, I think Vega will be an awesome gpu, only way I see a failure is if is 20-25% slower than 1080ti, for its rumoured die size that wouldn't be good, even taking in account it future proof uarch.
What do you mean "This time they have no competition?"I'd say the price point is nothing new except this time they have no competition that we can see and Nvidia are about as greedy as it's possible to be.
They price their cards to be as much as they can possibly get away with and based on how much more Titan XP cost vs Maxwell if they had no serious competition
I'm pretty sure 1080ti would have come in much nearer $1000 than $699. I just sold my two Titan XPs and will replace with single 1080ti and see what Vega brings.
To my understanding, Vega 10 comes first, then Vega 11.New day, and a new leak
https://videocardz.com/67275/amd-vega-spotted-with-4096-cores-and-8gb-2048-bit-memory
Just yesterday we found the first CompuBench result revealing 64 Compute Units in Vega. Today we share another leak, which confirms 64 CUs, but also memory subsystem configuration.
AMD Radeon RX Vega: 4096 Stream Processors and 8GB memory
SiSoft benchmark detected 64 Compute Units on 687F:C3 device (so not C1 like in the previous leak). This device has 8GB 2048-bit memory configuration, which means two HBM2 stacks, each 4GB and 1024-bit. When it comes to core clock (344 MHz) and L2 cache (16 kB), those are definitely wrong readings, so I wouldn’t pay attention to them.
Therefore, this leak would “confirm” that Vega has 4096 Stream Processors (64 * 64 CUs) and two stacks of HBM with 8GB in total. What we don’t know is whether 687F:C3 is the top model in RX Vega lineup, as there are clearly few different variants. So far there is no trace of cut-down chip or Vega 11.
Either way, it seems AMD engineers got really busy recently…
Here is the GTX 1080's scores for comparrison. And This Vega part is ahead by a reasonable amount, even if its only running by alpha drivers and clocks(1200mhz): http://imgur.com/z68xbLd
I'm somewhat curious what the die size for big Vega turns out to be. Some people have made estimates ranging from 470 - 520 mm^2 based on analysis of a picture where AMD held up a Vega die. Hopefully they've done a lot of work to improve their performance per SP as one of Fury's biggest problems was being able to utilize all of the hardware, but at the same time the move to 14 nm should allow for a much smaller chip so hopefully all of the special sauce is worth it. Yes there's more money to be had in the professional and HPC markets, but I'd like to see a competent gaming GPU out of AMD as well.
Also, I hope they're not using the same fabrication process as Polaris or Ryzen as it seems that the 14 nm LPP isn't going to allow for the higher clock speeds that AMD will likely need.
From a purely business look at it as I'm no tech expert, I don't see why Vega would have any massive flaw Polaris would have right?
I don't think it's Polaris that's flawed, but both Polaris and Ryzen don't have a lot of OC headroom and I think that's a result of the LPP process that they're fabbed on as much of anything else.
When I say flaw that's the massive one. Using gf is a flaw... Until gf can make produce something in which we all aren't worried it's of course going to probably be a flaw.I don't think it's Polaris that's flawed, but both Polaris and Ryzen don't have a lot of OC headroom and I think that's a result of the LPP process that they're fabbed on as much of anything else.