Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
AMD Vega GPUs will launch "over the next couple of months".
http://techreport.com/news/31948/amd-says-its-vega-cards-will-launch-over-the-next-couple-of-months
AMD has said that the Frontier Edition professional card will launch first, and is still on track for a Q2 launch. Gaming and compute cards will launch shortly thereafter, and all within a couple of months. So the Radeon RX Vega might not launch until July, but as of right now, should launch no later than that.
That's not a promise that all Vega cards that will ever launch will do so within two months of today. But it is a claim that within the next two months, AMD will launch at least one Vega-based GPU in each of their consumer graphics, professional graphics, and GPU compute lines. Whether those will be a hard launch, a soft launch, or a paper launch is yet to be confirmed.
Given that the consumer graphics cards typically launch well before the professional graphics cards, I expect that this is due to a shortage of HBM2. The ongoing scarcity of $9000 Nvidia Quadro GP100 cards that also rely on HBM2 points in this direction as well.
AMD has said that the Frontier Edition professional card will launch first, and is still on track for a Q2 launch. Gaming and compute cards will launch shortly thereafter, and all within a couple of months. So the Radeon RX Vega might not launch until July, but as of right now, should launch no later than that.
That's not a promise that all Vega cards that will ever launch will do so within two months of today. But it is a claim that within the next two months, AMD will launch at least one Vega-based GPU in each of their consumer graphics, professional graphics, and GPU compute lines. Whether those will be a hard launch, a soft launch, or a paper launch is yet to be confirmed.
Given that the consumer graphics cards typically launch well before the professional graphics cards, I expect that this is due to a shortage of HBM2. The ongoing scarcity of $9000 Nvidia Quadro GP100 cards that also rely on HBM2 points in this direction as well.
Comments
I'm interested in Vega, certainly. But hardly excited. I'm anticipating another release akin to Fiji - a whole lot of build-up for what turns out to be an otherwise decent card, but ~just~ decent in the same way vanilla ice cream is decent, not exciting, not earth shattering, not record setting, and not a real challenge to the competition.
Maybe Vega is all of those things - exciting, earth shattering, record setting, and a real challenge to the competition. Then maybe it will live up to the hype that's been generated around it. But I think we'll be lucky for it to draw even to a 1080 in most cases, in a few corner cases challenge a 1080Ti, and everyone will say "Just wait for more DX12 titles," because that's the excuse for coming out with a product months behind the competition and still not being able to catch them.
For GTX 1080 ti it's 4 months. Will Vega offer enough to 1070, 1080 and 1080 ti that they would update? Or will those wait for Volta? I mean those GP104 and GP102 cards will offer decent performance even after Vega has been launched. Too little to late most people already moved on to Nvidia. AMD should give up on the higher-end of the market, it's pretty obvious that they're shit at it.
I still find the pascal GPUs to be marketing turds. Hey guys buy our $500 GPU that plays modern titles only a little better than the competitions $200 GPU because we have failed to make our GPUs compatible with DX12 for 2 years.
There are always people who want a fast new gaming pc or an upgrade. Now they might go for 1070/1080, in a few months it could be a Vega card.
Competition is needed or NVIDIA keeps raising the price with each generation.
It's likely a problem of HBM2. Nvidia announced the Tesla P100 more than a year ago, and officially launched the Quadro GP100 in March. Good luck finding one in stock. If AMD and Nvidia cards using HBM2 are both delayed or scarce while AMD and Nvidia cards using other memory standards are plentiful and available, I'm inclined to blame HBM2.
Basically, Nvidia bet on GDDR5X and AMD bet on HBM2. While HBM2 is the better technology, Nvidia won that bet simply because Micron was able to deliver GDDR5X in large volumes well before Samsung and Hynix were able to deliver HBM2.
The RX cards are selling fast and when they were released, you heard similar comments 'too little too late'. But now it turns out they sell very well. Finally much needed competition in lower segment.
I also notice that many here never take pricing into account as reason to buy a card. It is possible that AMD will release cards in between 1070 and 1080 prices. Aiming at buyers for who the 1070/1080 cards were just out of budget.
Vega was going to be the high end counterpart to Polaris, and fill in that high performance gap as upgrades to Fiji and Hawaii.
Given that most of the Polaris lineup launched just about a year ago, and has already seen a refresh, and we are still waiting on the Vega counterpart, yet the competition was able to release (1080) and also refresh (Ti), - yeah, I agree Polaris was not a case of too little too late, but Vega is different entirely story.
The question is, has AMD made a card that's competitive against Volta? Because it looks like they have neither the time nor manufacturing capacity to make good profits before Volta's launch.
Agree. I was on a website the other day and they were hyping up someone starting to work on a script for some movie. Really? Just give us the release date.
Epic Music: https://www.youtube.com/watch?v=vAigCvelkhQ&list=PLo9FRw1AkDuQLEz7Gvvaz3ideB2NpFtT1
https://archive.org/details/softwarelibrary_msdos?&sort=-downloads&page=1
Kyleran: "Now there's the real trick, learning to accept and enjoy a game for what it offers rather than pass on what might be a great playing experience because it lacks a few features you prefer."
John Henry Newman: "A man would do nothing if he waited until he could do it so well that no one could find fault."
FreddyNoNose: "A good game needs no defense; a bad game has no defense." "Easily digested content is just as easily forgotten."
LacedOpium: "So the question that begs to be asked is, if you are not interested in the game mechanics that define the MMORPG genre, then why are you playing an MMORPG?"
Now, that isn't whatever the final "for gamers" card will be. The Frontier Edition is supposed to compete with Titan. Oddly enough, plenty of people buy Titans just for gaming, and they do well.
We are also seeing the new Vega FE product produce gaming results 25-45% faster than the aging R9 Fury X card. For an architecture to architecture leap that is impressive, but the market has shifted greatly since the Fury X took the stage, with Maxwell refreshes and Pascal more than a year into its reign at the top of the performance charts.
Not looking like the Good Ship Lollipop for Team Green, but it's also not totally looking like the Titanic.
The big question is how much is available to be had from driver improvements. Performance in the professional tests is far more promising, and if that's where AMD's driver focus has been so far, there could yet be a ton of improvement in gaming.
That said, if Vega really is an all-new architecture and not just a slightly modified GCN, then there likely is a ton of improvement left to be had from drivers. Remember that we haven't seen a really new GPU architecture since Maxwell, and even that started at the low end and worked its way up. Driver improvements over the first several months weren't so obvious, as they only applied to overpriced, low-end cards that hardly anyone cared about.
The real test will come with the launch of the more gaming-oriented Radeon RX Vega. When a gaming card launches, you can't use the "but they're not gaming drivers" excuse. If that card matches a GTX 1080 Ti, then no one will care about the initial Frontier Edition reviews. If it can only match a GTX 1070, then AMD's GPU division is going to be in about as bad of a position over the next few years as their CPU division was before the launch of Ryzen. That's not a good spot to be in, though it probably won't mean bankruptcy simply because the CPU market is so much bigger than the GPU market.
I think the consumer version will perform much better than this in gaming. 13.1 tflops compared to Fury X's 8.9 tflops. The Fury X is about on par with the 1070 on most games unless you do an unbalanced type of testing that typically is unfavorable for certain cards. Honestly, if the product reviewer doesn't have the Fury X about on par with the 1070, then I believe that they are doing some number fudging.
There is no actual difference whatsover between the 2 modes of drivers.
https://www.pcper.com/reviews/Graphics-Cards/Radeon-Vega-Frontier-Edition-16GB-Air-Cooled-Review
I think Vega will compete with the 1070 cards and will probably priced similarly.
Note that they don't show any Nvidia workstation cards in the gaming tests. Some of the Quadro cards might do rather poorly there as the drivers aren't optimized for gaming.
The question isn't so much whether there are driver improvements to come for Vega when they make gaming-focused drivers, but how big of improvements. If 5% is all they've got, then yeah, this is a disaster for AMD. But if driver improvements will typically add 70% to gaming performance, they're going to have quite a winner on their hands. And from the paper specs, the latter is plausible, though that is not at all similar to being probable.
If the only driver options in the Vega Frontier Edition are going to be which settings are visible, then either it doesn't support the professional drivers, it doesn't support the gaming drivers, or both. That wouldn't make any sense at all.
This cards looks too much like Fury cards: High computing power, and high power consumption, but I've got doubts whether AMD can translate that computing power into actual performance with any driver improvements.
Fury/Fiji XT isn't a bad card, and still isn't a bad card. It's limited in VRAM, although honestly I'm not sure exactly how much that hinders it's performance.
The only reason Fury looked bad upon release 2 years ago, is that the competition had typically higher performing cards available (Maxwell on the 980 Ti / TItan X), and upon release, Fury X was being sold at a premium price that didn't make a lot of sense in the price/performance.
It was a matter of too little too late, in my opinion. AMD had been gunning to take down the 980 with Fury, and did so. But nVidia beat them with the 980Ti in performance, and then dropped the price on the 980 to undercut in value, and it didn't go well for AMD. Had AMD priced the Fury X against the 980, they may could have come out better (not necessarily ahead), but given the price of HBM and the integrated water cooler, I don't know that they could really afford to.
Once Fury came down in price (like it is often found today), it's an excellent card, the price just needed to be right. The fact that it's still standing up 2 years later shows a lot, I think. But it really looks like history is about to repeat itself: Vega will probably match up very well against a 1080, but it probably won't (consistently) beat a 1080Ti - yet it will probably be priced about the same as the 1080Ti. Now that's a lot of conjecture on my part, but it's based on history and what data we have available today.
What makes the preliminary Vega benchmarks look so bad is not that it's losing to this or that Nvidia card. The problem is that it's performing markedly worse than you'd expect from a scaled-up Polaris. If you took a Polaris 10 and doubled everything except for the parts that you trivially only need one of (e.g., PCI Express), you'd have about the same power consumption as Vega 10, a substantially smaller die size, and meaningfully higher performance we're seeing. And that's in spite of Vega being built on a more mature version of the same process node as Polaris.
That would make Vega a worse architecture outright than GCN/Polaris. While that does occasionally happen (see Bulldozer), it's pretty for a new architecture to be worse than the same company's previous one. I see several possibilities in what happened:
1) Vega really is just that bad. Zen is good enough to keep AMD in business until Navi arrives, but Nvidia is going to dominate the GPU market for the next couple of years. Oh sure, AMD will have a product to compete outside of the high end, but if you need a 484 mm^2 die and 300 W to match the performance that your competitor gets in 314 mm^2 and 180 W, you just lost that generation and badly.
2) AMD botched something major in the Vega 10 die that will be fixed in subsequent iterations of Vega. This is pretty rare, so I'll have to cite the only example that comes to mind: the Radeon HD 2900 XT. It looked terrible as compared to Nvidia's Tesla (GeForce 8800 GTX), but a little over a year later, AMD demonstrated with the Radeon HD 4870 that they had the right idea and pretty handily won three consecutive generations of hardware until Nvidia copied their approach of more shaders clocked lower in Kepler.
3) Preliminary drivers don't perform well and that's all that we're seeing. By the time the GPU launches, the problems will be fixed and Vega will be a competitive architecture. Maybe a little better than Pascal or maybe a little worse, but at least competitive with a Titan Xp. It's quite possible that the reason AMD held off on the consumer version of Vega is that they knew the drivers simply weren't ready for gaming and there are huge performance improvements coming. There are always massive improvements to be had over the very early drivers on a new architecture. The public never sees those very early drivers, but might have seen an earlier version than most with Vega, as AMD has strong incentives to get Vega out there as soon as possible.