Here we go again…
Image credit: WCCFtech
Even more information has allegedly leaked out ahead of AMD’s official announcement of new 300-series Radeon GPUs, this time from rumor site WCCFtech. This information is totally unverified at least from any public source, but it is very specific regarding both price and GPU.
Here is the list published by WCCFtech in their report:
| Graphics Card | GPU | Price |
| R9 390X 8GB | Enhanced Hawaii XT | $389 |
| R9 390 8GB | Enhanced Hawaii Pro | $329 |
| R9 380X 3GB/6GB | Tonga XT (NOT CONFIRMED) | |
| R9 380 4GB | Tonga Pro | $235 |
| R9 380 2GB | Tonga Pro | $195 |
| R7 370 4GB | Pitcairn | $175 |
| R7 370 2GB | Pitcairn | $135 |
| R7 360 2GB | Bonaire | $107 |
As to whether this comes via leaked slides or is complete guesswork, we’ll likely have no answer until the official unveiling. Such an announcement is likely the purpose of the AMD gaming event at E3 which is now just days away. We can only hope that Fiji will in fact be making an appearance at the show as it does not appear on this list (again, if accurate).
Enhanced Hawaii for 300$? No,
Enhanced Hawaii for 300$? No, thanks.
Depends on the enhancement.
Depends on the enhancement. We already know that Hawaii is competitive with 970/980 at 4K. Add to that:
8GB confirmed
h.265 decode?
HDMI 2 for home theater?
Global Foundry 28nm process for leakage fix?
Aftermarket coolers on par with Tri-X OC?
It could most definitely be worth it.
sweet but if people want to
sweet but if people want to buy this amd gpu go with the tonga chip that r9 280 and 280x that chips is wonderful
hawaii > tonga
hawaii > tonga
Hawaii is missing some
Hawaii is missing some important features which tonga has (GCN 1.1 vs. 1.2).
The 1.2 design update includes (from Wikipedia) improved tessellation performance, lossless delta color compression in order to reduce memory bandwidth usage, an updated and more efficient instruction set, a new high quality scaler for video, and a new multimedia engine (video encoder/decoder).
Hawaii still has a lot more shaders though. The 1.1 parts (Hawaii) get you TrueAudio and power tune that is not present in 1.0. These don’t seem like they are really that important compared to the 1.2 features though. A high clocked Tonga variant may be a better choice at some price points depending on what they do with the Hawaii based part.
I’d really like H.265
I’d really like H.265 encoding so people can finally start enjoying the lower bitrate requirements compared to the now fairly dated h264 standard. Decoding isn’t a huge deal to me given its currently well possible on any recent desktop processor but will almost certainly be included if it will do hardware encoding.
On the other hand why worry about HDMI2.0? Anything thats able to display more information than you can carry on older HDMI spec already has display port input, so your not limited their. As for outputs even most current higher end cards aren’t suited to gaming on multiple displays that won’t run off older HDMI, so its not like your going to be able to hook up more displays than you have display port outputs by adding HDMI2.0. SO really I can’t see what benfit gaming cards get out of the new tech, yes its nice to have newer stuff since it’ll take longer to be out of date, but really it shouldn’t make any odds >99% of customers.
Also the Tri-X and Vapor-X are the coolers, which are avaible on the r9 280x and r9 290(x). The “OC” is just the factory clock-speed version, so its not really something you need to mention when talking about the non-reference cooler. But personally I’m not much of a fan of the huge triple fan designs, especially the trivapor-X since they use heaps more room (3 slots + long length) and don’t really provide much advantage over well made duel slot, duel fan coolers like the ACX2.0 found on EVGA’s FTW or MSI’s Frozer.
Sure, in an open air test be you’ll get better temps with the bigger cooler, but in a real mATX case or ATX where you have more than one card? The extra breathing room puts temps close enough tot he duel fan cards that its not going to have a noticeable effect on your OC or GPU life, so I’d rather have the more compatible product. Especially when its often cheaper and allows you to mount a water-block or other aftermarket cooling solution should you want to further down the line.
I’m really curious about what
I’m really curious about what they have to offer. I went with a R9 270X last year, and I would love a card with more preformance overhead. I would love a GTX 980 of 980Ti, but the price is just staggering. If this leak is showing new pricing, then the 390 series is very enticing, regardless if it’s a “rebrand” or not.
As long as there was some
As long as there was some modificaiton to the silicone on the “enhanced” Hawaii (and not what happened to Kaveri to Guvari(??) with just a small speed bump…
Either way this is not good for AMD. the Bonaire and Pitcarn should all be replaced with Tonga due to its newest GCN archtecture. I hate 3+ old silicone as a new product…
Ye, you just want something
Ye, you just want something newborn every time you visit the store.
And i bet you have precisely ZERO clue how a processor works at all.
Really inept and spoiled child thinks this way. Ignoring that price/performance is all that matters.
I understand you might have
I understand you might have missunderstood my orginial comment, and I”ll try to be more clear in the future to ensure that is avoided.
However having a major in Computer Engineering, does give me a smile when i read comments like the one you mentioned. I know very well how it works, and how process technology works. Build a few microprocessors myself (though not on that scale)
I’m also an AMD fan as being the underdog, but after what Intel did during the P4 days (corruption, etc to ensure companies buy Intel product even if AMD Athlon was far supperior) ensured AMD now had extremely limited funds to develop anything. Intel even payed AMD around 2billionusd + to drop the lawsuit which said it all.
As for the CPU, I’m eager to see what Zen has to offer and waiting till then to upgrade my AMD FX8350.
Same thing for Fiji, my R9 290x is also waiting for an upgrade. 🙂
As for seeing something newborn. Not every year mind you, but outside of Tonga which was released in September 2014, and Hawaii in October 2013, rest is 3+ year old tech. At this point the could have taken the Tonga (285) and refreshed all the way down. Ensuring that most of products are based on tech from at least last year.
Again, due to their limited funds I understand it. It does not mean I have to be happy with it
I hope at some point in your
I hope at some point in your classes they teach you the difference between silicon and silicone.
What does it matter? I want
What does it matter? I want both of them to go faster. Faster! GO FASTER!
^
HA, this dude gets it!
^
HA, this dude gets it!
They’ll be on sale after a
They’ll be on sale after a few months so the deals will be even better on the tweaked Hawaii based SKUs. I’m eager to see what Zen Based workstation APUs with HBM and Greenland Graphics are going to cost, and if the HBM can be had with 16/32 GB of memory then maybe there will be some interesting more affordable fire pro APU based workstations. On the gaming side once HBM becomes the standard the prices will come down, remember with a interposer based memory systems and each HBM memory stack getting an wide 1024 bit connection to each memory stack, all the memory trace complexity of the PCI card is moved onto the interposer, making the manufacture of the PCB much easier and less costly. We are talking about removing all of the memory traces from the PCB, as well as DDR mounts, for most consumer grade gaming systems, so once HBM comes down in price, the savings from moving complexity off of the PCB will become even more attractive, and include in that the savings(Space and complexity) removed from main/PCI based boards, so APUs with HBM will have an advantage in pricing relative to DDR based systems in not too log a time period, economy’s of scale will drive HBM’s cost down rapidly, and the added savings from motherboard/mainboard complexity and space savings will force all CPU/SOC manufactures to adopt HBM/other stacked interposer based memory for SOCs, as well as GPUs. There definitely is a new period of HBM based systems just beginning while the older DDR RAM based Memory based systems will begin to decline, except for the HPC/Server market which will use both HBM and DDR RAM for large in memory data base/analytics workloads.
Hawaii still has a lot more
Hawaii still has a lot more shaders than tonga. I think the 1.0 based parts will finally be going away and will be replaced with 1.2 parts based on Tonga. We will probably still have Hawaii based parts with low tesselation performance. From what I have heard about HairWorks, running at high tesselation does not seem to be worth it, even on Nvidia hardware. It also seems to be a bit buggy, so most people seem to be just shutting it off. I don’t think the tesselation performance is really a bottleneck yet. Regardless of how good your tesselator is, the rest of the gpu still needs to be able to handle all of the extra triangles produced by the tesselator.
Silicon.
Silicone is for one
Silicon.
Silicone is for one of my favorite parts of the female anatomy.
Josh, is that you?
Josh, is that you?
Bigger does not equal better.
Bigger does not equal better. Bigger often sacrifices too many other features for my taste.
Silicone? Someone thought I
Silicone? Someone thought I said that I lived in “silicone valley” once. I said, “Nope, that’s LA, not Santa Clara”.
I suspect that both companies have been hit with issues due to essentially cancelled process nodes. I don’t know how much resources they put into 20 nm planar designs before reverting back to 28 nm. There is some possibility that they will jump directly to 16 nm FinFet eventually.
This supposed leak doesn’t tell us anything that we want to know. What we want to know is if they have done any redesign/re-spin or ported Hawaii to a different process. A better process, even at 28 nm, would allow them to get the clock speed up and/or the power consumption down. Keep in mind that relatively small clock speed increases can make big differences in performance compared to CPUs. It is also possible that they are getting better yeilds but it is exactly the same part and process. They may still be able to increase clock speeds a bit. It probably will not be worth it to upgrade from a 290x to a 390x regardless of what they have done. That doesn’t mean that the 390x will be a bad product though. If you have an older card, upgrading to a 390x may still be the best deal in the price range.
It would be great if they could build essentially a big Tonga gpu; Hawaii numbers of shaders plus GCN 1.2 (at least) architectural features. I doubt that we will get this though. Hopefully they can at least EOL the GCN 1.0 based gpu variants and replace them with Tonga (GCN 1.2) based parts. A Hawaii variant will probably still be at 1.1, but since most games will be using DX11 for a while yet, that may not be that big if an issue. I don’t know if the HBM part will be considered a new GCN version beyond 1.2.
the new big performance card
the new big performance card is just proportionally more expensive.
Basically the PRICE/PERFORMANCE ratio… the ONLY thing that matters has stayed the same for the last two years… for both nvidia and amd.
So, in my humble opinion both these companies can go fuck themselves sideways.
It will be interesting to see
It will be interesting to see how HBM will figure into the PRICE/PERFORMANCE ratio, once all the economy of scale drives the cost of HBM down relative to the DDR RAM based systems currently in use. The Price/Per. Ratio is going to be disrupted initially by HBM, but the overall cost savings in moving the memory system’s complexity off of the mainboard/motherboard and onto the interposer will force the Price/Per. ratio to improve in the next few years as HBM supplants DDR RAM. Once All SOCs, and GPUs begin to take on HBM as standard, expect the motherboard/mainboard dimensions to shrink drastically on consumer/gaming systems, the space savings on the PCI based GPU cards will translate into space savings on the motherboard as SOCs/APUs with HBM begin to dominate the market. Those mini/micro ATX from factor systems are going to pack in a whole lot more processing power, once the high performance Zen based Gaming APUs hit the market, being derived from the AMD Zen based HPC/server APUs with HBM and Greenland graphics. It’s most likely that the Zen based gaming APUs will be made from binned versions of the Server/HPC APU SKUs with some units fused off.
Well said. I still believe
Well said. I still believe flagships should sell for around $350. We certainly can’t blame inflation for the doubling of this price point as O’dickhead and the fed tell us there is no inflation. It is just that there are a shitload of people willing to pay whatever Nvidia asks, as if everyone is swimming in money.
They have been stuck at 28 nm
They have been stuck at 28 nm for years, so the price for some specific amount of performance has remained relatively close. Also, there was a time when a high-end card was 100 W. Now you can get ~300 W monsters. These giant GPUs are still not really that bad compared to CPUs. If you want a CPU of similar die size to these GPUs, you will be paying probably more than $1000 for just the chip. With the gpu, you get the entire card and a lot of high-speed memory for that price. I don’t see too many people complaining about the price of Intel extreme edition parts.
Something is wrong with this
Something is wrong with this whole industry.
There are lots of 4K monitors on the market for some time, but graphic cards are strugling to deliver even in 1440p.
And the prices for that shitty performance are sky high.
So is my pocket struggling to
So is my pocket struggling to keep up with the costs to buying the monitors and cards to match 🙂
It’s a relatively new market.
It’s a relatively new market. If you’re trying to save money, wait for the dust to settle. Of course, by then, they’ll be pimping 8k so … hang onto your wallets then too.
It’s not the industry as both
It’s not the industry as both Nvidia and AMD want to come out with significantly better products, but they are limited by what can be produced. Since both TSMC and GloFo have had troubles getting new process tech working well that’s impacted new GPU production. Though clearly it’s hurt AMD more as they have a limited engineering budget. Hopefully next year will see a complete refresh from both AMD and Nvidia as the 16NM/14NM chips start being produced in mass quantities.
There seems to be a lot of
There seems to be a lot of problems going below 28 nm planar. Also, it seems like there may be problems with 14 nm FinFet. We are pushing against too many physical limitations. There is going to need to be some massive switch in process technology going forward. I think I have seen reports of Intel going with different technology for 10 nm.
Anyway, when we get GPUs using smaller than 28 nm processes and HBM 2 (allows 8 GB or more), then we should have plenty of power to run 4k displays. The gpu makers always want game developers to push the limits, otherwise, you wouldn’t have any reason to upgrade to a new card. We are getting to diminishing returns though. Better graphics cost a lot more money to develope and they do not necessarily make for better game play.
If everything goes well, by
If everything goes well, by the end of the year I will hopefully be able to get a VRR monitor and compatible GPU. That being said, I really don’t care which side has the better deal, as long as the VRR experience is good both in the VRR window and also below/above that same window. So yes AMD, I’m looking at you. Even if you have the better deal when I’m ready to spend the money, if that “below VRR window” isn’t fixed, i’ll have to go with the other side. And no, I won’t be able to wait either. When it comes to spending money in my house, I either spend it once I have it available, or else that oppurtunity is gone.
Gsync seems like a good short
Gsync seems like a good short term solution to push the industry forward, but I don’t think it is the best solution. I think the video card should be responsible for variable frame rate, including possibly multiplying frames to stay in the displays range. The video card should have a better idea when the next frame will be available than having the scaler guess. There should already be some functionality present in hardware or drivers to do frame pacing. If you have a display which supports, for example, 60 to 120 Hz, then I do not see why you wouldn’t be able to apply frame multiplication starting at just below 60 Hz. This would allow you to keep the display in a narrower range than 40 to 144 Hz we have seen on some panels. It should make it easier to keep the display parameters correct (color, overdrive, etc).
I think the overdrive should be handled on the display, and the video card should not have to know anything about it. This is why we have standards and interfaces. Without a proper standard, every panel manufacturer will need to do unnecessary work to make their display work with every video card manufacturer. The only reason we have ghosting on some free sync displays is that the scaler makers have not properly adjusted to running at variable refresh rates yet. Also, AMD has not implemented frame multiplication, so lowering the VRR window requires lowering the actual refresh on the panel. This will contribute to the ghosting issues. A frame buffer is unnecessary if the video card can resend the frame when necessary. The scaler would still need to store some frame rate history to predict when the next frame is coming so that overdrive can be adjusted properly. This is not complicated, and occasionally guessing wrong will not be that noticable.
Some of these issues may just go away with newer display technology anyway. Unless you really want a new display now, I would say wait a bit and just run with settings that keep the fame rate above 60 Hz.
As I said before, when the
As I said before, when the oppurtunity comes up, I either spend the money and get it at that time or have to wait an indefinite amount of time(meaning the wife will spend hte money on something else and then I’m SOL)
As of right now, Gsync has the better overall solution. There is no denying this. I also firmly believe that AMD will work with its partners to not only solve the ghosting issues, but also the below VRR window issue as well. It’s just a question of when really and the result will look and work fine. And when the time comes, hopefully by the end of the year and I have the budget, and if AMD has implemented the changes to take care of ghosting and frame multiplication and it stays cheaper than gsync, then i’d go Red team again.
In summary, and I’d like to think this applies to a lot of people, it’s not what will ultimately be the best implementation but what is best at the time I’m ready to buy.
I’d like to see something in
I’d like to see something in the $200 range. I just need something that will push 1080P at 60hz for the next few years until I can afford an ultrawide g-sync 120hz+ monitor 🙂 🙂
I’ll wait until 6/16 (I think
I’ll wait until 6/16 (I think that’s the date) for what will hopefully be a reveal, with products available, until I decide how good or bad this situation may be.
In the meantime, it seems like there will be one or two new enthusiast Fiji chips, and rebrand across the board. This bothers me, but if significant enough changes were made for Hawaii, or if Tonga plays a larger role in the lineup, I think I’ll be okay with it.
Their mid-range lineup should have the newest GCN tech, and if it doesn’t, I would want an explanation as to why not. From gamer cards of around $250, up through enthusiast class chips, I think consumers deserve the newest tech, or ‘at least’ something refined, ie, Fermi 400 series to Fermi 500 series, something akin to that. What I mean to say is, if what we’ve heard so far is true, the bulk of AMD cards being sold won’t even support the full feature set of DX12. This should frustrate everyone.
I don’t want to see AMD decide to ride out the status quo ad infinitum. 3 years now of basically cruising, it’s really beginning to worry me. I want a successful AMD, but I don’t think this is sustainable.
I probably sound very pessimistic, not trying to, for what it’s worth, I really want a successful launch, the competition benefits all of us.
What a lot of people don’t
What a lot of people don’t realize when they complain about 4K performance nowadays is that 4K is still relatively new. >95% of gamers out there do not have a 4K display. 1080p right now is the standard. Video cards in the last couple years have just started to kick butt at 1080p >60fps. To expect cards to be doing this at 4k (4x the resolution) is not very realistic.
Give it a couple years and there’ll be high end cards running every game out there at 60+fps max settings 4k.
Decades may be.TITANX can’t
Decades may be.TITANX can’t get 1600*1200 MAX 60fps in every game.1600*1200 was 4X normal res in late 1990s.
You don’t get it, do
You don’t get it, do you?
Gamers don’t spend on 4K displays because there is no dGPU on the market, which can deliver a proper visual experience.
1080p has been a standart res for 10 years. Now it should be 1440p, but it’s not. This is a realistic issue here.
Prices should go down if the hardware is not sufficient for a display.
Gamers don’t spend on 4K
Gamers don’t spend on 4K displays because 4K displays are still too expensive…
Most gamers out there cannot afford $500+ monitors.
4K is still a niche, it is still years from being the norm.
I bought at 28nm graphics
I bought at 28nm graphics card years ago (7970). It’s not getting replaced until AMD can push out something on a smaller node that is significantly faster. Rebadged Hawaii is just not going to cut the mustard.
Why not anyone ask if the
Why not anyone ask if the card has :
HDMI 2.0 ???
DP 1.3 ???
Is data bandwidth 18 GBPS ???