cancel
Showing results for 
Search instead for 
Did you mean: 

General Discussions

AMD fires back at 'Super' NVIDIA with Radeon RX 5700 price cuts

MD unveiled its new Radeon RX 5700 line of graphics cards with 7nm chips at E3 last month, and with just days to go before they launch on July 7th, the company has announced new pricing. In the "spirit" of competition that it says is "heating up" in the graphics market -- specifically NVIDIA's "Super" new RTX cards -- all three versions of the graphics card will be cheaper than we thought.

The standard Radeon RX 5700 with 36 compute units and speeds of up to 1.7GHz was originally announced at $379, but will instead hit shelves at $349 -- the same price as NVIDIA's RTX 2060. The 5700 XT card that brings 40 compute units and up to 1.9GHz speed will be $50 cheaper than expected, launching at $399. The same goes for the 50th Anniversary with a slightly higher boost speed and stylish gold trim that will cost $449 instead of $499.

That's enough to keep them both cheaper than the $499 RTX 2070 Super -- we'll have to wait for the performance reviews to find out if it's enough to make sure they're still relevant.

AMD fires back at 'Super' NVIDIA with Radeon RX 5700 price cuts 

1,953 Replies

The RX5700XT is "high end" compared to AMD Polaris GPU based RX480 4/8GB that was released >3 years ago on 14nm for 220-250 and replaced by the RX580 the following year which was factory overclocked AIB versions of RX480 with higher power consumption allowed in the BIOS, then RX590 late last year which was RX480 retargeted to 12nm process and tweaked for performance versus power consumption. Their Nvidia competition was GTX 1060 6GB AIB cards.

The die size of the Navi GPU in the RX5700/XT is ~ same as Polaris.

The RX Vega 64 Liquid was the high end card from AMD until the Radeon 7 16GB was released at the start of 2019.
The RX Vega 64 Liquid  is ~ same speed / slightly faster than a GTX 1080 8GB in benchmarks when you included DX12/Vulkan game titles in benchmark suite.
The RX Vega 64 Liquid is ~ 30% slower than a GTX1080Ti.
So it was ~ second place to Nvidia Pascal top end consumer GPU the GTX1080Ti.

The Radeon VII is Radeon Instinct M150 reject cards repurposed as a Prosumer / "Creator" GPU.
It is slower than RTX2080 for gaming, sitting somewhere between RTX2080 and RTX2070.

You may compare the specifications here:
https://www.amd.com/en/products/professional-graphics/instinct-mi50 
https://www.amd.com/en/products/graphics/amd-radeon-vii 
FP64 Perf and Separating Radeon VII from MI50 - The AMD Radeon VII Review: An Unexpected Shot At The... 


The Radeon VII is air cooled, on 7nm with 16GB of HBM2, single BIOS, better FP64 compute  and is ~ 8% faster than a stock clocked RX Vega 64 Liquid.
I can overclock RX Vega 64 Liquid to perform ~ same as a stock clocked Radeon VII in some game titles but overall it is a much better GPU. 
The review conclusion is here: Final Words - The AMD Radeon VII Review: An Unexpected Shot At The High-End 


I think the Radeon VII is now EOL.
I think it had a production run of about 6months.
I almost bought one, and I am still interested in it for compute.
This video may be interesting: Radeon VII — Final Review — Is It Worth Buying after EOL? - YouTube

The RX5700XT is a great advance for AMD in gaming, but the FP32 compute performance is lower than RX Vega 64.

0 Likes

Thats true but i read alot about Radeon VII its already a card that reached end of life status.......thats a bump cause its to expensive  to make, why they even bring it out than.....

0 Likes

Radeon VII are "Radeon Instinct M150 reject cards". That means they are Radeon Instinct M150 GPUs that did not pass all required QA testing to be sold as professional grade GPUs for scientific or engineering use in industry. They will not have full QA certified drivers or support. However they are good enough to be used as "Prosumer cards", in applications such as blender or perhaps OpenCL coding and non critical compute tasks. They are also good for gaming.

Rather than destroy the failed M150 GPUs they can be sold as Radeon VII GPUs and some money can be earned from them.

I think that is the point.

0 Likes

I also read about they were to expensive to make 16gb HBM2 memory. And thats why they made a few en put end of life status

0 Likes

The 16GB of HBM2 is an advantage for some Prosumer and Professional applications.
The 4x 4GB HBM2 inproves the memory bandwidth and it also improves the increase in performance if the main GPU clock is increased.
Smaller HBM memory modules are now in specification so 4x2GB HBM2 = 8GB total would also be an option in future for Vega 20 type architecture so smaller memory depth but same width would still give same or higher bandwidth. There is also HBCC on Vega cards, but I do not know what will happen with that in future.
 

0 Likes

FYI a comparison of RX Vega 64 Liquid specifications versus RX Vega 64 can be found here:
The AMD Radeon RX Vega 64 & RX Vega 56 Review: Vega Burning Bright 

I con no longer find the RX Vega 64 Liquid specifications the AMD product site.
It is a much better version of the RX Vega 64 than any of the AIB cards.
The RX Vega 64 Liquid  can now be picked up second hand on Ebay for ~ 300.

0 Likes

(In the voice of Gul Dukat) Are they insane?

No ray tracing, no AMD alternative to nVidia ReShade, higher power consumption, and, right now, the exact same price as the RTX 2070, a card which is effectively the same average speed as it...

I cannot understand why all reviewers are so favorable about the RX5700XT.
Whilst RX5700XT prices are down a little this weekend for Black Friday, I think they are still far too high.
Cheapest 2 slot card  I can find is PowerColor RX5700XT 8GB Reference for 350 - which I think is ~ 100 too expensive.
I am not conviced the AMD Drivers are even stable on the RX5700XT yet, and Gamers Nexus seem to point that out.

The AMD GPU drivers quality has been a general problem in my opinion. 

For example the following things are broken on R9 Fury X in Adrenalin 2019.
R9 Fury X is performing worse in 3D Mark compared to 16.12.2 Drivers in Single GPU and Dual DX11 Crossfire.
Triple & Quad DX11 Crossfire show negative scaling.
HBM overclocking is still locked down on R9 Fury X by AMD Drivers later than 16.12.2.
Radeon Performance Overlay CPU Utilization report is broken again - I previously reported it and it was fixed. - This is also a problem on RX Vega 56/64.
Radeon Chill is still pretty much useless since it kills keyboard only input FPS and breaks FreeSync - reported many times.  - Also a problem on RX Vega 56/64.

Would have been a completely different story if AMD had continued the practice of at least the last 7 years and priced a new generation card which directly replaced the previous generation card at or below the previous generation card's launch price, which would have put the 5700XT at $279, as it replaced the 590, but when they decided to pricematch nVidia with an inferior card and support, that was just a huge "F YOU" to every Radeon user, and that situation only got worse as the aftermarket cards climbed over $475, so you end up with an RX 5700XT matching up with the vastly superior 2070 Super for the same price.

We seem to be stuck on the fact that because the Navi GPUs have similar core counts to Polaris they must be the Polaris replacement.  However, meriam webster defines "replace" as "to take the place of especially as a substitute or successor"

"5700XT at $279, as it replaced the 590"

But it didn't right?  Because the 590 continues to be sold along side it.  What they did do, was replace Vega.  And yes, in that context, they are at or below the previous generations launch price.

0 Likes

Polaris is continuing to be sold, but that doesn't mean it's still being manufactured, and likely hasn't been manufactured in some time. There's likely a large glut of Polaris GPUs left over from the mining crash the same way that there was a large glut of GTX GPUs left over, so many nVidia had to buy them back and delay Turing for several months. There have been no announcements of AMD buying back GPUs the way nVidia did, and nVidia cards have a very large user share advantage which means the ancient (relatively) Polaris GPUs are much slower to shift than the GTX 1000 series.

The 5700/XT is the midrange Navi card, which replaced the midrange RX 580/90. The upcoming RX 5500/5600 will replace the RX 560/570, and next year's "nVidia killer" GPU will replace the high end Vega. This is the problem with AMD's strategy of split design teams, you have a range of cards which are always behind the other, and that causes both public confusion and aggravation when one company has a range of cards from entry level through very high end based on new architecture, and one company is competing with a mix of 3 different generations of cards, all of which support a differing feature set.

It doesn't matter that this year's mid range card, the RX 5700XT, has the performance of last year's high end card, Vega 64, anymore than this year's entry level gaming card from nVidia, the RTX 2060, outperforms last year's high end gaming card, the GTX 1080. It may have, and probably did, caused AMD to discontinue Vega production, but that's the fault of AMD's split strategy of letting high end cards lag behind by months, if they had released them all at once, like nVidia did, there would be no argument and confusion about the 5700/XT replacing the 580/90.

The 5700XT and 5700 clearly replaced Vega 56/64 in the current product stack. 

AMD even lists them as "enthusiast" level parts in their latest marketing material.

AMD.jpg

The performance segment (once occupied by the 590/580/570) is now solely occupied by the RX 5500, although that segment will by rounded out in January with the RX 5600.  So the RX 5600 and 5500 serve as RX 590/580/570 replacements in the performance segment, RX 5700 and 5700 XT occupy the enthusiast segment replacing Vega.  The "NVidia Killer" GPU will replace Radeon VII atop the enthusiast stack, while Polaris continues on in the mainstream and entry segments.

Seems like a stretch to call the 5700/XT "midrange" when exactly one GPU sits higher in the product stack and 7 will be below.  

I am more interested a substantial increase in compute units so that gaming at 4K is not so sluggish

0 Likes

Of course an RX 480 was an enthusiast level part when released too. It isn't so much that Navi replaces where the high end Polaris cards are at the time when Navi came out. It is that it goes into that intended slot where those Polaris cards were before they grew so long in the tooth.  I don't doubt that AMD are marketing Navi at a higher tier today than originally intended. Honestly I think they ended up releasing these things running with more power and speed than originally intended as they knew they had nothing better for the high end in short order to follow them up with. So yes we ended up with a card that wasn't really what we thought we were getting. I am not sure I would call any of this stuff Nvidia killers. In the time since Navi came out the Nvidia drivers have seen huge performance improvements that bridge if not surpass many of those around launch disparities of Navi being so much faster. For instance my 2060 super has gained about 20 fps in BF5 under DX12.  Competition is a great thing. It makes all the companies keep doing better for their customers. I am quite sure without the pressure Navi brought we would not have had those improvements or at least not as quickly. 

I know I remember several times where AMD even said that that Navi was a replacement for the mid level cards and that future top end cards were still to come.  This was in that year or two before the launch. With the absence of that better than current Navi cards actually materializing, maybe what we got wasn't either and something more in between. I would be very surprised to see AMD come out with anything on the GPU side for years to come that mimics what Ryzen did to Intel. Nvidia isn't Intel and I am quite sure that they are ready for what comes next at all times. Historically both companies have gone back and forth winning the performance crown but neither for too long and I just don't foresee that changing too much at least not with what we have seen in the past few years now. 

Regardless it doesn't much matter. If the card meets your needs and the price is right, you will buy it. 

0 Likes

Considering that you have moved over to NV (for some time now), are they now paying you to post on this forum.

And stop bumping other old threads to the top (that denigrate AMD), yes I know it is you.

Ryzen 5 5600x, B550 aorus pro ac, Hyper 212 black, 2 x 16gb F4-3600c16dgtzn kit, Aorus gen4 1tb, Nitro+RX6900XT, RM850, Win.10 Pro., LC27G55T..
0 Likes

I think AMD does have an edge here.  NVidia is looking to Multi-Chip modules for their post Ampere designs to continue scalability.  They published a fun paper on it.

https://research.nvidia.com/sites/default/files/publications/ISCA_2017_MCMGPU.pdf 

Essentially, it is just a series of smaller GPU base units linked together to form a single logical GPU.  This is essentially exactly the same approach Jim Keller developed with Ryzen over at AMD.  And seeing that AMD already has Infinity fabric, and a lot more experience doing this very thing on the CPU side, I imagine it is coming to their GPUs as well. 

That is likely why NVidia is looking at this approach, the 2080 Ti is already over 700mm^2, and continuing to grow these GPUs won't be possible (bad yields, high prices).  And I'm sure they are aware that AMD will port their CCX design over to GPUs to scale performance, so they are researching the same thing.  How far ahead is AMD?  Hard to say.  But the guy who implemented all this for AMD (Jim Keller) is now over at Intel working on their CPUs and GPUs with Intels massive budgets as well.

Things will likely start to get very interesting in 2021.

chiplet designs are going to become more common for a range of devices, RAM and a CPU can be side by side to have a more integrated unit for mobile machines etc.

0 Likes

I will be please as punch if they do leap ahead. It only helps consumers when this happens. It is a reality check for all companies to not get too complacent and to not rip their customers off too much. 

The thing I always look forward too in every generation is cards doing new tricks in their hardware and at the software level can be nice too. I hope AMD reaches feature parity soon and if they surpass the others too that is gravy. 

0 Likes

Not going to happen anytime soon sadly, between integer scaling and ReShade, nVidia has a large software feature advantage. There is speculation this year's yearly Crimson feature update would bring integer scaling, but I haven't seen anything concrete other than Frank Azor confirming a feature update would be coming December. Even if AMD were to somehow pull off that miracle, the chances of it extending beyond the Navi family are slim.

I think AMD is ahead in that regard.  Arcturus is already two Vega 64 CU GPUs linked by infinity fabric as one logical GPU.  Now, compute is different than gaming as the memory access patterns are different, but AMD is already bringing this stuff to market.

0 Likes

Why is the Radeon VII even on that graph. It is EOL.

0 Likes

Has AMD officially said it was EOL?  If they did and still put it in their marketing material that would be highly disingenuous indeed.

0 Likes

Yes it is EOLed months ago.

https://www.pugetsystems.com/labs/articles/DaVinci-Resolve-GPU-Roundup-NVIDIA-SUPER-vs-AMD-RX-5700-X... 
"Radeon VII is 100% EOL, we confirmed that directly with AMD before we started this round of GPU testing. Leftover supply does not mean it is still being manufactured."

You can also read the Gamers Nexus Article about it here:
Awards: Worst & Best Graphics Cards of 2019 | GPU Recap | GamersNexus - Gaming PC Builds & Hardware ... 

0 Likes

That is pretty hilarious.  The marketing material I posted just came out with the 5500 launch.  Guess the top of the chart was looking a bit sparse.

0 Likes

RE: That is pretty hilarious.  The marketing material I posted just came out with the 5500 launch.  Guess the top of the chart was looking a bit sparse.

It is either " highly disingenuous indeed" or "pretty hilarious"  in your own words.
The top of the chart, i.e. Enthusiast = something that can beat an RTX2080 or RTX2080Ti is actually empty.  

Those Radeon VII are still getting sold at high prices and no official "EOL" on most retailer sites I see.
If you didn't know you might buy one expecting driver support /development/fixes for a few years at least.
I think that is unlikely.

If it is "pretty hilarious" the joke is on the consumer, and not for the first time.
The same happened with the Vega Frontier Edition "Prosumer" GPUs.

0 Likes

To be honest though Radeon Technologies Group is still about as bare bones as it gets. AMD's financial situation after the disaster that was Bulldozer left resources thin, and the design team, led by Raja, failed on multiple fronts with the Vega architecture. Remember Vega was supposed to have AI and deep learning features, which it doesn't have, to compete against nVidia and why the Vega Cube, aimed at vehicles, never materialized, and it's exceedingly inefficient, both in terms of power and capabilities, not to mention extremely expensive due to the reliance on HBM2, not helped by the well known packaging fiasco which limited custom editions greatly. Navi was supposed to only be entry level through mid range, as Navi stems from the semi custom GPUs powering the next XBOX and PS, but with Lisa Su dedicating 80% or more of RTG resources to semi custom work, which we know from their own announcement, they had to repurpose Navi again to make a high end card, and I bet most people think, like I do, that the high end Navi based card coming next year to provide real competition to the 2070 Super and above, is going to be horrendously hot, inefficient, expensive, and will not be on par with the features of the competition, especially given how it will be competing against Ampere, which is rumored to be quite a bit more powerful (likely, typical generational improvement) and less expensive to produce.

AMD can't release a prosumer card because they don't have a card capable of prosumer work, as exemplified by the review Puget Systems did with Adobe Premiere Pro, where the prosumer Vega VII was, at best, a competitor to the entry level GTX 1660, and that's probably the biggest reason the were discontinued being manufactured, which is not the same as EOL'd, as technically any card not under legacy support is not EOL'd despite not receiving any new features or performance improvements.

https://www.pugetsystems.com/labs/articles/Premiere-Pro-CC-2019-AMD-Radeon-VII-vs-NVIDIA-GeForce-RTX-1395/

I would not buy a Vega based product when it came out and can't imagine doing so now. It is IMHO a card that does not even have feature parity with the older GCN cards that came before it. Since it's introduction it has been riddled with issues that still plague it today. They don't work at default settings for most anyone. Until AMD has a new architecture out that truly is new and has feature parity with the competition AMD might as well just concentrate on game consoles. They sure are not going to win over PC gamers in large numbers. I agree, they don't even have a better price anymore. They just price a little below the competition with most products buy not enough to be worth the trade-offs. I really do look forward to a new product from them that I will be happy with and hope it will happen, but another Navi based product isn't going to be it. If they push this architecture further is will be more of the same. As already said run hot and need too much power. AMD would be better to concentrate on fixing their drivers and selling their cards cheaper at the points they do compete and focus on a new architecture beyond that. These recent cards (last 2 years) and the Wattman drivers have tarnished what was a company that provided GOOD gpus. 

0 Likes

What RTG needs is Lisa Su to funnel some of the millions upon millions in profits from the processor division into them so they can develop an architecture as revitalizing to RTG as Zen is to AMD. Realistically AMD can't keep adapting semi-custom designs, and if they let Intel get the advantage of them with Xe, they risk slipping to third place in the graphics market, and that affects not only AICs, but any machine with integrated graphics, be they consoles, notebooks, tablets, desktops, even extending to more professional and specialized applications. Software definitely needs improvement, these "once per year feature updates" are inadequate, the focus on just the latest GPUs is a slap in the face, and the delay between WHQL drivers is borderline unacceptable.

And really, as far as "GOOD" GPUs go, AMD hasn't made one of those since the HD 7000 series, ever since then they've just been "decent", relegated to value alternatives to nVidia's cards, with their price to performance ratio keeping them in the game, even though Polaris, but now with the GTX 1660 delivering RX 590 performance for the same price point, but along with much lower power consumption and more features.

AMD has not put much focus on graphics for several years. All the development has been with the CPU which has been very successful at taking market share from Intel. AMD has lost market share to rival nVidia.

0 Likes

Sure, I think I understand the back story to it all.
For me EOL = when the card is no longer manufactured, because once that happens, the focus is on other GPUs.
I do think retail sites should put EOL notice on them and drop prices though.
I just looked around today and there are some good condition used Radeon VII selling second hand for 300-400. 

0 Likes

Never thought about doing it, but I did it tonight. There's an actual JEDEC standard for product discontinuance,J-STD-048. Companies issue EOL and LOD notices all the time, but the consumer rarely sees them unless they're pointed out in a tech site review, most recently exemplified by the fact that Intel, for whatever reason, has reascended the EOL notification for a certain garbage 22nm Haswell chip.

Would be nice if companies were required by law to list on product pages the last dates of mainstream support and extended support on the first date of sale, much like Microsoft dues, so consumers are better informed.

https://www.jedec.org/standards-documents/docs/j-std-048

now that is a standard that has been ignored widely by the tech sector

this has to do mostly due to demand in the market place which is very transient

0 Likes

0 Likes

Yea, they came online last night. Not going to download it for a while though, most of the features aren't applicable to pre-Polaris cards, so little reason to use them for me.

https://community.amd.com/community/gaming/blog/2019/12/10/change-the-way-you-game-with-amd-radeon-s... 


The user interface is a mess in my opinion. I filed an AMD Reporting form about it. 
One advantage AMD had over Nvidia was the Adrenalin GUI and the Radeon Overlay. 
Now it looks like the old Raptr AMD Gaming Evolved Team is back designing the GUI. 
FRTC has been removed so it is very likely Chill will completely break FreeSync now.
Hialgo Boost has been implemented for Vega. 
However the Driver GUI settings is such a mess I can't  be bothered to try it out. 
I am sticking to Adrenalin 2019 19.12.1.

I've always disliked AMD's GUI, I think it's unnecessarily clunky and lacking in basic options, such as being able to prevent creation of game profiles, as well as the inability to provide accurate basic information about your GPU, causing confusion among those not technically inclined and requiring the use of other programs, such as GPUz. I also dislike the very small level of control AMD gives compared to nVidia, and the fact they seem to be focusing on what could be described as niche features, such as AMD Link and now some integrated web browser in the overlay, instead of focusing resources on performance improvements for all "supported" GPUs and extending feature support to all "supported" GPUs instead of just the most current.

I never force games with a driver profile, I use in game settings case by case so that each is optimal

0 Likes

RE: being able to prevent creation of game profiles
That does not bother me too much if there were only a few games in my library but I agree that option was needed to allow the user to prevent auto creation or group games under tabs or search for games in more locations rather than the C: drive etc.

RE: the inability to provide accurate basic information about your GPU
Agreed that needed / needs more work. I have seen incorrect amount of VRAM used, ridiculously high GPU frequencies reported etc on Vega GPUs at least, however things seemed to be getting better. The one that has me stumped ts why CPU utilization reporting for AMD Ryzen processors is broken again, after getting fixed. The CPU utilization report has consistently worked on Intel processors. 

RE: I also dislike the very small level of control AMD gives compared to nVidia
I use Nvidia control panel, and I can think of some cases where they offer more options such as Dynamic Super Resolution settings versus Virtual Super Resolution with AMD. I think the Nvidia Control Panel GUI needs work to modernise it so it can work better at 2K and 4K resolution, and it is unresponsive / slow to respond at times. I think Nvidia GeForce ecperience is o.k.

RE: and the fact they seem to be focusing on what could be described as niche features, such as AMD Link and now some integrated web browser in the overlay
AMD Link has one use for me. To report GPU stats if the Radeon Performance Overlay is broken or does not work on a particular game.
I tired using a smartphone as a game controller to run BF1 and BFV just to see how it worked. Unplayable.
I do not know why this new integrated web browser is needed either. Basic things like cut/ copy/ paste with mouse and menu do not work. I do not know how the browser security is set, if such an option even exists.

RE:   instead of focusing resources on performance improvements for all "supported" GPUs and extending feature support to all "supported" GPUs instead of just the most current.

Absolutely.
I think the way AMD Vega users had to protest to get Radeon Image Sharpening into RX Vega GPUs and the fact that it is still not supported on earlier top of the range AMD cards such as R9 FuryX/Nano/Fury or R9 390X or R9 290X is ridiculous.

Nvidia Game Filter is supported on cards all the way down to GTX780Ti Kepler based cards. 

The entire focus has been on RX480/580/590/RX5700/XT.

I better post this now. I am running Adrenalin 2020 19.12.2 driver on a powercolor RX Vega 56 Red Dragon GPU and the screen is intermittently blacking out.

0 Likes

The release notes said they fixed the blackout issue. Clean up with DDU and try the driver fresh.

0 Likes