Thursday, August 4th 2016

Sapphire Announces the Radeon RX 470 Platinum Edition

Sapphire announced the Radeon RX 470 Platinum Edition OC graphics card. This card is the closest it gets to a market-available reference design AMD board for the RX 470. It sticks to the AMD reference design PCB, with a negligibly higher 1216 MHz engine boost clock (vs. 1206 MHz reference); but a significant 7.00 GHz (GDDR5-effective) memory clock (vs. 6.60 GHz reference). Sapphire customized the cooler design to feature a silvery finish on the cooler shroud, and a matching silver+black backplate covering the length of the PCB. This card could be priced close to the $179 MSRP set for the RX 470.
Add your own comment

30 Comments on Sapphire Announces the Radeon RX 470 Platinum Edition

#1
xvi
Seems odd to me how some cards come as 6-pin and some come as 8-pin. Looks pretty nice though, tbh.
Posted on Reply
#2
Chaitanya
xvi said:
Seems odd to me how some cards come as 6-pin and some come as 8-pin. Looks pretty nice though, tbh.
I think many of the 8-pins cards are due to the manufacturers being lazy. Most of these cards seem to reuse the PCB layout and components from their higher end 480 cards.
Posted on Reply
#3
Captain_Tom
Chaitanya said:
I think many of the 8-pins cards are due to the manufacturers being lazy. Most of these cards seem to reuse the PCB layout and components from their higher end 480 cards.
Hey if it's cheaper to make it better why not?!
Posted on Reply
#4
INSTG8R
My Custom Title
Rather handsome for a blower card.
Posted on Reply
#5
Jetster
GTX 760 was like that some were 6 pin some 8 pin
Posted on Reply
#6
Captain_Tom
Jetster said:
GTX 760 was like that some were 6 pin some 8 pin
I find it odd they would ever put just a 6-pin on that...
Posted on Reply
#7
Grings
£165 on ocuk, its at least cheaper than 480's by a big enough margin to not be pointless, every other card ive seen the cost is too close to the 480
Posted on Reply
#10
Casecutter
Captain_Tom said:
Although I will say that this 470 you linked to really should be $180.
I could swear I saw this yesterday on Egg and thinking well that a nice (gussied up) reference for the entry price of $180.
I didn't know at that time it had backing plate, but did notice it had a modest OC.. Thought there might be hope for volume of the 470 to hold to some reasonable pricing. But then when you see the prices of 970's Nvidia is maintaining, most often $250 on up, one can only deduce the market is still buying and willing to ante up, so why not take it (Newegg) from to pool if there's the willing and grab it while they can!
Posted on Reply
#11
Captain_Tom
Casecutter said:
I could swear I saw this yesterday on Egg and thinking well that a nice (gussied up) reference for the entry price of $180.
I didn't know at that time it had backing plate, but did notice it had a modest OC.. Thought there might be hope for volume of the 470 to hold to some reasonable pricing. But then when you see the prices of 970's Nvidia is maintaining, most often $250 on up, one can only deduce the market is still buying and willing to ante up, so why not take it (Newegg) from to pool if there's the willing and grab it while they can!
Exactly. I don't even think AMD saw this much demand coming, but it makes complete sense. Nvidia is price gouging so much that people are happy to sacrifice some performance if it means getting a phenomenal value.

Consider that for $600 you can get 3 x 470's that will dominate the 1080 and even match the $1200 Titan XP. Now I am staying away from crossfire right now, but if the games you play support it it really is a crazy example of how bonkers Nvidia is right now.


Personally I think AMD should just discontinue both the 8GB 470 and 4GB 480. Then tell AIB's to aim for $200 for the 470 and $260 for the 480.
Posted on Reply
#12
Casecutter
Captain_Tom said:
Personally I think AMD should just discontinue both the 8GB 470 and 4GB 480. Then tell AIB's to aim for $200 for the 470 and $260 for the 480.
My thinking is these are short lived, I want to consider GloFo (and this is more conjecture and hope) didn't hit the mark with Polaris out of the gate. The shortages are because AMD/RGT took the initial wafers then turn off the spigot, while GloFo is looking to re-vamp and re-spin their process (kind of like what happen in the early 28nm).

Could RTG just drop in such improved chips and re-name with a 475/485 designation, while then hold to these what seem right now oddly extortionate prices? If they just tweak the clocks a little more, hold/reduce TDP, and see a slightly more OC headroom these cards would really hit the mark at these prices.
Posted on Reply
#13
Captain_Tom
Casecutter said:
My thinking is these are short lived, I want to consider GloFo (and this is more conjecture and hope) didn't hit the mark with Polaris out of the gate. The shortages are because AMD/RGT took the initial wafers then turn off the spigot, while GloFo is looking to re-vamp and re-spin their process (kind of like what happen in the early 28nm).

Could RTG just drop in such improved chips and re-name with a 475/485 designation, while then hold to these what seem right now oddly extortionate prices? If they just tweak the clocks a little more, hold/reduce TDP, and see a slightly more OC headroom these cards would really hit the mark at these prices.
There are certainly a lot of things could speculate on right now. My current theory is that 14nm is behind 16nm by at least a few months, but that 14nm is also a lot cheaper that 16nm. Therefore these first group of cards from AMD are the rough early yields that while imperfect in performance, are dirt cheap to manufacture. Furthermore while they can't seem to get any yields of larger dies they seem to be able to spit out a ton of the smaller ones.

Also notice that the 470 is a lot more efficient than the 480, and I expect the 460 to be WAY more efficient than the 470 (Thus providing evidence for my claim).


There are other rumors too that Vega may bring a full overhaul to the architecture with tiled-based rendering (What Maxwell did to get such huge gains). In fact if you look on their roadmaps Vega replaces EVERYTHING from top to bottom including Polaris. This rumor goes hand-in-hand with what I said. AMD is just spitting out the cheapest chips they can while they prepare their new arch that requires GloFlo's 14nm process to mature.
Posted on Reply
#15
Captain_Tom
Casecutter said:
Heck I'm thinking Polaris 11 might be still at TSMC because AMD needs it to be perfect.
AMD has said it will go with the Foundry / Process that make sense for the product delivery and price.
http://www.extremetech.com/computing/232493-amd-has-built-hardware-at-samsung-could-tap-foundry-for-future-products
Nah they will go with 14nm. Again, the efficiency of the 14nm process seems to increase quite a lot the smaller you make a chip. I expect the 460 to use 50w or less, and if it does we have a great example of what big-14nm can be once larger dies get decent yields.

I wouldn't be surprised if they launch a 4096-sp 512-bit card on 16nm though. After all they can't just let Nvidia have the entire enthusiast market until Vega comes out in December (If we are lucky).
Posted on Reply
#16
danbert2000
Captain_Tom said:
Consider that for $600 you can get 3 x 470's that will dominate the 1080 and even match the $1200 Titan XP. Now I am staying away from crossfire right now, but if the games you play support it it really is a crazy example of how bonkers Nvidia is right now..
I'm sorry, but there is no way that crossfire 470's get even close to the GTX 1080 except for Ashes of the Singularity, an RTS game that is built as sort of a tech demo for multi-adapter explicit mode and asynchronous compute, and they still fail to catch up with the 1080 before you even start overclocking it. So, that's your best case, that even two 480's are weaker than a 1080. Now consider that most games have poor Crossfire implementation and don't get nearly as good of scaling as AotS. Scaling with three-way Crossfire is even more rare and less effective, so there is no way in hell a three-way 470 build will ever reach within even 80% of Titan XP performance. Your suggestion that somehow Crossfire 470's are anything but a gimmick is insulting to critical thinkers and is dangerous.

https://www.techpowerup.com/reviews/AMD/RX_480_CrossFire/19.html

Just look at those performance averages. Non-scaling games pull the average way down compared to a stock 1080, discounting the 10%+ guaranteed overclock you get on them. The RX 480 Crossfire setup does worse than 970 SLI at Fallout 4. Add on the fact that you only get good scaling on the games that do work after a couple months of patches, and any Polaris crossfire setup is hot, loud, uses twice as much electricity, and is inconsistent. A 1070 would give you much better performance 90% of the time for the same cost or less. Get your ridiculous assertions out of here.
Posted on Reply
#17
Captain_Tom
danbert2000 said:
I'm sorry, but there is no way that SLI 480's get even close to the GTX 1080 except for Ashes of the Singularity, an RTS game that is built as sort of a tech demo for multi-adapter explicit mode and asynchronous compute, and they still fail to catch up with the 1080 before you even start overclocking it. So, that's your best case, that the 480's are weaker than a 1080. Now consider that most games have poor Crossfire implementation and don't get nearly as good of scaling as AotS. Scaling with three-way Crossfire is even more rare and less effective, so there is no way in hell a three-way 480 build will ever reach within even 80% of Titan XP performance. Your suggestion that somehow Crossfire 480's are anything but a gimmick is insulting to critical thinkers and is dangerous.

https://www.techpowerup.com/reviews/AMD/RX_480_CrossFire/19.html

Just look at those performance averages. Non-scaling games pull the average way down compared to a stock 1080, discounting the 10%+ guaranteed overclock you get on them. The RX 480 Crossfire setup does worse than 970 SLI at Fallout 4. Add on the fact that you only get good scaling on the games that do work after a couple months of patches, and the 480 crossfire setup is hot, loud, uses twice as much electricity, and is inconsistent. A 1070 would give you much better performance 90% of the time for the same cost or less. Get your ridiculous assertions out of here.
Clearly not reading anything I said lol. I said 3 x 470's, and I said in scenarios where people play mostly or all scaling games.
Posted on Reply
#18
danbert2000
Captain_Tom said:
Clearly not reading anything I said lol. I said 3 x 470's, and I said in scenarios where people play mostly or all scaling games.
I just fixed it. I guess you didn't want to reply to any of my points, or couldn't. Just admit 3-way crossfire is pointless and AMD has nothing that can even touch 1070 levels, let alone a Titan XP. Can you even point out a modern game that scales well? I guess years old games will sort of benefit, but we can max those out with any old card now.
Posted on Reply
#19
Captain_Tom
danbert2000 said:
I just fixed it. I guess you didn't want to reply to any of my points, or couldn't. Just admit 3-way crossfire is pointless and AMD has nothing that can even touch 1070 levels, let alone a Titan XP. Can you even point out a modern game that scales well? I guess years old games will sort of benefit, but we can max those out with any old card now.
Reply to what point? LMAO if you can't read what I said, why would I bother reading your reply to the thing I didn't even say?


Now you have read it, but it seems you have still completely missed the point of what I was saying. At no point did I say I would recommend 3-way crossfire, and in fact I wouldn't considering how many of the latest games aren't using it. However Nvidia's pricing is so insane right now that people are considering some ridiculous alternatives - that's my point.

And frankly, I can't blame them either. Let's break down why:

-3 x 470's for $180 (Cheapest option I saw on newegg at launch) costs a total of $540
-In games that support crossfire these should match a Titan XP.
-In games that don't scale the end-user still only paid 45% as much!

Thus even if only half of the games a user plays support it, it almost doesn't matter. They paid so much less! That is how ridiculous the current high-end is right now, and that is why the 470 and 480 are constantly sold out.


P.S. The Fury X beats the 1070 in true DX12 / Vulkan games already lol.
Posted on Reply
#20
danbert2000
Captain_Tom said:
P.S. The Fury X beats the 1070 in true DX12 / Vulkan games already lol.
Hrmm... not really, not even before an overclock. I like your suggestion that there are "true" DX12 games. They just happen to be the ones that do better on AMD, right? Like Hitman and DOOM and those are the only games people play, right? Feel free to click around this review, they have a bunch of benches that detail why the Fury X is a dead end card even more so now than when it was bitchslapped at launch by the 980 Ti:

https://www.guru3d.com/articles_pages/evga_geforce_gtx_1070_sc_superclocked_gaming_review,15.html

I guess you can hold on for dear life to DOOM Vulkan performance, but seeing as Nvidia hasn't updated their Vulkan drivers to implement async yet, it would be foolish to suggest that those results will be the same for all Vulkan games. Plus, the 1070 is the same price (within $50 at least) and literally half the power draw. And you don't have to find a spot for a radiator or deal with pump noise and, frankly, pretty shitty frametimes:

https://techreport.com/review/30413/nvidia-geforce-gtx-1070-graphics-card-reviewed/7

Check the bottom of that page and others. The Fury X stutters more than the 1070 at every single game. So maybe it gets close to average frame times, but have fun with your fits and starts. Really worth saving the $50 that you'll spend on extra electricity in a year!

Captain_Tom said:
Nvidia's pricing is so insane right now that people are considering some ridiculous alternatives - that's my point.


You seem to think you're entitled to linear performance increases for dollar spent. That's just not how this works. AMD charges what they do because their cards are inferior. Nvidia charges what they want right now because most people are fine with a 1060 and the rest of us will buy good cards for extra money. The average person is not going to drop $1200 on a Titan XP, but isn't it nice that a 4K60 card finally exists? And Nvidia will sell it to you! AMD will give you a technically adequate solution that doesn't actually work half the time (or worse). Sure, if you can find 3 470's for MSRP (probably closer to $650-700 after tax and markup from the reference) and crossfire them, you'll get an approximation of maybe a 1080 (definitely not a Titan XP) in games that support crossfire, but the funny thing is no DX12 or Vulkan game supports crossfire at this point, except for AotS. Devs have said that they have to do all the work for multi-card now. Do you really think they're going to make three-way work, ever? And you say that's the future. So really, there is no situation where those cards are going to get a workout. That's why you can get the power so cheap by crossfiring them. Because they are going to spend most of the time stuttering and crashing in new games, and there is no work being done with multi-adapter support in DX12 games beyond AotS.

So you say you're not recommending them, but you sound pretty convinced that crossfire is useful for some odd reason. Or maybe you're just trying to justify why AMD can't even reach 75% of the Titan XP level of performance, and won't be able to until sometime next year.
Posted on Reply
#21
Captain_Tom
danbert2000 said:
Hrmm... not really, not even before an overclock. I like your suggestion that there are "true" DX12 games. They just happen to be the ones that do better on AMD, right? Like Hitman and DOOM and those are the only games people play, right? Feel free to click around this review, they have a bunch of benches that detail why the Fury X is a dead end card even more so now than when it was bitchslapped at launch by the 980 Ti:

https://www.guru3d.com/articles_pages/evga_geforce_gtx_1070_sc_superclocked_gaming_review,15.html

I guess you can hold on for dear life to DOOM Vulkan performance, but seeing as Nvidia hasn't updated their Vulkan drivers to implement async yet, it would be foolish to suggest that those results will be the same for all Vulkan games. Plus, the 1070 is the same price (within $50 at least) and literally half the power draw. And you don't have to find a spot for a radiator or deal with pump noise and, frankly, pretty shitty frametimes:

https://techreport.com/review/30413/nvidia-geforce-gtx-1070-graphics-card-reviewed/7

Check the bottom of that page and others. The Fury X stutters more than the 1070 at every single game. So maybe it gets close to average frame times, but have fun with your fits and starts. Really worth saving the $50 that you'll spend on extra electricity in a year!



You seem to think you're entitled to linear performance increases for dollar spent. That's just not how this works. AMD charges what they do because their cards are inferior. Nvidia charges what they want right now because most people are fine with a 1060 and the rest of us will buy good cards for extra money. The average person is not going to drop $1200 on a Titan XP, but isn't it nice that a 4K60 card finally exists? And Nvidia will sell it to you! AMD will give you a technically adequate solution that doesn't actually work half the time (or worse). Sure, if you can find 3 470's for MSRP (probably closer to $650-700 after tax and markup from the reference) and crossfire them, you'll get an approximation of maybe a 1080 (definitely not a Titan XP) in games that support crossfire, but the funny thing is no DX12 or Vulkan game supports crossfire at this point, except for AotS. Devs have said that they have to do all the work for multi-card now. Do you really think they're going to make three-way work, ever? And you say that's the future. So really, there is no situation where those cards are going to get a workout. That's why you can get the power so cheap by crossfiring them. Because they are going to spend most of the time stuttering and crashing in new games, and there is no work being done with multi-adapter support in DX12 games beyond AotS.

So you say you're not recommending them, but you sound pretty convinced that crossfire is useful for some odd reason. Or maybe you're just trying to justify why AMD can't even reach 75% of the Titan XP level of performance, and won't be able to until sometime next year.
Hahaha I just don't know even where to start.


1) I think it is pretty obvious "True" DX12 games are those that use full Async compute: AoS, Warhammer, and Tombraider (Now that its updated). I have no problem with you ignoring Hitman, just like no one should look at Project Cars benchmarks.

2) I love how you keep talking about this imaginary person who thinks Crossfire is a good idea right now, I would love to meet him. I put the text in bold, but I suppose I could underline it too if you can't read: The fact that people are considering 3-way crossfire over a single card shows you how much people hate Nvidia's pricing. Next time should I put it in all caps?

3) Nobody can predict the future. But what we can do is look at what we know now:

1. AMD's cards have been proven to gain 1-30% more performance in games that utilize Async.

2. Nvidia has a LONG history of building architectures that age horribly. Just look at how a 7970 beats a 780 and you can see why some people refuse to buy high-end Nvidia at this point.

3. DX12 implements systems that basically allow for hardwired crossfire support, but it is new and like all new API's in the past mGPU isn't working right away. Furthermore, AMD is also clearly preparing to make a big push for better crossfire implementation (Although it is far from successful right now lol). Also Nvidia basically told everyone to never SLI again, and frankly their mGPU plan was completely unsustainable years ago. In the future AMD will likely have well-working mGPU, and Nvidia just won't have it. But right now I wouldn't recommend it to most people, and as always nobody really knows what the future holds.


P.S. Can't believe you took that Fury X flamebait :P
Posted on Reply
#22
danbert2000
So, Captain_Tom, you're basically relying on a hope and a prayer and "historical proof" that AMD will somehow claw its way back to relevancy, even though Time Spy and gaming benchmarks show that Pascal async works just fine, Maxwell performance is still fine, and the 1060 is actually winning benchmarks for Rise of the Tomb Raider, AotS, Warhammer, and at Time Spy before an overclock that trounces whatever Polaris has been able to produce. You're also relying on AMD cards to be better than Nvidia cards a couple years after release. Never mind the fact that the long burn for AMD to finally optimize the GCN architecture is over, so there won't be any incredible leaps in AMD performance any more (that came from piss-poor drivers for years being fixed). You poor AMD fans justify this total ownage of performance segments by Nvidia like an abused spouse or cultist. You even made up imaginary friends that want 3-way crossfire and then detailed the whole justification for it. Yes, no one even mentioned it until you did. No one is actually considering 3-way crossfire, yet you say it's a fact like somehow everyone on this forum knows an idiot that's considering crossfire at all this gen.

That's fine, you can do whatever you want. I'll just treat you like I treat any religious person. Smile and nod.

P.S. Don't have much to say about frame times, do ya? *stutters intensify*
Posted on Reply
#23
dyonoctis
danbert2000 said:
So, Captain_Tom, you're basically relying on a hope and a prayer and "historical proof" that AMD will somehow claw its way back to relevancy, even though Time Spy and gaming benchmarks show that Pascal async works just fine, Maxwell performance is still fine, and the 1060 is actually winning benchmarks for Rise of the Tomb Raider, AotS, Warhammer, and at Time Spy before an overclock that trounces whatever Polaris has been able to produce. You're also relying on AMD cards to be better than Nvidia cards a couple years after release. Never mind the fact that the long burn for AMD to finally optimize the GCN architecture is over, so there won't be any incredible leaps in AMD performance any more (that came from piss-poor drivers for years being fixed). You poor AMD fans justify this total ownage of performance segments by Nvidia like an abused spouse or cultist. You even made up imaginary friends that want 3-way crossfire and then detailed the whole justification for it. Yes, no one even mentioned it until you did. No one is actually considering 3-way crossfire, yet you say it's a fact like somehow everyone on this forum knows an idiot that's considering crossfire at all this gen.

That's fine, you can do whatever you want. I'll just treat you like I treat any religious person. Smile and nod.

P.S. Don't have much to say about frame times, do ya? *stutters intensify*
mmmh. I'm aggree with some point, but disagree with other.
The fact that Pascal only get 5-6 % more performance with async on while GCN get 10-12% isn't working just fine since nvidia async compute is twice less efficient. It's true that it doesn't really matter right know since they still manage to be faster, but that's not a reason. Something is never "just fine" when the competitor get the lead, even for thing as trivial as that.



Next (it's not really toward you) the whole 7970 beat a 780 most likely come from the fact that amd is using GCN for quite a while now, while Nvidia released different architecture. If a recent GCN get gain with a driver, the older are likely to get some gain too. While Nvidia got better stuff to do than going on a hunt for +3 years old gpu driver tweak.

According to guru3D frametime review, it's fine on all card. Nothing that is visually noticable if it's not game engine related.

My conclusion is: AMD doesn't have any money right now so they can't just bring a new architecture out of the blue like that. Navi (2018) seems to be the real huge improvement , but we don't know jacks***t about what it is . All we can do is wait and pray that the gap don't become any larger for us to see a GTX 2060 FE at 423$ because Amd can't compete.

I'm not one of those guy that can afford to put 300-800$/€ on a gpu like it's nothing, I'm in the sub 200 $/€ market, and there isn't a lot of stuff interesting for me at the moment. So I actually enjoy seeing actual competition rather than total ownage. Let's be honest. Who wasn't happy when the Radeon HD 4000 where so fast and so cheap that 250 €/$-160€/$ was enough to get a gaming experience that was worth 500€/$-240€/$ a few month ago ? (and i'm not talking about different gen but actual competitor)
Posted on Reply
#24
Captain_Tom
dyonoctis said:
mmmh. I'm aggree with some point, but disagree with other.
The fact that Pascal only get 5-6 % more performance with async on while GCN get 10-12% isn't working just fine since nvidia async compute is twice less efficient. It's true that it doesn't really matter right know since they still manage to be faster, but that's not a reason. Something is never "just fine" when the competitor get the lead, even for thing as trivial as that.



Next (it's not really toward you) the whole 7970 beat a 780 most likely come from the fact that amd is using GCN for quite a while now, while Nvidia released different architecture. If a recent GCN get gain with a driver, the older are likely to get some gain too. While Nvidia got better stuff to do than going on a hunt for +3 years old gpu driver tweak.

According to guru3D frametime review, it's fine on all card. Nothing that is visually noticable if it's not game engine related.

My conclusion is: AMD doesn't have any money right now so they can't just bring a new architecture out of the blue like that. Navi (2018) seems to be the real huge improvement , but we don't know jacks***t about what it is . All we can do is wait and pray that the gap don't become any larger for us to see a GTX 2060 FE at 423$ because Amd can't compete.

I'm not one of those guy that can afford to put 300-800$/€ on a gpu like it's nothing, I'm in the sub 200 $/€ market, and there isn't a lot of stuff interesting for me at the moment. So I actually enjoy seeing actual competition rather than total ownage. Let's be honest. Who wasn't happy when the Radeon HD 4000 where so fast and so cheap that 250 €/$-160€/$ was enough to get a gaming experience that was worth 500€/$-240€/$ a few month ago ?
Jesus finally I talk to someone else with common sense. You are spot on! Right now AMD made a calculated defeat in the high end to spit out cheap midrange cards because marketshare is FAR more important than prestige right now. We all know Vega is coming with a ton of features, and apparently AMD sees no point in competing in the enthusiast market until they have the perfect card.
Posted on Reply
#25
fullinfusion
Vanguard Beta Tester
Dam nice looking card...
Posted on Reply
Add your own comment