ASUS ROG Crosshair VIII HERO (Wi-fi) review
MSI MEG X570 Godlike review
Corsair NightSword RGB gaming mouse review
Promo: URCDKey Summer Sale Windows 10 Pro for $11
Corsair MP600 PCIe 4.0 NVMe SSD review
Palit GeForce RTX 2070 SUPER JetStream review
MSI GeForce RTX 2070 SUPER Gaming X TRIO review
Palit GeForce RTX 2060 SUPER JetStream review
ASUS GeForce RTX 2060 SUPER Strix review
MSI GeForce RTX 2060 SUPER Gaming X review
AMD Radeon RX rumor: 12nm Polaris 30 in the 4th quarter
A fairly credible source in the past at the Chiphell forums is now mentioning that AMD might be releasing a new (another) version of Polaris. Polaris 30 would be manufactured at 12 nm and would be 15 percent faster compared to the current RX 570 and 580, Radeon RX 680?
This new info opens up a whole new ballpark of speculation, will there be a Radeon RX 600 series of GPUs based on Polaris? The current generation is built on 14 nm FinFET LPP+. So if AMD moves to a 12nm part, the clocks could indeed be bumped up a bit more, much like what Zen+ is doing for Ryzen 2000.
It's not the first time that this rumor now has surfaced and for now, it is just that, a rumor from Chiphells forum user wjm47196, via computerbase. So take it with a grain of salt.
AMD Radeon Vega 12 and Vega 20 Wave Hello From Ashes Of The Singularity Database - 06/19/2018 05:43 PM
It is getting a little too obvious who is planting small Vega leaks into the Ashes Of The Singularity database, I mean repetitive pattern each and every year. And what company is all over who is all ...
Resident Evil 2 Remake to be optimized for AMD Radeon - 06/15/2018 03:26 PM
AMD announced some new partnerships with Rebellion, Ubisoft and Capcom at E3. According to Scott Herkelman, VP and GM at AMD Radeon Gaming, The Division 2, Resident Evil 2 Remake and Strange Brigade w...
AMD Radeon Pro V340 with two Vega 10 GPUs surfaces online - 06/15/2018 08:49 AM
Some specs have been spotted on the AMD Radeon Pro V340, a graphics card intended for the professional segment of the market, interesting fact: it has 32GB HBM2 and has been fitted with two Vega GPU'...
AMD Radeon Pro Software Enterprise Edition 18.Q2 - 05/10/2018 08:36 AM
AMD released the new Radeon Pro Software Enterprise Driver 18.Q2, continuing to protect the investment of IT professionals and buyers by providing significant performance improvements in leading profe...
Download AMD Radeon Software Adrenalin Edition 18.4.1 drivers - 04/30/2018 06:40 PM
the Radeon Technologies Group has reelased Radeon Software Adrenalin Edition 18.4.1. This release contains initial support for the launch of Windows 10 April 2018 update today. This release also cont...
Embra
Senior Member
Posts: 864
Joined: 2014-11-19
Senior Member
Posts: 864
Joined: 2014-11-19
#5562031 Posted on: 07/02/2018 06:39 PM
AMD needs volume for these GPUs. Lots of them and at good prices. Bread and butter GPUs.
AMD needs volume for these GPUs. Lots of them and at good prices. Bread and butter GPUs.
Evildead666
Senior Member
Posts: 1026
Joined: 2003-09-14
Senior Member
Posts: 1026
Joined: 2003-09-14
#5562034 Posted on: 07/02/2018 06:45 PM
It would be rather interesting if it will be at last more efficient perf/w and worthy perf/$ from its preview generation models.
I mean look at vega. It failed. Newer generation never meant to be more power hungry and more expensive AT THE SAME TIME than previous one. Vega would be a success if it would be more eco friendly than fury, this was never supposed to happen.
Let's see if AMD learned something from the last year's fiasco.
They sold everything they could produce.
That is not failure. Its not a big win, its middling, but certainly not a failure.
On topic, Its going to be a very fun Autumn-Winter this year
This nm shift will give the same clocks, but lower power, or higher clocks, at the same power. Going from the past, it would be higher clocks at the same power.
So AMD doing a re-spin of Polaris on 12nm (with GDDR6?), and Nvidia could be putting Turing (Probably = Volta - (Tensor+FP64)) with GDDR6 out to dry.
No new tech until Xmas/New year at least....
It would be rather interesting if it will be at last more efficient perf/w and worthy perf/$ from its preview generation models.
I mean look at vega. It failed. Newer generation never meant to be more power hungry and more expensive AT THE SAME TIME than previous one. Vega would be a success if it would be more eco friendly than fury, this was never supposed to happen.
Let's see if AMD learned something from the last year's fiasco.
They sold everything they could produce.
That is not failure. Its not a big win, its middling, but certainly not a failure.
On topic, Its going to be a very fun Autumn-Winter this year
This nm shift will give the same clocks, but lower power, or higher clocks, at the same power. Going from the past, it would be higher clocks at the same power.So AMD doing a re-spin of Polaris on 12nm (with GDDR6?), and Nvidia could be putting Turing (Probably = Volta - (Tensor+FP64)) with GDDR6 out to dry.
No new tech until Xmas/New year at least....
sammarbella
Senior Member
Posts: 3931
Joined: 2014-12-09
Senior Member
Posts: 3931
Joined: 2014-12-09
#5562036 Posted on: 07/02/2018 06:48 PM
"Volume"?
Like in "Sells volume"?
Good luck with that.
Mining craze is over (for now (?)) so AMD "gaming" GPUs must compete with Nvidia gaming GPUs in performance and performance per dollar in a strange market: PC gaming GPUs used for GAMING.
The same market they failed to compete since 4-5 years...
AMD needs volume for these GPUs. Lots of them and at good prices. Bread and butter GPUs.
"Volume"?
Like in "Sells volume"?
Good luck with that.
Mining craze is over (for now (?)) so AMD "gaming" GPUs must compete with Nvidia gaming GPUs in performance and performance per dollar in a strange market: PC gaming GPUs used for GAMING.
The same market they failed to compete since 4-5 years...
Silva
Senior Member
Posts: 873
Joined: 2013-06-04
Senior Member
Posts: 873
Joined: 2013-06-04
#5562045 Posted on: 07/02/2018 07:03 PM
I would only buy an used GPU for less than half the price (not what they are asking for), so I'd gladly buy the RX670 over any GPU on the used marked that was used for mining.
I wonder if they'll get GDDR6 too. Although more expensive, it uses less power and has more density I think so less chips or bigger memory. Or a smaller bus and same speed as GDDR5 equivalent because of higher clocks.
I would only buy an used GPU for less than half the price (not what they are asking for), so I'd gladly buy the RX670 over any GPU on the used marked that was used for mining.
I wonder if they'll get GDDR6 too. Although more expensive, it uses less power and has more density I think so less chips or bigger memory. Or a smaller bus and same speed as GDDR5 equivalent because of higher clocks.
RzrTrek
Senior Member
Posts: 2217
Joined: 2012-04-16
Senior Member
Posts: 2217
Joined: 2012-04-16
#5562060 Posted on: 07/02/2018 07:29 PM
They could begin with releasing a new driver, free from stutter in major DX11 games and fully functional OSD, but perhaps that's asking too much.
They could begin with releasing a new driver, free from stutter in major DX11 games and fully functional OSD, but perhaps that's asking too much.
schmidtbag
Senior Member
Posts: 4113
Joined: 2012-11-10
Senior Member
Posts: 4113
Joined: 2012-11-10
#5562078 Posted on: 07/02/2018 08:18 PM
I will only see this as good news if this is 15% improvement is specifically above the 580 (the mentioning of the 570 is a little off-putting), and, if this is sold as the 670 or lower. If this is going to be the 680, that's looking kind of bleak.
I don't get any stuttering. No comment on the OSD though - I never use it (but, maybe that's what your problem is?).
Too bad the open source Linux drivers don't support DX, because they're overall better than the Windows drivers. To my knowledge, they can be ported to Windows.
I will only see this as good news if this is 15% improvement is specifically above the 580 (the mentioning of the 570 is a little off-putting), and, if this is sold as the 670 or lower. If this is going to be the 680, that's looking kind of bleak.
They could begin with releasing a new driver, free from stutter in major DX11 games and fully functional OSD, but perhaps that's asking too much.
I don't get any stuttering. No comment on the OSD though - I never use it (but, maybe that's what your problem is?).
Too bad the open source Linux drivers don't support DX, because they're overall better than the Windows drivers. To my knowledge, they can be ported to Windows.
Celcius
Senior Member
Posts: 156
Joined: 2018-06-21
Senior Member
Posts: 156
Joined: 2018-06-21
#5562091 Posted on: 07/02/2018 08:49 PM
Carping about the energy efficiency of any GCN-based video card in mid-2018 is very much like flogging the proverbial dead horse. GCN has never been about energy effiency, and likely never will be. We're seven years into this architecture; it's not likely to suddenly go Prius on us. I accepted this reality quite some time ago, and, obviously only speaking for myself, am quite comfortable with my (sometimes) 200W Radeon. The neighbor down the road is probably using more than ten times that much on his dinky mining setup, 24/7, and I'm supposed to agonize over my few hours of gaming per week? No...
I've been unpacking and installing the Windows 7 Adrenaline drivers for my Windows 8.1 rig without a single issue, and haven't been plagued with any performance problems. There is always room for improvement, of course, but I can't say I've much of anything to complain about, aside from RTG dropping Windows 8.1 support like a hot, radioactive potato. But, I've gotten over that, too.
Discussing the pricing of ANY video card while in the shadow of crypto-currency mining and VRAM pricing issues is, at best, nebulous. I'd like to see the launch-date MSRP return on every AMD and Nvidia card, for sure. Not holding my breath waiting for that to happen.
If AMD decides to take a third bite of the Polaris apple, I feel the only thing they need to accomplish is to give the buyer the level of performance that he or she requires, at a price they are comfortable paying. Done. The specifics of whatever contrivance is pushing the pixels around on my screen isn't something I obsess over while immersed in work, or while gaming.
Carping about the energy efficiency of any GCN-based video card in mid-2018 is very much like flogging the proverbial dead horse. GCN has never been about energy effiency, and likely never will be. We're seven years into this architecture; it's not likely to suddenly go Prius on us. I accepted this reality quite some time ago, and, obviously only speaking for myself, am quite comfortable with my (sometimes) 200W Radeon. The neighbor down the road is probably using more than ten times that much on his dinky mining setup, 24/7, and I'm supposed to agonize over my few hours of gaming per week? No...
I've been unpacking and installing the Windows 7 Adrenaline drivers for my Windows 8.1 rig without a single issue, and haven't been plagued with any performance problems. There is always room for improvement, of course, but I can't say I've much of anything to complain about, aside from RTG dropping Windows 8.1 support like a hot, radioactive potato. But, I've gotten over that, too.
Discussing the pricing of ANY video card while in the shadow of crypto-currency mining and VRAM pricing issues is, at best, nebulous. I'd like to see the launch-date MSRP return on every AMD and Nvidia card, for sure. Not holding my breath waiting for that to happen.
If AMD decides to take a third bite of the Polaris apple, I feel the only thing they need to accomplish is to give the buyer the level of performance that he or she requires, at a price they are comfortable paying. Done. The specifics of whatever contrivance is pushing the pixels around on my screen isn't something I obsess over while immersed in work, or while gaming.
RooiKreef
Senior Member
Posts: 353
Joined: 2016-06-08
Senior Member
Posts: 353
Joined: 2016-06-08
#5562119 Posted on: 07/02/2018 09:57 PM
Since I have a Polaris card, I always wondered about the architecture if it had similar clock frequencies as Nvidia Pascal +- 2000Mhz and then AMD add some HBM2 memory into the equation. I would love too see that combo.
Since I have a Polaris card, I always wondered about the architecture if it had similar clock frequencies as Nvidia Pascal +- 2000Mhz and then AMD add some HBM2 memory into the equation. I would love too see that combo.
Fox2232
Senior Member
Posts: 9359
Joined: 2012-07-20
Senior Member
Posts: 9359
Joined: 2012-07-20
#5562123 Posted on: 07/02/2018 10:29 PM
Price, that's only thing that matters. Polaris was not expensive to make (w/ 5.7B transistors). And pricing was good before mining. If this can be those 15% faster, then it will be mostly between Fury and Fury X in performance. Sometimes above, sometimes below this range.
That's very good performance. RX 580 is good GPU, but it misses that tiny bit of performance to max out game like Witcher 3 on 1080p and keep 60fps+ at most of the times.
So, I think this refresh (if real) will be just OK for AAA titles and people who do not want to spend fortune on gaming PC.
But I do question background... that untold story. If this comes out, it means AMD took officially step back from Vega. That would be final confirmation about Raja's "leaving".
Price, that's only thing that matters. Polaris was not expensive to make (w/ 5.7B transistors). And pricing was good before mining. If this can be those 15% faster, then it will be mostly between Fury and Fury X in performance. Sometimes above, sometimes below this range.
That's very good performance. RX 580 is good GPU, but it misses that tiny bit of performance to max out game like Witcher 3 on 1080p and keep 60fps+ at most of the times.
So, I think this refresh (if real) will be just OK for AAA titles and people who do not want to spend fortune on gaming PC.
But I do question background... that untold story. If this comes out, it means AMD took officially step back from Vega. That would be final confirmation about Raja's "leaving".
Seikon
Member
Posts: 78
Joined: 2016-02-09
Member
Posts: 78
Joined: 2016-02-09
#5562142 Posted on: 07/02/2018 11:32 PM
Well my Rx480 is chocked by memory bandwidth , using better timings gives me e better performance boost then going from 1303 to 1420 on the core , so if they manage to bring a 1600mhz polaris +gddr6 it should beat a gtx1070 if you ask me :/
Well my Rx480 is chocked by memory bandwidth , using better timings gives me e better performance boost then going from 1303 to 1420 on the core , so if they manage to bring a 1600mhz polaris +gddr6 it should beat a gtx1070 if you ask me :/
AMDfan
Member
Posts: 48
Joined: 2017-12-21
Member
Posts: 48
Joined: 2017-12-21
#5562146 Posted on: 07/02/2018 11:56 PM
Why would they shrink Polaris to 12 Nm when Vega 20 on 7Nm is in full production (as we speak), launch Q4 2018 Q1 2019....
I say there will be no Polaris 30 cards
Why would they shrink Polaris to 12 Nm when Vega 20 on 7Nm is in full production (as we speak), launch Q4 2018 Q1 2019....
I say there will be no Polaris 30 cards
BReal85
Senior Member
Posts: 297
Joined: 2016-10-25
Senior Member
Posts: 297
Joined: 2016-10-25
#5562159 Posted on: 07/03/2018 01:45 AM
Sorry, what's your problem? Nvidia is releasing worse and worse drivers maybe since last summer. They should get something done with those since they are the leading gamer GPU brand.
They could begin with releasing a new driver, free from stutter in major DX11 games and fully functional OSD, but perhaps that's asking too much.
Sorry, what's your problem? Nvidia is releasing worse and worse drivers maybe since last summer. They should get something done with those since they are the leading gamer GPU brand.
Fox2232
Senior Member
Posts: 9359
Joined: 2012-07-20
Senior Member
Posts: 9359
Joined: 2012-07-20
#5562187 Posted on: 07/03/2018 06:13 AM
Why would they shrink Polaris to 12 Nm when Vega 20 on 7Nm is in full production (as we speak), launch Q4 2018 Q1 2019....
I say there will be no Polaris 30 cards
You may be quite right, this may never come. But main reason for going Polaris again would be number of SP/TMU/ROPs per 1B of transistors.
If you take parts of GPU which are there out of picture (display controller, cache, ... let's presume 0.5B transistors), then Polaris delivers good 30% more SP/TMU/ROPs per transistor. That's huge.
Secondly, Clock to Clock comparison of Vega vs. Fiji did not end up well for Vega. While much smaller Polaris touched Fiji at times and should not been able to do so without some good improvements in architecture.
Polaris with 2304SP/144TMU/32ROPs, multiplied by 1.35 for clock advantage over Fiji => 3110SP/194.4/43.2. That's nowhere near Fiji's 4096/256/64. With this in mind, I expect that some important improvement in Polaris did not carry to Vega.
Polaris is simply much cheaper and seems to have some advantages.
Why would they shrink Polaris to 12 Nm when Vega 20 on 7Nm is in full production (as we speak), launch Q4 2018 Q1 2019....
I say there will be no Polaris 30 cards
You may be quite right, this may never come. But main reason for going Polaris again would be number of SP/TMU/ROPs per 1B of transistors.
If you take parts of GPU which are there out of picture (display controller, cache, ... let's presume 0.5B transistors), then Polaris delivers good 30% more SP/TMU/ROPs per transistor. That's huge.
Secondly, Clock to Clock comparison of Vega vs. Fiji did not end up well for Vega. While much smaller Polaris touched Fiji at times and should not been able to do so without some good improvements in architecture.
Polaris with 2304SP/144TMU/32ROPs, multiplied by 1.35 for clock advantage over Fiji => 3110SP/194.4/43.2. That's nowhere near Fiji's 4096/256/64. With this in mind, I expect that some important improvement in Polaris did not carry to Vega.
Polaris is simply much cheaper and seems to have some advantages.
warlord
Senior Member
Posts: 2151
Joined: 2012-10-22
Senior Member
Posts: 2151
Joined: 2012-10-22
#5562191 Posted on: 07/03/2018 06:42 AM
Polaris 30 aka rx 680 have a reason to exist only if it will be absolutely equal to gtx 1070 or always better than 980ti/furyX. It should be just the gpu below 1070ti/vega56. Anything else is for the trashcan.
Polaris 30 aka rx 680 have a reason to exist only if it will be absolutely equal to gtx 1070 or always better than 980ti/furyX. It should be just the gpu below 1070ti/vega56. Anything else is for the trashcan.
Click here to post a comment for this news story on the message forum.
Senior Member
Posts: 2151
Joined: 2012-10-22
It would be rather interesting if it will be at last more efficient perf/w and worthy perf/$ from its preview generation models.
I mean look at vega. It failed. Newer generation never meant to be more power hungry and more expensive AT THE SAME TIME than previous one. Vega would be a success if it would be more eco friendly than fury, this was never supposed to happen.
Let's see if AMD learned something from the last year's fiasco.