Well, AMD may be releasing the RX4xx Refresh next month - RX5xx possibly incoming in April.

[Zion Halcyon]

Grain of salt time:

https://videocardz.com/66359/amd-radeon-rx-500-series-to-launch-in-april

That said, on April 4th Radeon RX 580 and Radeon RX 570 will be available for sale. A week later, on April 11th AMD will unveil its Radeon RX 560 and Radeon RX 550 series. All cards, except RX 550, are most likely rebranded from Radeon RX 400 series.

AMD Radeon RX Vega will not be released in April, which somehow aligns with our initial information, that Radeon 500 series are expected in May. So in other words, expect Radeon RX 500 rebrands in April and Vega a month or two later.​

 

[NKD]

Grain of salt time:

https://videocardz.com/66359/amd-radeon-rx-500-series-to-launch-in-april

That said, on April 4th Radeon RX 580 and Radeon RX 570 will be available for sale. A week later, on April 11th AMD will unveil its Radeon RX 560 and Radeon RX 550 series. All cards, except RX 550, are most likely rebranded from Radeon RX 400 series.

AMD Radeon RX Vega will not be released in April, which somehow aligns with our initial information, that Radeon 500 series are expected in May. So in other words, expect Radeon RX 500 rebrands in April and Vega a month or two later.​

You mean "Radeon 500 series is expected in April" and Vega in May.

 

[Zion Halcyon]

You mean "Radeon 500 series is expected in April" and Vega in May.

That was just a quite from the article.

Like I said, take with a grain of salt.

 

[Schmave]

Hmm, so the 580 and 570 will just be reused Polaris 10 chips with small clock speed bumps? Seems a little weak to me, especially since many of the AIB RX 480s can already do 1340 MHz boost clocks. They need to get Vega into their entire product stack ASAP.

 

[SeymourGore]

I dunno, if true, I'm not sure why Raja couldn't have revealed this information yesterday during the GDC AMD conference. Could've really blown some minds:

"The RX 500 series launches next month, but for everyone in the audience today who owns a RX 400 series Radeon, you will get the chance to try a Radeon 500 series GPU today!"

Then Raja would give RX 500 GPU stickers to the audience.

 

[Algrim]

Now that they have more experience with the node they could also be more efficient and boost higher.

 

[NKD]

Hmm, so the 580 and 570 will just be reused Polaris 10 chips with small clock speed bumps? Seems a little weak to me, especially since many of the AIB RX 480s can already do 1340 MHz boost clocks. They need to get Vega into their entire product stack ASAP.

Yea but the problem is rx 480 was very iffy with clocks. It does not sustain 1266 and to sustain 1340 it has to be very well tweaked. Mine runs at 1288 stock, I can run it at 1340 with lower voltage but it dips down if i add any voltage to get more clock speed but I choose not to overclock it.

If new rx580 can sustain 1340 and amd has made some better power efficiency it could end up being 15-20% faster just with that. Now it depends on if amd has left some overclocking headroom with these. We will find out if this is truly improved revision or just they slapped a better cooler to get those speeds and better power delivery. But I don't think it will received too well by the reviewers if its just not improved in any other regard and still throttles.

But I do think that this will be a better revision since its almost been year they have had the rx480 out. I am sure they have a better revision here.

 

[razor1]

I dunno, if true, I'm not sure why Raja couldn't have revealed this information yesterday during the GDC AMD conference. Could've really blown some minds:

"The RX 500 series launches next month, but for everyone in the audience today who owns a RX 400 series Radeon, you will get the chance to try a Radeon 500 series GPU today!"

Then Raja would give RX 500 GPU stickers to the audience.

They don't want to talk about a rebrand, that will take focus away from Vega..... And actually hurt Vega's possible impression.

 

[Schmave]

Yea but the problem is rx 480 was very iffy with clocks. It does not sustain 1266 and to sustain 1340 it has to be very well tweaked. Mine runs at 1288 stock, I can run it at 1340 with lower voltage but it dips down if i add any voltage to get more clock speed but I choose not to overclock it.

If new rx580 can sustain 1340 and amd has made some better power efficiency it could end up being 15-20% faster just with that. Now it depends on if amd has left some overclocking headroom with these. We will find out if this is truly improved revision or just they slapped a better cooler to get those speeds and better power delivery. But I don't think it will received too well by the reviewers if its just not improved in any other regard and still throttles.

But I do think that this will be a better revision since its almost been year they have had the rx480 out. I am sure they have a better revision here.

That's true. I guess I wasn't really considering the power consumption. My Sapphire Nitro card stays at 1342 MHz boost clock but it sure as hell doesn't use < 150W at that speed. If they could make a new revision of the chip that stays under 150W while boosting up to 1340 it would definitely be an improvement.

 

[Anarchist4000]

Yea but the problem is rx 480 was very iffy with clocks. It does not sustain 1266 and to sustain 1340 it has to be very well tweaked. Mine runs at 1288 stock, I can run it at 1340 with lower voltage but it dips down if i add any voltage to get more clock speed but I choose not to overclock it.

If new rx580 can sustain 1340 and amd has made some better power efficiency it could end up being 15-20% faster just with that. Now it depends on if amd has left some overclocking headroom with these. We will find out if this is truly improved revision or just they slapped a better cooler to get those speeds and better power delivery. But I don't think it will received too well by the reviewers if its just not improved in any other regard and still throttles.

But I do think that this will be a better revision since its almost been year they have had the rx480 out. I am sure they have a better revision here.

Conceivably those capacitors they designed for Zen, and likely Vega, could be added to the design if they weren't already in there. That should stabilize signals a bit bringing down voltages/power. At the very least Zen appears to be clocking better than originally expected. Throw on some faster RAM and a significant bump shouldn't be unreasonable.

 

[razor1]

Yea but the problem is rx 480 was very iffy with clocks. It does not sustain 1266 and to sustain 1340 it has to be very well tweaked. Mine runs at 1288 stock, I can run it at 1340 with lower voltage but it dips down if i add any voltage to get more clock speed but I choose not to overclock it.

If new rx580 can sustain 1340 and amd has made some better power efficiency it could end up being 15-20% faster just with that. Now it depends on if amd has left some overclocking headroom with these. We will find out if this is truly improved revision or just they slapped a better cooler to get those speeds and better power delivery. But I don't think it will received too well by the reviewers if its just not improved in any other regard and still throttles.

But I do think that this will be a better revision since its almost been year they have had the rx480 out. I am sure they have a better revision here.

It's not even a 10% increase on the clock speed, how do you expect it to be 15-20% faster. And getting 10% more clocks as it is right now, The power usage goes through the roof. Its not a 10% increase in power more like a 20%, so yeah a respin is definitely required (and more), metal and mask respin. Transistor layout changes cause that is what affects target clocks and power usage. This will not be a simple respin. If they want something like a 10% increase in clocks at the same power usage, more like a revision of the chip.

And as you stated you are getting vdroop cause the chip/board wasn't made to do the frequency and clocks that you were trying for, its maxing out, so better cooling isn't really going to help that much. Well it might depending... but two of the three variables are maxing out, something is got to give.

 

[NKD]

It's not even a 10% increase on the clock speed, how do you expect it to be 15-20% faster. And getting 10% more clocks as it is right now, The power usage goes through the roof. Its not a 10% increase in power more like a 20%, so yeah a respin is definitely required (and more), metal and mask respin. Transistor layout changes cause that is what affects target clocks and power usage. This will not be a simple respin. If they want something like a 10% increase in clocks at the same power usage, more like a revision of the chip.

And as you stated you are getting vdroop cause the chip/board wasn't made to do the frequency and clocks that you were trying for, its maxing out, so better cooling isn't really going to help that much. Well it might depending... but two of the three variables are maxing out, something is got to give.

I was comparing it to more of the Reference design. Which had reports of going down to like 1160 under heavy gaming. Thats what I was referring to if the new design can sustain 1340 across the board it will considerably improve gameplay.

 

[razor1]

I was comparing it to more of the Reference design. Which had reports of going down to like 1160 under heavy gaming. Thats what I was referring to if the new design can sustain 1340 across the board it will considerably improve gameplay.

well we have to look at what the voltage, is at that frequency, and the temps to get an idea. cause if temps aren't going crazy then its a problem with different parts of the chip and how they are being pushed in different parts of a game. Its a pretty complex analysis. Vdroop is hard to figure out.

 

[noko]

You look at Polaris and it seems to have a very large variations in ability form OC, voltage and power. Compared that to Pascal which are virtually all identical in spec or performance from each other. Pascal is a chip done extremely well or right. That also means Polaris has potential to be improved upon or tightened up to less variations and ultimately a better performer with that. It would be hard for me to believe AMD would spend money on a major revision for Polaris unless it was a sure thing and they want to dominate the less then $300 market if Nvidia could not really respond quickly. Polaris 11 fails against Pascal 107 big time, if AMD can get that straighten out at least they maybe able to get more the mobile market over time. Another wait and see in the end.

 

[Deleted member 89018]

Really hoping it doesn't wind up a simple rebrand, regardless of naming conventions they use for Radeon Vega. If nothing else getting a full Polaris 11 in there would be nice.

 

[Jiv_au]

It would be hard for me to believe AMD would spend money on a major revision for Polaris unless it was a sure thing and they want to dominate the less than $300 market if Nvidia could not really respond quickly.

Are you speculating that AMD might be going after the GTX 1070 and possibly 1080 for laptops?

 

[Shintai]

You look at Polaris and it seems to have a very large variations in ability form OC, voltage and power. Compared that to Pascal which are virtually all identical in spec or performance from each other. Pascal is a chip done extremely well or right. That also means Polaris has potential to be improved upon or tightened up to less variations and ultimately a better performer with that. It would be hard for me to believe AMD would spend money on a major revision for Polaris unless it was a sure thing and they want to dominate the less then $300 market if Nvidia could not really respond quickly. Polaris 11 fails against Pascal 107 big time, if AMD can get that straighten out at least they maybe able to get more the mobile market over time. Another wait and see in the end.

GTX1060 got announced it would get 9Ghz as a go. Polaris needs up the performance or dip even lower in price.

Its simply damage control.

 

[NKD]

GTX1060 got announced it would get 9Ghz as a go. Polaris needs up the performance or dip even lower in price.

Its simply damage control.

I don't see why nvidia would just out of no where announce memory speed bump on gtx 1060. I mean does it really need more bandwidth? not like its lacking any for that performance segment. It seems like Nvidia knows about Amd refresh cards and adjusting the 1060 anyway they can.

What is to gain by just faster memory chips? I am sure tiny benefit but AMD has nothing to worry about there. Polaris doesn't need to do shit. It can already be had for cheaper than 1060.

 

[razor1]

I don't see why nvidia would just out of no where announce memory speed bump on gtx 1060. I mean does it really need more bandwidth? not like its lacking any for that performance segment. It seems like Nvidia knows about Amd refresh cards and adjusting the 1060 anyway they can.

What is to gain by just faster memory chips? I am sure tiny benefit but AMD has nothing to worry about there. Polaris doesn't need to do shit. It can already be had for cheaper than 1060.

Not really with the DX12 performance drivers if its all true, the 1060 is going to have the advantage again. Not sure about the 1060 with bandwidth but I'm thinking over clocked versions will need more bandwidth.

 

[razor1]

You look at Polaris and it seems to have a very large variations in ability form OC, voltage and power. Compared that to Pascal which are virtually all identical in spec or performance from each other. Pascal is a chip done extremely well or right. That also means Polaris has potential to be improved upon or tightened up to less variations and ultimately a better performer with that. It would be hard for me to believe AMD would spend money on a major revision for Polaris unless it was a sure thing and they want to dominate the less then $300 market if Nvidia could not really respond quickly. Polaris 11 fails against Pascal 107 big time, if AMD can get that straighten out at least they maybe able to get more the mobile market over time. Another wait and see in the end.

There are variations with Pascal too, but when you hand layout transistors, you have better control on things like that..... Not that its done right, it costs a ton more to do it that way and a lot more time.

 

[pandora's box]

And people say its Nvidia holding back the GPU industry.

 

[sparks]

call a 400 a 500 and charge more. Damn they are a sh#$ company.

 

[oldmanbal]

low/midrange mass market videocard strategy = NOT [H]

 

[sparks]

and all the NOT talk about vega.
must be as good as the new ryzen cpu.
yes that's sarcasm.

 

[Syntax_Error]

If they are offering the 5xx rebrands in April, it is extremely unlikly that they would be launching Vega much before Oct. or Nov.

 

[Algrim]

That may be true, or they simply don't have parts for Vega that are going to be in the Polaris price/performance brackets. IMO, it would be silly for RTG to completely replace the RX 480/470/460 parts with Vega counterparts due to all the money sunk into Polaris development.

 

[Anarchist4000]

That may be true, or they simply don't have parts for Vega that are going to be in the Polaris price/performance brackets. IMO, it would be silly for RTG to completely replace the RX 480/470/460 parts with Vega counterparts due to all the money sunk into Polaris development.

Raja did say they had limited engineering resources. With all the Vega parts we've seen likely having HBM2 it makes sense. Those Polaris low to mid tiers are probably destined to be replaced by APUs which could feature Vega cores. Making small Vega designs probably wasn't worth it.

 

[Shintai]

Raja did say they had limited engineering resources. With all the Vega parts we've seen likely having HBM2 it makes sense. Those Polaris low to mid tiers are probably destined to be replaced by APUs which could feature Vega cores. Making small Vega designs probably wasn't worth it.

APUs wont replace Polaris parts. Not even remotely close.