• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

AMD Readies Radeon RX 490 for December?

I really hope it is Vega, and probably is but it still could be Polaris using the highest binned chips

There is definitely plenty of variance in the chips for them to do that. I wouldn't be surprised if better yields could have them easily into the 1600mhz range that with GDDR5X could make for a 1070 contender.
 
I don't think RX480 core with 1600MHz could rival GTX 1070. They'd have to add few more shaders as well imo. I mean, GTX 1070 runs at what, 1.9 GHz ? I don't think RX480 could make up that much with extra 300-400 MHz alone.
 
I don't think RX480 core with 1600MHz could rival GTX 1070. They'd have to add few more shaders as well imo. I mean, GTX 1070 runs at what, 1.9 GHz ? I don't think RX480 could make up that much with extra 300-400 MHz alone.

Correct. RX480 has been clocked to 1450 stable and it's nowhere even near, not even same ballpark by any margin.

Just sticking GDDR5X on RX480 isn't gonna be RX 490, I'll give you that as a guarantee right now. RX480 is already well treated VRAM wise, well balanced overall, and core OC's deliver while memory OC's don't do much for most games.

Given the stream processor count I really don't see a 1070 competitor. And I also think we are forgetting an extremely important little detail: the real 1070 contender is already out and its called Fury X. AMD still wants to get rid of that stock and they go for cheap. In addition, a SMART AMD would position their high end-entry card a noticeable margin above the 1070 at a similar price point, because it will both be pushing the 1080 out of the arena and at the same time make 1070 obsolete. The 1070 and 1080 are about 25% apart in performance, if they were to position it somewhere round the middle @ stock clocks that would allow a 'OC to 1080 perf' card for a 1070 price, which is going to be a big winner in this segment, because everyone does that little OC. Matter of fact if they go down that route (wishful thinking I know, but still, it makes some sense) they may well have a '970 sales killer' on their hands. AMD has got the momentum PR-wise right now.
 
Last edited:
Correct. RX480 has been clocked to 1450 stable and it's nowhere even near, not even same ballpark by any margin.

Just sticking GDDR5X on RX480 isn't gonna be RX 490, I'll give you that as a guarantee right now. RX480 is already well treated VRAM wise, well balanced overall, and core OC's deliver while memory OC's don't do much for most games.

Given the stream processor count I really don't see a 1070 competitor. And I also think we are forgetting an extremely important little detail: the real 1070 contender is already out and its called Fury X. AMD still wants to get rid of that stock and they go for cheap. In addition, a SMART AMD would position their high end-entry card a noticeable margin above the 1070 at a similar price point, because it will both be pushing the 1080 out of the arena and at the same time make 1070 obsolete. The 1070 and 1080 are about 25% apart in performance, if they were to position it somewhere round the middle @ stock clocks that would allow a 'OC to 1080 perf' card for a 1070 price, which is going to be a big winner in this segment, because everyone does that little OC. Matter of fact if they go down that route (wishful thinking I know, but still, it makes some sense) they may well have a '970 sales killer' on their hands. AMD has got the momentum PR-wise right now.

You are referring to consumer binned Polaris that does 1450Mhz.
I'm talking commercial binned chips that go into Embedded chipsets in small form factor equipment where there is not much cooling. Those chips are capable of much higher clockspeeds, but the silicon process is new and there weren't enough yields of high quality chips to supply the consumer market.
 
I'd put my money on this being dual Polaris designed for VR (which uses dual FHD anyway, and dual medium GPUs makes more sense than one fast single). I don't expect Vega this year, rather about the same time as Zen (which January launch is questionable). Just a guess...
 
You are referring to consumer binned Polaris that does 1450Mhz.
I'm talking commercial binned chips that go into Embedded chipsets in small form factor equipment where there is not much cooling. Those chips are capable of much higher clockspeeds, but the silicon process is new and there weren't enough yields of high quality chips to supply the consumer market.

But so far we haven't seen a single AMD GCN card go over 1500, and SFF stuff +enterprise market usually gets lower clocks rather than higher ones because longevity and reliability.. Highly doubtful they can just pop open a bin of super high clocking cards on the same or similar arch. Remember, AMD is still on the GCN train that is being finetuned. Polaris is not so very different from previous iterations and I doubt AMD has the R&D to 'do a Pascal' already, not to mention an architecture that just doesn't work the same way.
 
But so far we haven't seen a single AMD GCN card go over 1500, and SFF stuff +enterprise market usually gets lower clocks rather than higher ones because longevity and reliability.. Highly doubtful they can just pop open a bin of super high clocking cards on the same or similar arch. Remember, AMD is still on the GCN train that is being finetuned. Polaris is not so very different from previous iterations and I doubt AMD has the R&D to 'do a Pascal' already, not to mention an architecture that just doesn't work the same way.

They didn't have the numbers for consumers, but with low TDP it should be possible to get extreme OC on the commercial binned chips.
 
i thought HBM2 was 1TB/s?
HBM is 512GB/s as in current Fiji aka Fury/Nano
 
Presumably 50+ PowerPoint slides incoming!


All showing "great new features" that AMD partners will release or could make or that may happen... if the wind stays fairing and the sails are set true to escape this damn VAPOR!!!!

I'm excited more for what HBM and Zen will bring to the CPU side.
 
Correct. RX480 has been clocked to 1450 stable and it's nowhere even near, not even same ballpark by any margin.

3dmark while not an end all be all puts them in the same bracket with a stock 1070 vs an RX480@16xx/2250, both score roughly 18K. More memory bandwidth would only help.
 
We already know that HBM provides more than enough bandwidth to drive a modern high-end card. The real question is how high will the core be clocked? If we're talking about a stable 1.25Ghz at stock, it might not be too bad.
 
My best guess is that this is a larger GPU built off of polaris 10. Perhaps a 3072 or 4096 core GPU with either a 512 or 384 bit interface, perhaps GDDR5X memory. Looking at the newer 480s compared to the launch models, 14nm today can make a far superior GPU to what was given to us at launch. The XFX 480 GTR at 1338 MHz pulls less power then a stock launch 480. wouldnt surprise me if they have a 1070 competitor for the christmas season.

It is also possible that this is vega, but I doubt it. I doubt that HBM2 is ready for production, and I doubt AMD was able to accelerate VEGA that quickly.
 
By all rumors circling around, this card will be between 1060 and 1070, both on price and performance ;)
 
Whatever it is it will stink. It's the end of November and AMD isn't talking. That's a massive red flag, and the norm for AMD launches. Meh.
 
Correct. RX480 has been clocked to 1450 stable and it's nowhere even near, not even same ballpark by any margin.

Just sticking GDDR5X on RX480 isn't gonna be RX 490, I'll give you that as a guarantee right now. RX480 is already well treated VRAM wise, well balanced overall, and core OC's deliver while memory OC's don't do much for most games.

Given the stream processor count I really don't see a 1070 competitor. And I also think we are forgetting an extremely important little detail: the real 1070 contender is already out and its called Fury X. AMD still wants to get rid of that stock and they go for cheap. In addition, a SMART AMD would position their high end-entry card a noticeable margin above the 1070 at a similar price point, because it will both be pushing the 1080 out of the arena and at the same time make 1070 obsolete. The 1070 and 1080 are about 25% apart in performance, if they were to position it somewhere round the middle @ stock clocks that would allow a 'OC to 1080 perf' card for a 1070 price, which is going to be a big winner in this segment, because everyone does that little OC. Matter of fact if they go down that route (wishful thinking I know, but still, it makes some sense) they may well have a '970 sales killer' on their hands. AMD has got the momentum PR-wise right now.

"they may well have a '970 sales killer' on their hands."
Yea, it's called the RX 480 and it's no 970 killer if you look at AMD's loss of market share.... to nVIDIA.
 
My best guess is that this is a larger GPU built off of polaris 10. Perhaps a 3072 or 4096 core GPU with either a 512 or 384 bit interface, perhaps GDDR5X memory. Looking at the newer 480s compared to the launch models, 14nm today can make a far superior GPU to what was given to us at launch. The XFX 480 GTR at 1338 MHz pulls less power then a stock launch 480. wouldnt surprise me if they have a 1070 competitor for the christmas season.

It is also possible that this is vega, but I doubt it. I doubt that HBM2 is ready for production, and I doubt AMD was able to accelerate VEGA that quickly.

Very well could be since the naming scheme would suggest that VEGA is Fury.
 
Vega 11 should be in Scorpio. That would fit with RX490 naming scheme as I think Vega 10 (bigger vega) will have the fury brand.
Isn't Vega opposite from Polaris? Vega 11 should be larger/faster than 10.
 
Last edited:
I loaded up on AMD stock last winter @ $2.25 a share holding 5000 shares for Zen&RX490
 
I loaded up on AMD stock last winter @ $2.25 a share holding 5000 shares for Zen&RX490
I'm kicking myself for not doing it when it was ~1.50 a share. I knew I should and I should have but, life got in the way. :(
 
I'm kicking myself for not doing it when it was ~1.50 a share. I knew I should and I should have but, life got in the way. :(

it's about $8.50 I'm still going to buy a little more before Zen.
 
it's about $8.50 I'm still going to buy a little more before Zen.
I wouldn't buy more at this point. It's a lot more riskier with it being this high considering if Xen isn't "the CPU of the year", it stands to lose value. The time to buy was a year or two ago when it was the lowest it had ever been, not now.
 
Back
Top