Radeon RX Vega Discussion Thread

Peppercorn said:
Bullshit. A big loose steamy pile of bullshit.
No matter what AMD says about Vega, it will be considered as hype. Is this your first rodeo?
Click to expand...

Huh, you basically agreed with me indirectly. Not saying anything leads to no hype and at this close to sale and with Nvidia just releasing a good card at a reasonable price, is not a good thing.

AMD needs to say something and that is for one reason, to build hype. This translations into desire to buy which translates into people not buying a Nvidia card and hopefully buying an AMD card when it does get released. And at this point, people deserve performance figures which is the best type of hype and best tool to get people to wait. Building hype is not a bad thing by itself, it's not leading up to expectations by overpromising that is the bad thing and even this is not that bad.

Just be honest in what your trying to hype and it will lead to little to no backlash. It's better than what AMD is doing now.

Honestly, you indirectly agreeing with me is the only none-shit argument you have had and that was piled with your usual shit of a response.
 
Presbytier said:
I don't know why AMD does not just borrow from Nvidia's playbook. When Nvidia launches a card they have ONE event they show they card announce its price tell us what it can do and tells us when we can buy it. AMD on the other hand announced Vega over a year ago has talked about it in several events has had 3 separate events just this year to hype Vega without releasing any concrete info on it. So which strategy to you seems to work better for both the company and the consumer?
Click to expand...

Gotta say I agree. I've pretty much been an ATI/AMD graphics guy since the 9700/9500. AMD is about to lose me as a graphics customer over the way they have handled this whole vega fiasco. I'm sick of waiting and they are giving me no reason to. That crap and cream deal really pissed me off and the 1080ti looks sweet.

P.S. Incase you haven't heard vega is going to be called RX Vega... now wait at least 3 more months bitches...
 
RAutrey said:
Gotta say I agree. I've pretty much been an ATI/AMD graphics guy since the 9700/9500. AMD is about to lose me as a graphics customer over the way they have handled this whole vega fiasco. I'm sick of waiting and they are giving me no reason to. That crap and cream deal really pissed me off and the 1080ti looks sweet.

P.S. Incase you haven't heard vega is going to be called RX Vega... now wait at least 3 more months bitches...
Click to expand...
That whole big surprise being the name is where they lost me. I bought a monitor with freesync as GSYNC is way to expensive I currently use a GTX 1080 but I have been paying attention to Vega because I would not mind using some of the vsync features, but they keep talking endlessly about nothing and can't seem to give any concrete details.
 
Presbytier said:
That whole big surprise being the name is where they lost me. I bought a monitor with freesync as GSYNC is way to expensive I currently use a GTX 1080 but I have been paying attention to Vega because I would not mind using some of the vsync features, but they keep talking endlessly about nothing and can't seem to give any concrete details.
Click to expand...

I don't know if it is that they are trying to get every drop of performance out of Vega to compete with the 1080Ti. But it sure is turning into a real pain just waiting.
 
Maddness said:
I don't know if it is that they are trying to get every drop of performance out of Vega to compete with the 1080Ti. But it sure is turning into a real pain just waiting.
Click to expand...
I think it is literally not ready and they are struggling to either A beef up production or B driver issues.
 
I keep checking up on this thread specifically hoping to one day to see something I can bite. I have always been a little partial to AMD/ATI (mostly for image quality reasons) but I don't know if I can do it anymore. The 1080ti is looking awesome right now and if the Zotac amps are out before I have anything tangible to grasp on from AMD, I think thats going to be it. Vega is being handled worse than Fiji at this point and I didn't think that was possible.
 
Bottom line is judge Vega once it is released to see if it is worth it for you. If something is already on the market that will fit your wants/needs and is priced ever so sweet - go for it. The 1080Ti to me is the best buy right now for a high end GPU and I do not see Vega being able to significantly beat it and most likely it will be less. If AMD sells it for $499 and it is within 10% of the 1080Ti then I would not see much regrets in buying a 1080Ti if one is shopping for a top end card now.
 
Yeah, I was originally planning on waiting for Vega, but the 1080 Ti is too nice (so I got one). Fully plan to sell the Ti and get 2 Vegas, provided it's even remotely competitive.
 
noko said:
The 1080Ti to me is the best buy right now for a high end GPU and I do not see Vega being able to significantly beat it and most likely it will be less.
Click to expand...

Nothing will get close to 1080 Ti until Volta. You can take that to the bank.

If AMD sells it for $499 and it is within 10% of the 1080Ti then I would not see much regrets in buying a 1080Ti if one is shopping for a top end card now.
Click to expand...

$499 for +/- 10% of 1080 Ti will never happen.
 
noko said:
Bottom line is judge Vega once it is released to see if it is worth it for you. If something is already on the market that will fit your wants/needs and is priced ever so sweet - go for it. The 1080Ti to me is the best buy right now for a high end GPU and I do not see Vega being able to significantly beat it and most likely it will be less. If AMD sells it for $499 and it is within 10% of the 1080Ti then I would not see much regrets in buying a 1080Ti if one is shopping for a top end card now.
Click to expand...
if vega is within 10% of 1080ti it will not sell for $499. Just simply not happening. It will definitely land at 599. Will probably still be a good buy 100 less than 1080ti. I am definitely sure Vega will be atleast $549 at its absolute minimum. But Amd is likely shooting for 599 to 650 if it can get within 10% of ti.
 
NKD said:
if vega is within 10% of 1080ti it will not sell for $499. Just simply not happening. It will definitely land at 599. Will probably still be a good buy 100 less than 1080ti. I am definitely sure Vega will be atleast $549 at its absolute minimum. But Amd is likely shooting for 599 to 650 if it can get within 10% of ti.
Click to expand...
I agree with everything you said, now if it comes within 10% or beats the 1080Ti in general then a $499 price tag would probably make the 1080Ti folks (at least a few) regret not waiting in that case. That was the jest of previous comment as unclear as it probably was.

If it is 10% slower and comes in at $599 - I don't think many of the 1080Ti buyers would even blink an eye. They would be playing on it for up to 3 months getting all the benefits of that added performance during that time.
 
If you think Vega will be anywhere near 1080TI performance you are just going to disappoint yourself. Vega is essentially a 250W 1080 card with packed math.

Its very clear from leaks, AMDs news and press where the card performs. Just as with Polaris, Ryzen and so on. So stop fooling yourself (again).
 
noko said:
If you have a poor setup or case yes - otherwise no.
Click to expand...

Sorry to say but it does not reach anywhere near its rated boost clock of 1000MHz.
Really worth considering its TDP is 100W less than the Fury X that has a boost clock of 1050MHz; these are the AMD spec rather than measured.

One example here:
01-Clock-Rate.png


They measured temps in a tiny closed case and even then temps were not the issue (topped out at 77c).
Note they compared results to their open bench and concluded:
Let’s take a look at the clock rate results in the closed mini-ITX PC case, which, we might add, don’t really differ from those of the open configuration.
Click to expand...
Cheers
 
Last edited:
Presbytier said:
I think it is literally not ready and they are struggling to either A beef up production or B driver issues.
Click to expand...

I think they are trying to find a way to sell an economic viable product that can compete on the $300-700 range. Vega may cost too much to compete against NVIDIA products on this range, and will therefore never see the light of the day.

DPI said:
$499 for +/- 10% of 1080 Ti will never happen.
Click to expand...

Given Ryzen price, i would not be surprised at all.

Shintai said:
Its very clear from leaks, AMDs news and press where the card performs
Click to expand...

It must be very comfortable to life in a world with so much certainty. 1 month ago you probably were pretty certain that Ryzen would not kill X99, but today there are only 2 CPUs on everyone's radar, and neither uses X99.
 
You know the biggest problem I see for both manufactures is they are approaching the overkill point fro 1080p and 1440p. Only 4k will continue to drive sales of cards in great numbers as the 1080 and 1440 people will be easily above 60 fps on current generation hardware. Next generation of cards the mid grade usually is equal to last generations high end and I see a sales slump occurring after that as people just wont feel a need to upgrade anymore. I may turn out to be wrong, guess will see in the next 2 years or so.
 
geok1ng said:
I think they are trying to find a way to sell an economic viable product that can compete on the $300-700 range. Vega may cost too much to compete against NVIDIA products on this range, and will therefore never see the light of the day.



Given Ryzen price, i would not be surprised at all.



It must be very comfortable to life in a world with so much certainty. 1 month ago you probably were pretty certain that Ryzen would not kill X99, but today there are only 2 CPUs on everyone's radar, and neither uses X99.
Click to expand...

Don't feed the troll! It was a nice conversation until he comes punching on Vega. There are ways to word things, to criticize AMD without any insults is something he is just not capable of.
 
Gideon said:
You know the biggest problem I see for both manufactures is they are approaching the overkill point fro 1080p and 1440p. Only 4k will continue to drive sales of cards in great numbers as the 1080 and 1440 people will be easily above 60 fps on current generation hardware. Next generation of cards the mid grade usually is equal to last generations high end and I see a sales slump occurring after that as people just wont feel a need to upgrade anymore. I may turn out to be wrong, guess will see in the next 2 years or so.
Click to expand...
Yeah at 1080 unless you are going for high refresh rates (and even then it becomes very CPU bound) these cards are overkill. The problem with 4k isn't just the fact these cards have a hard time hitting 60fps at max settings it's that 4k monitors are still extremely expensive $500+ items. There are also a lot of competing factors now to when you include these super widescreen monitors plus so few games actually take advantage of 4k with such low resolution textures. I think there needs to be more standardization on 4k and what features are important like HDR.
 
Gideon said:
people just wont feel a need to upgrade anymore.
Click to expand...

The market is receiving the first quantum dot displays with native HDR support and higher refresh rates. So we are living a multi rail arms race on the gaming industry:

- some gamers are upgrading for higher resolutions
-others are upgrading for higher refresh rates
-content industry is pushing a larger color gamut and bit depth

I do agree that 720p gaming is possible with integrated solutions, low cost discrete graphics can game at 1080p and 1440p is possible in the $200+ range. But all of it does not mean that there will not be a reason to upgrade. Quite the opposite, pick my example: over the next months i expect to get a 4k 120Hz 10 bit monitor. There is nothing on the market capable of driving that, i will likely spend a lot on graphics updates just to get half decent frame rates at abysmal quality settings. And this is without considering the hardware needs of the next gen VR kits.
 
geok1ng said:
The market is receiving the first quantum dot displays with native HDR support and higher refresh rates. So we are living a multi rail arms race on the gaming industry:

- some gamers are upgrading for higher resolutions
-others are upgrading for higher refresh rates
-content industry is pushing a larger color gamut and bit depth

I do agree that 720p gaming is possible with integrated solutions, low cost discrete graphics can game at 1080p and 1440p is possible in the $200+ range. But all of it does not mean that there will not be a reason to upgrade. Quite the opposite, pick my example: over the next months i expect to get a 4k 120Hz 10 bit monitor. There is nothing on the market capable of driving that, i will likely spend a lot on graphics updates just to get half decent frame rates at abysmal quality settings. And this is without considering the hardware needs of the next gen VR kits.
Click to expand...
I just bought the Samsung CFG70 which is a Quantum Dot 1080 144hz monitor it is gorgeous. We have a tremendous amount of choice right now when it comes to monitors and unfortunately I think that is part of the problem without some defined standards Devs don't know what to design for but this isn't really an issue for GPU makers because as long as they just push for more powerfull hardware they are likely to stay ahead of the game.
 
Those are nice monitors, but the issue I still see is there are very few that will get that kind of stuff. Look at the steam survey and 1080 dominates by a large margin, 4k is tiny in comparison. I just see the same problem coming for GPU's that has come for CPU's, it's good enough. Used to be every 2 years I upgraded my GPU as frame rate would be a issue, with a Vega or 1080 I just dont see me needing to upgrade for far longer which is not good for AMD and Nvidia. But I just dont see me ever getting a 4k monitor, just dont have the space for it to make sense for me. But I never doubt software programmers to be able to bloat their program and make it need more power as well. Guess will find out soon.
 
geok1ng said:
I think they are trying to find a way to sell an economic viable product that can compete on the $300-700 range. Vega may cost too much to compete against NVIDIA products on this range, and will therefore never see the light of the day.



Given Ryzen price, i would not be surprised at all.



It must be very comfortable to life in a world with so much certainty. 1 month ago you probably were pretty certain that Ryzen would not kill X99, but today there are only 2 CPUs on everyone's radar, and neither uses X99.
Click to expand...


The problem with trying to carry over ryzen's surprising price/performance is the the GPU department is mutually exclusive from their CPU depart. Also the situations are vastly different.

Ryzen is largely a 200mm2 die competing against Intels overpriced 232mm2.

Hence at 500 dollars it is still making a giant profit. Intel was just making crazy money over the years, particularly because that 232mm2 die is a salvage die.

AMD ryzen can look like a bargain because we have been paying so much for Intel's chips over the years. Far more than excess of the GPU market.

On the other hand, ryzen is a 500mm2 die, with very expensive HBM2. On top off that, you have to add the board cost, the power circuitry, the cooling solution, the licences for display technology and the partner cut to make the profit on the videocard. Hence, AMD can't afford to undercut Nvidia too much, particularly when Nvidia is using a salvage die and AMD is using a full chip.

More over pricing vega too low causes collateral damage for the rest of AMD's lineup. Nvidia will cut the pricing of their cards because they are cheaper to make than AMD and they would want to prevent the loss of marketshare.

E.g If VEGA performed 10% better than the gtx 1080 ti and priced it at 499(impossible price). Nvidia would drop the priced of the gtx 1080 down to 350 and a gtx 1070 would have to go down to 250-230 dollars. A gtx 1060 would have to go down to 150-130 and The rest of the polaris lineup would have to sell at 100 dollars.

This would be a catastrophe for AMD because they would be selling at a huge loss and essentially making no money on Vega, particularly when you include the R and D expenditure.

In short vega is too big, has too many components and has to many negative repercussions to underprice it if it has the performance. One more thing to add is Ryzen was the result of AMD pouring all its resources into the CPU division with Jim Keller heading it. Vega was designed from a skeleton of it's original crew at AMD graphic division with almost all of the workforce coming from a new division in China(why it was designed in Shanghai). Lack of resources + new talent force is not the recipe for success.

However, from the current narrative AMD is projecting, it seems like they are already projecting VEGA to be more of a gtx 1080 level card than a 1080 ti card. This is because they are talking about bringing affordable 4k performance. This would be fine and dandy if VEGA was a small die with cheaper memory. But Vega looking at the cost of production was designed to be an enthusiast card through and through. What this seems to indicate is AMD is repurposing the original goals of Vega to control the original expectations of the card.

Also, the performance demonstrated on Vega seems to indicate gtx 1080 performance as well. Remember when AMD was demonstrating Polaris samples in games like Hitman and the performance was indicative of r9 390 performance, but rumors to get the hype going and to get news hits were saying early drivers, expect gtx 980 ti/fury x performance when released? Where did the real performance lie?

With AMD giving pretty explicit hints that this is a card targeted at 1080 and the affordability angle, I think it is pretty safe to say the performance of Vega will be more along the lines of the gtx 1080. If Vega performed at gtx 1080 ti level and beyond, from a marketing standpoint, it would make sense to drop more VEGA performance figures now to slow down gtx 1080 ti and 1080 sales. Vega also at the gtx 1080 ti and beyond performance is not marketed as an affordable 4k solution and shown to be an enthusiast level card to prevent cannibalization of Polaris sales. The current silence this close to launch means they are having a difficult time deciding how to market this card, particularly with the price shifts and the release of the gtx 1080 ti.
 
tajoh111 said:
1) More over pricing vega too low causes collateral damage for the rest of AMD's lineup. Nvidia will cut the pricing of their cards because they are cheaper to make than AMD and they would want to prevent the loss of marketshare.
E.g If VEGA performed 10% better than the gtx 1080 ti and priced it at 499(impossible price). Nvidia would drop the priced of the gtx 1080 down to 350 and a gtx 1070 would have to go down to 250-230 dollars. A gtx 1060 would have to go down to 150-130 and The rest of the polaris lineup would have to sell at 100 dollars.
This would be a catastrophe for AMD because they would be selling at a huge loss and essentially making no money on Vega, particularly when you include the R and D expenditure.
.........................................................
2)With AMD giving pretty explicit hints that this is a card targeted at 1080 and the affordability angle, I think it is pretty safe to say the performance of Vega will be more along the lines of the gtx 1080. If Vega performed at gtx 1080 ti level and beyond, from a marketing standpoint, it would make sense to drop more VEGA performance figures now to slow down gtx 1080 ti and 1080 sales. Vega also at the gtx 1080 ti and beyond performance is not marketed as an affordable 4k solution and shown to be an enthusiast level card to prevent cannibalization of Polaris sales. The current silence this close to launch means they are having a difficult time deciding how to market this card, particularly with the price shifts and the release of the gtx 1080 ti.
Click to expand...

Those are very logical arguments in my opinion, and from logical point of view i agree 100% ;)
 
CSI_PC said:
Sorry to say but it does not reach anywhere near its rated boost clock of 1000MHz.
Really worth considering its TDP is 100W less than the Fury X that has a boost clock of 1050MHz; these are the AMD spec rather than measured.

One example here:
01-Clock-Rate.png


They measured temps in a tiny closed case and even then temps were not the issue (topped out at 77c).
Note they compared results to their open bench and concluded:

Cheers
Click to expand...
A closed mini-itx case - geee I wonder why. Hey Kyle after launch in a well thought out configuration was able to do it, I do it, Presbytier did it at 1100mhz! I've even showed it in this thread here with VR on video:
https://hardforum.com/threads/the-thrill-of-the-fight-vr-testing-on-radeon-nano.1926453/
Look at the clock speed, fan speed etc. Nano was displaying grahics on the monitor, graphics (each eye is separate) in the Vive and recording using Relive at the same time. Let me know what you think. Now I heard Tom's Hardware is looking for guy's/gal's like you :sneaky:.

Now I am wondering if AMD can do the same or better with Vega, having a kick ass SFF design?
 
tajoh111 said:
The problem with trying to carry over ryzen's surprising price/performance is the the GPU department is mutually exclusive from their CPU depart. Also the situations are vastly different.

Ryzen is largely a 200mm2 die competing against Intels overpriced 232mm2.

Hence at 500 dollars it is still making a giant profit. Intel was just making crazy money over the years, particularly because that 232mm2 die is a salvage die.

AMD ryzen can look like a bargain because we have been paying so much for Intel's chips over the years. Far more than excess of the GPU market.

On the other hand, ryzen is a 500mm2 die, with very expensive HBM2. On top off that, you have to add the board cost, the power circuitry, the cooling solution, the licences for display technology and the partner cut to make the profit on the videocard. Hence, AMD can't afford to undercut Nvidia too much, particularly when Nvidia is using a salvage die and AMD is using a full chip.

More over pricing vega too low causes collateral damage for the rest of AMD's lineup. Nvidia will cut the pricing of their cards because they are cheaper to make than AMD and they would want to prevent the loss of marketshare.

E.g If VEGA performed 10% better than the gtx 1080 ti and priced it at 499(impossible price). Nvidia would drop the priced of the gtx 1080 down to 350 and a gtx 1070 would have to go down to 250-230 dollars. A gtx 1060 would have to go down to 150-130 and The rest of the polaris lineup would have to sell at 100 dollars.

This would be a catastrophe for AMD because they would be selling at a huge loss and essentially making no money on Vega, particularly when you include the R and D expenditure.

In short vega is too big, has too many components and has to many negative repercussions to underprice it if it has the performance. One more thing to add is Ryzen was the result of AMD pouring all its resources into the CPU division with Jim Keller heading it. Vega was designed from a skeleton of it's original crew at AMD graphic division with almost all of the workforce coming from a new division in China(why it was designed in Shanghai). Lack of resources + new talent force is not the recipe for success.

However, from the current narrative AMD is projecting, it seems like they are already projecting VEGA to be more of a gtx 1080 level card than a 1080 ti card. This is because they are talking about bringing affordable 4k performance. This would be fine and dandy if VEGA was a small die with cheaper memory. But Vega looking at the cost of production was designed to be an enthusiast card through and through. What this seems to indicate is AMD is repurposing the original goals of Vega to control the original expectations of the card.

Also, the performance demonstrated on Vega seems to indicate gtx 1080 performance as well. Remember when AMD was demonstrating Polaris samples in games like Hitman and the performance was indicative of r9 390 performance, but rumors to get the hype going and to get news hits were saying early drivers, expect gtx 980 ti/fury x performance when released? Where did the real performance lie?

With AMD giving pretty explicit hints that this is a card targeted at 1080 and the affordability angle, I think it is pretty safe to say the performance of Vega will be more along the lines of the gtx 1080. If Vega performed at gtx 1080 ti level and beyond, from a marketing standpoint, it would make sense to drop more VEGA performance figures now to slow down gtx 1080 ti and 1080 sales. Vega also at the gtx 1080 ti and beyond performance is not marketed as an affordable 4k solution and shown to be an enthusiast level card to prevent cannibalization of Polaris sales. The current silence this close to launch means they are having a difficult time deciding how to market this card, particularly with the price shifts and the release of the gtx 1080 ti.
Click to expand...
I really don't think we can tell much of anything on performance when it will be launched. Just take a Fiji chip and if you can clock it to 1500mhz that would give a rather decent jump in performance. Now add in cleaning up the front end, better efficiency etc. -> AMD should know about know rough clock speeds wherever that will fall. Clock speed which will also be dictated by power as well. Putting out Vega with 10% performance advantage over a 1080 but consuming 300w would be somewhat comical. Another reason why you keep it secret is because you really don't know yet where you will fall or what the AIBs are willing to go with etc. Plus putting most attention on their next generation cpu is probably 100:1 more important for their survival.
 
Shintai said:
If you think Vega will be anywhere near 1080TI performance you are just going to disappoint yourself. Vega is essentially a 250W 1080 card with packed math.

Its very clear from leaks, AMDs news and press where the card performs. Just as with Polaris, Ryzen and so on. So stop fooling yourself (again).
Click to expand...

I think you need to up your game a little, provide some compelling reasoning vice trying to slide in something at the end that makes you look bad. For example you could use this argument, 250/180 ~ 1.39 ratio or 39%
  • The 1080Ti consumes by spec 39% more power over the 1080
  • The 1080Ti performance I believe is ~ 35% greater than the 1080
  • Nvidia has almost perfect scaling with perf/w between P104 and P102 -> That is F_ _ _ing amazing!
  • AMD would first have to at least equal perf/w of Pascal to have a 250w part to beat the 1080Ti
  • Polaris consumes 30%+ more power than Pascal abouts for the same performance
  • AMD using HBM buys them around 30w at most in savings and more like 20w (you can correct me here if you know a more accurate number)
  • So Vega would have to beat Polaris by a significant amount in perf/w but does not have to equal Pascal efficiency - they have ~ 20w to play with
  • I've not seen anything to indicate AMD will leap frog Nvidia with efficiency at higher frequencies than what they used before, which leads me to think AMD would have to put out a card that consumes more power, like 275w to beat a 1080Ti if at all. Unless AMD pulls a rabbit out of a hat this seems unlikely. So now it becomes perf/$ if they want to sell a lot of Vega's.
My prediction at this time, wild ass guess is: Vega will sit between the 1080Ti and 1080, consume 250w+, be priced reasonable and will have unique features that will entice many to buy.
 
noko said:
I think you need to up your game a little, provide some compelling reasoning vice trying to slide in something at the end that makes you look bad. For example you could use this argument, 250/180 ~ 1.39 ratio or 39%
  • The 1080Ti consumes by spec 39% more power over the 1080
  • The 1080Ti performance I believe is ~ 35% greater than the 1080
  • Nvidia has almost perfect scaling with perf/w between P104 and P102 -> That is F_ _ _ing amazing!
  • AMD would first have to at least equal perf/w of Pascal to have a 250w part to beat the 1080Ti
  • Polaris consumes 30%+ more power than Pascal abouts for the same performance
  • AMD using HBM buys them around 30w at most in savings and more like 20w (you can correct me here if you know a more accurate number)
  • So Vega would have to beat Polaris by a significant amount in perf/w but does not have to equal Pascal efficiency - they have ~ 20w to play with
  • I've not seen anything to indicate AMD will leap frog Nvidia with efficiency at higher frequencies than what they used before, which leads me to think AMD would have to put out a card that consumes more power, like 275w to beat a 1080Ti if at all. Unless AMD pulls a rabbit out of a hat this seems unlikely. So now it becomes perf/$ if they want to sell a lot of Vega's.
My prediction at this time, wild ass guess is: Vega will sit between the 1080Ti and 1080, consume 250w+, be priced reasonable and will have unique features that will entice many to buy.
Click to expand...

AOTS, AMD Demo, nothing shows anywhere near 1080TI performance but rather 1080.

If you only look at power consumption, Pascal got a 2x advantage over Polaris. The RX480 pretty much use the same power as a GTX1080. If you want a more equal one you can use the 1070 vs 480 since it removes the GDDR5X part as well as cutdown. Then Pascal uses 18W less and performs 41% better. or 59% perf/watt at 1440p.

Remember when AOTS for Polaris 10 and Ryzen got shown and the fanboys was in complete denial?

HBM isn't buying AMD 30W at all. I know HBM is another mega hype and buzzword. GDDR5X and soon GDDR6 killed any power advantage for now that HBM had and its down to ECC being its last main advantage. And how did HBM do for Fiji? That's right, a 384bit GDDR5(Not X) part using Maxwell with 50% more memory had no issues there.

upload_2017-3-16_10-29-13.png

perfrel_2560_1440.png

perfwatt_2560_1440.png
 
Last edited:
noko said:
...................................
My prediction at this time, wild ass guess is: 1) Vega will sit between the 1080Ti and 1080, 2)consume 250w+, 3)be priced reasonable and 4)will have unique features that will entice many to buy.
Click to expand...

1) If they launch their highly-hyped product, and this is proved to be between GTX1080 & GTX1080Ti, then this means : Even worst results for AMD, compared to FuryX Vs 980Ti history. (*FuryX was at least competitive with NVidia's flagship, while now, NV's flagship will remain unchallenged, so that will be no good news for AMD's future lines of GPUs)
2) 250+ power consumption is nothing worth mentioning now is it?
3)AMD's recent history with FuryX's price has already proven this estimation wrong, or premature at best !!.(from the point that it was priced equally to NV's card)
4) If a card fails in performance, then no matter how many unique features it may has (*or hasn't), no one will pay attention, since the GPU failed at the most important factor for a flagship GPU: performance ;)
 
noko said:
A closed mini-itx case - geee I wonder why. Hey Kyle after launch in a well thought out configuration was able to do it, I do it, Presbytier did it at 1100mhz! I've even showed it in this thread here with VR on video:
https://hardforum.com/threads/the-thrill-of-the-fight-vr-testing-on-radeon-nano.1926453/
Look at the clock speed, fan speed etc. Nano was displaying grahics on the monitor, graphics (each eye is separate) in the Vive and recording using Relive at the same time. Let me know what you think. Now I heard Tom's Hardware is looking for guy's/gal's like you :sneaky:.

Now I am wondering if AMD can do the same or better with Vega, having a kick ass SFF design?
Click to expand...
You did see the part where they compared it to their open bench and they were the same in behaviour?
Also 77c is a great temp if it is being pushed, look at temps of AMD's other GPUs even in open bench.

Anyway just in case you missed it this is where Tom's Hardware said it is the same between open and mini ITX.
Let’s take a look at the clock rate results in the closed mini-ITX PC case, which, we might add, don’t really differ from those of the open configuration.
Click to expand...
So that 77c is not throttling, which tbh all of us would say is a pretty good temp.
And with a TDP 100W officially lower than FuryX there is no way it is designed to get close to its boost clock of 1000MHz, and this is what Tom's results confirm.
Cheers
 
Last edited:
Shintai said:
AOTS, AMD Demo, nothing shows anywhere near 1080TI performance but rather 1080.

If you only look at power consumption, Pascal got a 2x advantage over Polaris. The RX480 pretty much use the same power as a GTX1080. If you want a more equal one you can use the 1070 vs 480 since it removes the GDDR5X part as well as cutdown. Then Pascal uses 18W less and performs 41% better. or 59% perf/watt at 1440p.

Remember when AOTS for Polaris 10 and Ryzen got shown and the fanboys was in complete denial?

HBM isn't buying AMD 30W at all. I know HBM is another mega hype and buzzword. GDDR5X and soon GDDR6 killed any power advantage for now that HBM had and its down to ECC being its last main advantage. And how did HBM do for Fiji? That's right, a 384bit GDDR5(Not X) part using Maxwell with 50% more memory had no issues there.

View attachment 19407
perfrel_2560_1440.png

perfwatt_2560_1440.png
Click to expand...
Now that was damn good! Thanks for the work and feedback.

I pretty much think AMD would have to pull a magic act with Vega to beat a 1080ti. That does not mean they can't have some very cool cards. Price will probably be key for any kind of success, and some market share.
 
CSI_PC said:
You did see the part where they compared it to their open bench and they were the same in behaviour?
Also 77c is a great temp if it is being pushed, look at temps of AMD's other GPUs even in open bench.

Anyway just in case you missed it this is where Tom's Hardware said it is the same between open and mini ITX.

So that 77c is not throttling, which tbh all of us would say is a pretty good temp.
And with a TDP 100W officially lower than FuryX there is no way it is designed to get close to its boost clock of 1000MHz, and this is what Tom's results confirm.
Cheers
Click to expand...
Looks like their methods corrupted their results, bad card or bad setup. Those numbers reflect more with PowerTune at 0% with card strictly limiting itself to 175w. Then again maybe some of us got mermaid cards or something. TomsHardware results no way reflects my results or my experience. Also there was a bios update after that review and some key drivers increasing the OC or stability at least in my experience.
 
noko said:
Looks like their methods corrupted their results, bad card or bad setup. Those numbers reflect more with PowerTune at 0% with card strictly limiting itself to 175w. Then again maybe some of us got mermaid cards or something. TomsHardware results no way reflects my results or my experience. Also there was a bios update after that review and some key drivers increasing the OC or stability at least in my experience.
Click to expand...
Also needs to be balanced against that they get a power/TDP of 185W for their measurements, a site known for doing accurate power measurements.
Remember the official TDP is 175W, ergo you are not going to get higher frequency without breaking TDP even more.
The corrected 480 ended up with a TDP of around 160W compared to its official 150W while also needing good fans to cool if even slightly overclocked.

But yeah OC would give a different result, depends how long it can be sustained for extended durations, as some mentioned it did not work out long for them.
Cheers
 
Last edited:
noko said:
I really don't think we can tell much of anything on performance when it will be launched. Just take a Fiji chip and if you can clock it to 1500mhz that would give a rather decent jump in performance. Now add in cleaning up the front end, better efficiency etc. -> AMD should know about know rough clock speeds wherever that will fall. Clock speed which will also be dictated by power as well. Putting out Vega with 10% performance advantage over a 1080 but consuming 300w would be somewhat comical. Another reason why you keep it secret is because you really don't know yet where you will fall or what the AIBs are willing to go with etc. Plus putting most attention on their next generation cpu is probably 100:1 more important for their survival.
Click to expand...

I think at this point we can tell the performance a little bit.

AMD is suggesting 4k gaming which suggest somewhere between 1080 to gtx 1080 ti level.

What tell you that it's closer to the former is that AMD is talking about 4k to the masses and making it democratic what ever that means. Generally what this means is lower pricing.

However if you look at at the design of VEGA, there is nothing to suggest it's a value card or even a performance card. It has the DNA of a balls to the wall enthusiast card, more so than the gtx 1080 ti. It is 500mm2 or so in size and it is using HBM2. and it has a tdp between 225 and 300watts. What this translate to is a high cost to manufacture

If AMD's original goal was to bring 4k gaming to the masses, their design would look alot more like the gtx 1080 in size. That being a card that would be cheaper to produce, they could sell at a lower cost and still make a profit. To me it looks like AMD was trying to make an enthusiast card but by the time it hit the market or will hit the market, competition has become to tough so it has to market this card as a 4k card with a low price.

Vega's die size, the fact it is manufactured on finfet(remember cost per transistor didn't drop this generation) and HBM2 make it incredibly expensive chip to produce. The most expensive die from AMD's graphic division ever. And if we look at the premarketing of the fury x, it does not parallel this launch. This launch in some ways feels more like polaris than fiji.

Before AMD officially launched the fury X, Amd pre-marketed it as an enthusiast card from the get go. The fact that Vega is not being marketed that way and there is a big emphasis on value indicates AMD is not setting it's sights on the gtx 1080 ti, but the gtx 1080. Hence why even AMD's performance comparisons have been against the gtx 1080 which AMD had every chance to compare it to titan x pascal.

Lets stop talking about the value Ryzen brings as it has nothing on vega, because vega is more than double the size, close to 3x the transistors, which means it costs more than double the cost for the die alone compared to ryzen. So add the HBM2 on top of it, the PCB and power components and the partner cut and AMD has to sell it for more to make a profit. The fact that AMD is suggesting it is a value card when it is not is suggesting alot about its performance.
 
tajoh111 said:
I think at this point we can tell the performance a little bit.

AMD is suggesting 4k gaming which suggest somewhere between 1080 to gtx 1080 ti level.

What tell you that it's closer to the former is that AMD is talking about 4k to the masses and making it democratic what ever that means. Generally what this means is lower pricing.

However if you look at at the design of VEGA, there is nothing to suggest it's a value card or even a performance card. It has the DNA of a balls to the wall enthusiast card, more so than the gtx 1080 ti. It is 500mm2 or so in size and it is using HBM2. and it has a tdp between 225 and 300watts. What this translate to is a high cost to manufacture

If AMD's original goal was to bring 4k gaming to the masses, their design would look alot more like the gtx 1080 in size. That being a card that would be cheaper to produce, they could sell at a lower cost and still make a profit. To me it looks like AMD was trying to make an enthusiast card but by the time it hit the market or will hit the market, competition has become to tough so it has to market this card as a 4k card with a low price.

Vega's die size, the fact it is manufactured on finfet(remember cost per transistor didn't drop this generation) and HBM2 make it incredibly expensive chip to produce. The most expensive die from AMD's graphic division ever. And if we look at the premarketing of the fury x, it does not parallel this launch. This launch in some ways feels more like polaris than fiji.

Before AMD officially launched the fury X, Amd pre-marketed it as an enthusiast card from the get go. The fact that Vega is not being marketed that way and there is a big emphasis on value indicates AMD is not setting it's sights on the gtx 1080 ti, but the gtx 1080. Hence why even AMD's performance comparisons have been against the gtx 1080 which AMD had every chance to compare it to titan x pascal.

Lets stop talking about the value Ryzen brings as it has nothing on vega, because vega is more than double the size, close to 3x the transistors, which means it costs more than double the cost for the die alone compared to ryzen. So add the HBM2 on top of it, the PCB and power components and the partner cut and AMD has to sell it for more to make a profit. The fact that AMD is suggesting it is a value card when it is not is suggesting alot about its performance.
Click to expand...
That is one way at looking at it. Companies do various different things on promoting or discussing upcoming products. For example, ATI, the R300 series - 9700pro came out of the blue, no one was expecting it to be as powerful as it was and blow away the anemic FX5800 vacuum cleaner which was launched late. Then you have AMD and Polaris launch with oddball comparisons that really doesn't show much, hype wording etc. AMD does much better when they keep their mouths shut at times with video cards. This time around for their cpu launch they did a really good job. Anyways I would not read to much into how they are delivering any messaging, it will most likely be delivered differently (more hype, less, different people in charge). Vega 11 should be more their GP104 equivalent except I am not sure how those two will compare, if Vega 11 top version is 1070 level or better that would be rather a big success - 1080 level??? When Vega 10 is only compared to the 1080? So AMD could surprise us again or come out with compellingly priced products for the performance. I am also hoping AMD cards are not big huge cards for size sake but well designed cards (like the Nano) with good AIB offerings.
 
noko said:
......................... 1) 9700pro came out of the blue, no one was expecting it to be as powerful as it was and blow away the anemic FX5800 vacuum cleaner which was launched late.
................................ 2) I am also hoping AMD cards are not big huge cards for size sake but well designed cards (like the Nano) with good AIB offerings.
Click to expand...

1) :LOL: !! i couldn't stop laughing with this comment!! You traveled so back at the past, when AMD hadn't even purchased ATI yet !!:D Different companies thus different strategies when 9700pro was released ( P.S.: FX5xxx line of GPUs had been the worst time in Nvidia's history in my opinion!!)
2) Take a small sneak-peek about the card's design :https://www.techpowerup.com/231591/amds-upcoming-rx-vega-card-pictures-surface
 
noko said:
That is one way at looking at it. Companies do various different things on promoting or discussing upcoming products. For example, ATI, the R300 series - 9700pro came out of the blue, no one was expecting it to be as powerful as it was and blow away the anemic FX5800 vacuum cleaner which was launched late. Then you have AMD and Polaris launch with oddball comparisons that really doesn't show much, hype wording etc. AMD does much better when they keep their mouths shut at times with video cards. This time around for their cpu launch they did a really good job. Anyways I would not read to much into how they are delivering any messaging, it will most likely be delivered differently (more hype, less, different people in charge). Vega 11 should be more their GP104 equivalent except I am not sure how those two will compare, if Vega 11 top version is 1070 level or better that would be rather a big success - 1080 level??? When Vega 10 is only compared to the 1080? So AMD could surprise us again or come out with compellingly priced products for the performance. I am also hoping AMD cards are not big huge cards for size sake but well designed cards (like the Nano) with good AIB offerings.
Click to expand...

Your making a mistake in your analysis because you have to look at the marketing at the time.

The launch of the r300 came during a different marketing era where viral marketing, social media didn't exist and when print media was far more relevant than today. Also ATI and Nvidia were much smaller companies then. They didn't have the big press releases we have today. Hence why pre launch hype marketing did not really exist.

So it was difficult to get the hype ball going. It was basically at a time where most of the hype came at launch. Remember the gtx 8800, it was another card with surprising performance at launch. But there was little to no marketing prior to it.

Particularly because there was not nearly as many leakers because design was simpler and the time from tapeout to launch was much shorter.

I would say today's era of marketing came during the gtx 280 launch. Nvidia just grew massively due to the profits from the 88xx series and they greatly increased their marketing budget.

Back to analysis, Vega 10 will be the only Vega chip coming out this year I suspect. This is because Vega 11 is a replacement for polaris 10. AMD must recoup it's investment on Polaris 10 before they release a direct successor that obsoletes the polaris design. AMD is not rich enough to replace it's own products so quickly like Nvidia. And even Nvidia is going to try to get 2 years our of pascal. AMD tries to get as much out of a design and milks it for as long as possible because it spends much less money on R and D. Hence why most of the designs for the rx5xx series will be rebrands. Finfet designs are 2 to 3x more expensive than 28nm designs. It's why AMD only released two GPU designs last year and skipped the high end market.

The only way Vega 11 comes out this year is in the form of a console and Microsoft goes back into putting discrete chips into scorpio.

I suspect Vega 10 will come out with 10% better performance than a gtx 1080, come at the 550 dollar price point. with a cutdown chip coming at 399-450 and outperforming the gtx 1070 by 15-20% and nipping at the heels of the gtx 1080.
 
tajoh111 said:
I suspect Vega 10 will come out with 10% better performance than a gtx 1080, come at the 550 dollar price point. with a cutdown chip coming at 399-450 and outperforming the gtx 1070 by 15-20% and nipping at the heels of the gtx 1080.
Click to expand...

I would love to see this. This is exactly what I'm expecting from Vega. It would be nice to have more, but this is kind of AMD's niche in the market.
 
I have no idea what those pictures are showing, unless there's some weird fancy new packaging for Vega.
 
Shintai said:
AOTS, AMD Demo, nothing shows anywhere near 1080TI performance but rather 1080.

If you only look at power consumption, Pascal got a 2x advantage over Polaris. The RX480 pretty much use the same power as a GTX1080. If you want a more equal one you can use the 1070 vs 480 since it removes the GDDR5X part as well as cutdown. Then Pascal uses 18W less and performs 41% better. or 59% perf/watt at 1440p.

Remember when AOTS for Polaris 10 and Ryzen got shown and the fanboys was in complete denial?

HBM isn't buying AMD 30W at all. I know HBM is another mega hype and buzzword. GDDR5X and soon GDDR6 killed any power advantage for now that HBM had and its down to ECC being its last main advantage. And how did HBM do for Fiji? That's right, a 384bit GDDR5(Not X) part using Maxwell with 50% more memory had no issues there.

View attachment 19407
perfrel_2560_1440.png

perfwatt_2560_1440.png
Click to expand...

We get it, you hate AMD no matter what. Ryzen is good and give it time, it will get better. Fiji and Polaris is disappointing but I can't take your word for this stuff, you are going over the top.
 
All guesses in the end and someone might guess right and think they somehow knew what was coming fooling themselves in the end. Couple of months more is all.
 
You must log in or register to reply here.
Back
Top