Nvidia thinks "Pascal is just unbeatable" and already decided not to move Volta ahead

[Taf the Ghost]

6 minutes ago, leadeater said:

It's the halo effect or anti-halo effect, if the top card is shit then the ones below it must be even if they mostly aren't.

Halo effect can matter, but it's much more how the "guy you ask for advice" is thinking. Nvidia already owns that space, so that's not a huge issue for AMD at the moment.

5 minutes ago, snortingfrogs said:

Perhaps.

But the RX lineup performs bad, are expensive and is almost 1½ year late to the party, also you could use them to heat up a Siberian Mansion during winter.

RX performs fine. Vega 56 looks to be generally above the 1070 in most games before FineWine kicks in. If you do any GPU-based render stuff, the RX Vega line is far & away better than Nvidia's offering.

 

Plus, as I said, people do have 1440p FreeSync monitors already, so there's a very real market for the Fury/Fury X users that want to upgrade.

[cj09beira]

6 minutes ago, Trixanity said:

Source?

Very true but AMD shouldn't have to clock the GPU past its efficiency curve to keep pace. That's the problem. They even did work to add pipeline stages to make it clock higher but it almost seems like it created new problems so it wasn't worth the trouble.

Speaking of too much voltage. It seems they have a problem with validating their cards to run with optimal voltage. I think it would be wise (if possible) to have the Radeon Software run a 'calibration' of the card which will find the sweet spot for voltage while maintaining stability instead of running extra high voltage through it to make sure even the worst chip can hit the desired clock speed.

i think nvidea does something similar, i would do it on silicon (i have though about it before), it wouldn't be a easy task but when done would make their cards a lot more efficient out of the box, 

[i_build_nanosuits]

4 minutes ago, PCGuy_5960 said:

Makes sense to me, Vega 64 performs like a 1080 while drawing as much power as a 1080 SLI. Vega 56 is more impressive, as it beats the 1070 while costing the same, but all Nvidia has to do to remain competitive is drop the 1070's price to $350

nvidia has some wiggle room there, 350$ for a ''X70'' level card is still pretty damn expensive if you look at how much was a GTX 570, 670, 770, 970...these cards usually cost around 299$ a few months after launch...GTX 1070 has been out for 15 months already and it's still the same price as when it launched.

[samcool55]

Oh well, if we need to wait for AMD to bring out a competitor we'll have a long way to go.

They should have enough money now tho, that's the only good thing i can think about the mining sh*¨tness that has been going on.

[Trixanity]

Just now, ravenshrike said:

Navi is going to be Vega with improvements and maybe a MCM design shrunk to 7nm. According to GloFo their 7nm design can either boost performance by 40% or reduce power consumption by 60%. It's possible that they could go for the 40% performance boost, but I would think they would go for power efficiency.

I don't believe we've heard anything of the sort about Navi. We know it'll be 7nm, we know it'll have "next gen memory", we know it will be "scalable" which probably indicates MCM but other than that it sounds like speculation to me. Unless MCM will be their saving grace, it doesn't look good if Navi is more of the same; it clearly isn't working for anything but being a jack of all trades with relatively poor efficiency. Navi is supposedly an RTG product versus Polaris and Vega already being in the pipeline at the time - whatever that means.

[cj09beira]

2 minutes ago, Trixanity said:

I don't believe we've heard anything of the sort about Navi. We know it'll be 7nm, we know it'll have "next gen memory", we know it will be "scalable" which probably indicates MCM but other than that it sounds like speculation to me. Unless MCM will be their saving grace, it doesn't look good if Navi is more of the same; it clearly isn't working for anything but being a jack of all trades with relatively poor efficiency. Navi is supposedly an RTG product versus Polaris and Vega already being in the pipeline at the time - whatever that means.

most of the efficiency problem is controlling the voltages given to the gpu better, right now amd ships their gpus with way to much voltage which ruins efficiency, they having to overvolt hbm2 also doesn't help 

[Taf the Ghost]

@snortingfrogs

 

There was one other little thing about Vega that LTT came across that I'm quite curious to see tested more.

 

There's quite a lot of room between the 1080 and the Vega 64 in CS:GO. This matters a lot. I also want to see Dota 2 testing, especially against the 1080 Ti. If Vega 64 = 1080 Ti in two of the most popular games available, suddenly Vega's value completely changes.

 

This needs further testing.

[2FA]

8 minutes ago, ravenshrike said:

Navi is going to be Vega with improvements and maybe a MCM design shrunk to 7nm. According to GloFo their 7nm design can either boost performance by 40% or reduce power consumption by 60%. It's possible that they could go for the 40% performance boost, but I would think they would go for power efficiency.

source?

[Trixanity]

Just now, cj09beira said:

most of the efficiency problem is controlling the voltages given to the gpu better, right now amd ships their gpus with way to much voltage which ruins efficiency, they having to overvolt hbm2 also doesn't help 

Do we know if proper voltage brings the power consumption in line with Nvidia? From what I've seen undervolting still gives Nvidia a considerable advantage but that still leaves the performance gap as in no 1080 Ti competitor and trading blows with 1080 a year after its release.

[ravenshrike]

1 minute ago, DeadEyePsycho said:

source?

The fact that Raja Koduri isn't insane enough to try and design a brand new architecture for a brand new process that's nearing the limit of silicon capability?

[snortingfrogs]

2 minutes ago, Taf the Ghost said:

@snortingfrogs

 

There was one other little thing about Vega that LTT came across that I'm quite curious to see tested more.

 

There's quite a lot of room between the 1080 and the Vega 64 in CS:GO. This matters a lot. I also want to see Dota 2 testing, especially against the 1080 Ti. If Vega 64 = 1080 Ti in two of the most popular games available, suddenly Vega's value completely changes.

 

This needs further testing.

People playing CS:GO and/or Dota 2 are not going to fork up $800 for a GPU and $100+ for a new PSU to be able to use said GPU.

[The Benjamins]

10 minutes ago, snortingfrogs said:

People playing CS:GO and/or Dota 2 are not going to fork up $800 for a GPU and $100+ for a new PSU to be able to use said GPU.

Why not? I am willing to bet most heavy CS:GO (or Dota 2) Players also play other games like AAA titles that would also use the GPU power.

[leadeater]

15 minutes ago, Trixanity said:

Do we know if proper voltage brings the power consumption in line with Nvidia? From what I've seen undervolting still gives Nvidia a considerable advantage but that still leaves the performance gap as in no 1080 Ti competitor and trading blows with 1080 a year after its release.

Look at Gamers Nexus for that type of information, they only have a 56 review last time I check but they did the undervolt testing and got the 56 within 20W-30W of Pascal.

[cj09beira]

18 minutes ago, Trixanity said:

Do we know if proper voltage brings the power consumption in line with Nvidia? From what I've seen undervolting still gives Nvidia a considerable advantage but that still leaves the performance gap as in no 1080 Ti competitor and trading blows with 1080 a year after its release.

just undervolting would help a lot is what i am saying but there are things they can do in silicon to reduce it even more, the thing is that with amd you are buying better compute perf and even if you aren't using it it will still consume its power, games more and more use compute for some effects but amd needs that more of them use it and that dx11 dies, as they are in disadvantage on dx11 thanks to nvidea being able to force multitreading on (amd cant)

[snortingfrogs]

10 minutes ago, The Benjamins said:

Why not? I am willing to bet most heavy CS:GO (or Dota 2) Players also play other games like AAA titles that would also use the GPU power.

 

Most gamers who play these games are "Low-Budget Gamers", of course there are gamers who play CS:GO, BF1, Fallout and so on, but from my experience the wast majority of CS:GO/Dota 2 players got an Pentium/i3 CPU and 960/1050 card and that's well good enough for them.

[The Benjamins]

2 minutes ago, snortingfrogs said:

 

Most gamers who play these games are "Low-Budget Gamers", of course there are gamers who play CS:GO, BF1, Fallout and so on, but from my experience the wast majority of CS:GO/Dota 2 players got an Pentium/i3 CPU and 960/1050 card and that's well good enough for them.

Bad word use on my part but I meant, hardcore CS:GO and dota 2 players. the people that care about 144hz gaming. would also play other games.

[linustouchtips]

noooo vega why you got to suck so much oh well at least ill have the fastest card for a little while

[TacoSenpai]

Of course they say this and Nvidia will release a new Titan XP2.5 card only to be followed up by a high end gaming card that beats/keeps up with it...

 

Deja Vu.

 

No but I really want AMD Radeon to get their shit together on the GPU front but I think they are spread thin with trying to improve CPU architectures like Ryzen and make GPUs at the same time that said Polaris seemed pretty good. 

[Taf the Ghost]

18 minutes ago, snortingfrogs said:

 

Most gamers who play these games are "Low-Budget Gamers", of course there are gamers who play CS:GO, BF1, Fallout and so on, but from my experience the wast majority of CS:GO/Dota 2 players got an Pentium/i3 CPU and 960/1050 card and that's well good enough for them.

You literally don't get that both Dota 2 & CS:GO are billion dollar Revenue titles... per year. Each. Huge marketing potential there. If they perform at 1080 Ti levels in *just* those two titles, suddenly a ~$500 USD card performing like a ~$800 USD card changes the entire proposition for the uArch. This matters, quite a lot. Which is why it needs a LOT more testing.

[TVwazhere]

54 minutes ago, i_build_nanosuits said:

970...these cards usually cost around 299$ a few months after launch

The 970 was still $380 three months after it launched. I know because thats when I bought it! (Dec 2014)

56 minutes ago, i_build_nanosuits said:

GTX 1070 has been out for 15 months already and it's still the same price as when it launched.

This is also partially due to that stupid crytocurrency mining craze when hey ran out of AMD cards.

 

[AlTech]

1 hour ago, snortingfrogs said:

He is not wrong, RX Vega is pretty damn shit so why would people buy it?

VEGA 56 is a good performer. It's just a matter of price in my opinion.

 

VEGA 64 is problematic for AMD since it doesn't appear to perform significantly faster than the VEGA 64 for the $100 price difference.

[AlTech]

50 minutes ago, snortingfrogs said:

People playing CS:GO and/or Dota 2 are not going to fork up $800 for a GPU and $100+ for a new PSU to be able to use said GPU.

As other have said, why not?

 

I play CS:GO occasionally. Does that mean I should use an RX 560 or a GTX 1050? Not necessarily.

 

I want incredibly high performance on most games I play. If i happen to play CS:GO in addition to demanding AAA titles then I'll use a high end GPU for both.

[Flavio hc 16]

1 hour ago, ravenshrike said:

Navi is going to be Vega with improvements and maybe a MCM design shrunk to 7nm. According to GloFo their 7nm design can either boost performance by 40% or reduce power consumption by 60%. It's possible that they could go for the 40% performance boost, but I would think they would go for power efficiency.

They will go for power effinciency, but for a good reason: NAVI as all the rumors had said and also some hints from amd slides is going to be the ryzen of gpu: multiple small gpus "glued" with infinity fabric to work and be recognised  from the sistem as only one gpu. If you take the vega 56,, shrink it to the 7nm process and let it work at his peak efficiency  you are looking at a gpu that uses 90-100w ( from the data of power savings thanks to the new manufacturing  process) and is someting like 2-5% slower than an rx56, now stick 3 of them on a single die and you have a 270-300w gpu that is 50-60% more powerfull than a 1080ti, so that should be on par with the hypotetical gtx 1180ti/2080ti, while having way better die yellids  and scalability ( even though  under heavy oc it could draw someting retarded like 750w alone).

 

I hope that you get my point, and i also hope that amd is going to do someting like this and i'm not daydreaming, because if I  am, and gpu department  is screwed

[Misanthrope]

1 hour ago, porina said:

AMD's main weapon has never been technical, it is in pricing. They knew they had to aim competitively in pricing. 

At least it is working out better for them on the CPU side... 

It is now and it has been for a while, but no. Some of their past processors were actually quite better than contemporary intel offerings for example. Goes for some generations under ATI as well.

[Misanthrope]

1 hour ago, leadeater said:

It's the halo effect or anti-halo effect, if the top card is shit then the ones below it must be even if they mostly aren't.

Well it's regrettable, but it works: AMD has had better midrange offerings for a while now: Polaris was at a better price point when launched, the 380x was also fairly solid and the 7950/280x were also far better for what they cost, etc.

 

Yet even if this are the majority of the cards they only seem to claw back so much market away from Nvidia: people fall for the prestige of saying "They have the fastest card" even if realistically most of them don't buy that card and AMD cards are either competitive or flat-out better in some games.