[EDIT]hmmm.....whats going on at AMD? (and history of 3Dfx Rampage)

[SteelCrysis]

R9 480 reportedly performs close to R9 390X: http://www.techpowerup.com/222156/amd-ra...to-r9-390x

[SickBeast]

I will believe it when I see it when it comes to AMD hardware. Thanks for sharing though.

[ocre]

It's my prediction, Polaris 10 will not be anywhere near as fast as a 980ti. The big polaris 10 will be more like 390-390x.

I am not sure if it will even match a 390x. The will try to sell it on power efficiency and price. But idk. Nvidia will have smaller pascal chips and I fully expect that pascal could be ultra efficient as well. Especially the small chips in the lower end of the spectrum, where I think polaris 10 will be.

We really might see the gap grow this generation, with amd starting to be left behind.

Perhaps I am pessimistic. I am, really really feeling pessimistic on this. But, the clock speed is not looking very high from what we have seen when it comes to polaris. On top of the specs, the most recent specs. This is a gcn tweak, I don't see a major per core improvement. There should be some ipc improvement, I am hopeful.
But it is gcn, not a new architecture...not a major shift. It's hard for me to think that the Performance per core will change significantly. It could go up, it should actually go up. But it would have to be a major major improvement. Look at the specs, it's performance based on specs is 385ish to r9 290 at best. The improvements to GCN, I would put it at best 390x range but the clock speed disadvantages make that so hard to believe. Which may have 290x for best case performance.

And this may sound low but I think the lower the bar the better on polaris 10. Just think, it's will be much better if the chip exceeds a lower expectation rather than fall way short of a high one.

We will see where it stands. But I feel like amd may be holding back this one. Cause they know how the play this card...it's really gonna be critical . I don't think they are confident at all.
There is a lot of speculation on my part. A whole whole lot. And, I honestly wouldn't mind to be 100% wrong here. But, just look at what they have saide, the charts we have seen. They are not promoting or branding about performance. This is unique. AMD has a history of promoting performance ahead of time, in their charts. Even then, those estimates seldom aligned when the new architecture launched. This time, it's been Performance per watt and banging thus bring the cost down drum. This is the stage they are trying to set and have been testing around with. It is very telling all by itself.

[SteelCrysis]

Looks like Zen is going to be using 8 cores again: http://www.techpowerup.com/222231/amd-su...-initially

[RolloTheGreat]

Looks like Zen is going to be using 8 cores again: http://www.techpowerup.com/222231/amd-su...-initially

Moar corez! to sit idle while all the barely threaded apps run on 1-4.

[SickBeast]

Actually it looks like AMD is using the same logic I have been saying Intel should use. They are putting more CPU cores and removing the integrated graphics.

Rollo there are tons of games and programs that benefit from more than 4 cores. The CPUs in the Xbox One and the PS4 both have 8 cores. We are going to see more and more games that can use a lot of cores.

[ocre]

Any performance leaks?

[SteelCrysis]

Actually it looks like AMD is using the same logic I have been saying Intel should use.  They are putting more CPU cores and removing the integrated graphics.

Rollo there are tons of games and programs that benefit from more than 4 cores.  The CPUs in the Xbox One and the PS4 both have 8 cores.  We are going to see more and more games that can use a lot of cores.

That would have been believable 2 years ago. It's not believable now.

[ocre]

Well...

Either this site reads our forum and ran with it or there is a very sad launch coming soon.

http://www.fudzilla.com/news/graphics/40...erformance

It's almost exactly as I was projecting. Sad sad sad.
Maybe, somehow this will all be a false alarm. But reading the post and comments, it seems amd fanatic advocates are out trying to spin this one big time. Claiming 390 performance for less watts is exactly what everyone in the pc market has always wanted. This is ironic....I remember reading thread after thread, pro amd force screaming and yelling how 100 extra watts is meaningless. It barely increases your electric bill, like long drawn out post with formulas and hypothetical usage based on some of the lowest kw per hr rates in the world. But now...here we are seeing them preach that 390 performance for leww watts is the 2nd comming... ..

News flash!!!!
We have had 390 performance for a fraction of the power consumption for years now. It was found in the gm204. Seriously, we had the performance for around 150 watts. If AMD best it at all, by how much realisticly. I am thinking 125watts is too much to ask. But even it is, your looking at 25, 35 watt improvement over a card that will perform in the gm204 performance range. This is incredible...for how freaking sad it's looking.

Geez, thus can't be real....can it?

[SteelCrysis]

Rumor says Polaris launches late this month: http://www.techpowerup.com/222347/amd-to...y-late-may

[BoFox]

INSTGR over at TPU commenting in the above link says

"Please don't make my Fury irrelevant....Please don't make my Fury irrelevant...:("

Geez, AMD shills are a joke.

When I first heard of AMD launching HD 3870 at such tiny size, even though it was the first card with GDDR4 RAM, I was NOT excited at all - rather quite the opposite of excited.

It's the same thing all over again now. Tiny overpriced chip. I sold my Radeon R9 290 for maybe $300 (can't remember exactly) a couple years ago, and this will probably not be any faster or any cheaper, hardly. Teh suckage at its finest.

[SteelCrysis]

AMD's AIB partners are frustrated with AMD: http://www.techpowerup.com/222398/radeon...ted-at-amd

[ocre]

So.....
Anyone else notice how low the clock speed have been on all the Polaris cards we have heard of?

I have been drawn to this and know it just didn't seem right....well, at the very least it didn't look pretty to me.
Knowing that it is GCN and the consoles are updating to it, it just doesn't seem logical there will be any major gcn changes. Polaris most likely will be very similar to gcn chips of the past, I don't expect much of an ipc improvement.

Seeing the clock speed and the core count, it seems likely polaris 10 will be pretty slow. If they don't clock the chip running at least 1000mhz, then it probably will be about as fast as hawaii. The original 290(x).

I have been stuck on this clock speed for awhile. Why have we seen these clocks so freaking low?

Well....
A little gossip has occured and it really must be looked at so you guys can draw your own conclusion.

I expect this thread will be locked and possibly deleted but I will post the source as well.

[SteelCrysis]

What are you talking about? This thread hasn't done anything wrong.

[ocre]

http://forums.anandtech.com/showthread.php?t=2472724

And the info comes from here
http://www.overclock.net/t/1599305/vc-fi...marks/1290

I think it really seems logical on many levels.
For starters, the low clocks we have seen..that absurdity.

Then the fact that we know there will be no polaris cards on the market when amd has their even later this month. There are no cards in the channels, we would have seen a lot of leaks from vendors if there were. Boxes and card pics, at the very least.
Why and how are there no cards yet when amd supposedly demoed polaris last yr. Claimed to be a real deal chip..they showed it off like they were well on their way with polaris....yet still it's not around

There are signs that something isn't right....signs besides the low clock speed that has intrigued me.

Discuss

[ocre]

That was supposed to be part of the post above but in my phone and prematurely posted. So it is split

[SteelCrysis]

Oh, OK.

[RolloTheGreat]

So.....
Anyone else notice how low the clock speed have been on all the Polaris cards we have heard of?

I have been drawn to this and know it just didn't seem right....well, at the very least it didn't look pretty to me.
Knowing that it is GCN and the consoles are updating to it, it just doesn't seem logical there will be any major gcn changes.  Polaris most likely will be very similar to gcn chips of the past, I don't expect much of an ipc improvement.

Seeing the clock speed and the core count, it seems likely polaris 10 will be pretty slow.  If they don't clock the chip running at least 1000mhz, then it probably will be about as fast as hawaii. The original 290(x).

I have been stuck on this clock speed for awhile.  Why have we seen these clocks so freaking low?

Well....
A little gossip has occured and it really must be looked at so you guys can draw your own conclusion.  

I expect this thread will be locked and possibly deleted but I will post the source as well.

That would be a huge bump for consoles though and make them legit 1080p boxes.

Consoles are probably 90% of AMDs GPU business these days, so I get why they would focus on that first.

[ocre]

That thread on ATF was closed pretty much immediately.



The only reason I was reading their forum was for a good laugh. The amd red maggots have been in full force, non stop bashing of pascal. It's just one post after another after another.

But they are not restricted, it's actually pretty bad on almost every site these days. Somehow, +80% of posters on any random tech site are anti nvidia. Statistically improbable yet it's generally the case these days. Going by the echoes, I believe it's mostly the same people who have endless time to shill.

It's so funny now though, seeing that amd has no new cards at all. Yet they have a million bad things to say about the gp104. AMD has given us ntohi6at this point yet they still think bashing nvidia and pascal is a winning strategy.

See, there was some alternative before. Some amd card to try to promote along with the nvidia bashinh. This time......
Yeah, nothing.

See, I don't see how they think this will get them anywhere. Through all their efforts, there was little effect when it came to persuading buyers to amd. Nvidia gained marketshare and gained it. But this time, there is no amd cards at all. Nothing there, nothing to try to get potential gp104 buyers to buy instead. But, man....they sure are putting a lot of energy out these days.

When you can't compete, obstruct.

But honestly, there just aren't many buyers roaming tech forums anymore. They are just a few old guys left in what has become a toxic waste land. It's mostly the same people shilling with multiple accounts and then a few old timers who stuck with forums from back in the day.

[RolloTheGreat]

 

But honestly, there just aren't many buyers roaming tech forums anymore.  They are just a few old guys left in what has become a toxic waste land.  It's mostly the same people shilling with multiple accounts and then a few old timers who stuck with forums from back in the day.

When Apoppin started this site X years ago, I told him the same thing.

He gave me some crazy claptrap about how it was going to be different here, because it's "metaphysical" and he was going to build a 3d forum. (whatever the hell that meant)

Well here we are X years later, and what I said then is just as true when you say it today.

Tech forums are now infomercial tv, as are tech reviews.

[SteelCrysis]

GTX 970 Mini doesn't throttle when airflow is low, unlike the R9 Nano:

[SteelCrysis]

More info on Polaris: http://www.techpowerup.com/222450/more-p...s-revealed

[RolloTheGreat]

More info on Polaris: http://www.techpowerup.com/222450/more-p...s-revealed

It might just be me, but I'm not sure a die shrunk Hawaii is what the market has been waiting for.

(except ATVF posters who are mostly waiting for AMD, according to their poll)

It seems odd that NVIDIA now has 80% of the discrete market and the poster on ATVF will be avoiding Pascal at a 2:1 ratio. Very mysterious......

[SickBeast]

More info on Polaris: http://www.techpowerup.com/222450/more-p...s-revealed

It might just be me, but I'm not sure a die shrunk Hawaii is what the market has been waiting for.

(except ATVF posters who are mostly waiting for AMD, according to their poll)

It seems odd that NVIDIA now has 80% of the discrete market and the poster on ATVF will be avoiding Pascal at a 2:1 ratio. Very mysterious......

The only bright spot for AMD right now is that one rumor that said that Polaris 10 "Ellesmere" is as fast as a GTX 980 Ti.  If it really is that fast and it's priced well, then a card like that will sell like hotcakes for AMD and it will be very successful for them.  Most of the rumors, however, are saying that it's only as fast as a 390/390x which is disappointing IMO.  Most serious gamers already have a GPU in that ballpark.  We want something faster.  It's not exciting to have the performance of a flagship from 2 generations ago.  I guess it will be nice in the sub $200 segment but it's not nearly as exciting as Pascal IMO.  Vega is too far off and I would be shocked to see it compete with Big Pascal when it finally comes out as the GTX 1080 Ti.

[ocre]

It might just be me, but I'm not sure a die shrunk Hawaii is what the market has been waiting for.

(except ATVF posters who are mostly waiting for AMD, according to their poll)

It seems odd that NVIDIA now has 80% of the discrete market and the poster on ATVF will be avoiding Pascal at a 2:1 ratio. Very mysterious......

It looks more like a die shrink tonga. Sadly

[SickBeast]

It might just be me, but I'm not sure a die shrunk Hawaii is what the market has been waiting for.

(except ATVF posters who are mostly waiting for AMD, according to their poll)

It seems odd that NVIDIA now has 80% of the discrete market and the poster on ATVF will be avoiding Pascal at a 2:1 ratio. Very mysterious......

It looks more like a die shrink tonga.  Sadly

Yes, albeit an overclocked Tonga. Here is a link:

http://wccftech.com/amd-r9-480x-470x-spe...-revealed/

If AMD is only able to hit 1350mhz on their cores while nVidia can hit 2.1ghz+ they are in big trouble. The pricing of the Polaris cards makes no sense either. Spending $70 extra on a GTX 1070 is a no brainer. This looks like it's going to be a terrible launch for AMD. The sense I'm getting is that the main point of these new GPUs is to create a new APU for the PS4.5 and the new 4k Xbox.

[SteelCrysis]

AMD is sending out invitations to the Polaris 10 launch at Computex: http://videocardz.com/59947/amd-to-launc...t-computex

[SickBeast]

That will be really great news if AMD does indeed launch Polaris at Computex.

[RolloTheGreat]

More info on Polaris: http://www.techpowerup.com/222450/more-p...s-revealed

It might just be me, but I'm not sure a die shrunk Hawaii is what the market has been waiting for.

(except ATVF posters who are mostly waiting for AMD, according to their poll)

It seems odd that NVIDIA now has 80% of the discrete market and the poster on ATVF will be avoiding Pascal at a 2:1 ratio. Very mysterious......

The only bright spot for AMD right now is that one rumor that said that Polaris 10 "Ellesmere" is as fast as a GTX 980 Ti.  If it really is that fast and it's priced well, then a card like that will sell like hotcakes for AMD and it will be very successful for them.  Most of the rumors, however, are saying that it's only as fast as a 390/390x which is disappointing IMO.  Most serious gamers already have a GPU in that ballpark.  We want something faster.  It's not exciting to have the performance of a flagship from 2 generations ago.  I guess it will be nice in the sub $200 segment but it's not nearly as exciting as Pascal IMO.  Vega is too far off and I would be shocked to see it compete with Big Pascal when it finally comes out as the GTX 1080 Ti.

Depends who you mean. For AMD stockholders and employees, locking down the next two consoles is probably a big deal as they survive to fight another day.

For guys like me with a custom cooled 290 in my 1600p pc and a 980Ti (at least until I sold it for $520 a few weeks ago) in my 4K pc, 290X performance at any price isn't going to even elicit a yawn.

The spambots on the forums will talk about how AMD is bringing all the performance you need at a great price and talk up these parts like they are some "new" technical marvel, but the true mission is indeed consoles as that is AMDs remaining stronghold market.

The future really was Fusion.

[BoFox]

Now that there's a rumor that AMD might launch Vega way ahead of schedule (originally early 2017, now October?)..  it's probably getting Nvidia ready and prepared for this.  Nvidia's plan:  Launch Titan Pascal and possibly also GTX 1080 Ti (based on GP100) very shortly after AMD launches Vega.  At least NV would paper-launch GP100 without any trouble at all, since GP100 is already in production (quoted from Tomshardware:

The Tesla P100 is in volume production today and will ship “soon." Jen-Hsun Huang said that really means it will show up in the cloud first, and ship via OEMs in Q1 2017.

Traditionally, the flagship Titan / GTX xx80 Ti cards are clocked maybe 15% higher than Tesla variants.  I'd expect it to be clocked almost just as high as GP104 GTX 1080 - at least 1500Mhz core with 1600MHz boost for sure.  HBM2 memory might be clocked way higher than 1.4GHz as it's being refined, but I'd try to be conservative with that one - probably not more than 1.6GHz compared to 1.4GHz on the P100.

Notice how the GP100 actually has MORE than 2x as many trannies as GTX 1080?

Only 7200 million trannies, whereas GP100 has 15.3 billion.
http://www.pcgamer.com/nvidia-pascal-p10...deep-dive/

What's interesting is that GP100 can do twice as many half-precision (FP16) textures as single-precision FP32 (unlike previous generations).  Would that really help with games if they have simpler operations that could use half-precision?

The GP100 based P100 has 3584 CUDA Cores. The chip itself has a turbo clock of 1480 Mhz. The total peak single precision performance of the P100 GPU therefore is 10.6 TFLOPs (3584 CUDA Cores x 2 Operations Per Second x 1480 Megahertz). Since we are talking about the exact same Pascal architecture here, we can directly compare performance of the GP104 based GTX 1080 to the P100.

The GTX 1080 on the other hand only has 2560 CUDA Cores. However, the card can reach clock rates of 2.1 Ghz conservatively and 2.5 Ghz on a liquid cooled setup. At a clock rate of 2.1 Ghz, the GTX 1080 is actually 10.7 TFLOPs of single precision performance! At a clock rate of 2.5 Ghz, the GTX 1080 will blow past the P100 GPU giving performance levels of 12.8 TFLOPs! That is an absolutely insane figure and roughly 20% faster than the P100. To put this into perspective that is almost twice the theoretical single precision performance of the GTX 980 Ti.

http://wccftech.com/geforce-gtx-1080-blo...d-edition/

2.5GHz!?!???   This is insane, and unbelievable!

[SickBeast]

OMGWTFBBQ!

Thanks BoFox! That's amazing news if it's even remotely true!

[SteelCrysis]

http://videocardz.com/59947/amd-to-launc...t-computex
This story has been updated, the June 1 event will take place at 10 AM local time.

[ocre]

Bofox, it's gonna take vega to compete with the gp104. Polaris is significantly weak compared to the 1080. I imagine that vega will be more in the 1080 range, the gp100 will be in another league.

It seems logical that amd could notch above the 1080, they will know exactly where it stands and it would be really pathetic if they can't beat it.

Vega looms to be about the same core count as fury x. If they can't get the clock speed up, it would be an epic failure. I don't think they will reach anywhere near the gp104 boost speeds but it doesn't have to on order to win. I figure 1300mhz would be enough.

Ultimately, overclocked 1080s may end up faster. And by that time, nvidia might have have better results with the new node. They may could do a 1085 and be done with it.

I don't think vega will be massively more powerful than the gp104. Sadly, amd is slipping way behind

[SickBeast]

Even if Polaris is only as fast as a GTX 980 Ti I would love that and I would buy one at the right price. I doubt it's that fast though. It's the nVidia cards that look like they are going to be real monsters.

[ocre]

There are perhaps some indicators.

1st- Amd comments. The specific words they have chosen when talking up Polaris are interesting. If these chops were capable of 980ti performance at 120 watts, I just don't think Amd would have a much different tone. Amd would not hold something like that back. They were demoing and showing off early early silicon last year, when the chip was no where close to ready for retail. Yet Amd is trying so hard for relevance they had staged some press showing with a supposed Polaris vs a low power maxwell solution. This wouldn't be bad except for the fact that Polaris was no where near launching. It's bizarre they would pull such a stunt at that time. It screams of desperation. Showing off preproduction chip on a new node vs one that was created long ago, of course they should be able to beat an old chip on an old node. But showing it off when your not even launching for at leat another 6 months....
See, if it was beating or matching the 980ti, surely this would be much more to brag about.

2-
Fury core 2 duo.
If Polaris is as fast as the 980ti, then why would they have rushed out the fury core 2 duo? The timing says a lot. The fury duo came just ahead of the gp104 and the 1080. Amd has already spoke about the gp104 saying that everything they know about it points to it being an a high end card, suggesting that Polaris was in a different bracket. Amd knew pascal was coming and it was high end. They have contacts, moles, and even share Aib partners. Nvidia can keep exact performance close but Amd wasn't clueless. I belive the fury core 2 duo was AMDS play, timed directly for the gp104. Try to hold the line somewhat until vega.
See, think back and it looks like a move from their playbook. Maxwell launch for example, gm204 was walking all over the 290x but you would often find post on forums and tech sites denouncing maxwell and claiming the 295X2 as the fastest card out. It didn't matter that it was a dual chip or the cf issues, just having something to come back with...just to keep them part of the discussion.

When maxwell launched, AMD had no response for a very long time. Eventually, the consoles contracts helped, GCN performance improved as newer games came out, and then finally the 300 series/fury. But for the time in between, there was this period of the 295x2.

Fast forward and it seems to me Amd has a similar play they are pulling from the play book.

I don't see why Amd would have such a solution, such a large chip with hbm. If Polaris 10 was as fast as the 980ti, such a tiny chip with a small buss...two of those on a board. Just think about it, the fury is 4gb ram. The duo will be restricted. A Polaris 10 could be 8 or more.

I think A lot points to Polaris not being a huge performer. I would say 290x ish.

[SteelCrysis]

Fury X running out of VRAM at 1080p in Rise of the Tomb Raider benchmark with 1x SMAA, but this is due to the Rise of the Tomb Raider benchmark using over 9 GB of VRAM if given the chance:

[SteelCrysis]

Next APUs and Polaris "updates" are coming to Computex: http://techreport.com/news/30160/bristol...putex-show

[SickBeast]

Notice how they said "update" and not "launch". AMD is really sneaky.

[RolloTheGreat]

Notice how they said "update" and not "launch".  AMD is really sneaky.

AMD can't do anything but what they are doing.

When you've got the parts, you let them do the talking.

When you've got the vaporware, you spread the FUD and hints.

[SteelCrysis]

Zen chip sighted with 8 cores in 2 blocks of 4, and it's not from CharLIE: http://semiaccurate.com/2016/05/22/38688/