Quantcast
(UPDATE): AMD Unveil RX 5000 (Navi) with their new architecture design, Interview with Lisa SU, Ray Tracing detail on another SKU's variant, PCIe 4.0 help fast loading on PS5

Forums - Gaming Discussion - (UPDATE): AMD Unveil RX 5000 (Navi) with their new architecture design, Interview with Lisa SU, Ray Tracing detail on another SKU's variant, PCIe 4.0 help fast loading on PS5

Console gamer (Playstation fans especially ) how will you response

Yes 3 14.29%
 
No 4 19.05%
 
It's happening 3 14.29%
 
i dont know how to response 1 4.76%
 
i am not interested in powah 0 0.00%
 
Yes bring it on 3 14.29%
 
Playstation 5 will be a Super Beast 7 33.33%
 
Total:21
haxxiy said:

Strange Brigade demo? It's a known AMD-friendly title. That's kind of shady. This sounds like GCN rebranded with some new features. Like Maxwell -> Pascal. I also wonder about the power consumption. If this is anywhere close to 200W, that's... very bad.

At least Ryzen is looking pretty good.

And there is plenty of NVidia friendly games, some with paid for better performance.



duduspace11 "Well, since we are estimating costs, Pokemon Red/Blue did cost Nintendo about $50m to make back in 1996"

http://gamrconnect.vgchartz.com/post.php?id=8808363

Mr Puggsly: "Hehe, I said good profit. You said big profit. Frankly, not losing money is what I meant by good. Don't get hung up on semantics"

http://gamrconnect.vgchartz.com/post.php?id=9008994

Around the Network

Guys some terrible news, I think navi sucks, look at this tweet:

It's 50% effiency gain with 7nm process and architectural improvements, just doing simple math we are absolutely not getting any crazy Teraflops numbers. The card AMD demoed is at best a 225watt gpu, probably 250watt and has about 10 Teraflops based on the 25% ipc gain. A $400 dollar console (PS5) need a gpu that pulls 120watt and a premium $500 console (Xbox anaconda) about 150-160watt.

Both console will almost certainly be below 10 Teraflops, based on the gonzalo codenamn showing PS5 having a 1,8ghz navi gpu it will likely just have 32CU's with 2-4 disabled CU's giving 6,4-6,9TF. Xbox anaconda will probably have 40CU's but 4 disable for yield improvements so 36CU's clocked at 1,9ghz is my guess and will result in a 8,7TF gpu.

Navi went from super-good to good to terrible. Look at this quote from Lisu Su at computex:

"And then, when you put that together, both the architecture – the design capability – as well as the process technology, we're seeing 1.5x or higher performance per watt capability on the new Navi products"

She mentions process technology!!!



"Donald Trump is the greatest president that god has ever created" - Trumpstyle

6x master league achiever in starcraft2

Beaten Sigrun on God of war mode

Beaten DOOM ultra-nightmare with NO endless ammo-rune, 2x super shotgun and no decoys on ps4 pro.

1-0 against Grubby in Wc3 frozen throne ladder!!

Trumpstyle said:

Both console will almost certainly be below 10 Teraflops, based on the gonzalo codenamn showing PS5 having a 1,8ghz navi gpu it will likely just have 32CU's with 2-4 disabled CU's giving 6,4-6,9TF. Xbox anaconda will probably have 40CU's but 4 disable for yield improvements so 36CU's clocked at 1,9ghz is my guess and will result in a 8,7TF gpu.

Please stop posting more nonsense every 10 minutes.

The PS4Pro already has 36CUs. There is no way a next gen chip has fewer CUs.

Also your 1.8Ghz clock is way too high for consoles, unless we get monster cooling solutions. There has been an image of a navi board floating around for a few weeks now, it has two 8-pin connectors. That alone tells you all about the problems AMD STILL has with power consumption, even at 7nm.



drkohler said:
Trumpstyle said:

Both console will almost certainly be below 10 Teraflops, based on the gonzalo codenamn showing PS5 having a 1,8ghz navi gpu it will likely just have 32CU's with 2-4 disabled CU's giving 6,4-6,9TF. Xbox anaconda will probably have 40CU's but 4 disable for yield improvements so 36CU's clocked at 1,9ghz is my guess and will result in a 8,7TF gpu.

Please stop posting more nonsense every 10 minutes.

The PS4Pro already has 36CUs. There is no way a next gen chip has fewer CUs.

Also your 1.8Ghz clock is way too high for consoles, unless we get monster cooling solutions. There has been an image of a navi board floating around for a few weeks now, it has two 8-pin connectors. That alone tells you all about the problems AMD STILL has with power consumption, even at 7nm.

Dont reply or quote to him, he is  a stealth Sony hatter. 



Trumpstyle said:
shikamaru317 said:

If this thing is $400 at retail I will be highly amazed if PS5 uses it. PS4 essentially used a desktop GPU that cost $180 at retail at the time PS4 released. Sony would be losing money if they put a $400 at retail GPU into PS5 (probably $300 for them since they buy in bulk and get special deals), and I recall them saying that they don't want to sell hardware at a loss anymore. 

What happened to the rumor from about 2 month's ago that the 5800 would cost $280-320 with the 5700 coming in around $220-250 (right in the price range range for potential PS5 usage)? 

Sony doesn't pay retail prices, they pay wafer costs + margins to AMD, now we know why the gonzalo codename said it was a navi 10 lite. Amd and Sony has probably done some customizations to Navi 10 to reduce the die size, they need to get it down to about 200-210mm2 to fit 8core zen2 cpu in a 350mm2 apu/soc.

Not sure what rumors you talking about, I'm assuming adoredtv, they were fake dude. He got tricked, stuff like that can happen.

About gpu prices for PC, I expect them to come down big when Nvidia launches their 7nm gpu lineup.

I know they don't pay retail, but a $400 at retail GPU in PS5 still seems highly unlikely to me when PS4's GPU was much cheaper than that.

I think I saw that $280-320 pricing rumor for the highest end Navi chipset on Anandtech or another big site, but I could be wrong. 

Last edited by shikamaru317 - on 27 May 2019

Around the Network

As a reminder...



                                                                                                             

CGI-Quality said:

As a reminder...

If not to stir up trouble why would anyone come to a thread?

To engage in friendly conversation? Eww.



Straffaren666 said:

Unfortunately the information is pretty scarce at the moment, but according to WCCF the 64 CU limit is gone and according to a leak by KOMACHI there are 8 SEs in Navi 10, which would be very good news if true. That would mean we will see performance improvements in the geometry/rasterizer/zROP bound workloads, which becomes more prominent when scaling beyond 40 CUs (10CU/SE). Most likely the 64 ROP limit has been increased as well even though the midrange cards to be introduced in July probably won't feature more than 64 ROPs.

It will be interesting to see where the 25% IPC comes from and under what conditions they have measured it. Some will come from the improved cache and most likely widened data paths into the CUs as well as some new instructions, but it's unlikley that would yield a 25% IPC. Introducing separate integer units in the CUs similar to Turing could be one explanation but I'm sure AMD would have touted that if that was the case, so the IPC increase probably comes from several minor improvments.

All in all, at least some notable improvements from AMD, even though they still lag nVidia by quite a bit, and good news for the PS5, in particular the performance/watt improvements. Still no confirmation that Navi supports VRS, which when utilized properly can give a significant performance boost (I'm pretty sure PS5 will support it though, but that's pure speculation on my behalf). Hopefully AMD will be more informative at E3.

Finally, someone speaking English!

Navi is certainly a re-balanced uArch, the doubling of SE's over Vega is a very strong hint towards that... GCN was always very compute heavy.
It's similar to what AMD did when they took VLIW5 and made VLIW4, re-balanced the functional units for better overall utilization in games.

I think Variable Rate Shaders will be in, Microsoft is baking it into Direct X, so should be a uniform standard going forward with some luck.

haxxiy said:

Strange Brigade demo? It's a known AMD-friendly title. That's kind of shady. This sounds like GCN rebranded with some new features. Like Maxwell -> Pascal. I also wonder about the power consumption. If this is anywhere close to 200W, that's... very bad.

At least Ryzen is looking pretty good.

AMD is likely driving up the clockrates hard at the expense of power I think. I could be wrong.
But I would expect power consumption around the RX 480/580 levels, it's not super high.

vivster said:
I wonder how scalable this will be. Will it be possible to produce a higher end chip with this that's closer to the 2080? If not I see dark times for people who expect a 10+TFLOPS console.

The increase in real-world gaming performance will be substantially larger than the increase in TFLOPS.

Trumpstyle said:

I don't know the die size, Lisa Su said it was small on stage and I assume it's a midrange card putting it about 200mm2.

Anandtech has done some basic rough math and pegged it at almost 275mm2.



There will be some deviation from that of course, it's still a large die for 7nm, to put that into perspective... Vega 7 was 331mm2.

In time though, it will be an extremely cost effective chip to manufacture... But I doubt they will have a costing advantage over the 2070, which despite being 445mm2... It is built on an extremely high-yield, mature and cheap 16nm TSMC based process. AKA. 12nm.

Bofferbrauer2 said:

Polaris, when it came out as the RX 480, was pretty good - just not the high-end chip everybody was hoping for, but a mid-range chip. It did keep up with Geforce 970 in performance/watt and performance in games. It just couldn't keep up with the 980/980Ti/Titan, and Pascal destroyed it in terms of performance per watt one year later.

I think your vision of Polaris got skewed by Maxwell and Pascal, where NVidia leaped so far forward that AMD couldn't nearly keep up anymore, but kept rehashing the chip as they had nothing else apart from Vega to show off.

And while 25% IPC increase is not 25% increase in performance, it should come pretty close at the same clock speed and CU amount. Also, don't forget the 50% reduction in power draw at the same time.

So yeah, it will be like Polaris, a good mid-range card that can keep up with it's peers from NVidia at it's release - it's just not the high-end card to compete with them at the top.

I never expected Polaris to hit the high-end, so my expectations were more aligned to reality back then.

The RX 480 was a good chip, don't get me wrong... And it has aged well. But a high-end part it never was... I do own a Polaris GPU, so I am talking from an owners perspective as well.

Maxwell was just a solid part and nVidia refined it with Pascal.

The reduction in power is probably what Navi needs to bring to the table the most, AMD pushed up the clocks and voltages with the RX 580 and 590, so power consumption got a little out of control...

But it was the same with Vega 64, it's actually an extremely efficient GPU... At lower clocks and lower voltages, AMD pushed it super hard to try and keep pace with nVidia.

At the end of the day though, I found it extremely difficult to recommend AMD's GPU's for gaming towards the end of Pascals life because nVidia was simply the superior option in almost every regard, even sometimes price if you shopped around.
The RTX does lean things towards AMD a little, but not definitively... And certainly not in the high-end.

Trumpstyle said:

The tweet is probably correct, those people are experts, we probably looking at a Navi card with 275mm2 die size with very close performance to geforce 2070, so it's a bit worse than I expected ):

Anandtech can still get shit wrong (And have done so in the past), but it's probably a good ballpark number.
Either way, we will know definitely in the coming months.

I will probably be an early Navi adopter, I game at 1440P, so it should be the perfect GPU for my needs.

Trumpstyle said:

Sony doesn't pay retail prices, they pay wafer costs + margins to AMD, now we know why the gonzalo codename said it was a navi 10 lite. Amd and Sony has probably done some customizations to Navi 10 to reduce the die size, they need to get it down to about 200-210mm2 to fit 8core zen2 cpu in a 350mm2 apu/soc.

Not sure what rumors you talking about, I'm assuming adoredtv, they were fake dude. He got tricked, stuff like that can happen.

About gpu prices for PC, I expect them to come down big when Nvidia launches their 7nm gpu lineup.

Rumor has it that nVidia may be updating RTX with faster GDDR6 memory rather than drop their parts to 7nm... So that may make things more difficult for AMD and will probably allow nVidia to continue to charge their nVidia tax.

fatslob-:O said:

AMD aging better is probably just down to benchmarking new games transitioning to new gfx APIs ... 

Being moderately faster tham a 2070 is easy enough when Vega 64 matched it but the hardest part for AMD is going to be finding another tier of performance so that it can get closer to the 2080 ...

That is exactly it.

AMD still has a faster Vega 7 part they can throw into the fight as well with a full 64CU part...

Trumpstyle said:

Guys some terrible news, I think navi sucks, look at this tweet:

It's 50% effiency gain with 7nm process and architectural improvements, just doing simple math we are absolutely not getting any crazy Teraflops numbers. The card AMD demoed is at best a 225watt gpu, probably 250watt and has about 10 Teraflops based on the 25% ipc gain. A $400 dollar console (PS5) need a gpu that pulls 120watt and a premium $500 console (Xbox anaconda) about 150-160watt.

Both console will almost certainly be below 10 Teraflops, based on the gonzalo codenamn showing PS5 having a 1,8ghz navi gpu it will likely just have 32CU's with 2-4 disabled CU's giving 6,4-6,9TF. Xbox anaconda will probably have 40CU's but 4 disable for yield improvements so 36CU's clocked at 1,9ghz is my guess and will result in a 8,7TF gpu.

Navi went from super-good to good to terrible. Look at this quote from Lisu Su at computex:

"And then, when you put that together, both the architecture – the design capability – as well as the process technology, we're seeing 1.5x or higher performance per watt capability on the new Navi products"

She mentions process technology!!!

It's not going to "suck". - It's just not going to be some revolutionary improvement that some people are making it out to be.

It's the same with Zen 2... There is a jump in IPC, but the bulk of the performance improvements is actually thanks to an uptick in clockrates thanks to the  headroom that the smaller manufacturing process brings... And then AMD takes the IPC and Clockrate and pegs it as a "50%" improvement over the old part... Which whilst true, clock for clock the needle movement was actually smaller when comparing architecture against architecture, clock for clock.

drkohler said:
Trumpstyle said:

Both console will almost certainly be below 10 Teraflops, based on the gonzalo codenamn showing PS5 having a 1,8ghz navi gpu it will likely just have 32CU's with 2-4 disabled CU's giving 6,4-6,9TF. Xbox anaconda will probably have 40CU's but 4 disable for yield improvements so 36CU's clocked at 1,9ghz is my guess and will result in a 8,7TF gpu.

Please stop posting more nonsense every 10 minutes.

The PS4Pro already has 36CUs. There is no way a next gen chip has fewer CUs.

Also your 1.8Ghz clock is way too high for consoles, unless we get monster cooling solutions. There has been an image of a navi board floating around for a few weeks now, it has two 8-pin connectors. That alone tells you all about the problems AMD STILL has with power consumption, even at 7nm.

CU's and Clockrate play a role hand in hand. You can have 18 CU's and have more performance than 36 CU's.

I think about 40CU's is a good place to be with a high clockrate, 7nm does bring with it a slew of advantages in regards to die size and power consumption remember.

shikamaru317 said:

I know they don't pay retail, but a $400 at retail GPU in PS5 still seems highly unlikely to me when PS4's GPU was much cheaper than that.

I think I saw that $280-320 pricing rumor for the highest end Navi chipset on Anandtech or another big site, but I could be wrong. 

There is going to be a larger emphasis on the CPU next gen though, it's not going to be all about the GPU. Actually excited for the Playstation 5 and what it will mean in CPU based tasks and how that is going to translate over to simulation quality of the worlds being built.



--::{PC Gaming Master Race}::--

HollyGamer said:
drkohler said:

Please stop posting more nonsense every 10 minutes.

The PS4Pro already has 36CUs. There is no way a next gen chip has fewer CUs.

Also your 1.8Ghz clock is way too high for consoles, unless we get monster cooling solutions. There has been an image of a navi board floating around for a few weeks now, it has two 8-pin connectors. That alone tells you all about the problems AMD STILL has with power consumption, even at 7nm.

Dont reply or quote to him, he is  a stealth Sony hatter. 

He's just  unique. lol



CGI-Quality said:
OdinHades said:
So my PC will essentially be good enough for the whole next console generation.

Nice.

Specs?

Got myself an i7-9900K with an RTX 2070 and 32 Gigs of Ram. So nothing too crazy, but it should be in the ballpark of the new consoles.



Official member of VGC's Nintendo family, approved by the one and only RolStoppable. I feel honored.