By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Gaming Discussion - Rumor:PS5 & Anaconda Scarlet GPU on par with RTX 2080, Xbox exclusives focus on Cross gen, Developer complain about Lockhart.UPDATE: Windows Central said Xbox Anaconda target 12 teraflop

 

What do you think

I am excited for next gen 22 61.11%
 
I cannot wait to play next gen consoles 4 11.11%
 
I need to find another th... 2 5.56%
 
I worried about next gen 8 22.22%
 
Total:36
drkohler said:
Xxain said:
What is the performance difference between 10-11 and 12TF?

The difference is high enough that one particular company will shout it from every rooftop that "We have the fastest console on earth", to everyone that mostly isn't interested at all.

In reality, it's games that count, and nothing else. We will have umpteen DigitalFoundry analyses that this game looks better on "We have the fastest console on earth" because pixels 128 to 133 in scanline 326 are sharper than on the competitor. For the rest of the world, the difference is precisely irrelevant. If anything has told us the past years, t's games that count.

Yep. I stopped caring once I saw the Digital Foundry video, where they zoomed in on the freaking mud on the ground to show the difference between "4K Textures". Like I'm really going to look straight down and zoom in while playing the game. -_-



Around the Network
Trumpstyle said:
We now have Jason Schreier, FLUTE leak, Microsoft E3 video, The verge saying Anaconda is above 10TF, a verified insider saying both PS5 and Xbox anaconda is above 10TF. Add 3 people who have contacts with game developers (Kleegamfan, Colin Mccarthy, Andrew Reiner) saying PS5 is above Anaconda, what conclusion can we draw?

That the discussion if PS5 or Anaconda is below 10TF is over, the people who thought below 10TF are defeated. The next battle is whether next-gen consoles are using TSMC N7P or 7nm+.

I'm on 7nm+, who's with me?

10TFlops at full precision? I very much doubt that.

Navi gets up to 8.5 TFlops with 225W, Vega VII up to 11 with 300W TDP. Even just the 225W of the Navi are way too much for a console, that's about the TDP budget of the entire console, CPU, RAM, SSD, etc... included. Did I mention that both are produced in 7nm already?

At half precision, 10 TFlops are entirely possible. And I'm pretty sure the engineers did tell this numbers since Navi gets the same performance as Vega with about 35% less TFlops in raw power, so the increase over Pro/X would look rather low despite a sizable performance boost.

The only way they could make them work at full precision is to make the boxes a lot bigger to allow for larger coolers and fans, otherwise they RRoD themselves in no time. But then also don't expect such lowly prices as $499 for such a console, even by next year that would be too expensive.



Xxain said:
What is the performance difference between 10 and 12TF?

2 TF, I guess.



Bofferbrauer2 said:
Trumpstyle said:
We now have Jason Schreier, FLUTE leak, Microsoft E3 video, The verge saying Anaconda is above 10TF, a verified insider saying both PS5 and Xbox anaconda is above 10TF. Add 3 people who have contacts with game developers (Kleegamfan, Colin Mccarthy, Andrew Reiner) saying PS5 is above Anaconda, what conclusion can we draw?

That the discussion if PS5 or Anaconda is below 10TF is over, the people who thought below 10TF are defeated. The next battle is whether next-gen consoles are using TSMC N7P or 7nm+.

I'm on 7nm+, who's with me?

10TFlops at full precision? I very much doubt that.

Navi gets up to 8.5 TFlops with 225W, Vega VII up to 11 with 300W TDP. Even just the 225W of the Navi are way too much for a console, that's about the TDP budget of the entire console, CPU, RAM, SSD, etc... included. Did I mention that both are produced in 7nm already?

At half precision, 10 TFlops are entirely possible. And I'm pretty sure the engineers did tell this numbers since Navi gets the same performance as Vega with about 35% less TFlops in raw power, so the increase over Pro/X would look rather low despite a sizable performance boost.

The only way they could make them work at full precision is to make the boxes a lot bigger to allow for larger coolers and fans, otherwise they RRoD themselves in no time. But then also don't expect such lowly prices as $499 for such a console, even by next year that would be too expensive.

You looking at the wrong gpu, look at radeon 5700, not radeon 5700XT

Add some undervolting and you can hit good numbers, around 1,7ghz+ on tsmc 7nm is the magic number and on tsmc N7P or 7nm+ it should be 1,8ghz+. But we are getting indications they going for 2ghz.

I don't know where you getting your TF number from but radeon 5700 is 7,8TF+ and radeon 5700XT 9,5TF+.

Based on numbers I stolen from someone else, a console with 44CU's at 1,8ghz should pull about 195W on tsmc N7P or tsmc 7nm+ and maybe 230W with 44CU's at 2ghz, it will depend on when exactly Navi power consumption goes out of whack on Tsmc N7P or Tsmc 7nm+.



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!!

Cerebralbore101 said:
drkohler said:

The difference is high enough that one particular company will shout it from every rooftop that "We have the fastest console on earth", to everyone that mostly isn't interested at all.

In reality, it's games that count, and nothing else. We will have umpteen DigitalFoundry analyses that this game looks better on "We have the fastest console on earth" because pixels 128 to 133 in scanline 326 are sharper than on the competitor. For the rest of the world, the difference is precisely irrelevant. If anything has told us the past years, t's games that count.

Yep. I stopped caring once I saw the Digital Foundry video, where they zoomed in on the freaking mud on the ground to show the difference between "4K Textures". Like I'm really going to look straight down and zoom in while playing the game. -_-

I've done that in VR, except zoom in means banging my head off the real floor.

What I'm more interested in is loading times and patch times. It currently takes close to 2 minutes from turning on the ps4 pro to being signed up for the next online race. No chance to make it with only 1 minute left until the next one starts. And it takes over 35 minutes to update the game after a patch (copying 100 GB over)

I wouldn't mind playing at 1080p60 with 2K textures if that would mean faster loading. Faster start up, faster game loads, faster patches, then maybe better graphics. In some games it felt faster to simply run somewhere than to use 'fast' travel...



Around the Network
SvennoJ said:

What I'm more interested in is loading times and patch times. It currently takes close to 2 minutes from turning on the ps4 pro to being signed up for the next online race. No chance to make it with only 1 minute left until the next one starts. And it takes over 35 minutes to update the game after a patch (copying 100 GB over)

I wouldn't mind playing at 1080p60 with 2K textures if that would mean faster loading. Faster start up, faster game loads, faster patches, then maybe better graphics. In some games it felt faster to simply run somewhere than to use 'fast' travel...

In theory, with a superfast ssd (like a ReRam solution), you can dump the entire system state onto the ssd in a second, quit, and restart the game exactly where you left it within a second or two. Whether that is OS supported and developers will use it is anyone's guess.

If you have to patch 100GBytes, that is entirely a question on how fast you can suck bytes over the wire. No amount of next-gen console will change that.



Trumpstyle said:

Add some undervolting and you can hit good numbers, around 1,7ghz+ on tsmc 7nm is the magic number and on tsmc N7P or 7nm+ it should be 1,8ghz+. But we are getting indications they going for 2ghz.

Hardly anyone is currently using 7nm+, almost all are in the N7P camp. What is clear is that you can't do 2GHz on 7nm, that would result in a jet engine taking off, all the time.



Trumpstyle said:
Bofferbrauer2 said:

10TFlops at full precision? I very much doubt that.

Navi gets up to 8.5 TFlops with 225W, Vega VII up to 11 with 300W TDP. Even just the 225W of the Navi are way too much for a console, that's about the TDP budget of the entire console, CPU, RAM, SSD, etc... included. Did I mention that both are produced in 7nm already?

At half precision, 10 TFlops are entirely possible. And I'm pretty sure the engineers did tell this numbers since Navi gets the same performance as Vega with about 35% less TFlops in raw power, so the increase over Pro/X would look rather low despite a sizable performance boost.

The only way they could make them work at full precision is to make the boxes a lot bigger to allow for larger coolers and fans, otherwise they RRoD themselves in no time. But then also don't expect such lowly prices as $499 for such a console, even by next year that would be too expensive.

You looking at the wrong gpu, look at radeon 5700, not radeon 5700XT

Add some undervolting and you can hit good numbers, around 1,7ghz+ on tsmc 7nm is the magic number and on tsmc N7P or 7nm+ it should be 1,8ghz+. But we are getting indications they going for 2ghz.

I don't know where you getting your TF number from but radeon 5700 is 7,8TF+ and radeon 5700XT 9,5TF+.

Based on numbers I stolen from someone else, a console with 44CU's at 1,8ghz should pull about 195W on tsmc N7P or tsmc 7nm+ and maybe 230W with 44CU's at 2ghz, it will depend on when exactly Navi power consumption goes out of whack on Tsmc N7P or Tsmc 7nm+.

The TFlops you're quoting are at peak boost performance, which the GPU can't hold without consuming much more than those values you're posting below. For instance, in the Techpowerup test the GPU only reached 1672 Mhz on average, which results into almost exactly 7.7 TFlops. Undervolting will simply not be enough to get to 10 TFlops, especially since 160W is already the maximum a console can take. And Navi translates that into more clock speed anyway by itself unless you lock the speed - but that only gives like 30-40 Mhz in general while it would need more like 200-300 Mhz to make it work for your example.

However, I agree that I did mess up a bit, I took consumption at max boost and used the TFlops at base clock. Still, my point stands that 10 TFlops are still out of reach



drkohler said:
SvennoJ said:

What I'm more interested in is loading times and patch times. It currently takes close to 2 minutes from turning on the ps4 pro to being signed up for the next online race. No chance to make it with only 1 minute left until the next one starts. And it takes over 35 minutes to update the game after a patch (copying 100 GB over)

I wouldn't mind playing at 1080p60 with 2K textures if that would mean faster loading. Faster start up, faster game loads, faster patches, then maybe better graphics. In some games it felt faster to simply run somewhere than to use 'fast' travel...

In theory, with a superfast ssd (like a ReRam solution), you can dump the entire system state onto the ssd in a second, quit, and restart the game exactly where you left it within a second or two. Whether that is OS supported and developers will use it is anyone's guess.

If you have to patch 100GBytes, that is entirely a question on how fast you can suck bytes over the wire. No amount of next-gen console will change that.

The patch can be as small as 100MB, however the ps4 copies the entire game to apply the patch to the copy, then deletes the original if successful. So you need enough free space to copy the whole game and wait for the stuff to copy. 'Luckily' it starts making a copy while downloading the patch so the size of the patch is completely irrelevant. Copying takes much longer than the download.

SSD would certainly help and a smarter patch system.



SvennoJ said:
drkohler said:

In theory, with a superfast ssd (like a ReRam solution), you can dump the entire system state onto the ssd in a second, quit, and restart the game exactly where you left it within a second or two. Whether that is OS supported and developers will use it is anyone's guess.

If you have to patch 100GBytes, that is entirely a question on how fast you can suck bytes over the wire. No amount of next-gen console will change that.

The patch can be as small as 100MB, however the ps4 copies the entire game to apply the patch to the copy, then deletes the original if successful. So you need enough free space to copy the whole game and wait for the stuff to copy. 'Luckily' it starts making a copy while downloading the patch so the size of the patch is completely irrelevant. Copying takes much longer than the download.

SSD would certainly help and a smarter patch system.

Yeah the PS4 patching system is a pain. First there's "Preparing" then the actual download then it starts "Copying". It's really annoying, and I have no idea why they do it that way. Xbox doesn't have that problem. I don't think you need twice the size on your HDD to download and install a game on Xbox either.

But yeah, even if they don't change the system and you still have to copy, at least an SSD will make that a whole lot less painful.