John Carmack:You need a 3.68 Teraflop GPU to match PS4's GPU

 

yep, overall GTX 480 is faster

Memory Bandwidth

GTX 480 : 177408 MB/sec

7850 : 153600 MB/sec

Texel Rate ( texture filtering)

GTX 480 :   42000 Mtexels/sec

7850 :  55040 Mtexels/sec

Pixel Rate ( resolution & AA)

GTX 480 : 33600 Mpixels/sec

7850 : 27520 Mpixels/sec

04dcarraher

Oh look who joined the one who claimed the PS4 would not use more than 4GB of ram for games because no PC game did it,yet Killzone SF a launch game already use more than 4GB..:lol:

So how is that crow.?

[QUOTE="04dcarraher"]

 

yep, overall GTX 480 is faster

Memory Bandwidth

GTX 480 : 177408 MB/sec

7850 : 153600 MB/sec

Texel Rate ( texture filtering)

GTX 480 :   42000 Mtexels/sec

7850 :  55040 Mtexels/sec

Pixel Rate ( resolution & AA)

GTX 480 : 33600 Mpixels/sec

7850 : 27520 Mpixels/sec

tormentos

Oh look who joined the one who claimed the PS4 would not use more than 4GB of ram for games because no PC game did it,yet Killzone SF a launch game already use more than 4GB..:lol:

So how is that crow.?

[QUOTE="04dcarraher"]

 

yep, overall GTX 480 is faster

Memory Bandwidth

GTX 480 : 177408 MB/sec

7850 : 153600 MB/sec

Texel Rate ( texture filtering)

GTX 480 :   42000 Mtexels/sec

7850 :  55040 Mtexels/sec

Pixel Rate ( resolution & AA)

GTX 480 : 33600 Mpixels/sec

7850 : 27520 Mpixels/sec

tormentos

Oh look who joined the one who claimed the PS4 would not use more than 4GB of ram for games because no PC game did it,yet Killzone SF a launch game already use more than 4GB..:lol:

So how is that crow.?

OMG
I said the PS4 wont use more then 4gb for video use not the whole system usage. also KZ SF is only using 3 gb for video and 1.5 gb for system use
Its pointless to throw more then 4gb on a mid teir gpu thats only rendering at 1080 with AA only running at 30 fps.

I understand his point. Not sure about the accuracy behind his claim.

To tone down the 40k models hype,  KZ SF will not have an always true 40,000 polygon characters. The way they are going to allow 40k npc models is when there is only one npc within the distance of 0-2, the more npc's on screen within a certain distance the LOD quality count lowers. So for example You have three npc's at a distance of 1 their polygon count will be about a 3rd each.  The main reason behind this is that the gpu does not have enough processing power to render multiple high polygon count models.

You can see object LOD suffering at further distances

It will take 5 years for consoles to relatively and visually match a 3.68 TFLOP PC GPU.  And "optimization" to me means anything from dedicated code to metal programming to heavy LOD management.  The LOD management in console games is typically really awful and noticeable, which doesn't translate in comparison screenshots like it does in actual high res game footage.  This coming generation will benefit from flexible tessellation which should mitigate noticeable LOD changes on many game objects, especially high polygon characters and detailed models.  

The Assassin's Creed games come to mind when I think of horrid LOD pop-in.  Bleh.

PC_Otter

The PS4/Xbox One will never match a 3.68 TFLOP PC GPU.

Games will look better at time goes by but that is becuase devs find more efficient ways to render things.

Just like how my 6 year old 8800gt still trashes any current console many times over the same will apply to the 3.68 TFLOP PC GPU.

To tone down the 40k models hype,  KZ SF will not have an always true 40,000 polygon characters. The way they are going to allow 40k npc models is when there is only one npc within the distance of 0-2, the more npc's on screen within a certain distance the LOD quality count lowers. So for example You have three npc's at a distance of 1 their polygon count will be about a 3rd each.  The main reason behind this is that the gpu does not have enough processing power to render multiple high polygon count models.

You can see object LOD suffering at further distances

04dcarraher

[QUOTE="PC_Otter"]

It will take 5 years for consoles to relatively and visually match a 3.68 TFLOP PC GPU.  And "optimization" to me means anything from dedicated code to metal programming to heavy LOD management.  The LOD management in console games is typically really awful and noticeable, which doesn't translate in comparison screenshots like it does in actual high res game footage.  This coming generation will benefit from flexible tessellation which should mitigate noticeable LOD changes on many game objects, especially high polygon characters and detailed models.  

The Assassin's Creed games come to mind when I think of horrid LOD pop-in.  Bleh.

RyviusARC

The PS4/Xbox One will never match a 3.68 TFLOP PC GPU.

Games will look better at time goes by but that is becuase devs find more efficient ways to render things.

Just like how my 6 year old 8800gt still trashes any current console many times over the same will apply to the 3.68 TFLOP PC GPU.

[QUOTE="RyviusARC"]

[QUOTE="PC_Otter"]

It will take 5 years for consoles to relatively and visually match a 3.68 TFLOP PC GPU.  And "optimization" to me means anything from dedicated code to metal programming to heavy LOD management.  The LOD management in console games is typically really awful and noticeable, which doesn't translate in comparison screenshots like it does in actual high res game footage.  This coming generation will benefit from flexible tessellation which should mitigate noticeable LOD changes on many game objects, especially high polygon characters and detailed models.  

The Assassin's Creed games come to mind when I think of horrid LOD pop-in.  Bleh.

GioVela2010

The PS4/Xbox One will never match a 3.68 TFLOP PC GPU.

Games will look better at time goes by but that is becuase devs find more efficient ways to render things.

Just like how my 6 year old 8800gt still trashes any current console many times over the same will apply to the 3.68 TFLOP PC GPU.

You're hilarious!

[QUOTE="04dcarraher"]

To tone down the 40k models hype,  KZ SF will not have an always true 40,000 polygon characters. The way they are going to allow 40k npc models is when there is only one npc within the distance of 0-2, the more npc's on screen within a certain distance the LOD quality count lowers. So for example You have three npc's at a distance of 1 their polygon count will be about a 3rd each.  The main reason behind this is that the gpu does not have enough processing power to render multiple high polygon count models.

You can see object LOD suffering at further distances

GioVela2010

i own a titan, and my pc is better

Sweenix

 

OMG
I said the PS4 wont use more then 4gb for video use not the whole system usage. also KZ SF is only using 3 gb for video and 1.5 gb for system use
Its pointless to throw more then 4gb on a mid teir gpu thats only rendering at 1080 with AA only running at 30 fps.

 

04dcarraher

Is using 4.5 GB idiot and it was on the unfinish kits..:lol:

3GB already and was done on 4GB kits remember that the demo is not running on final hardware,so they were working with 4GB the 4.5 reside on the fact that dev kits have more ram than retail consoles.

Respawn Entertainment may have been the ones lobbying for the RAM bump. This would actually make sense, as Respawn has been very clear and vocal about how the 5GB of RAM allotted for gaming on the Xbox One has already caused issues in development of TitanFall.

http://www.examiner.com/article/xbox-one-rumors-point-to-a-higher-clock-speed-but-doubtful-on-increase-ram

The problem with you is that you think that ram is directly linked to how good something look or how powerful something most be..

So how is that crow...:lol:

Eltormentos guessing, that KZ SF will use even more ram....

Come on now

Current dev kits are the same ones from the Feb demo and they only have 6 cores of the CPU open to games (7 is the goal, may or may not happen), no dedicated audio hardware so the CPU has to do that, the compute customizations aren't on the GPU yet so the CPU is also doing those the the game is only using a total of 4.5Gb of RAM (3gb is for video).

Just common sense...

A 7850 can only address so much Vram, the bus width is a large indication, 2GB is already the 7850's maximum, it may address more with a tighter coded game and a locked down framerate.

GTX 680 cannot address the full 4GB Vram until put in 3-way SLi, where there is enough GPU muscle to output the frame rate and data.

256-Bit is the real killer here.

Tormentos, you read things like it is gospel, from what i read, 1.5GB is being used on the GPU, the rest is all stored in system ram in order for the GPU to have time to render.

This is why 30FPS, and still lagging.

[QUOTE="GioVela2010"]

To tone down the 40k models hype,  KZ SF will not have an always true 40,000 polygon characters. The way they are going to allow 40k npc models is when there is only one npc within the distance of 0-2, the more npc's on screen within a certain distance the LOD quality count lowers. So for example You have three npc's at a distance of 1 their polygon count will be about a 3rd each.  The main reason behind this is that the gpu does not have enough processing power to render multiple high polygon count models.

You can see object LOD suffering at further distances

04dcarraher

Why cant we have games that look like this without the shooting part?

I miss Myst and Riven...

Just common sense...

 

A 7850 can only address so much Vram, the bus width is a large indication, 2GB is already the 7850's maximum, it may address more with a tighter coded game and a locked down framerate.

GTX 680 cannot address the full 4GB Vram until put in 3-way SLi, where there is enough GPU muscle to output the frame rate and data.

256-Bit is the real killer here.

 

Tormentos, you read things like it is gospel, from what i read, 1.5GB is being used on the GPU, the rest is all stored in system ram in order for the GPU to have time to render.

 

This is why 30FPS, and still lagging.

AMD655

Max 32-bit address 4 294 967 296, 64-bit is 18 446 744 073 709 551 616. (also IPv4 is a 32-bit #, and IPv6 is a 64-#)

The max theoretical bandwidth, of the PS4, at 30fps, is 5.87GB/frame. (does not mean it can effectively use that much memory 'per frame')

Having more GPU's in series doesn't increase their ability to "address" VRAM. When you run multi-GPU setups, they have to duplicate all the memory necessary for the full scene (3 SLI GPU's, each with 4GB of RAM, only effectively have 4GB of RAM). I'm not sure how much extra VRAM is necessary to run SLI/Crossfire, but the power of the GPU has nothing to do with the VRAM use; though depending on the bandwidth and processing power, you may not be able to use all of the memory in one frame.

----

I doubt the PS4 is having bandwidth issues tbh, I'm betting the frame-rate drops are related to a job on the CPU or GPU just not getting done quickly enough (not enough processing power, or inefficient use of).

[QUOTE="AMD655"]

Just common sense...

 

A 7850 can only address so much Vram, the bus width is a large indication, 2GB is already the 7850's maximum, it may address more with a tighter coded game and a locked down framerate.

GTX 680 cannot address the full 4GB Vram until put in 3-way SLi, where there is enough GPU muscle to output the frame rate and data.

256-Bit is the real killer here.

 

Tormentos, you read things like it is gospel, from what i read, 1.5GB is being used on the GPU, the rest is all stored in system ram in order for the GPU to have time to render.

 

This is why 30FPS, and still lagging.

Inconsistancy

Max 32-bit address 4 294 967 296, 64-bit is 18 446 744 073 709 551 616. (also IPv4 is a 32-bit #, and IPv6 is a 64-#)

The max theoretical bandwidth, of the PS4, at 30fps, is 5.87GB/frame. (does not mean it can effectively use that much memory 'per frame')

Having more GPU's in series doesn't increase their ability to "address" VRAM. When you run multi-GPU setups, they have to duplicate all the memory necessary for the full scene (3 SLI GPU's, each with 4GB of RAM, only effectively have 4GB of RAM). I'm not sure how much extra VRAM is necessary to run SLI/Crossfire, but the power of the GPU has nothing to do with the VRAM use; though depending on the bandwidth and processing power, you may not be able to use all of the memory in one frame.

----

I doubt the PS4 is having bandwidth issues tbh, I'm betting the frame-rate drops are related to a job on the CPU or GPU just not getting done quickly enough (not enough processing power, or inefficient use of).

Eltormentos guessing, that KZ SF will use even more ram....

Come on now

Current dev kits are the same ones from the Feb demo and they only have 6 cores of the CPU open to games (7 is the goal, may or may not happen), no dedicated audio hardware so the CPU has to do that, the compute customizations aren't on the GPU yet so the CPU is also doing those the the game is only using a total of 4.5Gb of RAM (3gb is for video).

04dcarraher

WTF..:lol:

No Guerilla them self stated that the demo was done using old kits that did not even have the sound block,and sound was been done on CPU when it will be done on the sound block..

So no it was different hardware they did not had 8GB of ram to use for the game,and the hardware was unfinish forcing them to use the CPU for sound.

CPU Load

• 60 AI characters
• 940 entities, 300 active
• 8200 physics objects (1500 key-framed, 6700 static)
• 500 particle systems
• 120 sound voices
• 110 ray casts
• 1000 jobs per frame

http://www.eurogamer.net/articles/digitalfoundry-inside-killzone-shadow-fall

The PS4 has a sound chip,and for the demo the sound chip wasn't present and the CPU was been use to emulate it..

If funny that you some how think it will not use more ram,because the video of Killzone SF release on february clearly show,pop in detail which is atribute in almost all cases to ram starvation,smoke suddenly became more black which show the game need more rams.

[QUOTE="AMD655"]

Just common sense...

 

A 7850 can only address so much Vram, the bus width is a large indication, 2GB is already the 7850's maximum, it may address more with a tighter coded game and a locked down framerate.

GTX 680 cannot address the full 4GB Vram until put in 3-way SLi, where there is enough GPU muscle to output the frame rate and data.

256-Bit is the real killer here.

 

Tormentos, you read things like it is gospel, from what i read, 1.5GB is being used on the GPU, the rest is all stored in system ram in order for the GPU to have time to render.

 

This is why 30FPS, and still lagging.

Inconsistancy

Max 32-bit address 4 294 967 296, 64-bit is 18 446 744 073 709 551 616. (also IPv4 is a 32-bit #, and IPv6 is a 64-#)

The max theoretical bandwidth, of the PS4, at 30fps, is 5.87GB/frame. (does not mean it can effectively use that much memory 'per frame')

Having more GPU's in series doesn't increase their ability to "address" VRAM. When you run multi-GPU setups, they have to duplicate all the memory necessary for the full scene (3 SLI GPU's, each with 4GB of RAM, only effectively have 4GB of RAM). I'm not sure how much extra VRAM is necessary to run SLI/Crossfire, but the power of the GPU has nothing to do with the VRAM use; though depending on the bandwidth and processing power, you may not be able to use all of the memory in one frame.

----

I doubt the PS4 is having bandwidth issues tbh, I'm betting the frame-rate drops are related to a job on the CPU or GPU just not getting done quickly enough (not enough processing power, or inefficient use of).

Probablyn the last this is a launch game,which tend to be rush to get them to the finish line.

Resistance look nothing like Killzone 2.

Just common sense...

 

A 7850 can only address so much Vram, the bus width is a large indication, 2GB is already the 7850's maximum, it may address more with a tighter coded game and a locked down framerate.

GTX 680 cannot address the full 4GB Vram until put in 3-way SLi, where there is enough GPU muscle to output the frame rate and data.

256-Bit is the real killer here.

 

Tormentos, you read things like it is gospel, from what i read, 1.5GB is being used on the GPU, the rest is all stored in system ram in order for the GPU to have time to render.

 

This is why 30FPS, and still lagging.

AMD655

Such a joke..

No is 3GB on the GPU learn to read,and was for unsinish hardware.

And the game is 30 FPS because that is the target,on consoles 60 FPS get rarely push,it could have a Titan GPU and the power will be use more for detail that  for frames.

[QUOTE="AMD655"]

Just common sense...

 

A 7850 can only address so much Vram, the bus width is a large indication, 2GB is already the 7850's maximum, it may address more with a tighter coded game and a locked down framerate.

GTX 680 cannot address the full 4GB Vram until put in 3-way SLi, where there is enough GPU muscle to output the frame rate and data.

256-Bit is the real killer here.

 

Tormentos, you read things like it is gospel, from what i read, 1.5GB is being used on the GPU, the rest is all stored in system ram in order for the GPU to have time to render.

 

This is why 30FPS, and still lagging.

tormentos

Such a joke..

No is 3GB on the GPU learn to read,and was for unsinish hardware.

And the game is 30 FPS because that is the target,on consoles 60 FPS get rarely push,it could have a Titan GPU and the power will be use more for detail that  for frames.

I think you should just shut up and go to bed pal.

Just common sense...

A 7850 can only address so much Vram, the bus width is a large indication, 2GB is already the 7850's maximum, it may address more with a tighter coded game and a locked down framerate.

GTX 680 cannot address the full 4GB Vram until put in 3-way SLi, where there is enough GPU muscle to output the frame rate and data.

256-Bit is the real killer here.

Tormentos, you read things like it is gospel, from what i read, 1.5GB is being used on the GPU, the rest is all stored in system ram in order for the GPU to have time to render.

This is why 30FPS, and still lagging.

AMD655

AMD GCN can address more than 2 GB of RAM e.g. it supports 64bit pointers.

AMD GCN is IOMMU is based on X86-64 IP.

Professional real time 3D applications can use the entire 4GB of VRAM on AMD FirePro W7000 (based on AMD Pitcairn ASIC).

------------

This video answers many technical questions about the DX11.2's tiled resources, massive textures (e.g. 10 GB) and smaller VRAM pool.

[QUOTE="AMD655"]

Just common sense...

 

A 7850 can only address so much Vram, the bus width is a large indication, 2GB is already the 7850's maximum, it may address more with a tighter coded game and a locked down framerate.

GTX 680 cannot address the full 4GB Vram until put in 3-way SLi, where there is enough GPU muscle to output the frame rate and data.

256-Bit is the real killer here.

 

Tormentos, you read things like it is gospel, from what i read, 1.5GB is being used on the GPU, the rest is all stored in system ram in order for the GPU to have time to render.

 

This is why 30FPS, and still lagging.

Inconsistancy

Max 32-bit address 4 294 967 296, 64-bit is 18 446 744 073 709 551 616. (also IPv4 is a 32-bit #, and IPv6 is a 64-#)

The max theoretical bandwidth, of the PS4, at 30fps, is 5.87GB/frame. (does not mean it can effectively use that much memory 'per frame')

Having more GPU's in series doesn't increase their ability to "address" VRAM. When you run multi-GPU setups, they have to duplicate all the memory necessary for the full scene (3 SLI GPU's, each with 4GB of RAM, only effectively have 4GB of RAM). I'm not sure how much extra VRAM is necessary to run SLI/Crossfire, but the power of the GPU has nothing to do with the VRAM use; though depending on the bandwidth and processing power, you may not be able to use all of the memory in one frame.

----

I doubt the PS4 is having bandwidth issues tbh, I'm betting the frame-rate drops are related to a job on the CPU or GPU just not getting done quickly enough (not enough processing power, or inefficient use of).

[QUOTE="AMD655"]

 

Shows how brainless you really are.

 

With a GPU such as the GTX 480 and 7850, the GPU will run out of performance before memory becomes a bottleneck, the 480 has the same bandwidth as a GTX 680 when you push the memory to 4ghz.

 

Also take note that i can play BF3 on Ultra with 2xMSAA at 2560x1440 with no memory bottleneck.

 

 

You do not murder anyone other than yourself trying to sound smart, when all you are is a foolish inexperienced person looking up stuff on the internet, then throwing it around like it is gospel.

tormentos

On your PC yeah on the PS4 not at all,Killzone SF already use 4GB incredible draw distances,port all that code to run on your 480 GTX,and either resolution,frames,image quality or AA had to be lower or drop,because you can't magicall fit 4GB of data into 1.5Gb.

So does the 7850,that doesn't mean more memory will not benefit the game.

You are an idiot that was proved wrong and started a hude damage control your 480 is less powerful than the PS4 and a hell of allot less efficient to.

This topic may be old but I want to reiterate on wasdie's beautiful post that PC's have a lot of bullshit overhead to deal with so they gimp their games to the lowest specs. From the legend Carmack himself. Hermits lining up for a Titan as we speak.

Now this "My PC is better" bullshit can rest, unless you own a Titan.

DrTrafalgarLaw

From http://semiaccurate.com/2013/07/17/keynotes-for-amd-developer-summit-2013-announced/

---

PC has "AMD HSA" supported 3D engines e.g. EA DICE's Frostbite 3 and Unity3D.

http://www.slideshare.net/zlatan4177/gpgpu-algorithms-in-games

Smaller VRAM issues was addressed by DX11.2's tiled resource i.e. converts the smaller/faster VRAM into a cache. The key point is the just-in-time texture loads on smaller/faster VRAM based on user's view port. On AMD GCN, this feature is hardware accelerated i.e. AMD PRT. On other GPUs, this feature would be done via software e.g. NVIDIA CUDA or CPU.ronvalencia

Just in time textures is way older than GCN.

And that advantage that state there also apply to the PS4 which has even more ram.

So the advantage ram wise should be even bigger,you talk as if PRT was not supported on PS4 stop trying to act like that.

Any advantage in ram usage those GPU get the PS4 wouold also get them,hell even more.

[QUOTE="ronvalencia"] Smaller VRAM issues was addressed by DX11.2's tiled resource i.e. converts the smaller/faster VRAM into a cache. The key point is the just-in-time texture loads on smaller/faster VRAM based on user's view port. On AMD GCN, this feature is hardware accelerated i.e. AMD PRT. On other GPUs, this feature would be done via software e.g. NVIDIA CUDA or CPU.tormentos

Just in time textures is way older than GCN.

And that advantage that state there also apply to the PS4 which has even more ram.

So the advantage ram wise should be even bigger,you talk as if PRT was not supported on PS4 stop trying to act like that.

Any advantage in ram usage those GPU get the PS4 wouold also get them,hell even more.

AMD GCN done it on per polygon basis and its hardware accelerated (i.e. nearly zero performance cost).

DX11.2's tiled resource has the similar spec as AMD PRT i.e. both has 64KB texture tile and 'etc'.

AMD PRT doesn't change the GPU's potential i.e. it fixes the smaller VRAM pool issue.

AMD PRT can be applied from persistent storage devices e.g. hard disk or Blu-ray, but remember the tile stream size would be smaller from hard disk and Blu-ray sources. ZLib decompression could increase effective tile stream size.

PS; DX11.2 enables the CPU to directly access GPU's memory areas.

[QUOTE="killzowned24"][QUOTE="cdragon_88"]

So are we really putting words in John Carmack's mouth? If so here's mine: 

GioVela2010

Topic head by TC: "John Carmack:You need a 3.68 Teraflop GPU to match PS4's GPU" :? I'm sure Carmack never said that. Who's the idiot now?

[QUOTE="GioVela2010"][QUOTE="killzowned24"] https://twitter.com/ID_AA_Carmack/status/50277106856370176cdragon_88

Topic head by TC: "John Carmack:You need a 3.68 Teraflop GPU to match PS4's GPU" :? I'm sure Carmack never said that. Who's the idiot now?

.

.

wis3boi

For the PC, EA DICE's Frostbite 3 and Unity3D has AMD HSA support (HSAIL software).

i think my I7 3960x and 690gtx will be more than enough to quiet the PS4........period.

[QUOTE="wis3boi"]

.

ronvalencia

For the PC, EA DICE's Frostbite 3 and Unity3D has AMD HSA support (HSAIL software).

you really cant read a goddamn thing, can you? Stop posting random charts to posts that have zero to do with your crap :|

.

wis3boi

Winner.

[QUOTE="ronvalencia"]

[QUOTE="wis3boi"]

.

wis3boi

For the PC, EA DICE's Frostbite 3 and Unity3D has AMD HSA support (HSAIL software).

you really cant read a goddamn thing, can you? Stop posting random charts to posts that have zero to do with your crap :|

[QUOTE="wis3boi"]

[QUOTE="ronvalencia"] http://semiaccurate.com/2013/07/17/keynotes-for-amd-developer-summit-2013-announced/

 

For the PC, EA DICE's Frostbite 3 and Unity3D has AMD HSA support (HSAIL software).

 

 

ronvalencia

you really cant read a goddamn thing, can you? Stop posting random charts to posts that have zero to do with your crap :|

nothing to do with the picture i posted, keep being an irrelevant spammer

I don't really get what the issue is, even if Carmack is correct - who cares?

Just wait 6 months after the consoles are released and then buy a GPU that will blow them out of the water for the next 7 years.

This topic may be old but I want to reiterate on wasdie's beautiful post that PC's have a lot of bullshit overhead to deal with so they gimp their games to the lowest specs. From the legend Carmack himself. Hermits lining up for a Titan as we speak.

Now this "My PC is better" bullshit can rest, unless you own a Titan.

DrTrafalgarLaw

titan runs games at maxed out setting at resolutions of 2560x1600 and 60fps, ps4 cant run killzone at 1920x1080p/60fps. Seems legit. LMAO, ouch

console peasants can't accept the facts

[QUOTE="tormentos"]

[QUOTE="ronvalencia"] Smaller VRAM issues was addressed by DX11.2's tiled resource i.e. converts the smaller/faster VRAM into a cache. The key point is the just-in-time texture loads on smaller/faster VRAM based on user's view port. On AMD GCN, this feature is hardware accelerated i.e. AMD PRT. On other GPUs, this feature would be done via software e.g. NVIDIA CUDA or CPU.ronvalencia

Just in time textures is way older than GCN.

And that advantage that state there also apply to the PS4 which has even more ram.

So the advantage ram wise should be even bigger,you talk as if PRT was not supported on PS4 stop trying to act like that.

Any advantage in ram usage those GPU get the PS4 wouold also get them,hell even more.

AMD GCN done it on per polygon basis and its hardware accelerated (i.e. nearly zero performance cost).

DX11.2's tiled resource has the similar spec as AMD PRT i.e. both has 64KB texture tile and 'etc'.

AMD PRT doesn't change the GPU's potential i.e. it fixes the smaller VRAM pool issue.

AMD PRT can be applied from persistent storage devices e.g. hard disk or Blu-ray, but remember the tile stream size would be smaller from hard disk and Blu-ray sources. ZLib decompression could increase effective tile stream size.

PS; DX11.2 enables the CPU to directly access GPU's memory areas.

Nothing of what you say there prove me wrong.

PRT and JIt compression can be done on PS4 period.

[QUOTE="wis3boi"]

.

ronvalencia

For the PC, EA DICE's Frostbite 3 and Unity3D has AMD HSA support (HSAIL software).

You really don't read what people post..:lol:

You just hit reply quoting what ever you feel like it,look at what you quote and tell me it has something to do with what you posted there..:lol:

.

wis3boi

[QUOTE="wis3boi"]

.

DrTrafalgarLaw

I guess carmack would be more credible IF he weren't paid a fat stash of benjamins to spout BS to console fanboys. Not to mention, all his games are usually console exclusive. He knows who butters his bread and who doesn't. This would be like asking Gabe Newell, founder of Steam, what his thoughts are on the specs of console hardware. Face it, your weak hardware will NEVER even come close to even a 7870, let alone a Titan lol.

I guess carmack would be more credible IF he weren't paid a fat stash of benjamins to spout BS to console fanboys. Not to mention, all his games are usually console exclusive. He knows who butters his bread and who doesn't. This would be like asking Gabe Newell, founder of Steam, what his thoughts are on the specs of console hardware. Face it, your weak hardware will NEVER even come close to even a 7870, let alone a Titan lol.

blaznwiipspman1

carmack is a casual now he's been building too many rockets lately and forgot how to make games.

John Carmack:lol: