Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [TT] NVIDIA should launch its next-gen Pascal GPUs with HBM2 in 2H 2016
New Posts  All Forums:Forum Nav:

[TT] NVIDIA should launch its next-gen Pascal GPUs with HBM2 in 2H 2016 - Page 66

post #651 of 724
Quote:
Originally Posted by Mand12 View Post

Yes, it is the point really.

Because all that this generated was bad publicity. This was a mistake, and only a mistake. Not a con, not a cheat, not a deception, not a lie. The ONLY thing this did was make Nvidia look bad. As you said, deservedly so, because it was a foolish error they should have caught. But that's why I make the claim that the forum-riot impact of this issue is way, way WAY out of proportion with the actual facts of the issue. And as for a motive, the only thing I can think of is people want Nvidia to be bad, so they latch onto anything that they can find that's bad and don't let go, don't see other perspectives, don't consider any mitigating possibilities or likelihoods, and instead fixate on how evil Nvidia must be because they make us pay for things they design.

Quote:
Originally Posted by Mand12 View Post

rolleyes.gif

Why bother with facts, when a conspiracy will do, eh?

There is no upside to them trying this. There is ZERO chance this wouldn't have been found out. And then it blows up in their faces.

A mistake is rectified as soon as possible, not 4 months after the fact. Are you seriously going to claim that nobody, not a single person at nVidia, be it the engineers or the marketing people noticed something off with the spec sheets in a timespan of over 4 months? And given the fact that both the 550 Ti and 660 Ti had segmented memory and was disclosed at launch, while the 970's segmented memory wasn't until 4 months post launch, this substantially weakens your "it's just a mistake" stance.

As for facts, LOL. rolleyes.gif Please, do you have access to internal memos and emails from nVidia? You don't? Well then you don't know any more "facts" than we do, so get off your high horse and stop being condescending.

And finally...
Quote:
Originally Posted by Mand12 View Post

What's more condescending? My position, which is that a minor technical difference would be weighed against superb benchmark performance and not affect sales much, or your position, which is that droves and droves of 970 buyers look only at the press-release spec sheet and wouldn't bother with reading any detailed investigation on card performance?

That you keep repeating this ad nauseam, and your ignoring of my posts regarding opportunity cost here and here proves to me you either don't understand opportunity cost, or you're deliberately being obtuse about it because you have no arguments.
Edited by magnek - 3/1/16 at 12:39pm
post #652 of 724
Quote:
Originally Posted by Mahigan View Post

It works smile.gif

http://m.hardocp.com/article/2016/02/29/rise_tomb_raider_graphics_features_performance/4#.VtWl9OTS9oM

That's evidence that it works. I've asked him to post evidence to the contrary and he doesn't have any.

Without evidence to the contrary, he can't refute the claim that it works.

A.k.a he's wrong.

End of discussion

I'll preface this by saying I know next to nothing about the way that drivers are written or optimised, but I am curious, if you'll humour me.

I read the H article earlier this afternoon and saw their VRAM results and I had thought something like this : As I understand it, AMD/RTG are spending a large amount of time picking through each game and deciding which resources are most necessary to game performance to be stored directly in VRAM, due to the fact that the Furies are relatively limited on VRAM. They had said something like this would be happening on the release of the Furies when people had mentioned the relative lack of VRAM IIRC.

What happens later on when Polaris is out with possibly 8 or 16GB of HBM2. Will they spend the same amount of time fine tuning for the older cards, or doesnt it work like that and it's for some reason easier for AMD on their tech to do then Nvidia.
post #653 of 724
Quote:
Originally Posted by Mahigan View Post

It works smile.gif

http://m.hardocp.com/article/2016/02/29/rise_tomb_raider_graphics_features_performance/4#.VtWl9OTS9oM

That's evidence that it works. I've asked him to post evidence to the contrary and he doesn't have any.

Without evidence to the contrary, he can't refute the claim that it works.

A.k.a he's wrong.

End of discussion


I never said anywhere that AMD don't have working optimizations in place. In fact I said the direct opposite. All I'm contesting, blatantly, is HBM's uniqueness or the relevance of its bandwidth. And yes, you very much did imply so.

Quote:
Originally Posted by Mahigan View Post

It's possible to do on GDDR5 but GDDR5 isn't as fast as HBM..

Like here, where you very much imply the speed of HBM vs GDDR5 is some aiding factor.

Quote:
I said "HBM is different than GDDR5". <--- notice the period? Then I mentioned "Its bandwidth is used like an embeded cache". <--- notice the period.


And here. If we're picking at writing structure, a period does not inherently signal the end and isolation of a train of thought... especially not when the sentence it ends parrots a statement made in the preceding quote and quite significantly affects the meaning of the following sentences. If a period completely closed off individual thoughts from one another, then coherent essays and paragraphs would be impossible unless they were just one, nonstop sentence. Notice the quote from AMD you were repeating immediately before making that statement. The implication of HBM's difference to GDDR5 making size comparisons murky is clearly there whether you intended it or not.

I mean what, you just decided to randomly throw it out there that HBM is different than GDDR5 as the opening sentence of a new paragraph with no meaning behind it whatsoever? Okay...

Quote:
I never made that claim. Where in my quotation do I mention its capacity? Nowhere. What you fail to grasp is that you misunderstood what I said and have created this enormous strawman that you then write 5 paragraphs slaying.


Of course you're talking about working around capacity limitations. That's the entire point of these driver optimizations: dealing with limited VRAM capacity and you've implied more than once that HBM's speed or bandwidth is part of that in the quotes above. Unless you're just randomly throwing facts out there in the middle of your paragraphs, for no reason, that coincidentally imply very specific meanings. In which case, a misunderstanding is perfectly reasonable.

Quote:
Is HBM different than GDDR5? Yes, it ia stacked DRAM with many connection points over an interposer. These connection points or "pins" are in greater number than GDDR5 and thus allow for a wider memory interface (4,096 bit). This translates into higher bandwidth. So yes HBM is different than GDDR5.


Again, random point thrown into a sentence with no regard for context? AMD said and you paraphrased:
Quote:
Main point being that HBM can be thought of as a giant embedded cache, and is not directly comparable to GDDR5 sizes.

That sentence of yours that immediately followed makes absolutely no sense in a forced vacuum. It clearly stemmed from the AMD quote.




But on the matter of how effective these optimizations are regardless, HardOCP's take on RoTR is a very limited sample to base an overall, conclusive judgment of the tech on. HardOCP themselves have had mixed things to say about Fiji's VRAM (or, well, mixed things to say in general) depending on when, such as their interpretation of Dying Light's VRAM demands way back at Fiji's launch (of course, before the supposed optimizations):

http://www.hardocp.com/article/2015/07/26/amd_radeon_r9_fury_x_4k_video_card_review/8#.VtXtW_krLIU


And there are these videos showing what appears to be VRAM-related stuttering/freezes back then as well:

https://www.youtube.com/watch?v=8hnuj1OZAJs

https://www.youtube.com/watch?v=k_wNEOviaeE



If these are from last year, games are only getting more demanding. And if they also demand game-specific optimizations from AMD, then Fiji's already walking on thin ice given the investment required for individual tuning. Not to mention that the most increasingly-large graphical assets continue to be things like textures that absolutely tend to demand bandwidth (system RAM through PCI-e has very harsh limits; especially in a theoretical Xfire scenario as you mentioned both due to double the power with the same VRAM and additional PCI-e limitations). But most importantly, that "hand-tuning" that needs to be done for specific games has its limits in implementation time and scope (i.e. which games get the treatment and when?). HardOCP's latest review showing VRAM not to be an issue in RoTR is very different from their previous one from nearly 3 weeks after the game's release:

http://www.hardocp.com/article/2016/02/15/rise_tomb_raider_video_card_performance_review/11#.VtXxTfkrLIU


And HardOCP noted in their latest RoTR that it was the first time they actually saw AMD's claimed optimizations in action:
Quote:
If you remember a good while ago that AMD was making claims to this effect, but this is the first time we have actually been able to show results in real world gaming.


Game specific, takes time, first time HardOCP have actually seen it in action, limited testing so far, etc.; the approach obviously works to some degree, but has some severe drawbacks and it's all obviously because Fiji's VRAM isn't enough under standard, non-tuned scenarios. That's no myth given the attention AMD themselves have publicly given the issue.
Edited by Serandur - 3/1/16 at 12:07pm
Ice-nine
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K Asus Z170-DELUXE Gigabyte G1 GTX 980 Ti Corsair Vengeance LPX DDR4 3000 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 500 GB Seagate Barracuda 7200 RPM 2 TB Asus DVD Drive DRW-24B1ST/BLK/B/AS Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 Pro AOC Q2770PQU Corsair K95 RGB - Cherry MX Blue EVGA Supernova G2 750W 
CaseMouseMouse PadAudio
NZXT Phantom 410 Gunmetal Edition Logitech G502 Proteus Core Razer Goliathus Speed Edition Small Cyber Acoustics Satellite CA-3602 
Audio
Oppo PM-3 Closed Back Planar Magnetic Headphones 
  hide details  
Reply
Ice-nine
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K Asus Z170-DELUXE Gigabyte G1 GTX 980 Ti Corsair Vengeance LPX DDR4 3000 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 500 GB Seagate Barracuda 7200 RPM 2 TB Asus DVD Drive DRW-24B1ST/BLK/B/AS Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 Pro AOC Q2770PQU Corsair K95 RGB - Cherry MX Blue EVGA Supernova G2 750W 
CaseMouseMouse PadAudio
NZXT Phantom 410 Gunmetal Edition Logitech G502 Proteus Core Razer Goliathus Speed Edition Small Cyber Acoustics Satellite CA-3602 
Audio
Oppo PM-3 Closed Back Planar Magnetic Headphones 
  hide details  
Reply
post #654 of 724
Warning: Spoiler! (Click to show)
Quote:
Originally Posted by Serandur View Post

I never said anywhere that AMD don't have working optimizations in place. In fact I said the direct opposite. All I'm contesting, blatantly, is HBM's uniqueness or the relevance of its bandwidth. And yes, you very much did imply so.

Like here, where you very much imply the speed of HBM vs GDDR5 is some aiding factor.
And here. If we're picking at writing structure, a period does not inherently signal the end and isolation of a train of thought... especially not when the sentence it ends parrots a statement made in the preceding quote and quite significantly affects the meaning of the following sentences. If a period completely closed off individual thoughts from one another, then coherent essays and paragraphs would be impossible unless they were just one, nonstop sentence. Notice the quote from AMD you were repeating immediately before making that statement. The implication of HBM's difference to GDDR5 making size comparisons murky is clearly there whether you intended it or not.

I mean what, you just decided to randomly throw it out there that HBM is different than GDDR5 as the opening sentence of a new paragraph with no meaning behind it whatsoever? Okay...
Of course you're talking about working around capacity limitations. That's the entire point of these driver optimizations: dealing with limited VRAM capacity and you've implied more than once that HBM's speed or bandwidth is part of that in the quotes above. Unless you're just randomly throwing facts out there in the middle of your paragraphs, for no reason, that coincidentally imply very specific meanings. In which case, a misunderstanding is perfectly reasonable.





Again, random point thrown into a sentence with no regard for context? AMD said and you paraphrased:
That sentence of yours that immediately followed makes absolutely no sense in a forced vacuum. It clearly stemmed from the AMD quote.




But on the matter of how effective these optimizations are regardless, HardOCP's take on RoTR is a very limited sample to base an overall, conclusive judgment of the tech on. HardOCP themselves have had mixed things to say about Fiji's VRAM (or, well, mixed things to say in general) depending on when, such as their interpretation of Dying Light's VRAM demands way back at Fiji's launch (of course, before the supposed optimizations):

http://www.hardocp.com/article/2015/07/26/amd_radeon_r9_fury_x_4k_video_card_review/8#.VtXtW_krLIU


And there are these videos showing what appears to be VRAM-related stuttering/freezes back then as well:

https://www.youtube.com/watch?v=8hnuj1OZAJs

https://www.youtube.com/watch?v=k_wNEOviaeE



If these are from last year, games are only getting more demanding. And if they also demand game-specific optimizations from AMD, then Fiji's already walking on thin ice given the investment required for individual tuning. Not to mention that the most increasingly-large graphical assets continue to be things like textures that absolutely tend to demand bandwidth (system RAM through PCI-e has very harsh limits; especially in a theoretical Xfire scenario as you mentioned both due to double the power with the same VRAM and additional PCI-e limitations). But most importantly, that "hand-tuning" that needs to be done for specific games has its limits in implementation time and scope (i.e. which games get the treatment and when?). HardOCP's latest review showing VRAM not to be an issue in RoTR is very different from their previous one from nearly 3 weeks after the game's release:

http://www.hardocp.com/article/2016/02/15/rise_tomb_raider_video_card_performance_review/11#.VtXxTfkrLIU


And HardOCP noted in their latest RoTR that it was the first time they actually saw AMD's claimed optimizations in action:
Game specific, takes time, first time HardOCP have actually seen it in action, limited testing so far, etc.; the approach obviously works to some degree, but has some severe drawbacks and it's all obviously because Fiji's VRAM isn't enough under standard, non-tuned scenarios. That's no myth given the attention AMD themselves have publicly given the issue.

I did not imply anything other than "HBM is different than GDDR5" and "HBM is faster than GDDR5". That's what I wrote. That's what I implied.

And by HBM being different, I stressed its speed. Then I said that you coud also use this feature on GDDR5. Go back to my posts and re-read them because you clearly missed a lot of info.

When I said that HBM is faster, that means that its fast enough to warrant the feature. Imagine if AMD spent all of their time optimizing games for 4GB GDDR5 cards when they could have just used 8GB GDDR5 and called it a day?

You make assumptions, that you pull out of your own mind, and then write enormous walls of text for nothing. You strawman. You really do. You make up, out of thin air, assumptions based on what I've stated and then proceed to attack those assumptions. You're basically attacking your own concoctions.

If I say, Intel CPUs are different than AMDs, that doesn't mean that AMD CPUs aren't x86 compliant (see a made up concoction or a strawman). If I then add Intel CPUs are faster, that doesn't mean that Intel CPUs are capable of floating point operations while AMD CPUs aren't.

But ironically, that's exactly what you have been doing in your walls of text. Attacking non-existant arguments that you attribute to me and then proceed to slay.
Edited by Mahigan - 3/1/16 at 12:58pm
Kn0wledge
(20 items)
 
Pati3nce
(14 items)
 
Wisd0m
(10 items)
 
Reply
Kn0wledge
(20 items)
 
Pati3nce
(14 items)
 
Wisd0m
(10 items)
 
Reply
post #655 of 724
You quoted AMD:
Quote:
Main point being that HBM can be thought of as a giant embedded cache, and is not directly comparable to GDDR5 sizes.

And immediately followed by paraphrasing it in agreement;
Quote:
HBM is different than GDDR5. Its bandwidth is used like an embeded cache.


I pulled nothing out of my mind, it's very blatant. If it's not what you meant, I'm sorry, but that's what it said. If you can't understand why that gives off the meaning it does, then I really can't help here but it does.
Ice-nine
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K Asus Z170-DELUXE Gigabyte G1 GTX 980 Ti Corsair Vengeance LPX DDR4 3000 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 500 GB Seagate Barracuda 7200 RPM 2 TB Asus DVD Drive DRW-24B1ST/BLK/B/AS Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 Pro AOC Q2770PQU Corsair K95 RGB - Cherry MX Blue EVGA Supernova G2 750W 
CaseMouseMouse PadAudio
NZXT Phantom 410 Gunmetal Edition Logitech G502 Proteus Core Razer Goliathus Speed Edition Small Cyber Acoustics Satellite CA-3602 
Audio
Oppo PM-3 Closed Back Planar Magnetic Headphones 
  hide details  
Reply
Ice-nine
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K Asus Z170-DELUXE Gigabyte G1 GTX 980 Ti Corsair Vengeance LPX DDR4 3000 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 500 GB Seagate Barracuda 7200 RPM 2 TB Asus DVD Drive DRW-24B1ST/BLK/B/AS Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 Pro AOC Q2770PQU Corsair K95 RGB - Cherry MX Blue EVGA Supernova G2 750W 
CaseMouseMouse PadAudio
NZXT Phantom 410 Gunmetal Edition Logitech G502 Proteus Core Razer Goliathus Speed Edition Small Cyber Acoustics Satellite CA-3602 
Audio
Oppo PM-3 Closed Back Planar Magnetic Headphones 
  hide details  
Reply
post #656 of 724
Quote:
Originally Posted by Serandur View Post

You quoted AMD:
And immediately followed by paraphrasing it in agreement;
I pulled nothing out of my mind, it's very blatant. If it's not what you meant, I'm sorry, but that's what it said. If you can't understand why that gives off the meaning it does, then I really can't help here but it does.

Yes, because GDDR5 chips can be scaled to much larger quantities of DRAM than HBM1.

So it makes no sense to hire a driver team to optimize games for GDDR5 when you can just use a larger quantity of DRAM.

And yes you did pull it out of your mind because you assumed something I didn't say. AMD uses HBM differently than they do GDDR5. Why? Because HBM is different than GDDR5. HBM is faster but only scaled to 4GB in its first iteration. So AMD used its bandwidth advantage like an embeded cache.

See I also said all of this in several other postings (some before you started to strawman):
Quote:

As for 4GB NVIDIA cards suffering, yes, they would. They don't have these optimizations. The Hawaii cards I'm not sure about. These are per game optimizations by AMD.

It's about what your graphics driver stores in the framebuffer vs what it stores in the system memory. If the graphics driver stores bandwidth constrained data into your frame buffer and non constrained into the system memory, you won't notice a difference in-game. And you also won't notice a difference when benchmarking. In other words, aside from using more system memory, there is no performance penalty.

So you get the full benefit of HBMs increased memory bandwidth. So what's different about HBM? Only that it is wider and thus offers more bandwidth than GDDR5. And as I stated above, nothing is stopping AMD from using the same optimizations for Hawaii and its 4GB GDDR5.

Let me break this down for you:

1. Drivers control what goes into the framebuffer and what doesn't.
2. Driver optimisations can be used to store bandwidth sensitive data into the frame buffer and non-sensitive data into system memory.
3. HBM can thus be compared to a sort of embeded cache in the way AMDs drivers treat it.
4. The result is thus a 0 performance penalty from using system memory for non bandwidth sensitive data.
5. You get the full performance benefit of using HBM even if you're buffering more data than its capacity would allow.
6. You can do this with GDDR5 memory as well but of course it will only offer you the full benefit of your GDDR5 memory bandwidth.

These all explained what I mean't in greater detail. You conveniently ignored them and clung to your strawman which was your interpretation of what I was saying. What YOU think I implied.
Kn0wledge
(20 items)
 
Pati3nce
(14 items)
 
Wisd0m
(10 items)
 
Reply
Kn0wledge
(20 items)
 
Pati3nce
(14 items)
 
Wisd0m
(10 items)
 
Reply
post #657 of 724
If you can't, you can't see it. There's no reason to continue here. Back to Pascal.
Ice-nine
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K Asus Z170-DELUXE Gigabyte G1 GTX 980 Ti Corsair Vengeance LPX DDR4 3000 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 500 GB Seagate Barracuda 7200 RPM 2 TB Asus DVD Drive DRW-24B1ST/BLK/B/AS Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 Pro AOC Q2770PQU Corsair K95 RGB - Cherry MX Blue EVGA Supernova G2 750W 
CaseMouseMouse PadAudio
NZXT Phantom 410 Gunmetal Edition Logitech G502 Proteus Core Razer Goliathus Speed Edition Small Cyber Acoustics Satellite CA-3602 
Audio
Oppo PM-3 Closed Back Planar Magnetic Headphones 
  hide details  
Reply
Ice-nine
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K Asus Z170-DELUXE Gigabyte G1 GTX 980 Ti Corsair Vengeance LPX DDR4 3000 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 500 GB Seagate Barracuda 7200 RPM 2 TB Asus DVD Drive DRW-24B1ST/BLK/B/AS Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 Pro AOC Q2770PQU Corsair K95 RGB - Cherry MX Blue EVGA Supernova G2 750W 
CaseMouseMouse PadAudio
NZXT Phantom 410 Gunmetal Edition Logitech G502 Proteus Core Razer Goliathus Speed Edition Small Cyber Acoustics Satellite CA-3602 
Audio
Oppo PM-3 Closed Back Planar Magnetic Headphones 
  hide details  
Reply
post #658 of 724
Quote:
Originally Posted by criminal View Post

You are still limited by the PCI-e bus no matter how fast the "cache". That's what he is saying.

And what he is saying is that that limitation isn't important if the only data being stored on the system memory isn't bandwidth sensitive (ie data that must be accessed quickly by the GPU). All I know is that SOMETHING is going on with HBM as compared to GDDR5 when you consider how well Fiji cards do at 4k, a resolution that kills 4GB GDDR5 cards...
post #659 of 724
Quote:
Originally Posted by Majin SSJ Eric View Post


And what he is saying is that that limitation isn't important if the only data being stored on the system memory isn't bandwidth sensitive (ie data that must be accessed quickly by the GPU). All I know is that SOMETHING is going on with HBM as compared to GDDR5 when you consider how well Fiji cards do at 4k, a resolution that kills 4GB GDDR5 cards...

 

Yep. It could very well be that my experience is actually a fever-dream and that in reality my games are sputtering along at unplayable FPS, crushed 'neath the weight of VRAM deficiency. I'm no engineer - chemistry and biology are my stock in trade. I haven't got the time to make a dozen graphs in Excel plotting my usage or to FRAPS videos of my untainted experience. Maybe soon. But purely dealing with the empirical, my Fiji crossfire works wonderfully at 60Hz 4k. I regularly play games at settings which are nominally over my VRAM limit, and it doesn't seem to have a deleterious effect on my playable experience. Perhaps once 4k gets up into the higher refresh rates there will be an issue; I couldn't say with any certainty. For now I'm very happy with 4Gb of HBM.

     
CPUMotherboardGraphicsRAM
AMD Ryzen 7 1700X @ 4.0 GHz 1.37v o/s  Biostar X370 GT7 Sapphire Radeon RX Vega 64 8GB HBM2 G.SKILL TridentZ RGB Series 16GB (2 x 8GB) @ 32... 
Hard DriveHard DriveHard DriveHard Drive
Biostar M200-240 240GB m.2 SATA SSD (Win) Samsung 850 Pro 256GB SSD (Linux) SanDisk Ultra II 480GB SSD (Games) Seagate 2TB SSHD (Storage) 
Optical DriveCoolingCoolingCooling
LG DVD RW Swiftech H220-X2 Swiftech Komodo RX-ECO Vega Water Block XSPC EX240 Rad 
CoolingOSOSMonitor
Gentle Typhoon 120mm x4 Ubuntu 17.10 w/ custom kernel Windows 10 Pro Pixio PX277 27" 1440p 144Hz DP-Adaptive Sync w... 
KeyboardPowerCaseMouse
JAR-OWL J1 Rainbow LED Mechanical SeaSonic X-Series 1050w Gold+ Corsair 760T Zowie FK1 
Mouse PadAudioAudioAudio
Razer Firefly Plantronics Rig 500e Gaming Headset Marantz Pro MPH-1 Monitoring Headphones S.M.S.L SD-793II Audio DAC + Headphone Amplifier 
Other
Xbox One Controller 
CPUMotherboardGraphicsRAM
Intel Core i7-4790k @ 4.6GHz Asrock Z97 Extreme6 EVGA GTX 1070 SC Corsair Vengeance 16GB (2x8GB) @ 2133 10-10-10-... 
Hard DriveHard DriveOptical DriveCooling
Intel 530 Series 480GB SSD Seagate 2TB SSHD LG DVD RW Cooler Master Hyper 212 Plus 
OSMonitorKeyboardPower
Windows 10 40" Vizio HDTV HP Generic EVGA SuperNOVA NEX 650w 
CaseMouseAudio
Corsair Carbide 300R Logitech Generic HyperX Cloud 
CPUMotherboardGraphicsRAM
AMD FX-6300 @ 4.3GHz 1.35v Biostar TA970 v5.3 MSI HD 6950 2GB G.Skill Ripjaws Z 16GB (4x4GB) 1600 
CoolingCoolingCoolingOS
Cryorig H5 Ultimate Corsair AF120 (White LED) x3 Corsair AF140 (White LED) x2 Windows 10 Pro 
MonitorKeyboardPowerCase
Dell S2440L Microsoft Ergonomic 4000 EVGA B2 850w Corsair 300R 
Mouse
SteelSeries Rival 300 
  hide details  
Reply
     
CPUMotherboardGraphicsRAM
AMD Ryzen 7 1700X @ 4.0 GHz 1.37v o/s  Biostar X370 GT7 Sapphire Radeon RX Vega 64 8GB HBM2 G.SKILL TridentZ RGB Series 16GB (2 x 8GB) @ 32... 
Hard DriveHard DriveHard DriveHard Drive
Biostar M200-240 240GB m.2 SATA SSD (Win) Samsung 850 Pro 256GB SSD (Linux) SanDisk Ultra II 480GB SSD (Games) Seagate 2TB SSHD (Storage) 
Optical DriveCoolingCoolingCooling
LG DVD RW Swiftech H220-X2 Swiftech Komodo RX-ECO Vega Water Block XSPC EX240 Rad 
CoolingOSOSMonitor
Gentle Typhoon 120mm x4 Ubuntu 17.10 w/ custom kernel Windows 10 Pro Pixio PX277 27" 1440p 144Hz DP-Adaptive Sync w... 
KeyboardPowerCaseMouse
JAR-OWL J1 Rainbow LED Mechanical SeaSonic X-Series 1050w Gold+ Corsair 760T Zowie FK1 
Mouse PadAudioAudioAudio
Razer Firefly Plantronics Rig 500e Gaming Headset Marantz Pro MPH-1 Monitoring Headphones S.M.S.L SD-793II Audio DAC + Headphone Amplifier 
Other
Xbox One Controller 
CPUMotherboardGraphicsRAM
Intel Core i7-4790k @ 4.6GHz Asrock Z97 Extreme6 EVGA GTX 1070 SC Corsair Vengeance 16GB (2x8GB) @ 2133 10-10-10-... 
Hard DriveHard DriveOptical DriveCooling
Intel 530 Series 480GB SSD Seagate 2TB SSHD LG DVD RW Cooler Master Hyper 212 Plus 
OSMonitorKeyboardPower
Windows 10 40" Vizio HDTV HP Generic EVGA SuperNOVA NEX 650w 
CaseMouseAudio
Corsair Carbide 300R Logitech Generic HyperX Cloud 
CPUMotherboardGraphicsRAM
AMD FX-6300 @ 4.3GHz 1.35v Biostar TA970 v5.3 MSI HD 6950 2GB G.Skill Ripjaws Z 16GB (4x4GB) 1600 
CoolingCoolingCoolingOS
Cryorig H5 Ultimate Corsair AF120 (White LED) x3 Corsair AF140 (White LED) x2 Windows 10 Pro 
MonitorKeyboardPowerCase
Dell S2440L Microsoft Ergonomic 4000 EVGA B2 850w Corsair 300R 
Mouse
SteelSeries Rival 300 
  hide details  
Reply
post #660 of 724
i'm not sure what you guys are even trying to argue about.

the access latency from PCI-e and system RAM is a few hundred nanoseconds, well within a single-frame's render-time.
the bandwidth of DDR4 has already exceeded the 32GB/s of PCI-e 3.0 x16, and with a 32GB/s of bi-directional bandwidth, you can swap a 100MB worth of data in under 4milliseconds.
the less bulky the data is, the faster you can swap VRAM to system RAM, effectively all will be within a single-frame's render-time.

or in other words, the PCI-e and system RAM is fast enough to transact a few hundred MB worth of data that wouldn't heavily affect the GPU's performance.
you guys have seen how AMD's APUs work well enough with a bunch of slow DDR3 RAM right? so why can't the GPUs do the same?
Edited by epic1337 - 3/1/16 at 10:40pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Rumors and Unconfirmed Articles
Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [TT] NVIDIA should launch its next-gen Pascal GPUs with HBM2 in 2H 2016