Overclock.net › Forums › Industry News › Hardware News › [Guru3D]Nvidia Slide reveals numbers on Single and Double precision for Flagship Pascal GPU
New Posts  All Forums:Forum Nav:

[Guru3D]Nvidia Slide reveals numbers on Single and Double precision for Flagship Pascal GPU - Page 16

post #151 of 168
Quote:
Originally Posted by Mahigan View Post

Yes because the 60% number didn't jive with what RTG said. They said 70/30 split.
I'm going to try this with focused responses. Where did RTG say 70% of the 2.5x performance/watt figure came from the process node and 30% came from uarch improvements? This is Ryan Shrout from pcper:

Quote:
Even RTG staff were willing to admit that the move to 14nm FinFET process tech was the majority factor for the improvement we are seeing here, something on the order of a 70/30 split.


The majority factor part came from RTG, but then Shrout adds, following a comma, "something on the order of a 70/30 split" and that's it. We don't know if he got the number from RTG or just gave a rough example/guess, we don't know if he misconstrued something, we don't know how they determined those numbers (against which part of their lineup, which differs in efficiency these days given the mix of three GCN generations).

Basically, we don't even know if RTG actually said that or what it specifically means. Additionally, this was following a discussion of the little Polaris efficiency results RTG demoed against the 950:

Quote:
It is likely that this is the first Polaris GPU being brought up (after only 2 months I’m told) and could represent the best improvement in efficiency that we will see. I’ll be curious to see how flagship GPUs from AMD compare under the same conditions.

How is Polaris able to achieve these types of improvements? It comes from a combination of architectural changes and process technology changes. Even RTG staff were willing to admit that the move to 14nm FinFET process tech was the majority factor for the improvement we are seeing here, something on the order of a 70/30 split.


This GPU is performing on the same level as Pitcairn and will be a replacement for the GCN1-based Pitcairn at best (everything or below this point is now extremely old). Therefore, it will have an accumulation of improvements since Pitcairn. Shrout thinks this could be the absolute peak scenario of Polaris's efficiency (makes sense, given a natural comparison to Pitcairn) and If the 70/30 split is, as proximity within the article suggests, referring to this, then it is not referring to changes over GCN2 or GCN3, but rather the original GCN1 and even then, not all GCN1 but in the lower-end range designed for efficiency.

Among those improvements include a lot of power gating ones and given the small size of such a GPU and the additional space afforded by 14nm, AMD may also be counting going for a somewhat wider and lower-clocked design to increase efficiency as non-14nm improvements in their figure, assuming the figure actually did come from them.


It doesn't have to refer to across-the-board uarch changes from GCN3 (or any GCN iteration) even if it did come from RTG, but instead could refer to this very specific and specialized low-power design also meant to replace GCN1 (maybe even in comparison to higher-clocked Pitcairns as well, which further reduces baseline efficiency).



And then all this could be a moot point because the 70/30 split could just be something Ryan Shrout (or an RTG employee for that matter) made up. Maybe even just something RTG made up based on the amount of changes they've made to the chip design and not indicative of actual perf/watt gains, let alone outright performance. Hence why I'd say it's a vague figure unless properly detailed and actually documented as coming from an RTG employee.

Quote:
Originally Posted by Mahigan View Post

Also it is up to 60%, this varies on clock speed as an increase in clock speed will reduce that figure.

Assuming the architecture, and individual chip specs remain the same and the new process node alone is actually sufficient to increase clock speeds by any substantial amount then yes, but it's pretty naive to think (with marketing materials being extremely vague on actual performance increases and instead hyping up perf/watt) that AMD aren't going to have lower-clocked/optimally-efficient parts and that they won't base their perf/watt claims based on their best perf/watt scenario (the aforementioned lower-clocked parts).

Only an incredibly foolish company wouldn't advertise using figures from their better products (for any particular metric, that is). I highly doubt, despite AMD themselves mentioning 60% reductions as a possibility and hyping up perf/watt, they are basing their numbers on a significantly less-efficient, higher-clocked performance chip instead.
Edited by Serandur - 2/24/16 at 2:59pm
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 #152 of 168
Quote:
Originally Posted by zalbard View Post

FLOPS indicate theoretical best-case performance and mean very little.

Fury X has much higher SP performance than GTX 980 Ti (8.60 TFLOPS vs 5.63 TFLOPS) yet significantly trails the latter in games.
Because it's being hampered by DX11;



http://www.anandtech.com/show/10067/ashes-of-the-singularity-revisited-beta/5
post #153 of 168
Quote:
Originally Posted by NightAntilli View Post

Because it's being hampered by DX11;



http://www.anandtech.com/show/10067/ashes-of-the-singularity-revisited-beta/5

Anyone else find it odd that the gtx 680 gained 21% while the 780 Ti went negative despite both being Kepler?
Upstairs Rig
(11 items)
 
  
CPUMotherboardGraphicsRAM
4770k Asus Maximus VI Hero evga 1080 Ti with Hybrid mod Corsair Vengeance Pro 2133 mhz 
Hard DriveHard DriveCoolingOS
Samsung 850 EVO 500gb WD Caviar Black Corsair h100i GTX Windows 8.1 64bit 
MonitorPowerCase
xb280hk EVGA Supernova 1000 G2 Corsair Carbide Air 540 
  hide details  
Reply
Upstairs Rig
(11 items)
 
  
CPUMotherboardGraphicsRAM
4770k Asus Maximus VI Hero evga 1080 Ti with Hybrid mod Corsair Vengeance Pro 2133 mhz 
Hard DriveHard DriveCoolingOS
Samsung 850 EVO 500gb WD Caviar Black Corsair h100i GTX Windows 8.1 64bit 
MonitorPowerCase
xb280hk EVGA Supernova 1000 G2 Corsair Carbide Air 540 
  hide details  
Reply
post #154 of 168
Quote:
Originally Posted by mcg75 View Post

Anyone else find it odd that the gtx 680 gained 21% while the 780 Ti went negative despite both being Kepler?

Everyone, including the author of the article.
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
post #155 of 168
Imagine if the GTX 680 can accidentally do async compute xD LOL

Or... Has nVidia hampered the GTX 600 series performance through software when the 700 series was released....?
Edited by NightAntilli - 2/24/16 at 3:37pm
post #156 of 168
Quote:
Originally Posted by criminal View Post

Yep. I typically want performance now when I purchase a gpu, but the 7970 and 290/290x have changed my tune a bit. But I still need for the upgrade to be worth the price. Again, that's why I was disappointed with Fury on release. The Fury wasn't any faster than a 980 and was $50 premium.

that is totally understandable.

Polaris/Pascal is going to be very hard to decide what route to go. I am really damn excited about those GPUs and I will basically do as much "research" as I can do when I am ready to make the purchase.

Let's wait and see who gets my money this round smile.gif
current rig
(4 items)
 
  
CPUGraphicsGraphicsKeyboard
2500k R9 290X Tri-X GTX 680 Filco Majestouch 2 Brown Ninja 
  hide details  
Reply
current rig
(4 items)
 
  
CPUGraphicsGraphicsKeyboard
2500k R9 290X Tri-X GTX 680 Filco Majestouch 2 Brown Ninja 
  hide details  
Reply
post #157 of 168
OMG will have the same prices and perf. Both GP104 and amd's equivalent.
PC
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel® Core™ i7 3770K @4.5Ghz (Multiplier OC) 1... MSI Z77A-GD65 [10.10 BIOS] MSi GTX 980 GAMING + MX4 @1544mhz/1.262v; +450M... Geil Black Dragon 4x4Gb 1866Mhz [XMP] 
Hard DriveCoolingOSMonitor
Samsung 850 EVO 500Gb [AHCI] Noctua NH-D14 Windows 10 Home 64-Bit 10586.420 (FPP/retail key) BenQ XL2411T [Dual-Link DVI, 144Hz] AUO M24HW01... 
KeyboardPowerCaseMouse
Razer BlackWidow Ultimate 2014 [green switches] Enermax Revolution 87+ 750W and APC Back-UPS 950VA Phanteks Evolv Atx Tempered Glass Black + ATX 2... Razer Mamba 2015 Tournament Edition 
Mouse PadAudio
SteelSeries QcK Mini Asus Xonar DGX 
  hide details  
Reply
PC
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel® Core™ i7 3770K @4.5Ghz (Multiplier OC) 1... MSI Z77A-GD65 [10.10 BIOS] MSi GTX 980 GAMING + MX4 @1544mhz/1.262v; +450M... Geil Black Dragon 4x4Gb 1866Mhz [XMP] 
Hard DriveCoolingOSMonitor
Samsung 850 EVO 500Gb [AHCI] Noctua NH-D14 Windows 10 Home 64-Bit 10586.420 (FPP/retail key) BenQ XL2411T [Dual-Link DVI, 144Hz] AUO M24HW01... 
KeyboardPowerCaseMouse
Razer BlackWidow Ultimate 2014 [green switches] Enermax Revolution 87+ 750W and APC Back-UPS 950VA Phanteks Evolv Atx Tempered Glass Black + ATX 2... Razer Mamba 2015 Tournament Edition 
Mouse PadAudio
SteelSeries QcK Mini Asus Xonar DGX 
  hide details  
Reply
post #158 of 168
Quote:
Originally Posted by CalinTM View Post

OMG will have the same prices and perf. Both GP104 and amd's equivalent.

Sorry I don't understand your post. Could you elaborate?
current rig
(4 items)
 
  
CPUGraphicsGraphicsKeyboard
2500k R9 290X Tri-X GTX 680 Filco Majestouch 2 Brown Ninja 
  hide details  
Reply
current rig
(4 items)
 
  
CPUGraphicsGraphicsKeyboard
2500k R9 290X Tri-X GTX 680 Filco Majestouch 2 Brown Ninja 
  hide details  
Reply
post #159 of 168
Like, fury x and 980 has the aprox. same performance, and same debut prices. Like in the past as well. So, its not like amd or nvidia will crush one or another.
PC
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel® Core™ i7 3770K @4.5Ghz (Multiplier OC) 1... MSI Z77A-GD65 [10.10 BIOS] MSi GTX 980 GAMING + MX4 @1544mhz/1.262v; +450M... Geil Black Dragon 4x4Gb 1866Mhz [XMP] 
Hard DriveCoolingOSMonitor
Samsung 850 EVO 500Gb [AHCI] Noctua NH-D14 Windows 10 Home 64-Bit 10586.420 (FPP/retail key) BenQ XL2411T [Dual-Link DVI, 144Hz] AUO M24HW01... 
KeyboardPowerCaseMouse
Razer BlackWidow Ultimate 2014 [green switches] Enermax Revolution 87+ 750W and APC Back-UPS 950VA Phanteks Evolv Atx Tempered Glass Black + ATX 2... Razer Mamba 2015 Tournament Edition 
Mouse PadAudio
SteelSeries QcK Mini Asus Xonar DGX 
  hide details  
Reply
PC
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel® Core™ i7 3770K @4.5Ghz (Multiplier OC) 1... MSI Z77A-GD65 [10.10 BIOS] MSi GTX 980 GAMING + MX4 @1544mhz/1.262v; +450M... Geil Black Dragon 4x4Gb 1866Mhz [XMP] 
Hard DriveCoolingOSMonitor
Samsung 850 EVO 500Gb [AHCI] Noctua NH-D14 Windows 10 Home 64-Bit 10586.420 (FPP/retail key) BenQ XL2411T [Dual-Link DVI, 144Hz] AUO M24HW01... 
KeyboardPowerCaseMouse
Razer BlackWidow Ultimate 2014 [green switches] Enermax Revolution 87+ 750W and APC Back-UPS 950VA Phanteks Evolv Atx Tempered Glass Black + ATX 2... Razer Mamba 2015 Tournament Edition 
Mouse PadAudio
SteelSeries QcK Mini Asus Xonar DGX 
  hide details  
Reply
post #160 of 168
Quote:
Originally Posted by CalinTM View Post

Like, fury x and 980 has the aprox. same performance, and same debut prices. Like in the past as well. So, its not like amd or nvidia will crush one or another.

AMD can't afford to have the same price as Nvidia (if everything is basically the same: performance, noise, heat, power draw, VRAM, size etc.)

I am pretty sure there will be differences smile.gif
current rig
(4 items)
 
  
CPUGraphicsGraphicsKeyboard
2500k R9 290X Tri-X GTX 680 Filco Majestouch 2 Brown Ninja 
  hide details  
Reply
current rig
(4 items)
 
  
CPUGraphicsGraphicsKeyboard
2500k R9 290X Tri-X GTX 680 Filco Majestouch 2 Brown Ninja 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [Guru3D]Nvidia Slide reveals numbers on Single and Double precision for Flagship Pascal GPU