Overclock.net › Forums › Graphics Cards › NVIDIA › [AdoredTV] History of Nvidia GeForce
New Posts  All Forums:Forum Nav:

[AdoredTV] History of Nvidia GeForce - Page 13

post #121 of 140
Quote:
Originally Posted by 12Cores View Post

I think the main thing he pointed out here is that buyers need to look at die size(GTX 560 - 332mm vs GTX 1080 - 314mm) to get good gauge of how much they are truly getting for their money. The 1080 is an incredible card, but its clear that the price would have been a lot lower if there was some competition. Lets hope Ryzen can bring amd back profitability and we can get some competition. My 7970 is still the best card I have ever owned period, my crossfire 7970 build destroyed everything I threw at it at 2880x1620.
Quote:
Originally Posted by Adored View Post

Err, no this wasn't "the guy" who did anything of the sort, link to wherever this bs was said?

You want a harsh AMD video check out the previous one "Something Wrong at Radeon".


Laird AdoredTV,

I always work out how much I'm paying per billion transistors where as you focus on die size. I'm not a hardware man myself as such but a piece of silicon is just a bit of fancy sand with the transistors you fab a huge die the size of a haggis but it when they try to put the transistors in there that's when it becomes tricky.

For me anyway
2012
Radeon HD 7950 224.54GBP 4.31 billion transistors on 352mm2 die 52GBP per 1 biliions transistors or 0.63GBP per 1 mm2

2016
GTX 1070 399,95GBP 7.2 billion transistors on 314mm2 die 55.56GBP per 1 billion transistors or 1,27GBP per 1 mm2


So for me cost of office floor space has almost doubled (199%) in price but the signing on fee/golden hello to be paid for the transistor has only gone up 6% this has to be offset though by the power usage of the GTX 1070 which Í see as the weekly wage bill which has down. In my mind I see it like a three sided triangle akin to a football club 1 point transfer fee (die size), 2nd point player signing-on/agent fee (cost per transistor), 3rd point player salary & bonus (perf/WATT) = total outlay


Anyways you got to watch out for these nVidia Biatches & Huang Hoes spreading FUD claiming 100 TFLOPS in Volta V100. I put the smack-down on them http://www.overclock.net/t/1638396/tt-nvidia-tesla-v100-tested-near-unbelievable-gpu-power/100 as Ali G said keep it real...


PS

Vega is BEST PC card. Why? PC is a general purpose machine turns it's hand to everything and does it well. Like Range Rover Sport good on the road, fast as you like, air suspension lift it up, all wheel drive take it off-road. You never know when you need it, it's like PC otherwise buy console. Console low-level OS, 1 hardware specification means you get higher performance per GBP. Range Rover Sport is expensive but it can handle all things with a high level of performance, style, luxury and pedigree the other guys just can't match. So in synopsis you need revise your Vega conclusion to RTG Vega = PC GPU nVidia GTX = console GPU

Thanks
Laird Gill
Edited by HaiderGill - 9/21/17 at 3:22am
post #122 of 140
Quote:
Originally Posted by HaiderGill View Post


Laird AdoredTV,

I always work out how much I'm paying per billion transistors where as you focus on die size. I'm not a hardware man myself as such but a piece of silicon is just a bit of fancy sand with the transistors you fab a huge die the size of a haggis but it when they try to put the transistors in there that's when it becomes tricky.

For me anyway
2012
Radeon HD 7950 224.54GBP 4.31 billion transistors on 352mm2 die 52GBP per 1 biliions transistors or 0.63GBP per 1 mm2

2016
GTX 1070 399,95GBP 7.2 billion transistors on 314mm2 die 55.56GBP per 1 billion transistors or 1,27GBP per 1 mm2


So for me cost of office floor space has almost doubled (199%) in price but the signing on fee/golden hello to be paid for the transistor has only gone up 6% this has to be offset though by the power usage of the GTX 1070 which Í see as the weekly wage bill which has down. In my mind I see it like a three sided triangle akin to a football club 1 point transfer fee (die size), 2nd point player signing-on/agent fee (cost per transistor), 3rd point player salary & bonus (perf/WATT) = total outlay


Anyways you got to watch out for these nVidia Biatches & Huang Hoes spreading FUD claiming 100 TFLOPS in Volta V100. I put the smack-down on them http://www.overclock.net/t/1638396/tt-nvidia-tesla-v100-tested-near-unbelievable-gpu-power/100 as Ali G said keep it real...


PS

Vega is BEST PC card. Why? PC is a general purpose machine turns it's hand to everything and does it well. Like Range Rover Sport good on the road, fast as you like, air suspension lift it up, all wheel drive take it off-road. You never know when you need it, it's like PC otherwise buy console. Console low-level OS, 1 hardware specification means you get higher performance per GBP. Range Rover Sport is expensive but it can handle all things with a high level of performance, style, luxury and pedigree the other guys just can't match. So in synopsis you need revise your Vega conclusion to RTG Vega = PC GPU nVidia GTX = console GPU

Thanks
Laird Gill

Nvidia GTX = console GPU. Um, aren't all consoles actually using AMD parts?

Which company has the performance crown in PC gaming and has had it for three years running?
post #123 of 140
Quote:
Originally Posted by 8051 View Post

Nvidia GTX = console GPU. Um, aren't all consoles actually using AMD parts?

Which company has the performance crown in PC gaming and has had it for three years running?

By that I mean nVidia consumer level GPUs are only good for games not so good at anything else, it's like a young Ronaldo - just a one trick pony, no end product. AMD GPU is flexible can used for GPGPU.
Edited by HaiderGill - 9/24/17 at 1:47pm
post #124 of 140
Quote:
Originally Posted by HaiderGill View Post

By that I mean nVidia consumer level GPUs are only good for games not so good at anything else, it's like a young Ronaldo - just a one trick pony, no end product. AMD GPU is flexible can used for GPGPU.

Nvidia is the go-to choice for high performance computing. Something like two out of ten of the highest performance computing clusters in the world are using Nvidia GPU's.
post #125 of 140
Quote:
Originally Posted by HaiderGill View Post

By that I mean nVidia consumer level GPUs are only good for games not so good at anything else, it's like a young Ronaldo - just a one trick pony, no end product. AMD GPU is flexible can used for GPGPU.

What more do you want a GPU to do?

Technically the way nVidia went ahead and stripped their consumer GPU's of compute has actually made it better for us consumers as gaming cards vs what AMD doing. And I wouldn't doubt that when AMD does have the funding we'll also see a shift where they will strip compute abilities from consumer line and create what nVidia has. Consumer > ProSumer > Professional.
average
(11 items)
 
  
CPUMotherboardGraphicsRAM
i7-4790K Asus Z97 TUF Gryphon Sapphire Radeon RX Vega 64 G.Skill TridentX 
Hard DriveHard DriveCoolingOS
Samsung 850 EVO Samsung 850 EVO NZXT Kraken X61 Windows 10 
MonitorPowerCase
LG 24GM77 Corsair AX1200i Phanteks Enthoo EVOLV ATX 
  hide details  
Reply
average
(11 items)
 
  
CPUMotherboardGraphicsRAM
i7-4790K Asus Z97 TUF Gryphon Sapphire Radeon RX Vega 64 G.Skill TridentX 
Hard DriveHard DriveCoolingOS
Samsung 850 EVO Samsung 850 EVO NZXT Kraken X61 Windows 10 
MonitorPowerCase
LG 24GM77 Corsair AX1200i Phanteks Enthoo EVOLV ATX 
  hide details  
Reply
post #126 of 140
Many things, it like having a 4x4 with off-road pedigree. I don't plan to go off-roading but it's nice to know if I ever need to I have the ability, it's why I have a PC instead of a console.
post #127 of 140
Quote:
Originally Posted by HaiderGill View Post

Many things, it like having a 4x4 with off-road pedigree. I don't plan to go off-roading but it's nice to know if I ever need to I have the ability, it's why I have a PC instead of a console.

Nvidia GTX GPU's can still do compute. Double precision is limited to 1/32. If Nvidia didn't split off most of the compute ability they would also have a huge inefficient die that doesn't do anything for gaming, which is the main function of GTX GPU's.
Boinc Desktop
(14 items)
 
CrunchAholic
(8 items)
 
 
CPUMotherboardGraphicsRAM
AMD Ryzen 7 1700X ASRock X370 Taichi Gigabyte GTX 970 G1 HyperX Predator 
Hard DriveCoolingCoolingCooling
Hyper Predator M.2 Watercool Heatkiller IV PRO AM4 CPU Block EKWB GTX 970 GPU Block EKWB Coolstream XE 240 Radiator 
CoolingCoolingOSPower
EKWB Coolstream PE 360 Radiator Watercool Heatkiller 150mm Tube Res Gentoo Linux EVGA 850 G2 
CaseMouse
Thermaltake View 31 Tempered Glass RGB Edition Roccat Kone EMP 
CPUCPUMotherboardGraphics
Intel Xeon E5-2670 Intel Xeon E5-2670 Asrock Rack EP2C602 XFX RX 480 RS 
RAMHard DriveOSPower
64 GBs Samsung M939B1K70CHD-CH9 PC3-10600R Samsung HD322HJ Arch Linux Corsair AX1200 
  hide details  
Reply
Boinc Desktop
(14 items)
 
CrunchAholic
(8 items)
 
 
CPUMotherboardGraphicsRAM
AMD Ryzen 7 1700X ASRock X370 Taichi Gigabyte GTX 970 G1 HyperX Predator 
Hard DriveCoolingCoolingCooling
Hyper Predator M.2 Watercool Heatkiller IV PRO AM4 CPU Block EKWB GTX 970 GPU Block EKWB Coolstream XE 240 Radiator 
CoolingCoolingOSPower
EKWB Coolstream PE 360 Radiator Watercool Heatkiller 150mm Tube Res Gentoo Linux EVGA 850 G2 
CaseMouse
Thermaltake View 31 Tempered Glass RGB Edition Roccat Kone EMP 
CPUCPUMotherboardGraphics
Intel Xeon E5-2670 Intel Xeon E5-2670 Asrock Rack EP2C602 XFX RX 480 RS 
RAMHard DriveOSPower
64 GBs Samsung M939B1K70CHD-CH9 PC3-10600R Samsung HD322HJ Arch Linux Corsair AX1200 
  hide details  
Reply
post #128 of 140
The thing is that AMD seems to want to push games in the direction of being mostly compute. It is possible, but, developers don't really have that mindset yet. Things are still done mostly the traditional way, which is through the graphics pipeline, with minor compute.
post #129 of 140
Quote:
Originally Posted by Rei86 View Post

What more do you want a GPU to do?

Technically the way nVidia went ahead and stripped their consumer GPU's of compute has actually made it better for us consumers as gaming cards vs what AMD doing. And I wouldn't doubt that when AMD does have the funding we'll also see a shift where they will strip compute abilities from consumer line and create what nVidia has. Consumer > ProSumer > Professional.

What more? Everything, it's for IBM compatible Personal Computers. If it was in a games console then fine. I bought computer to handle games, run Virtualbox Redhat VMs/Docker, Oracle RDBMS and a few IDEs, children use it with graphics tablet to do drawings, word processing etc..., I also play games on it. PC = Flexibility, games console = games only...
post #130 of 140
Quote:
Originally Posted by Diffident View Post

Nvidia GTX GPU's can still do compute. Double precision is limited to 1/32. If Nvidia didn't split off most of the compute ability they would also have a huge inefficient die that doesn't do anything for gaming, which is the main function of GTX GPU's.

Card TFLOPS Transistors mm2 Lith Nm Power TFLOPS/W
Vega56 10,50 12,50 486,00 14,00 210,00 0,05
GTX1070 6,50 7,20 314,00 16,00 150,00 0,04

1.) Vega56 is more efficient by an extra 0.01 TFLOP per watt.

2.) GTX 1070 chip cheaper to manufacture. Looking at the GTX1070 chip 7.2B transistors on 314mm means it's cheaper to manufacture, so should be cheaper than Vega56. I can tell you my Inno3D GTX1070 Airboss X3 costs around GBP420 bought it about a year ago at GBP400. GTX 1070 Strix GBP470/Strix OC GBP490. Wait for the Vega56 Strix cards to come out but looking at 6.5 to 10TFLOP is 62 percent more so a Vega56 Strix comparable card from nVidia would be GBP900.

3.) It easy to consume less when you put less out my 140i does 0-60 mph in under 4.5s, kidsmobile does 0-60 8s, by that time in closing in on 100mph OBVIOUSLY kidsmobile is going to consume less fuel. Like i tell my wife when she's amazed with 50+mpg the 320d returns if she get the 320d kidsmobile to hit a 100mph in 8s whilst doing 50+mpg. Can't say GTX 1070 consumes less power, it consumes less power because it is only putting out just above half the grunt of Vega56.

Look forward to your reply in earnest...
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › [AdoredTV] History of Nvidia GeForce