Overclock.net › Forums › Industry News › Hardware News › [twitter] Titan V revealed.
New Posts  All Forums:Forum Nav:

[twitter] Titan V revealed. - Page 64

post #631 of 1303
Quote:
Originally Posted by Majin SSJ Eric View Post

How can you be "disappointed" with a performance increase of 15% in gaming over the Titan Xp? Obviously the thing is expensive as hell, but when you are talking about the unquestioned king of performance for a video card the price has always been out of whack. This is not the new gaming Titan on Volta, it is a repurposed workstation card. The fact remains that for the guy who has an unlimited wallet and simply wants the fastest single video card money can buy, the Titan V is the only option.

Its weird to be on OCN and see people disappointed with a card that is significantly (if not overly) faster than even the Titan Xp.

Oh majin you are here, i rmb how badly i was bashed by you for insisting titan volta will release this hear!

Look like this time i won the bet lol.
Project Frostbite
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core I7-6800K OC 4.2GHz @ 1.28 Vcore ASUS X99 Sabertooth Gigabyte Aorus GTX 1080 Ti 11G Corsair Vengeance White LED 4x8GB DDR4-3200MHz 
Hard DriveCoolingOSMonitor
Samsung Evo 850 1TB Corsair H100i V2 AiO Window 10 64-Bits Home Premium Acer Predator XB271HU 
KeyboardPowerCaseMouse
Tesoro Spectrum Mechanical Keyboard (Blue Switc... Thermaltake Toughpower 1000W Gold Phantek Enthoo Pro Full Acrylic Black Razer Diamondback 3G 
AudioAudioAudio
Audio Engine A2+ Speaker Audio Engine D1 DAC Audio Engine S8 Subwoofer 
  hide details  
Reply
Project Frostbite
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core I7-6800K OC 4.2GHz @ 1.28 Vcore ASUS X99 Sabertooth Gigabyte Aorus GTX 1080 Ti 11G Corsair Vengeance White LED 4x8GB DDR4-3200MHz 
Hard DriveCoolingOSMonitor
Samsung Evo 850 1TB Corsair H100i V2 AiO Window 10 64-Bits Home Premium Acer Predator XB271HU 
KeyboardPowerCaseMouse
Tesoro Spectrum Mechanical Keyboard (Blue Switc... Thermaltake Toughpower 1000W Gold Phantek Enthoo Pro Full Acrylic Black Razer Diamondback 3G 
AudioAudioAudio
Audio Engine A2+ Speaker Audio Engine D1 DAC Audio Engine S8 Subwoofer 
  hide details  
Reply
post #632 of 1303
Quote:
Originally Posted by unseen0 View Post

This might be the first time a GPU is actually capable of running 4k decently.

Haven't you heard? You need 8K now.
post #633 of 1303
Quote:
Originally Posted by superstition222 View Post

Haven't you heard? You need 8K now.

8K won't be worth it unless its on oled with low mouse lag hopefully 2019 tv's will fix it hdmi 2.1
post #634 of 1303
Quote:
Originally Posted by gamingarena View Post



Well this definitely is not "gaming" card specially for $3k USD you have to be out your mind to spend this kind of money for this underwhelming performance, 2x Titan Xp's will blow Titan V out of the water for much less money



This...is kind what some people in this thread have been saying already, yet some have already bought the damn thing. You know, because a Titan Xp (or several) isn't enough for "gaming" lol
post #635 of 1303
Quote:
Originally Posted by Jpmboy View Post

me biggrin.gif

*Cough* Screw loose

biggrin.gif

Quote:
Originally Posted by tajoh111 View Post

Yup. I got a gtx 1080 regular for about 550 cad which is equal to about 420-440usd in november of 2016. It was a good deal then and even now.
2050mhz is hardly a modest clock on a titan XP, that 15.74 tflops. Significantly more than Titan V.

Also all his results beside superposition and firestrike are using stock settings. Look under the part where he is referring to Deus Ex he says everything is stock for games and the rest of the benchmarks.

Compared to a stock gtx 1080 ti, it appears anywhere from 20-30% faster in games which is not impressive particularly, but it does show two things. Very modest IPC gains and the odd change in the architecture. The IPC benchmark I have used is always tflop vs game performance,

This card in some games has a tflop per performance slightly less than a gtx 1080 ti. At times it is a bit more. Considering how scaled up this card already is vs pascal and the bottleneck effect of the removed cache, ROPs and memory bandwidth, simply matching the tflop to performance of a gtx 1080 ti is already showing IPC gains.

Look at fiji/Vega for example of what happens when IPC are not achieved but clocks and components are scaled up. This is how could predict Vega, was going to be mediocre for gaming. This still scaling relatively linearly this early into drivers while being so scaled up is a decent but not great show.

I wouldn't say Titan V's results are good, but they do show promise considering Nvidia's architecture has been very software dependent on performance.

If nvidia can extend the pipeline to get clocks up another 200mhz by removing tensor and fp64, add in 128 rops vs the 96 here, and get drivers performance up(5-10%), this could extend this lead to close to 50% stock vs stock vs atleast a gtx 1080 ti. Take 1.25x current baseline vs gtx 1080ti * 1.13(clocks)*1.05(ROP bottleneck)*1.07(drivers) = 1.58x 1080 ti.


2050Mhz is an average, I'd say? Anything beyond 2088 is good with unconditional stability.
post #636 of 1303
Quote:
Originally Posted by epic1337 View Post

Warning: Spoiler! (Click to show)
Quote:
Originally Posted by looniam View Post

Quote:
Originally Posted by epic1337 View Post


no, i'm saying that their intentions on Titan V is half-baked, they gave it full functionality of Volta but left it with minimal support, on the other hand they had left the Quadro empty without any Volta cards in place.
you had mentioned that those researchers wouldn't be able to afford a DGX-1 but a Quadro would be cheap enough to fill that gap, Titan V on the other hand is way too "cheap" for that purpose.

hence we come to this point, the target of Titan V isn't those existing researchers or big companies but those of the new people, simply put they're like the Indie devs of the gaming industry.
and obviously these new people aren't so well versed in Volta's architecture to get by without professional support, nor would the community help out due to intense competition in a growing market.
yeah even industry experts are confused: rolleyes.gif
http://fortune.com/2017/12/08/nvidia-ai-titan-v/?
Quote:
Chip industry analyst Patrick Moorhead, president of Moor Insights & Strategy, said the Titan V offered an “impressive value proposition,” as in some ways it could keep up with the performance of Nvidia’s highest-end Tesla V100 card that sells for $10,000 and is aimed at server computers, especially in cloud data centers. Both cards include “Volta” chips and crunch AI calculations at almost the same rate. But the more expensive card has the capability to exchange data with other, surrounding chips at a much faster rate, making it a better choice for servers that often contain four or more graphics cards.

That’s a pretty clear market distinction, according to Moorhead. The Titan V could go in a high-powered PC used by an AI scientist on a local project, while the Tesla V100 will be good for servers that handle the immense app workloads that power the voice recognition abilities of digital assistants from Google or Amazon, for example. “I don’t expect a lot of cannibalization as Tesla is more for production workloads and Titan V is more for research,” Moorhead said.

so you know better than nvidia, an industry analyst and the press as you sit there, in your bubble at your keyboard typing utter and complete garbage?

since the topic is about research, why don't you try it sometime?

maybe you'll stop assuming you know so much.

we have the Quadro market segment for that, it practically invalidates his point.
like i said, those guys have enough capital to back their research, without a doubt can afford a Quadro, the ones Titan V are targeting are the ones new to the market.

furthermore you forget that Volta is a recent architecture, TensorRT 3 and CuDNN 7 isn't even thoroughly supported yet.
https://github.com/tensorflow/tensorflow/issues/14126

and now you expect them to do their research blind-sidedly without technical support from Nvidia?

you didn't do ANY research that i suggested when all you had to do was go to twitter when it was released:

https://twitter.com/nvidia/status/939200818137329664

and btw, the announce was made at the NIPS conferce a conference on neural information processing systems; where everybody who is anybody and many that are nobody meet in the AI filed like google, facebook . .elon musk and nvidia.

that tweeted link sends you to nvidia's IA site and if anyone can put forth the effort to scroll down and look at the FAQ . . . Warning: Spoiler! (Click to show)

i've said in most every post that nvidia will not release "the most powerful desktop gpu in the world" and then abandon it like an illegitimate child at someone's door step. that's just COMMON SENSE! that many seem to be missing. hopefully AI research can help them.(sponsored by NV)

and the second glaring mistake is not looking at who are the (potential) buyers; do look at some of those folks resumes. most of them lectured coding/programming that gave graduate CI/CS students nightmares and few/some wrote that code themselves. freaking geniuses in their field(s); about the only help they would need from nividia would be to confirm their observations.

also, stop comparing this to the quadro, though i have also made that mistake myself, the tesla product(s) is/are much more relevant and is used in AI, not quadros.

so yeah . .all these concerns you continue to express are from your own lack of research pal; pretty much just trolling or not very intelligent.

your choice, good luck with that. smile.gif
loon 3.2
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7-3770K Asus P8Z77-V Pro EVGA 980TI SC+ 16Gb PNY ddr3 1866 
Hard DriveHard DriveHard DriveOptical Drive
PNY 1311 240Gb 1 TB Seagate 3 TB WD Blue DVD DVDRW+/- 
CoolingCoolingOSMonitor
EKWB P280 kit EK-VGA supremacy Win X LG 24MC57HQ-P 
KeyboardPowerCaseMouse
Ducky Zero [blues] EVGA SuperNova 750 G2 Stryker M [hammered and drilled] corsair M65 
AudioAudio
SB Recon3D Klipsch ProMedia 2.1  
  hide details  
Reply
loon 3.2
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7-3770K Asus P8Z77-V Pro EVGA 980TI SC+ 16Gb PNY ddr3 1866 
Hard DriveHard DriveHard DriveOptical Drive
PNY 1311 240Gb 1 TB Seagate 3 TB WD Blue DVD DVDRW+/- 
CoolingCoolingOSMonitor
EKWB P280 kit EK-VGA supremacy Win X LG 24MC57HQ-P 
KeyboardPowerCaseMouse
Ducky Zero [blues] EVGA SuperNova 750 G2 Stryker M [hammered and drilled] corsair M65 
AudioAudio
SB Recon3D Klipsch ProMedia 2.1  
  hide details  
Reply
post #637 of 1303
Could you explain what the benefit of that would be to the user? Seems from their ToS, Nvidia has license to any results of using it.
Quote:
You hold all rights, title and interest in and to your modifications and derivative works of the NVIDIA sample and example source code software that you create as permitted hereunder (collective, “Derivatives”), subject to NVIDIA’s underlying intellectual property rights in and to the NVIDIA sample or example source code; provided, however that you grant NVIDIA and its affiliates an irrevocable, perpetual, nonexclusive, worldwide license to make, have made, use, have used, reproduce, license, distribute, sublicense, transfer and otherwise commercialize Derivatives without any obligation or compensation to you.
post #638 of 1303
Quote:
Originally Posted by guttheslayer View Post

Oh majin you are here, i rmb how badly i was bashed by you for insisting titan volta will release this hear!

Look like this time i won the bet lol.

Im surprised he hasnt told everyone "again" how his OG titans are still more than enough for any game to date.... he always loves to reassure himself that his trash gpu's are still good.... lmbo.
post #639 of 1303
Quote:
Originally Posted by looniam View Post

Quote:
Originally Posted by epic1337 View Post

Warning: Spoiler! (Click to show)
Quote:
Originally Posted by looniam View Post

Quote:
Originally Posted by epic1337 View Post


no, i'm saying that their intentions on Titan V is half-baked, they gave it full functionality of Volta but left it with minimal support, on the other hand they had left the Quadro empty without any Volta cards in place.
you had mentioned that those researchers wouldn't be able to afford a DGX-1 but a Quadro would be cheap enough to fill that gap, Titan V on the other hand is way too "cheap" for that purpose.

hence we come to this point, the target of Titan V isn't those existing researchers or big companies but those of the new people, simply put they're like the Indie devs of the gaming industry.
and obviously these new people aren't so well versed in Volta's architecture to get by without professional support, nor would the community help out due to intense competition in a growing market.
yeah even industry experts are confused: rolleyes.gif
http://fortune.com/2017/12/08/nvidia-ai-titan-v/?
Quote:
Chip industry analyst Patrick Moorhead, president of Moor Insights & Strategy, said the Titan V offered an “impressive value proposition,” as in some ways it could keep up with the performance of Nvidia’s highest-end Tesla V100 card that sells for $10,000 and is aimed at server computers, especially in cloud data centers. Both cards include “Volta” chips and crunch AI calculations at almost the same rate. But the more expensive card has the capability to exchange data with other, surrounding chips at a much faster rate, making it a better choice for servers that often contain four or more graphics cards.

That’s a pretty clear market distinction, according to Moorhead. The Titan V could go in a high-powered PC used by an AI scientist on a local project, while the Tesla V100 will be good for servers that handle the immense app workloads that power the voice recognition abilities of digital assistants from Google or Amazon, for example. “I don’t expect a lot of cannibalization as Tesla is more for production workloads and Titan V is more for research,” Moorhead said.

so you know better than nvidia, an industry analyst and the press as you sit there, in your bubble at your keyboard typing utter and complete garbage?

since the topic is about research, why don't you try it sometime?

maybe you'll stop assuming you know so much.

we have the Quadro market segment for that, it practically invalidates his point.
like i said, those guys have enough capital to back their research, without a doubt can afford a Quadro, the ones Titan V are targeting are the ones new to the market.

furthermore you forget that Volta is a recent architecture, TensorRT 3 and CuDNN 7 isn't even thoroughly supported yet.
https://github.com/tensorflow/tensorflow/issues/14126

and now you expect them to do their research blind-sidedly without technical support from Nvidia?

you didn't do ANY research that i suggested when all you had to do was go to twitter when it was released:

https://twitter.com/nvidia/status/939200818137329664

and btw, the announce was made at the NIPS conferce a conference on neural information processing systems; where everybody who is anybody and many that are nobody meet in the AI filed like google, facebook . .elon musk and nvidia.

that tweeted link sends you to nvidia's IA site and if anyone can put forth the effort to scroll down and look at the FAQ . . . Warning: Spoiler! (Click to show)

i've said in most every post that nvidia will not release "the most powerful desktop gpu in the world" and then abandon it like an illegitimate child at someone's door step. that's just COMMON SENSE! that many seem to be missing. hopefully AI research can help them.(sponsored by NV)

and the second glaring mistake is not looking at who are the (potential) buyers; do look at some of those folks resumes. most of them lectured coding/programming that gave graduate CI/CS students nightmares and few/some wrote that code themselves. freaking geniuses in their field(s); about the only help they would need from nividia would be to confirm their observations.

also, stop comparing this to the quadro, though i have also made that mistake myself, the tesla product(s) is/are much more relevant and is used in AI, not quadros.

so yeah . .all these concerns you continue to express are from your own lack of research pal; pretty much just trolling or not very intelligent.

your choice, good luck with that. smile.gif

thanks, thats a good news to Titan V users.

you could've done this from the start you know? you just dug yourself into a round-about argument.
Edited by epic1337 - 12/11/17 at 6:02am
post #640 of 1303
Quote:
Originally Posted by maltamonk View Post

Could you explain what the benefit of that would be to the user? Seems from their ToS, Nvidia has license to any results of using it.
Quote:
You hold all rights, title and interest in and to your modifications and derivative works of the NVIDIA sample and example source code software that you create as permitted hereunder (collective, “Derivatives”), subject to NVIDIA’s underlying intellectual property rights in and to the NVIDIA sample or example source code; provided, however that you grant NVIDIA and its affiliates an irrevocable, perpetual, nonexclusive, worldwide license to make, have made, use, have used, reproduce, license, distribute, sublicense, transfer and otherwise commercialize Derivatives without any obligation or compensation to you.

ikn, ask 3 lawyers and you'll likely get 3 answers, but from what i've seen . .that's common enough. nvidia wants a right to capitalize if you use their resources. how right that is pretty debatable depending on circumstances . . . i would believe

and i would be not surprise if sooner or later you're going to hear about nvidia's goal of:


ready? Warning: Spoiler! (Click to show)
Nvidia AIworks®
to be debated by the masses on how good/bad that is.
Quote:
Originally Posted by epic1337 View Post

thanks, thats a good news to Titan V users.

you could've done this from the start you know? you just dug yourself into a round-about argument.

though i do admit i get . .cranky, i would not stand in the way of someone helping themselves instead of spoon feeding them info.

graduate of the school of hard knocks here.

if you desire a personal apology . . we can PM about that. tongue.gif
Edited by looniam - 12/11/17 at 6:16am
loon 3.2
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7-3770K Asus P8Z77-V Pro EVGA 980TI SC+ 16Gb PNY ddr3 1866 
Hard DriveHard DriveHard DriveOptical Drive
PNY 1311 240Gb 1 TB Seagate 3 TB WD Blue DVD DVDRW+/- 
CoolingCoolingOSMonitor
EKWB P280 kit EK-VGA supremacy Win X LG 24MC57HQ-P 
KeyboardPowerCaseMouse
Ducky Zero [blues] EVGA SuperNova 750 G2 Stryker M [hammered and drilled] corsair M65 
AudioAudio
SB Recon3D Klipsch ProMedia 2.1  
  hide details  
Reply
loon 3.2
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7-3770K Asus P8Z77-V Pro EVGA 980TI SC+ 16Gb PNY ddr3 1866 
Hard DriveHard DriveHard DriveOptical Drive
PNY 1311 240Gb 1 TB Seagate 3 TB WD Blue DVD DVDRW+/- 
CoolingCoolingOSMonitor
EKWB P280 kit EK-VGA supremacy Win X LG 24MC57HQ-P 
KeyboardPowerCaseMouse
Ducky Zero [blues] EVGA SuperNova 750 G2 Stryker M [hammered and drilled] corsair M65 
AudioAudio
SB Recon3D Klipsch ProMedia 2.1  
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [twitter] Titan V revealed.