Monday, April 10th 2017

NVIDIA Beats AMD to Market On HBM2 - Announces Tesla P100

by
Raevenlord
 Discuss (105 Comments)
NVIDIA has announced availability of their latest data center accelerator, the Tesla P100, which is the world's first HBM2-powered add-in-card. this means that NVIDIA effectively beat AMD in time to market with HBM2 technology, which AMD pioneered (in its HBM form) with the Fury line of graphics cards.

NVIDIA naturally touts this as the world's most advanced data center accelerator, for workloads such as "Artificial intelligence for self-driving cars. Predicting our climate's future. A new drug to treat cancer." NVIDIA's green graphics show an almost 50x increase in computing power from 8x Tesla P100 accelerators when compared to a dual CPU server based on Intel's Xeon E5-2698 V3 (which isn't really all that surprising.) NVIDIA further brings in the PR talk with examples on how a single GPU-accelerated node powered by four Tesla P100s - interconnected with PCIe - can replace up to 32 commodity CPU nodes for a variety of applications - saving up to 70% in overall data center costs.
Source: NVIDIA

Related News

Add your own comment

105 Comments on NVIDIA Beats AMD to Market On HBM2 - Announces Tesla P100

#76
Relayer
NaitoAMD being as generous as they are, didn't keep it to themselves and it has since become a JEDEC standard for all to manufacture and use (you can't steal an open standard - proprietary implementations excluded ). It's a smart move by AMD because widespread adoption of the technology increases volume and an increase in volume generally leads to cheaper prices for all. For the consumer, this means more SKUs which implement HBM in the future. Leaving Nvidia out of this equation, means missing out and a large chunk of the market, thus slowing adoption. That's one of the great things about AMD - they understand that being a smaller player means they have to get creative.
It's also what made the PC great in the 1st place. Open standards and compatibility won out over exclusion and brand experience. Making things proprietary and exclusive is a giant step back and in the end self defeating.
Posted on Reply
#77
Naito
RelayerIt's also what made the PC great in the 1st place. Open standards and compatibility won out over exclusion and brand experience. Making things proprietary and exclusive is a giant step back and in the end self defeating.
And it's great that many companies are realizing this now (Microsoft, for example). Although, as counter-intuitive as it seems, one could argue that in some cases proprietary tech is needed to differentiate and monetize to help further develop other areas, otherwise there is not much incentive from a business point of view.
Posted on Reply
#78
Fluffmeister
There certainly is a degree of irony when one company champions open standards, but then fanboys of said company get mad when someone else uses it.
Posted on Reply
#79
Relayer
FluffmeisterThere certainly is a degree of irony when one company champions open standards, but then fanboys of said company get mad when someone else uses it.
Who's mad that nVidia is using it?
Posted on Reply
#80
Fluffmeister
RelayerWho's mad that nVidia is using it?
I ask myself the same question every day.
Posted on Reply
#81
RejZoR
RelayerWho's mad that nVidia is using it?
No one, because it benefits everyone. But no one will thank AMD for investing into development of HBM and being so nice to allow others to use/license it freely...
Posted on Reply
#82
Ruru
S.T.A.R.S.
Well, GDDR3 was also a technology developed by ATI and used by Nvidia also, so what's the matter who uses who's technology etc.?
Posted on Reply
#83
medi01
oxidizedWhat way is, "the other way around"?
The other way round is that Huang bends you over, but Lisa doesn't.

Overpriced cards, even more money grabbed for adaptive sync, quickly diminishing support for older cards.
Posted on Reply
#84
oxidized
medi01The other way round is that Huang bends you over, but Lisa doesn't.

Overpriced cards, even more money grabbed for adaptive sync, quickly diminishing support for older cards.
Yeah as always, AMD is just doing good for everyone, nvidia is stealing money, AMD is a charity foundation, nvidia is a lucrative corp, AMD are the good guys, nvidia the bad guys.
Please man, you're smarter than this. They're both huge multinationals making money off of us, one was luckier, and had the "privilege" of charging more because they could, the same goes for intel, actually intel is a better example, but i think neither nvidia nor intel made bad products and made you pay ridiculous prices for their products, not most of the time al least (titan is an absolute retarded joke, but ignore the price for a moment, is it a product you can call bad?), but AMD surely did more bad products than both of nvidia and intel, in the past, so at the end they both bend you over, just in different directions.
Posted on Reply
#85
Relayer
RejZoRNo one, because it benefits everyone. But no one will thank AMD for investing into development of HBM and being so nice to allow others to use/license it freely...
Well, I don't think AMD does it out of the kindness of their hearts. ;) It's the best way to get the industry to move in the same direction as they are.

Like Vulcan. They already invested in Mantle. By giving it to the Khronos group they just made it easily and cheaply accessible to the rest of the industry. And msft was likely never to develop a truly multi thread API if they didn't have to. Meanwhile, their hardware was designed with a lot of programming needed to get the most out of it.
Posted on Reply
#86
ratirt
RelayerWell, I don't think AMD does it out of the kindness of their hearts. ;) It's the best way to get the industry to move in the same direction as they are.

Like Vulcan. They already invested in Mantle. By giving it to the Khronos group they just made it easily and cheaply accessible to the rest of the industry. And msft was likely never to develop a truly multi thread API if they didn't have to. Meanwhile, their hardware was designed with a lot of programming needed to get the most out of it.
Well that's one point they did give it to Khronos group. It's just the group see potential in this and they asked for it? DX12 is nice but it lacks a lot. I read an article once comparing those 2. Even though it is made on mantle base coding the NV cards benefit from this as well. Simply Vulcan is just more advanced as an API and works better. :) Maybe it's also less complicated and that pushes companies to adapt this one instead. Making their products better.
Posted on Reply
#87
medi01
RelayerWell, I don't think AMD does it out of the kindness of their hearts.
Well, I remember when OpenGL was great.
You know, up until some company started "me, ME; MEEEE; ONLY MINE" thing and it ended up with Microsoft DirectX wiping the floor with that standard.

All companies are out there for profits, but not all companies get as low as it gets, Huang is particularly good at it.
Posted on Reply
#88
bug
medi01Well, I remember when OpenGL was great.
You know, up until some company started "me, ME; MEEEE; ONLY MINE" thing and it ended up with Microsoft DirectX wiping the floor with that standard.

All companies are out there for profits, but not all companies get as low as it gets, Huang is particularly good at it.
Actually, all companies do get as low as it gets. It's just that when you;re the underdog, you can't get as low as the top dog, that's the only difference.
E.g. When ATI was top dog (Radeon 9000 days), they didn't give a rat's ass for Linux. They had their fglrx (which they got through an acquisition, iirc) which offered a sorry-ass OpenGL implementation and would barely support newer kernel or X servers and that was it. But I don't call them evil or something because of that. Neither do I think Nvidia were angels for providing a better Linux driver for years. It's just business.
Posted on Reply
#91
Kanan
Tech Enthusiast & Gamer
jimmyz4202Old card, the new one is 3840 cuda cores and HBM2...
Still, the title is stupid, HBM2 @ Nvidia is nothing new. AMD was "beaten to it" last year, not now.

Edit: the "old card" had 3840 shaders as well - no idea why some people are trashtalking about GPU's they don't know jack about.

Utterly senseless "news" here.
Posted on Reply
#92
medi01
bugActually, all companies do get as low as it gets
No, they don't.
And indeed, while underdog is limited in harm it could do to the market, even Intel didn't go as low as nvidia did, in my opinion.
Posted on Reply
#93
ratirt
medi01No, they don't.
And indeed, while underdog is limited in harm it could do to the market, even Intel didn't go as low as nvidia did, in my opinion.
Well. In my opinion it's just marketing. NV is playing short stick here and will regret that soon. But that's their strategy. Can't blame them either if you put at the top people who care only about money. Like I said short stick, short term income which will definitely impact longer term investment. Guess they care about here and now. Or since Vega will be released soon maybe NV is putting all the stubs now to get some pennies out since vega may or may not threaten their strategy of getting as much as they can. The titan Xp(second edition) was to nail the customers good. especially those who bought the previous version. It is all clear that it is about money and it always is but not like this. Just not like this really.
Posted on Reply
#94
Relayer
bugActually, all companies do get as low as it gets. It's just that when you;re the underdog, you can't get as low as the top dog, that's the only difference.
E.g. When ATI was top dog (Radeon 9000 days), they didn't give a rat's ass for Linux. They had their fglrx (which they got through an acquisition, iirc) which offered a sorry-ass OpenGL implementation and would barely support newer kernel or X servers and that was it. But I don't call them evil or something because of that. Neither do I think Nvidia were angels for providing a better Linux driver for years. It's just business.
No, not all companies "get as low as it gets". That's the, but mommy, Billy did it too, defense. And that would have been ATI, not AMD back in the Radeon 9000 days. Besides, sometimes companies learn from the past. Or simply a new CEO changes the way they do business. You're going back all the way to 2002 for your example.
Posted on Reply
#95
jimmyz4202
The article you linked says :

The Tesla P100 is a GPGPU with the most powerful GPU in existence - the NVIDIA GP100 "Pascal," featuring 3,584 CUDA cores, up to 16 GB of HBM2 memory.


And yes I agree, it's annoying when people act like they know what's going on, even to the point of calling someone out, when actually a few seconds reading the link they provided would show they don't know "jack" as you put it.

Still nobody else mentioning Vega up and running liquid sky's gaming system.
Posted on Reply
#96
bug
RelayerNo, not all companies "get as low as it gets". That's the, but mommy, Billy did it too, defense. And that would have been ATI, not AMD back in the Radeon 9000 days. Besides, sometimes companies learn from the past. Or simply a new CEO changes the way they do business. You're going back all the way to 2002 for your example.
Well, ATI/AMD haven't been top dogs ever since. When they'll be, their behaviour won't be any different than Nvidia's or intel's is now.
But that won't happen for a long time, they've got to get out of a big hole first.

And yes, sometimes companies learn and CEOs change, but again, that holds for all companies, it's not something AMD specific. But at the end of the day, companies are held accountable by their shareholders, not their customers. You can't simply decree "we're going to pour $10mn into developing this standard and then offer it for free". You have to have a business case for that.
Another example: do not, for a moment, think AMD priced Ryzen the way they did because of goodness of their hearts. They have models and projections that tell them those price points are the sweetspot for recuperating their investment while also gaining marketing share.
Posted on Reply
#97
Relayer
bugWell, ATI/AMD haven't been top dogs ever since. When they'll be, their behaviour won't be any different than Nvidia's or intel's is now.
But that won't happen for a long time, they've got to get out of a big hole first.

And yes, sometimes companies learn and CEOs change, but again, that holds for all companies, it's not something AMD specific. But at the end of the day, companies are held accountable by their shareholders, not their customers. You can't simply decree "we're going to pour $10mn into developing this standard and then offer it for free". You have to have a business case for that.
Another example: do not, for a moment, think AMD priced Ryzen the way they did because of goodness of their hearts. They have models and projections that tell them those price points are the sweetspot for recuperating their investment while also gaining marketing share.
Never have I claimed AMD doing anything ouy of the kindness of their hearts. I've stated the opposite in this very thread. What they do is calculated. My point is that you can be calculated, you can do what's best for your company, and you don't have to sink to the depths of as low as you can go. You saying their behavior wouldn't be any different than nVidia and Intel is simply justification for them doing what they've done. You have no evidence to make this claim. Just your belief that it's the way anyone would act in a similar situation. And for the record I don't think Intel and nVidia are the least bit comparable.
Posted on Reply
#98
ratirt
bugWell, ATI/AMD haven't been top dogs ever since. When they'll be, their behaviour won't be any different than Nvidia's or intel's is now.
But that won't happen for a long time, they've got to get out of a big hole first.

And yes, sometimes companies learn and CEOs change, but again, that holds for all companies, it's not something AMD specific. But at the end of the day, companies are held accountable by their shareholders, not their customers. You can't simply decree "we're going to pour $10mn into developing this standard and then offer it for free". You have to have a business case for that.
Another example: do not, for a moment, think AMD priced Ryzen the way they did because of goodness of their hearts. They have models and projections that tell them those price points are the sweetspot for recuperating their investment while also gaining marketing share.
No company needs to be a top dog to release a good or I'd say competitive product. The only difference between underdog and top dog is that the underdog is trying. top dog doesn't give a damn and is ripping people of their money.
And there is no harm or disgrace in cheering for the underdog. I'd say it is reasonable to do so.
Posted on Reply
#99
medi01
bugWhen they'll be, their behaviour won't be any different than Nvidia's or intel's is now
This is repeating "all companies are the same".
No they aren't.
Intel isn't like nvidia either, and that despite having oh so much more dominant position.

And even back, with market being rather close to 50/50 between team red/green, green was going full throttle shit (effectively killing OpenGL), while red didn't.
Posted on Reply
#100
bug
RelayerNever have I claimed AMD doing anything ouy of the kindness of their hearts. I've stated the opposite in this very thread. What they do is calculated. My point is that you can be calculated, you can do what's best for your company, and you don't have to sink to the depths of as low as you can go. You saying their behavior wouldn't be any different than nVidia and Intel is simply justification for them doing what they've done. You have no evidence to make this claim. Just your belief that it's the way anyone would act in a similar situation. And for the record I don't think Intel and nVidia are the least bit comparable.
Actually, neither of us has any evidence supporting our claims, because AMD (or ATI) never had a lead in the market like Nvidia or Intel have now. So we're only stating our (educated?) guesses here.
medi01This is repeating "all companies are the same".
No they aren't.
Intel isn't like nvidia either, and that despite having oh so much more dominant position.

And even back, with market being rather close to 50/50 between team red/green, green was going full throttle shit (effectively killing OpenGL), while red didn't.
There may be differences between companies, but all public companies share a common goal: make as much money as possible. Otherwise they'd be charities or NGOs.

What you say is certainly true in markets with more competitors. But when only a handful are involved, they tend to act alike.
Posted on Reply
Add your own comment
Nov 25th, 2024 11:18 EST change timezone

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts