NVIDIA Beats AMD to Market On HBM2 - Announces Tesla P100

[RejZoR]

Same could be said about NV showing the prototype last year, or argued backwards that it was a dud. What matters that P100 is in retail and MI25 is not.


Newegg became a horrible place to look for enterprise stuff.
Here's one I found just a few days ago:
http://www.thinkmate.com/hardware/co-processors?a=YToxOntpOjE0Mzc7YToxOntpOjA7czoxMDoiVGVzbGEgUDEwMCI7fX0=

Supermicro also started to offer P40 and P100-based server solutions.

From where did you drag MI25 in? I was talking about CONSUMER card, the RX Vega.

 

[Relayer]

Uhm what is this, year old news?

Same could be said about NV showing the prototype last year, or argued backwards that it was a dud. What matters that P100 is in retail and MI25 is not.


Newegg became a horrible place to look for enterprise stuff.
Here's one I found just a few days ago:
http://www.thinkmate.com/hardware/co-processors?a=YToxOntpOjE0Mzc7YToxOntpOjA7czoxMDoiVGVzbGEgUDEwMCI7fX0=

Supermicro also started to offer P40 and P100-based server solutions.

Thanks. I was actually being sarcastic.

Surprised at the pricing. I thought it would be dearer.

 

[Nokiron]

And fp16 performance. If price does not matter, you actually can buy quadro GP100 now.

So much drama! These things have been up and running and available for months now, but only as part of their DGX-1.

This guy for example was playing with a rack of em last year: https://linustechtips.com/main/topic/688793-benchmark-10x-tesla-p100-gpu-nvtp100-16/

Point is if you have deep pockets you can buy em indivually now.

https://www.scan.co.uk/products/16g...u-tba-3584-cores-93-tflops-sp-47-tflops-dp-pa

As said, this can't be the news. You have been able to buy P100s seperatly for quite some time(in PCI-e form).

Maybe a month or two.

 

[TheinsanegamerN]

Don't bother, being AMD biased has become the new cool thing in the last years. Eventually people will realize how stupid it is to support something or someone, just because they're the underdogs

There is nothing wrong with supporting an underdog, if they are competitive with the big dog.

When a company is an underdog because they are not competitive, they are not an underdog, they are noncompetitive, full stop. And AMD has not been competitive GPU wise since (arguable) the 200/300 series. CPU wise, they just became competitive again.

The people rooting for vega need to realize how badly AMD is managing their GPU line. The 400s are competitive against the 900 line, from 2014. They have no GPU to perform against the 1070, 1080, 1080ti, titan pascal, or titan Xp. Vega is a year late, and much like fury X, no fancy HBM is going to fix an noncompetitive GPU.

 

[oxidized]

There is nothing wrong with supporting an underdog, if they are competitive with the big dog.

Supporting the underdog, only because they're the underdog, is one of the stupidest thing. This said, AMD did pretty good in the last years with their GPUs, they've been doing pretty good since 7xxx at least. CPUs wise well, as you said they just got competitive again, after a long time. Thing is, if there's a underdog which also makes bad products (so not AMD) what's the point in supporting them? So there's plenty of people supporting AMD only because they're the underdog, not actually because they make good products.

About VEGA, well, we'll see, if it's only slightly better than pascal, then it's a flop, total, at least from an architectural standpoint, if they place the cards with good prices is another thing, so they could be competitive on the price/performance, plus hbm2, which should be supposedly a pro over pascal.

 

[Aenra]

"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"

Someone inform this peasant about how this works/what they do with them/etc.. i cant even grasp it, i'd have thought this is the sole domain of a CPU?

edit:

Supporting the underdog, only because they're the underdog, is one of the stupidest thing

No offense dude, but you have a very particular style. You start by taking a bite out of peoples' asses, then take it a notch down to passive/aggressive+innuendos and the world be damned. You do this repeatedly, you do this the instant someone says something you do not agree with. Repeatedly enough to get you noticed.
Cool it. Or better yet, ammend.

 

[jabbadap]

As said, this can't be the news. You have been able to buy P100s seperatly for quite some time(in PCI-e form).

Maybe a month or two.

Quite bit longer(half a year or so)...
https://geizhals.de/pny-tesla-p100-pcie-tcsp100m-16gb-pb-nvtp100-16-a1501119.html
https://geizhals.de/pny-tesla-p100-pcie-tcsp100m-12gb-pb-nvtp100-12-a1501147.html

But yeah that sxm2 version which is in dxg-1 haven't been really available for separate sell. And why would it be, you would need special motherboard to take one. Funny fact is that those pcie taslas does not support nvlink but quadro gp100 does.

 

[silentbogo]

From where did you drag MI25 in? I was talking about CONSUMER card, the RX Vega.

For the same reason you are dragging a CONSUMER Vega into ENTERPRISE P100 discussion.
See, I'm also proficient with caps.

MI25 was the first officially announced AMD product based on VEGA and using HBM2.

Someone inform this peasant about how this works/what they do with them/etc.. i cant even grasp it, i'd have thought this is the sole domain of a CPU?

https://en.wikipedia.org/wiki/General-purpose_computing_on_graphics_processing_units

 

[Aenra]

@silentbogo thank you.. my mind is blown, no irony. Had no idea we've moved that far. Damn...

I'm growing old, lol

 

[TheinsanegamerN]

Supporting the underdog, only because they're the underdog, is one of the stupidest thing.

OK, you apparently didnt read what I wrote. There is nothing wrong with supporting them if it is COMPETITIVE. See, I dont like monopolies, I like competition. You dont just support them because they are the underdog, you support them if they are the underdog AND competitive. If you still think that is the "stupidest thing" then I cant help you here.

This said, AMD did pretty good in the last years with their GPUs, they've been doing pretty good since 7xxx at least.

You missed the 300 series, when nvidia owned 83% of GPU shipments at their height, and over 70% every quarter for over a year. the only reason that changed was the 400 series captured the low end low margin market, while leaving the mid range and lucrative high end and halo markets completely unopposed.

CPUs wise well, as you said they just got competitive again, after a long time. Thing is, if there's a underdog which also makes bad products (so not AMD) what's the point in supporting them?

See, again, apparently you glossed over my comment. I said there was no point in supporting a noncompetitive underdog.

So there's plenty of people supporting AMD only because they're the underdog, not actually because they make good products.

There are people supporting AMD in markets where they are competitive (low end CPU and GPU). And there are people who are supporting them based on promises and speculation. That second group is the issue.

About VEGA, well, we'll see, if it's only slightly better than pascal, then it's a flop, total, at least from an architectural standpoint, if they place the cards with good prices is another thing, so they could be competitive on the price/performance, plus hbm2, which should be supposedly a pro over pascal.

The issue with being competitive on price vs performance is that nvidia has had the market for a solid year, and has made mad bucko bucks as a result. They can simply cut the price across the line and starve AMD's high end from being competitive.

Performance wise, the vega peaks AMD has shown are around 1070 level, if not slightly higher. If AMD has mature drivers at launch (which, given their history, is highly unlikely) they MIGHT reach a 1080 in some games, and be between the 1070 and 1080 in others. Given the price for making a fury, I am not confident that AMD can release vega, with HBM2, cheaper then a 1080 and make anywhere near the money nvidia makes per 1080.

 

[ZoneDymo]

No need to take it so personally.

man you sure are all over the map...
You attack him personally on a bias, implying he is blind by fanboyism, then he provides simple facts supporting the claims and that is your reply...

Good stuff man, next time perhaps just apologize and take your words back.

 

[Vayra86]

You know history is full of discussion on 'who was first' to whatever subject you can make up.

In the end all that matters is who has the best business, you can be first or last it doesn't mean shit. Look at Apple, last in everything, had one good idea, has the highest margin on mobile of the entire globe. Look at AMD: wants to be first and pioneer in tech, worst sales figures in the tech business.

This topic discussion has awful likeness to news topics where the first commenter yells 'FIRST!'... and about the same value. Think on that for abit.

 

[Xajel]

This is only for compute.. for those asking about crysis. no.. really not just kidding.. there's no video output in this card.

 

[bug]

This is only for compute.. for those asking about crysis. no.. really not just kidding.. there's no video output in this card.

Yes, but without on output, how can you be sure the card isn't running Crysis in the background?

 

[jabbadap]

This is only for compute.. for those asking about crysis. no.. really not just kidding.. there's no video output in this card.

Well while true in tesla cards, quadro gp100 has video outputs and it has gp100 with hbm2 memory. So yes you can run crysis on it.

 

[qubit]

I know the P100 is architecturally a bit different to the P102, so could it play games if put into a graphics card and if so, would it be faster than a P102?

 

[TheGuruStud]

I'm thinking nvidia bought Hillary's shill team, b/c there sure is a guy in here posting like it...

I'm so glad he has full knowledge of an unreleased product. Whew, nothing to see here, move along.

 

[oxidized]

OK, you apparently didnt read what I wrote. There is nothing wrong with supporting them if it is COMPETITIVE. See, I dont like monopolies, I like competition. You dont just support them because they are the underdog, you support them if they are the underdog AND competitive. If you still think that is the "stupidest thing" then I cant help you here.



You missed the 300 series, when nvidia owned 83% of GPU shipments at their height, and over 70% every quarter for over a year. the only reason that changed was the 400 series captured the low end low margin market, while leaving the mid range and lucrative high end and halo markets completely unopposed.
See, again, apparently you glossed over my comment. I said there was no point in supporting a noncompetitive underdog.

There are people supporting AMD in markets where they are competitive (low end CPU and GPU). And there are people who are supporting them based on promises and speculation. That second group is the issue.


The issue with being competitive on price vs performance is that nvidia has had the market for a solid year, and has made mad bucko bucks as a result. They can simply cut the price across the line and starve AMD's high end from being competitive.

Performance wise, the vega peaks AMD has shown are around 1070 level, if not slightly higher. If AMD has mature drivers at launch (which, given their history, is highly unlikely) they MIGHT reach a 1080 in some games, and be between the 1070 and 1080 in others. Given the price for making a fury, I am not confident that AMD can release vega, with HBM2, cheaper then a 1080 and make anywhere near the money nvidia makes per 1080.

Ok then, maybe i missed something you said, supporting a competitive underdog is absolutely fine.

"There are people supporting AMD in markets where they are competitive (low end CPU and GPU). And there are people who are supporting them based on promises and speculation. That second group is the issue."

Totally agreed, the big problem is those people always try and make you look like the intel/nvidia fanboy or amd hater, so yeah.

In my opinion, if VEGA doesn't beat Pascal clean, it's a flop, there's no other way to say it, AMD in the latest years, was never a whole year behind nvidia, never, 2xx 3xx and also 4xx were/are extremely competitive, always 10% within nvidia cards, that's why i'd rate them a flop, in case they were even slightly faster than 1080

 

[hardcore_gamer]

I know the P100 is architecturally a bit different to the P102, so could it play games if put into a graphics card and if so, would it be faster than a P102?

It could play games as long as the drivers support it. The gaming performance would be a bit slower than Tixan Xp because of the lower clock speeds.

 

[kruk]

Didn't PR release articles usually have a " Press Release" tag?

 

[RejZoR]

For the same reason you are dragging a CONSUMER Vega into ENTERPRISE P100 discussion.
See, I'm also proficient with caps.

MI25 was the first officially announced AMD product based on VEGA and using HBM2.


https://en.wikipedia.org/wiki/General-purpose_computing_on_graphics_processing_units

How about you stop your wild horses and start actually reading the posts mkay?

 

[R0H1T]

But... Can it play Crysis 3? (sorry I had to XD)

Obviously yes, but I'm more interested in the soon(?) to be released HL3

 

[qubit]

Obviously yes

No it can't. Nothing can play Crysis no matter how powerful!

 

[Nokiron]

It could play games as long as the drivers support it. The gaming performance would be a bit slower than Tixan Xp because of the lower clock speeds.

It will be slower than that, more along the lines of a lower clocked Titan X Pascal. There are no GP100-chips with 3840 CUDA-cores, they are all at 3584.

 

[ssdpro]

How did this thread go so far off down the "no, I swear AMD was actually first" path? The whole point is NVIDIA beat AMD to market. It is one thing to have a concept and show a single sample, it is another to turn it into a functioning product for sale.

For the record, AMD stock is still down 7.5% in the last week alone. Better hope those April sales numbers aren't as bad as March.