Nvidia Volta Won't Use HBM2, But GDDR5X

[NvidiaIntelAMDLoveTriangle]

Well it's not like we didn't see this coming.

Quote

We've written a lot back and forth about GDDR5X versus HBM2 (and GDDR6 these days). Nvidia did not place all their eggs in one basket, AMD pretty much did so for high-end. That being the main and root cause of delays as HBM2 is difficult to fab and expensive. As it stands now, opcoming consumer GeForce cards based on Volta will not use HBM2.

This is a new claim that originates from Fudzilla, who states that this info came from well-informed sources. I tend to agree as nothing is pointing or indicating towards HBM2 from Nvidia at this moment in time. From the looks of it, small and compact HBM2 memory simply is too expensive for consumer products. Nvidia does use HBM2 on its GV100-gpu, but much like what happened with Pascal, here again they will revert to GDDR5X memory for the consumer parts.

 

Quote

While HBM2 might seem to offer better memory-bus width, latency and lower voltages, it seems harder to fab and implement and likely thus is more costly and complicated to use than expected. Just look at the massive AMD Radeon RX Vega (consumer) delay, this likely is due to the limited availability of HBM2. Meanwhile Nvidia has been plastering GDDR5X in their recent high-end SKUs. GDDR5X still rocks hard this year. Next to HBM2, GDDR6 is the new thing for 2018. The new graphics memory type will offer up to 16Gb (2GB) per IC. So eight ICs would already get you to 16 GB of graphics memory.

Nvidia could make use of SK Hynix and/or Micron GDDR5X memory. 

 

 

Source: http://www.guru3d.com/news-story/upcoming-geforce-gtx-cards-use-gddr5x-not-hbm2.html

Personally I don't think it's going to matter at all. While HBM is great, it's way over hyped.

 

This is my first news post, ever. If I did something wrong please don't crucify me. If I did something wrong, my apologies.

 

[BestPCBuilder2017]

wtf

[NumLock21]

So HBM2 is only for their Tesla

[agent_x007]

Well NV got burned on that (G)DDR2 deal with 5800 Ultra.
So, no wonder they don't like "newest memory type available" in consumer grade cards.

Other than that, they pump SOO MUCH HBM 2.0 into Quadros and Teslas it's no brainer they skipped GeForce line up (HBM 2.0 source isn't "bottomless").
This way they can also have higher margins (by lower cost) vs. Vega RX (which will have HBM 2.0 ?).

[21rkosta]

they may have tested a unit with both and found little difference

[LegendKillerRG]

AMD is able to offer HBM/2 for cheap..

[mynameisjuan]

If benchmarks show volta at the top after vega then I really dont care. They probably didnt see a need for it. If they keep releasing cards that are faster then why are people complaining?

[shadowbyte]

rip

[domandric034]

[vanished]

Is this surprising?  Really?  When did we hear they were going to use HBM?

[ARikozuM]

I'm fine with GDDR5X. No point in adding cost to a card that, at the top end, should cost 700-800.

[domandric034]

NVIDIA wake up !!

You don't see upcomming enemy rising from Red Team, you need to put god fucking HBM2 or rip your next architecture cause Vega (If everything goes correctly and shows that those demos were real) will destroy u.

[MoonSpot]

10 minutes ago, LegendKillerRG said:

AMD is able to offer HBM/2 for cheap..

Where?

[79wjd]

11 minutes ago, LegendKillerRG said:

AMD is able to offer HBM/2 for cheap..

Where?

 

EDIT: @MoonSpot sonofabitch

[Notional]

Gotta give it to NVidia. Very clever choice.

 

Look at Fiji (Fury) cards. Look at how delayed Vega is. And for what? With GDDR5x and GDDR6, the bandwidth is good enough. Now I do think HBM is a great thing, but AMD should have gone the Volta pro route, and only had them on Pro cards. The demand is not as high there, which coincides better with supply, and the sometimes limited capacity (4GB on Fury).

 

I love that AMD is pushing forward and innovating, but they are not cashing in on all the hard work. They are just suffering all the initial problems with nothing to show for it.

 

4 minutes ago, domandric034 said:

[domandric034]

2 minutes ago, Notional said:

Gotta give it to NVidia. Very clever choice.

 

Look at Fiji (Fury) cards. Look at how delayed Vega is. And for what? With GDDR5x and GDDR6, the bandwidth is good enough. Now I do think HBM is a great thing, but AMD should have gone the Volta pro route, and only had them on Pro cards. The demand is not as high there, which coincides better with supply, and the sometimes limited capacity (4GB on Fury).

 

I love that AMD is pushing forward and innovating, but they are not cashing in on all the hard work. They are just suffering all the initial problems with nothing to show for it.

 

 

When I see ur LTT profile.

[NvidiaIntelAMDLoveTriangle]

Why are some people upset that it won't have HBM?

I mean if the performance of the cards is good, then who cares if they used HBM or GDDR?

Shouldn't you care about the performance and price it's at, instead of what tech it uses?

[AnonymousGuy]

It'll probably still be faster than anything AMD puts out anyways, HBM or not.  So don't really care what the spec sheet says so long as its fast.

[Notional]

16 minutes ago, domandric034 said:

 

When I see ur LTT profile.

He aims! He shoots!

 

 

Aaaand he misses.

[YoloSwag]

Dang, I saw this the other day and forgot it was actual news  Silly me...

 

Signs of the seasons ladies and gentlemen.

 

HBM2 was supposed to be last year then got pushed to H2.

 

GDDR6 was supposed to be next year with Volta.

 

Here we are with corporate people saying all sorts of mumbo jumbo just raise their stocks' value for all the wrong reasons

[domandric034]

34 minutes ago, Notional said:

He aims! He shoots!

 

 

Aaaand he misses.

He wont miss when he sprays with AK 47

[StealthArsenal]

I don't recall reading anywhere that Nvidia was definitively going to use HBM2 on Volta.  There have been numerous rumors of this, but no concrete evidence.  With that said, Nvidia has been turning out some fast cards running on GDDR5X.  Why would they cut off their nose to spite their face and go with a more expensive process when they are competing just fine with GDDR5X or GDDR6 if they decided to use that.  If Vega comes out and still doesn't hang with Nvidia, then this makes even more sense and would make the argument that HBM2 in actuality might not be as fast as everyone thought. 

 

That's my way of looking at it.

[Hunter259]

Because all high end gpu's need HBM... /s

Faster GDDR5X or GDDR6 will be able to supply the GPU's just fine. Stop freaking out about it.

[DXMember]

brilliant, put HBM2 and NVLink in all your marketing materials, confuse the hell out of consumers, and then tell them that you're shipping with GDDR5X which in term is just overclocked GDDR5

[BuckGup]

1 hour ago, domandric034 said:

NVIDIA wake up !!

You don't see upcomming enemy rising from Red Team, you need to put god fucking HBM2 or rip your next architecture cause Vega (If everything goes correctly and shows that those demos were real) will destroy u.

Nah this is good. We want NVIDIA to take a big hit. If AMD kills it with threadripper, they certainly will in the server space, and with VEGA. Nvidia is going to have to actually innovate again. Rather than slapping on some memory and a dye shrink calling it new.