Quick News

Volta uses HBM as well so Nvidia have decided to go with a Pascal die shrink and DDR5x or 6 instead which makes sense also the rumour is that Volta's architecture isn't as good for regular use like gaming and stuff
 
Don't be silly; they've started work on Volta years ago when they had no idea about Vega.

I know. I used to work developing software so I know how it works. You remain at least three steps ahead so you've always got something to show ;)

I read your last post wrong. I didn't realise you were addressing Ecalabur's quote/rumour and not the rumour that Nvidia are not releasing Volta yet.

Crossed wire, sorry about that :)
 
Volta was ready before they even launched Pascal.....

Because at one point the release order was reversed (I seem to remember)/Pascal was inserted in-between (the majority seems to remember)? Is there any truth to this rumour of Volta being a Pascal die shrink/optimization? I'm having a hard time believing it (sorry Excalibur).

I know. I used to work developing software so I know how it works. You remain at least three steps ahead so you've always got something to show ;)

I read your last post wrong. I didn't realise you were addressing Ecalabur's quote/rumour and not the rumour that Nvidia are not releasing Volta yet.

Crossed wire, sorry about that :)

Makes sense, no worries! :cool:
 
Last edited:
Because at one point the release order was reversed (I seem to remember)/Pascal was inserted in-between (the majority seems to remember)? Is there any truth to this rumour of Volta being a Pascal die shrink/optimization? I'm having a hard time believing it (sorry Excalibur).



Makes sense, no worries! :cool:

I'm pretty sure Volta was after Maxwell. But they renamed stuff so Pascal became Volta.
 
Intel Hyperthreading Bug in KabyLake and Skylake Processors Was Addressed By BIOS Fix In April 2017

The following is what Intel had to say about the affair:

The issue has been addressed with a fix that started rolling out in April 2017. As always, we recommend checking to make sure your BIOS is up to date, but the chance of encountering this issue is low, as it requires a complex number of concurrent micro-architectural conditions to reproduce.

http://www.tomshardware.com/news/hyperthreading-kaby-lake-skylake-skylake-x,34876.html
 

It would have to be around the same performance jump as going from a 980 Ti to 1080 Ti to get me to say "Wow I want it", I'm still holding my money ready for Vega anyway ^_^

In other news, Nvidia quietly enabled DX12 support for GTX 400/500 series cards and upgraded the DX12 Resource Binding from Tier 2 to Tier 3 on Pascal cards as well as Netflix 4K support in the latest 384.76 driver.

https://www.reddit.com/r/nvidia/comments/6kib27/updates_to_fermi_maxwell_and_pascal_dx12/
 
Last edited:
It would have to be around the same performance jump as going from a 980 Ti to 1080 Ti to get me to say "Wow I want it", I'm still holding my money ready for Vega anyway ^_^

In other news, Nvidia quietly enabled DX12 support for GTX 400/500 series cards and upgraded the DX12 Resource Binding from Tier 2 to Tier 3 on Pascal cards as well as Netflix 4K support in the latest 384.76 driver.

https://www.reddit.com/r/nvidia/comments/6kib27/updates_to_fermi_maxwell_and_pascal_dx12/

You can expect Nvidia to put on the brakes now that they have seen Vega. And they will, and they are. Welcome to the land of Intel, where it's 5% every year or so.
 
You can expect Nvidia to put on the brakes now that they have seen Vega. And they will, and they are. Welcome to the land of Intel, where it's 5% every year or so.

As soon as Nvidia does that, AMD will take over. And it won't take them five years like it did with Ryzen. So your point is moot. I can't see Nvidia introducing 5% incremental improvements. That's completely illogical. I feel like your habit of doom-bringing is getting in the way of logic and economics.
 
As soon as Nvidia does that, AMD will take over. And it won't take them five years like it did with Ryzen. So your point is moot. I can't see Nvidia introducing 5% incremental improvements. That's completely illogical. I feel like your habit of doom-bringing is getting in the way of logic and economics.

AMD aren't doing crap for the foreseeable. Their next tech is Navi, and that is scale-able so they need to basically get multi GPU working. And that means it needs to happen to a console first, and that is way off. The whole idea of Navi is to use IF to connect more than one small core (like Polaris). They don't have anything else in the pipes.

I'm not doom bringing. Vega is what it is and if the price is right it will slot into the mid range nicely. However, if Nvidia dominate the high end like they've done for years now Vega will not change much.

I'm annoyed at AMD AGF. Very, very annoyed at AMD. They've had their chance to drop the kitchen sink for years yet still they have persisted with this GCN. Nvidia remarked that the 7970 was AMD's "Fermi" and AMD have just continued doing the same thing since.

They used HBM which drove up the price of the Fury X. Hardly any were sold (look around you at how many of us have them). It didn't have enough VRAM and within 6 months it was showing issues. Now they've basically gone and done the exact same thing with Vega. IE - made it expensive. That means that taking on the 1070 (which is what they are going to be looking to do) is going to be really hard because of the cost of that HBM2. Nvidia can simply drop the prices of the 1070 etc.

This is all deserved criticism. I absolutely love their CPU dept atm, but the GPU dept is almost like it's a completely different entity akin to the old AMD.
 
Would be lovely if Intel got involved with the discrete GPU market, at lest power efficiency wise their iGPUs arent half bad. But obviously you can't just scale those up.I see Intel as the only company which has the resources to challenge Nvidia. But Intel has shown absolutely no interest in that, so it's not like it would realistically happen.
 
AMD aren't doing crap for the foreseeable. Their next tech is Navi, and that is scale-able so they need to basically get multi GPU working. And that means it needs to happen to a console first, and that is way off. The whole idea of Navi is to use IF to connect more than one small core (like Polaris). They don't have anything else in the pipes.

I'm not doom bringing. Vega is what it is and if the price is right it will slot into the mid range nicely. However, if Nvidia dominate the high end like they've done for years now Vega will not change much.

I'm annoyed at AMD AGF. Very, very annoyed at AMD. They've had their chance to drop the kitchen sink for years yet still they have persisted with this GCN. Nvidia remarked that the 7970 was AMD's "Fermi" and AMD have just continued doing the same thing since.

They used HBM which drove up the price of the Fury X. Hardly any were sold (look around you at how many of us have them). It didn't have enough VRAM and within 6 months it was showing issues. Now they've basically gone and done the exact same thing with Vega. IE - made it expensive. That means that taking on the 1070 (which is what they are going to be looking to do) is going to be really hard because of the cost of that HBM2. Nvidia can simply drop the prices of the 1070 etc.

This is all deserved criticism. I absolutely love their CPU dept atm, but the GPU dept is almost like it's a completely different entity akin to the old AMD.

Vega is not released yet so how can you confidently say that "Vega is Vega"? If it beats a 1080 at the same price point, that's the largest market share in the high-end sector covered and will keep at least a few customers contented (like myself). For how long they'll be competitive in that sector we do not know. If you say you know then that's called speculation. Doom-bringing is speculation with a focus on the negative possibilities. Vega 10 is not a midrange product as far as we can tell. It's a 484mm2 chip. That's not 'midrange'.

And neither do we know for sure that Navi's only call to arms is its 'scaleability', ie using smaller chips connected via the Infinity Fabric. Navi could offer more than that. We don't know.

I'm annoyed at AMD too.

I'm disappointed they have taken so long to get the GPU out. The longer it takes the more people will think it's going to be totally amazeballs when it was only ever supposed to offer 1070 and 1080 level performance (in my eyes); the more people will give up waiting and buy Nvidia; and the closer it gets to Volta or whatever Nvidia has coming next.

I'm frustrated they let the words "poor VOLTA" creep into their promotional video. That was an arrogant and foolish move to make. Unless they know what Volta has to offer, that's speculation, or arrogance.

I'm annoyed they chose to focus on fiddly technology like HBM2 instead of getting solid GPU's out at the right time. However I was not one of the idiots who saw fit to hype Vega to be a Titan XP beater like AdoredTV. Well, he's not an idiot but that was a clear misstep, something that folks here and on other forums have mimicked. If Vega beats the Titan XP and subsequently the 1080Ti, AMD will have more than doubled the performance of Fiji AMD have never done that with any of their previous GPU architectures after only two years. The die shrink may help get them closer to that, but Vega was in my eyes never supposed to beat Nvidia's flagship. The biggest market is the 1070 and 1080 sector. If they can sneak their way into that market, good. They'll also hopefully make some money in the prosumer and professional market. That's not what I consider "doing crap for the foreseeable future".

And I disagree about the Fury X. I have had no issues with my Fury. It has largely gotten better with time with only a couple of notable exceptions to that fact. I would have preferred a cheaper GPU with 8GB of GDDR5, but it's done me well for the last two years. I wouldn't buy one now and if I could go back to when I was upgrading my monitor and GPU I'd strongly consider 980Ti/Gsync, but the Fury/Freesync setup was better value than the 980/Gsync setup and I would have really struggled to afford the 980Ti/Gsync setup. The 980 is now considered a midrange GPU because the 1060 often beats it, but rarely does a 1060 beat a Fury. And a Fury X regularly matches or beats a 980Ti.
 
If you really, honestly think that the desktop Vega is going to be that much better than the Vega already out there then good luck to you.

I just don't see how it is ever going to beat a 1080 with air cooling. If the reference pro card is already throttling by nearly 200mhz then it obviously needs a better cooler.

Who knows? maybe this time around they will allow AIBPs to make proper coolers for it, instead of putting a boat anchor on it.

Quickly, because I don't want to burn my dinner.. The Fury X runs out of VRAM a lot. Sure, they've bodged the drivers but the performance tanks.

And on that note I am off to shove gammon steaks and chips into my fat gob :D
 
If you really, honestly think that the desktop Vega is going to be that much better than the Vega already out there then good luck to you.

I just don't see how it is ever going to beat a 1080 with air cooling. If the reference pro card is already throttling by nearly 200mhz then it obviously needs a better cooler.

Who knows? maybe this time around they will allow AIBPs to make proper coolers for it, instead of putting a boat anchor on it.

Quickly, because I don't want to burn my dinner.. The Fury X runs out of VRAM a lot. Sure, they've bodged the drivers but the performance tanks.

And on that note I am off to shove gammon steaks and chips into my fat gob :D

I don't know. That's my point. I'm only guesstimating. We've seen Vega beat the 1080 before, we've seen Fiji and Polaris gain huge strides with driver updates, so it's within the realm of possibility in my opinion. Also we have the fact that it in some games the Fury X actually matches a GTX 1070. If AMD can't at least beat the 1080 then the jump from Fiji to Vega has only been a 10-20% increase. I find that very hard to believe. In fact, with the clock speed increase I find it borderline impossible. But I don't know that for sure.

Enjoy your steak. :)
 
Could you explain how 1080 performance at the same price will be good?
The 1080 is a year old. There are TONS of different 1080's for sale, lets se how many cards AMD can produce. And the 1080 does not use HBM2, so if they offer it at the same price, it will probably hit AMD on the profit margin.

It is ONLY for people who ONLY want AMD. (But as i mentioned earlier, mining will probably save AMD)
 
Back
Top