Vega/Navi Rumors (Updated)

[tential]

There's zero way that 1080 Ti is being sold at anything close to a loss.

This had nothing to do with pricing amd out of the market either. They had been doing this... For awhile before amd even had thought of pricing products this high.

This is why this annoys me so much on this forum. People think that Nvidia is playing a chess match vs amd. This isn't the case. Nvidia does this because it works....

They're a well run company. Management of a company isn't constantly talking like this. Nvidia had tried a higher price point if I'm not mistaken with the 780ti a lower with the 980ti and in the middle relatively with the 1080ti. Wow.... And next year when Nvidia does this again people will still be surprised...

This is why I was telling people to buy Nvidia stock instead of the gtx 1070 back when we had that conversation. Too bad I don't listen to my own stock picks. But hey sell your picks, index your portfolio.

 

[piesquared]

Very likely we will here more about Vega at AMD financial analyst day at the end of the month.

 

[tamz_msc]

If which connector on your card you connect a display to matters in this case, then the multi-monitor implementation is broken.

As for your numbers, I have no idea if the are correct or not.

Different reviewers had varying levels of success with multi-monitor power consumption due to the new memory state.
I can't understand why you'd call it broken.

 

[Mopetar]

Different reviewers had varying levels of success with multi-monitor power consumption due to the new memory state.
I can't understand why you'd call it broken.

If something isn't working all of the time as its intended to there's a word for that.

Oh yeah, broken (definition 2b).

I mean great if you get it working personally, but you wouldn't buy a car that had such erratic behavior and call it anything but broken. AMD will likely figure out what the bug is and get it fixed with a patch, but you don't patch something that isn't broken.

 

[Mopetar]

Sure, there is, because it limits what amd can sell vega for. It will hurt amd far more than nvidia. This move has been planned for a year. It was a preemptive strike that likely doesn't tell us anything.

This is monopoly building 101. You price your competition out of the market, even if you take a loss doing it. Though, I doubt nvidia is selling at a loss.

NVidia absolutely doesn't want AMD to fail. That would mean RTG gets sold off and guess who buys it. Intel.

Do you think NVidia wants to compete with an Intel that has a useful GPU division? NVidia might get some short term gain from being the only show in town, but Intel having better graphics IP and engineers would be a far worse threat to NVidia because Intel could fab the GPUs on their own, better process, devote an R&D budget that would allow for better products, and would absolutely want to get into the lucrative markets that NVidia has carved out for itself in HPC.

 

[tamz_msc]

If something isn't working all of the time as its intended to there's a word for that.

Oh yeah, broken (definition 2b).

I mean great if you get it working personally, but you wouldn't buy a car that had such erratic behavior and call it anything but broken. AMD will likely figure out what the bug is and get it fixed with a patch, but you don't patch something that isn't broken.

People with Asus ROG laptops were failing to install the latest NVIDIA driver that brought support for Win 10 Creator's Update. I'm curious to know to what degree the term 'broken' is applicable in this case to either Asus, NVIDIA, or Microsoft.

 

[guachi]

With freesync enabled my monitor OSD hz displayed between 40 and 62 which was LFC working to double the frames to bring them within my freesync range (30-75). I had no tearing, but it was stuttery (I mean it's only 20-30 fps after all ).

So no tearing, but I wouldn't want to game at 20-30 fps for any extended period of time (a single drop here or there maybe). Anything 40+ was super smooth though, even mid 30s weren't bad.

Thing I hate most about the 4k Freesync (and GSync) is that LFC is impossible as the Freesync range is only 40-60Hz.

Give me Vega and a better 4k screen up to 90 Hz and I'll be happy. Sure, I'd take 120 Hz or 144 Hz, but 90 Hz is enough for LFC to function.

 

[guachi]

Since the initial Vega talk at the beginning of the year we've had a Ryzen 7 release, a Ryzen 5 release, and a whole refresh of the Rx gpu line. Vega previews and spoilers in any of this would take media attention and hype away from those products.

When the media buzz about the Rx refresh dies down, BOOM VEGA.

This is the positive (and I hope correct) spin on AMD's strategy. Despite teething problems and no APU, I really like Ryzen. Like it enough to have bought one.

Though given how underwhelming the 500 series is, the media buzz won't take long to die down.

Technically, H1 ends in 2.5 months but I'd sure love to have something in May. I wonder if AMD, assuming they have good products, decides to do a staggered roll out like with Ryzen.

 

[SpaceBeer]

Thing I hate most about the 4k Freesync (and GSync) is that LFC is impossible as the Freesync range is only 40-60Hz.

Give me Vega and a better 4k screen up to 90 Hz and I'll be happy. Sure, I'd take 120 Hz or 144 Hz, but 90 Hz is enough for LFC to function.

Freesync LFC requires max freq is 2.5x higher than minimal. So 30-75 Hz range should be enough to have LFC

 

[Flapdrol1337]

Thing I hate most about the 4k Freesync (and GSync) is that LFC is impossible as the Freesync range is only 40-60Hz.

I don't have LFC on my 40-75 fps freesync screen, but it's fine, I have vsync turned off, tearing is hard to spot at low fps.

not great above 75, so you have to cap below that.

 

[Phynaz]

People with Asus ROG laptops were failing to install the latest NVIDIA driver that brought support for Win 10 Creator's Update. I'm curious to know to what degree the term 'broken' is applicable in this case to either Asus, NVIDIA, or Microsoft.

What does that have to do with anything Vega or Navi? Seriously, try to get out of AMD defender mode and stop attempting to deflect, this isn't an Nvidia thread.

Multi-Monitor has been broken on AMD cards for years, starting with the 6000 series in 2010.

 

[ElFenix]

If which connector on your card you connect a display to matters in this case, then the multi-monitor implementation is broken.

As for your numbers, I have no idea if the are correct or not.

Different reviewers had varying levels of success with multi-monitor power consumption due to the new memory state.
I can't understand why you'd call it broken.

If something isn't working all of the time as its intended to there's a word for that.

Oh yeah, broken (definition 2b).

I mean great if you get it working personally, but you wouldn't buy a car that had such erratic behavior and call it anything but broken. AMD will likely figure out what the bug is and get it fixed with a patch, but you don't patch something that isn't broken.

People with Asus ROG laptops were failing to install the latest NVIDIA driver that brought support for Win 10 Creator's Update. I'm curious to know to what degree the term 'broken' is applicable in this case to either Asus, NVIDIA, or Microsoft.

In case you weren't aware, the above is sarcasm.

AT Moderator ElFenix

 

[IntelUser2000]

NVidia absolutely doesn't want AMD to fail. That would mean RTG gets sold off and guess who buys it. Intel.

Nvidia is a better run company than Intel, which in turn is better than AMD. Plus Intel doesn't know graphics.

All their process advantage did not help them outside of their core competency. Heck, Nvidia has the market for Deep Learning and Automobiles that neither AMD or Intel has. AMD is nowhere at HPC either. Xeon Phi and Xeon is better there.

Other points: Yes it's possible that AMD knew what Nvidia was going to do with Ti and still be able to do little about it. If they were having some problems and that's the best they could do. Unless your product is decisively better you have to somewhat hope the worst doesn't happen. Which may not be the case here.

 

[Crumpet]

What does that have to do with anything Vega or Navi? Seriously, try to get out of AMD defender mode and stop attempting to deflect, this isn't an Nvidia thread.

Multi-Monitor has been broken on AMD cards for years, starting with the 6000 series in 2010.

I genuinely have no idea what this broken multi-monitor thing is..

I've had 2 and 3 monitor setups working flawlessly on HD7850, R9 290 4gb and an R9 390 8GB for years.

What exactly is it doing that's meant to be broken?

 

[SpaceBeer]

I would like to know that also

Nvidia is a better run company than Intel, which in turn is better than AMD. Plus Intel doesn't know graphics.

All their process advantage did not help them outside of their core competency. Heck, Nvidia has the market for Deep Learning and Automobiles that neither AMD or Intel has. AMD is nowhere at HPC either. Xeon Phi and Xeon is better there.

Other points: Yes it's possible that AMD knew what Nvidia was going to do with Ti and still be able to do little about it. If they were having some problems and that's the best they could do. Unless your product is decisively better you have to somewhat hope the worst doesn't happen. Which may not be the case here.

AMD didn't know graphics either until 2008. and they have still managed to make great products (first GCN architecture for example) while company was struggling, And both AMD and Intel are present in HPC. The fact that nVidia is leader at this moment, doesn't mean there are no other player. Intel has Xeon Phi and FPGA, in case they buy RTG (which is not going to happen), it won't take long for them to take the leadership.

Again, Vega project is going well. The product will be, if not the best, than good enough for most of the people and organizations
http://www.amd.com/en-us/press-releases/Pages/amd-vega-gpu-2017mar01.aspx

 

[Malogeek]

I genuinely have no idea what this broken multi-monitor thing is..

I've had 2 and 3 monitor setups working flawlessly on HD7850, R9 290 4gb and an R9 390 8GB for years.

What exactly is it doing that's meant to be broken?

The ability for memory clocks to reduce at idle.

 

[Crumpet]

The ability for memory clocks to reduce at idle.

Well that doesn't sound "broken" to me. What does that really do apart from slightly drive up power consumption?

 

[SpaceBeer]

Does the system RAM reduces clocks at idle? I always wondered why is that expected behaviour for GPUs, but not for system RAM. It might be quite useful on laptops and tablets

And this is not only AMD "issue"
https://forums.evga.com/FindPost/2358234

 

[KompuKare]

The ability for memory clocks to reduce at idle.

To be fair, the best known test of this is probably W1zzard's over at TPU, but Computerbase used to (past-tense as couldn't find an recent test) do this too. Except they tested with two monitors and three monitors. With two monitors Nvidia has a big advantage (for instance the GTX1060 has same power draw at idle with one and two monitors), but with three monitors, the difference is minimal.
Like I said they don't seem to do that test anymore, but here it is from the RX470 review
https://www.computerbase.de/2016-08...saufnahme-des-gesamtsystems-windows-desktop_2
HT4U.net do similar tests:
https://www.ht4u.net/reviews/2016/msi_radeon_rx_470_gaming_x_4gb_review/index13.php
Not going to post the images as this is anyway mostly off-topic for this Vega thread, but the point remains that for multi-monitor neither AMD or Nvidia are able to compete with Intel's IGP or even AMD's APUs when it comes to power usage. BTW, what the 500 series brings in terms of memory clocks should be possible to do with Wattman anyhow or even BIOS editing.

 

[ElFenix]

Well that doesn't sound "broken" to me. What does that really do apart from slightly drive up power consumption?

well, it's the "fix" for the increased power consumption of running ram at an intermediate level that may not be working in all cases on radeon 500 series cards, per the anandtech review.

however, considering the HBM-equipped fury didn't increase power consumption under multimonitor per the TPU review of it (largely because it just ran at 500 mhz all the time), this bit of conversation probably isn't germane to vega/navi and should probably stay in the polaris refresh thread.

 

[lobz]

In case you weren't aware, the above is sarcasm.

AT Moderator ElFenix

I just simply love the moderators' style here

 

[Arachnotronic]

They are easily making $500 profit on each. The production cost is nothing compared to the sales price.

That's too much, no way NVIDIA is making that much per card.

At a $699 selling price, $500 would imply a gross margin of ($500/$699) = 71.5%. Or, put another way, you're saying cost of goods sold is $199.

Think of all of the elements of the cost of goods sold associated with the sale of a GTX 1080 Ti FE:

- Raw die cost
- Packaging and test cost
- PCB cost (and all of the non-GDDR5X components on that PCB...)
- GDDR5X cost (don't underestimate this one!)
- Cooling shroud cost
- Shipping cost from contract manufacturer to distributor
- Cut that distributor receives for distribution

There is virtually NO way that all the costs associated with building and distributing the 1080 Ti (not including operating expenses like R&D, marketing, etc., just the cost of goods sold) is just $199.

 

[HumblePie]

Rumor has it that someone I know has his mitts on a couple Vega cards as of today. And that is all I can say about that rumor right now.

That someone being me I can neither confirm or deny.

 

[HumblePie]

Lol I will say my friend forgot one thing because I have no drivers for these things right now so I think installing these hardware goodies I recently received in the new 1800x ryzen system he gave me is going to be interesting.

Probably get a thumbdrive with it later.

 

[crisium]

Nvidia released a 520mm^2 GF110 variant for only $289 MSRP, so that was with the intent of profit. I think the cost of GPUs is higher now thanks to a much higher tranistor count, but the higher price is also due to a lack of competition and this affects our perception of their cost. I don't know if it's $200, but I would not be surprised.