Vega/Navi Rumors (Updated)

Page 109 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

Elfear

Diamond Member
May 30, 2004
7,126
738
126
36:07 if this is not posted yet.

So basically the same rumored specs that have been floating around for a while now. Wish we could have gotten more of a wink, wink, nod, nod from Linus as I imagine he has some insider info by now.
 

alcoholbob

Diamond Member
May 24, 2005
6,310
355
126
Isn't that not new info? Plus no clock speeds.

Yep, nothing matters until we get clock speeds. If all AMD can hit is 1200-1300MHz then it will be a 1070/1080 competitor. If they can hit 1600MHz then it's within striking distance of 1080 Ti.
 

Paratus

Lifer
Jun 4, 2004
17,121
14,489
146
Yep, nothing matters until we get clock speeds. If all AMD can hit is 1200-1300MHz then it will be a 1070/1080 competitor. If they can hit 1600MHz then it's within striking distance of 1080 Ti.
Even clock speed won't tell us everything. The new process means Vega can have up to 50% more transistors than fury for the exact same layout

  • 4096sp
  • 256TMU
  • 64ROP
That should mean clock for clock Vega is faster than Fury and clocks higher.
 

Guru

Senior member
May 5, 2017
830
361
106
Yep, nothing matters until we get clock speeds. If all AMD can hit is 1200-1300MHz then it will be a 1070/1080 competitor. If they can hit 1600MHz then it's within striking distance of 1080 Ti.
Clock speeds matter, but not in the way you think. They can be running it at 1000MHz and be at 1080ti level, it all depends how the core is build.

I mean AMD does have an advantage in DX12 and Vulkan and we've seen the 1080 or 1080ti don't really benefit a lot from async computing, from more parallel programming, so I think big Vega the one to compete with the 1080ti is going to be up to 15% faster in DX12/Vulkan and up to 15% slower in DX11.

I think there is going to be an incentive to buy Vega as it will be more future proof thanks to DX12 and Vulkan and getting more out of it as more games adopt DX12 and/or Vulkan.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
26,129
15,275
136
Even clock speed won't tell us everything. The new process means Vega can have up to 50% more transistors than fury for the exact same layout

  • 4096sp
  • 256TMU
  • 64ROP
That should mean clock for clock Vega is faster than Fury and clocks higher.
silly question maybe, but is 4096sp cuda cores ?
 

swilli89

Golden Member
Mar 23, 2010
1,558
1,181
136
Even if these Vega specs are true, we have no idea what the new architecture will perform on a per/sp basis.

Take Kepler to Maxwell for example, people thought the 980 chip would under perform because the specs looked like this

780 Ti:

Cuda Cores: 2880
Texture Units: 240
ROPs: 48
Clockspeed: 875mhz

980:

Cuda Cores: 2048
Texture Units: 120
ROPs: 64
Clockspeed: 1126mhz

Despite all that with Tile rendering Maxwell was way more efficient. Maybe AMD has applied this to Vega and its 4096 will be much faster and on an even faster process.
 

beginner99

Diamond Member
Jun 2, 2009
5,231
1,605
136
AMD needs some massive profits to be able to keep up with the crazy profits Nvidia has been enjoying, but I don't see them as a company who likes taking in tons of money, but would rather sell you a great product for very competitive price.

Problem they have on both fronts (CPU,GPU) is lack of market share. Once you are near irrelevant no one will optimize for your hardware. Ryzen isn't here to increase their margins, especially not in the consumer space but to get market share. While market share in GPU sector isn't great it for sure isn't as bad as it is on the CPU sector. I don't think big vega will be cheap and perform else AMD would have to lower price of RX 580 because if it delivers 1080 TI performance for say $550, Vega will have better perfromance/dollar than Polaris. Doesn't make much sense. So if it is within 1080 Ti reach, it will have to be $650. Also there supposedly is a small vega. So 2 chips and 4 skus that must fit between RX 580 and the top vega sku. That would not work if big vega were already as low as $499.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Woah woah woah, lets look at the actual FPS and we can tell these settings are not playable at all.



Yes and the 1080 doesn't even break 30 fps, heck even the 1080 Ti has 36 fps!



Ooooo this one is excellent, even the 1080 Ti can't hit over 32 FPS! Those look like playable settings to me



Mmmm yeah, that sweet 27.7 fps on a 1070 and 19.6 fps on a 980 Ti. Definite playable settings there too



Mmmmmm yet another one where a 1080 Ti isn't over 32 fps!



In actual playable settings it is fine. None of the cards, not even 1080 Ti were playable in those settings tested



You'll be turning down settings on a 1080 Ti to play those same games, unless you think $700+ GPU should only do 30 fps.
Finally someone. Thank you for sparring me the energy.

Sent from my VTR-L09 using Tapatalk
 

nad-

Junior Member
May 4, 2017
12
1
11
Those "specs" leaked in the driver have been known for awhile now, nothing new. For those who don't want to bother even watching that snippet:
  • 12.5tflops
  • 4096 steam processors
  • 2048 bit memory interface
  • 64 rops
  • 256 tmus
  • 8 ace

Wouldn't that TFLOP number and 4096 ALU count put it around 1530MHz? At least that's what the Radeon Instinct Mi25 should have, the consumer version should be higher clocked looking at AMD's Firepro/Firestream line.

I've heard that AMD's TFLOPs ratings are based on the base clock(<- Don't quote me on this) not the boost clock, boost should be higher in this case but that depends on the TDP of the chip.
 

maddie

Diamond Member
Jul 18, 2010
4,881
4,951
136
Problem they have on both fronts (CPU,GPU) is lack of market share. Once you are near irrelevant no one will optimize for your hardware. Ryzen isn't here to increase their margins, especially not in the consumer space but to get market share. While market share in GPU sector isn't great it for sure isn't as bad as it is on the CPU sector. I don't think big vega will be cheap and perform else AMD would have to lower price of RX 580 because if it delivers 1080 TI performance for say $550, Vega will have better perfromance/dollar than Polaris. Doesn't make much sense. So if it is within 1080 Ti reach, it will have to be $650. Also there supposedly is a small vega. So 2 chips and 4 skus that must fit between RX 580 and the top vega sku. That would not work if big vega were already as low as $499.
Why do you see both possibilities as mutually exclusive?

I know computing is binary, but do we have to be so blinkered?
 
Reactions: IEC

itsmydamnation

Platinum Member
Feb 6, 2011
2,923
3,549
136
I don't see how the can get Intel margins by selling Ryzen at the price they are selling it.

A lot of the costs are fixed, testing,packaging,transport etc so the only real difference is in the die size so something like 140-150mm for kaby lake vs 190-200mm for Zeppelin is just a few dollars (5-10) assuming a high wafer cost and only gets less if wafers are cheaper.

Intel also have much higher volume which helps to drive per unit cost down, but really even something like the 1600 @ 239 retail ( prob sold by amd to disty for something like $190) is probably in the 200-300% margin area. Its the low end that really drags margins down as even if you make a smaller chip there are still a lot of fixed costs to go with it so something like a $70 Pentium is probably only making 20-30 points of margin in retail.

What no one knows is what margin intel sells for to OEM's. So can AMD have Intel margins with the current Zen prices, they sure can, just give up the revenue of the low end of the market...........

But AMD have said over the longer term they are targeting 40-45% overall margin.
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,440
5,429
136
People continually forget two things when discussing market share and margins:
1) AMD's true market share includes console CPUs and GPUs. If looking to maximize revenue potential in a game, it is wise to optimize for GCN and get those cross-platform sales.
2) AMD has been surviving on very thin gross margins because their lack of competitive chips, especially on the CPU side, gave little leverage to demand more. As Maddie points out, because of how thin the margins were previously they can simultaneously increase margins (as they have with Ryzen 7) while also gaining market share. It should be self-evident that these things are not necessarily mutually exclusive.

Back on topic:
Even if AMD utterly failed and launched Vega with a draw stream binning rasterizer that doesn't work (i.e. drivers not ready at launch), it would still be a massive improvement just because the design appears to be far more balanced than Fiji, which is overly SP-heavy. Just the announced geometry pipeline improvements in Polaris and in Vega should fix the bottlenecks that are present in Fiji. Add higher clockspeeds and 8GB HBM2 vs 4GB HBM, and it would be a very capable GPU.

The real questions that will determine where Vega slots into the GPU heirarchy are:
1) What are the retail clockspeeds
and
2) Just how much performance does AMD's draw stream binning rasterizer and other software-dependent features give to Vega?
 

.vodka

Golden Member
Dec 5, 2014
1,203
1,537
136
Dunno.

According to WTTFtech, "AMD Taking The Covers Off Vega, Navi & Zen+ On May 16th". May 16th is AMD's Financial Analyst Day.


2017 Financial Analyst Day
Date: May 16, 2017
Time: 1:00 PM PT (4:00 PM ET)
Webcast: Webcast will be available prior to the event


Executive Presentations
Presenters Titles

Lisa Su President and Chief Executive Officer

Mark Papermaster Senior Vice President and Chief Technology Officer

Jim Anderson Senior Vice President and General Manager, Computing and Graphics Business Group

Raja Koduri Senior Vice President and Chief Architect, Radeon Technologies Group

Forrest Norrod Senior Vice President and General Manager Enterprise, Embedded, and Semi-Custom Business Group

Devinder Kumar Senior Vice President, Chief Financial Officer and Treasurer

So we might get some information, or this is another "quality" piece of news from that site.
 

Guru

Senior member
May 5, 2017
830
361
106
The biggest issue with AMD is that they forego the lucrative and expensive markets like servers and professional graphics and compute, which allowed Intel and Nvidia to swoop in and uncontested get all that money.

They can't gain any sort of significant profits with low and mid range cpu's and gpu's, which is what they've been doing for the past 4-5 years with garbage APU's for ultra low end, ultra low cost potato mashers.

I mean even though the Fury and Nano series weren't the best in terms of performance, it still gave them big profit boosts because of how expensive they were. The Nano for the niche market, Fury X with water cooling from start, it was something that enthusiasts and niche markets took to.
 

tamz_msc

Diamond Member
Jan 5, 2017
3,865
3,729
136
The biggest issue with AMD is that they forego the lucrative and expensive markets like servers and professional graphics and compute, which allowed Intel and Nvidia to swoop in and uncontested get all that money.

They can't gain any sort of significant profits with low and mid range cpu's and gpu's, which is what they've been doing for the past 4-5 years with garbage APU's for ultra low end, ultra low cost potato mashers.

I mean even though the Fury and Nano series weren't the best in terms of performance, it still gave them big profit boosts because of how expensive they were. The Nano for the niche market, Fury X with water cooling from start, it was something that enthusiasts and niche markets took to.
Why do you think that is the case? Because they straddle both the CPU and GPU market while not being a dominant force in either. Intel and NVIDIA stood no chance in mobile, what do you think will happen if they introduce high margin products all of a sudden? The ecosystem in entrenched, AMD has a big obstacle to overcome if they are to have the same levels of profitability as Intel or NVIDIA.
 
Status
Not open for further replies.
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |