AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

Agreed it seems a good way to get some of the more 'boring' items out of the way. I mean who really cares about 12nm low TDP CPUs? Sure it's great AMD is pushing for more of their low tdp cpus in laptops, but still boring. If these had been 7nm that would have been exciting. That said I tend to agree we won't see a lauch of the 7nm Zen 2 but rather more details about it and maybe a sneak peak at performance?

On the other topic AMD should have been providing those drivers all along. It should never have taken this long to get driver support. It's also still not available and who knows how long the growing pains and driver issues with plague them. Avoiding AMD low TDP laptops for now is the best option.

 

I agree until the part of avoiding those laptops. Primarily, many of the chips are in low power form factor laptops and Chromebooks. If the price is right, it makes sense as the Chromebook offerings are going against Intel atom CPUs. On the U and H series chips, it is a mix. Definitely need to price compare. If the price is lower than the six core Intel chips, them it may make sense. If price is the same, then Intel makes sense, depending on specs.

But, lots of Chromebooks and laptops using AMD mobile parts were released today, or announced today, so AMD likely put it out to control the brand, because otherwise it would be piecing together hardware from other manufacturers to get all the data.

As for the driver, agreed it should have been done long ago. With that said, Tim's critique on editing desktop drivers to get more performance is also seen on Nvidia drivers, although to a lesser degree. But, amd has been needing to not rely on hardware partners to push the updates, as they are very slow to certify new drivers.

Sent from my SM-G900P using Tapatalk

 

I agree with ajc9988.
Namely, if you're buying a low TDP AMD laptop, just make sure to check if it has a single or dual channel RAM configuration (and if its single, then check if you can install a second RAM stick inside for dual channel and obviously more RAM - if not, I'd advise in that case getting a similar system from a different/better OEM that didn't 'forget' to give a laptop dual-channel support or RAM upgradeability).

Drivers should be more or less settled now aren't they?
AMD made a commitment to release mobile APU drivers on a regular basis (which admittedly they should have done before - plus it shouldn't be that much of a hassle to do in the first place since all they would usually need to do is just make the desktop drivers mobile-compatible - and performance/stability-wise it should work which they can test on a similar system), so this kind of a problem shouldn't happen again.

Although I don't understand why is AMD reliant on hardware partners to push the updates for mobile APU's?
They don't seem to have these 'limitations' in case of desktop drivers (both GPU and chipset), though do correct me if I'm wrong.

 

So, for the drivers, due to integrating so many components, you can see driver conflicts. So system integrators often test with a set of drivers, certifying them so that they will all work and not conflict with each other. But drivers come out at different intervals for each component, making testing an ongoing process that requires resources. Because of this, AMD allows those manufacturers to control drivers for their clients. The problem comes when a manufacturer doesn't maintain regular driver updates for their systems. This is further compounded by reliance on windows driver updates to keep updating the system with what is certified for the OS. We regularly download graphics driver updates and chipset updates, so we don't think much of it, along with troubleshooting or rolling back drivers. But some users don't mess with all that. So there is a bit of a reason to hand it off to someone else, but due to the nature of some drivers, like graphics, users need to be able to get it and roll the dice, so to speak, because of software compatibility and performance.

This move just takes AMD in the right direction, doing something that should have been done long ago. Hopefully with increased revenue on the CPU side, some of those profits are being used to expand driver support, as well as getting people on site to help code for their hardware.

Sent from my SM-G900P using Tapatalk

 

Would waiting for OEM's be that much of a problem though?
I mean, take GL702ZC for example... it has desktop class hardware which was 'mashed' into a laptop effectively. And desktop chipset and gpu drivers work perfectly fine on the system.
The only issue I could find is that you can't clean install latest chipset/gpu drivers from AMD website directly if you want to preserve FreeSynch functionality on GL702ZC. For that, you need to first install original OEM drivers and then perform an 'Express install' to latest drivers as downloaded from AMD website.

All AMD really needs to do is release drivers for specific chipset and type of APU's (which would be based on desktop hardware anyway).

Unless we're talking about switching from an iGPU in the APU to the dGPU (that in itself might be tricky in regards to 'generic AMD drivers', but realistically, AMD should have the specs for APU's and possible pairings with mobile dGPU's anyway, so, again, I don't technically see too much problem on AMD doing it because same/similar desktop systems exist with APU's and dGPU's - so why not use that as a basis?).

OEM's are notoriously unreliable for software updates (that's something AMD cannot control and shouldn't be blamed for).
Asus effectively dropped GL702ZC support after the first quarter of 2018 (and its been officially released in late 2017 - so, roughly 6 months).
RAM support on GL702ZC is still limited to 2400 MhZ for example (even though 3000 MhZ and above for laptops have been on the market for a year now) , and there is 0 BIOS/UEFI support for Ryzen +, let alone Zen 2.
The original GPU/chipset drivers are still of 17.xx variety (released in April/May 2018 on Asus website, but based off older AMD drivers - I don't understand why ASUS hadn't used contemporary stable drivers as a basis when they released their own).

 

CES 2019: Lisa Su of Advanced Micro Devices I Fortune
Fortune Magazine
Published on Jan 8, 2019
AMD CEO Lisa Su joins Fortune’s Michal Lev-Ram at CES 2019.
http://forum.notebookreview.com/thr...ts-video-articles.826767/page-8#post-10844061

 

http://forum.notebookreview.com/thr...s-video-articles.826767/page-11#post-10844866



AMD CEO introduced Radeon VII during the company’s first CES keynote.
https://videocardz.com/newz/amd-introduces-radeon-vii-based-on-7nm-vega-architecture

AMD Radeon VII with 3840 Stream Processors
"The Vega II became VII. The card launched at CES features 60 Compute Units (3840 Stream Processors) and 16GB of HBM memory.

AMD Radeon VII features second generation Vega architecture based on the 7nm node, this is the world’s first 7nm gaming graphics card.

AMD has not confirmed whether Radeon VII is a brand name or just a single graphics card. This model has fewer cores than Radeon RX Vega 64.

The Radeon VII reference design is equipped with three fans and dual 8-pin power connectors.

The performance slides released by AMD, indicate that the Radeon VII is faster than GeForce RTX 2080 in selected games.

The card will be available on February 7th for 699 USD."

 

AMD Ryzen 3rd Gen 'Matisse' Coming Mid 2019: Eight Core Zen 2 with PCIe 4.0 on Desktop
by Ian Cutress on January 9, 2019 1:01 PM EST
https://www.anandtech.com/show/13829/amd-ryzen-3rd-generation-zen-2-pcie-4-eight-core

"...
+15% Performance Generation on Generation, Minimum.
During the keynote, AMD showed some performance numbers using the new Ryzen 3rd Generation (Matisse) processor. The test in question was Cinebench R15.

Our internal numbers show the 2nd Generation Ryzen 7 2700X scores 1754.

This new 3rd Generation Ryzen processor scored 2023.

This would mean that at current non-final clocks, the new parts give a 15.3% increase in performance generation on generation. Cinebench is an idealized situation for AMD, but this is not at final clocks either. It will depend on the workload, but this is an interesting data point to have.

Identical Performance to the Core i9-9900K, Minimum.
Our internal benchmarks show the 9900K with a score of 2032.

The 8-core AMD processor scored 2023, and the Intel Core i9-9900K scored 2042.

Both systems were running on strong air cooling, and we were told that the Core i9-9900K was allowed to run at its standard frequencies on an ASUS motherboard. The AMD chip, by contrast, was not running at final clocks. AMD said that both systems had identical power supplies, DRAM, SSDs, operating systems, patches, and both with a Vega 64 graphics card.

At Just Over Half The Power…?!
Also, in that same test, it showed the system level power. This includes the motherboard, DRAM, SSD, and so on. As the systems were supposedly identical, this makes the comparison CPU only. The Intel system, during Cinebench, ran at 180W. This result is in line with what we’ve seen on our systems, and sounds correct. The AMD system on the other hand was running at 130-132W.

If we take a look at our average system idle power in our own reviews which is around 55W, this would make the Intel CPU around 125W, whereas the AMD CPU would be around 75W.

This suggests that AMD’s new processors with the same amount of cores are offering performance parity in select benchmarks to Intel’s highest performing mainstream processor, while consuming a lot less power. Almost half as much power.

That is a powerful statement. ( ed: pun not intended)

There’s room for a little something extra in there. There’s not much room for a little something extra, but I’m sure if AMD wanted to, there’s just enough space for another CPU chiplet (or a GPU chiplet) on this package. The question would then be around frequency and power, which are both valid.

There's also the question of lower core count processors and the cheaper end of the market. This processor uses silicon from TSMC, made in Taiwan, and GlobalFoundries, made in New York, then packaged together. We have heard some discussion from others not in the industry that this makes cheaper processors (sub $100) less feasible. It is entirely possible that AMD might address that market with future GPU.

What AMD has plans for in the future, I don’t know. I don’t have a crystal ball. But it does look like AMD has some room to grow in the future if they need to."

 

Didn't I say a long time ago (when the rumoured engineering samples started showing up) that a 65W TDP 8c/16 th Zen 2 should basically achieve performance parity with i9-9900k if not exceed it, and that a 95W TDP version should handily beat it?

Interesting.
It appears that this new data seemingly confirms that estimation.

Oh well.

 

VEGA IS BACK! Radeon VII... 1080Ti Performance 2 Years Later! AMD CES 2019
The Good Old Gamer
Published on Jan 9, 2019
Today AMD had their much anticipated CES 2019 Keynote. Well it didn't go quite how we thought, but there was some Good, Bad, and some Ugly Truths to be gleaned from this. Vega is BACK, and Rebranded as the Radeon VII. Ryzen 3000 Showed up and Kicked Intel's BUTT!

 

50 W power difference isn't that big difference at all if they can push the clocks further and overclockable units must perform similar to Intel otherwise most of them go for Intel laptops.

 

Hi @Vasudev, I am responding here because I didn't want to start up a technical discussion about Vega II / Vega VII in the CES 2019 news thread...

I'm busy right now doing just that, posting CES 2019 news and other things, so let's talk about this after CES.

To prepare for discussion, please watch the AMD Vega release videos and articles here in this thread and in the CES 2019 thread, some cover the history of the AMD GPU's, and in short the Vega 7nm Professional and Gaming GPU's really are just process node shrinks + some rejiggering of the ROP / Cores vs the last few generations of AMD Radeon GPU's.

There are no new functional units like RT / Tensor in Vega.

Navi architecture is new, probably not out till after the CPU / APU's release in mid 2019, and we don't know what it will have.

I doubt AMD is going to approach DXR ray-tracing in the real-time sense any time soon. Their plan was always to provide creator tools so that developers can prepare for the future, even using the hardware for content creation now, and to delivery actual DXR Ray-tracing as a replacement for traditional rasterization in total - not just tacked on shiny crap - in future hardware when feasible.

Nvidia has jumped the gun IMHO, and screwed up. It's going to cost them and gamers in years of frustration and disappointment.

DLSS is just a tool to fix problems caused by RT, it's double filtering the video - RT slows things down - DLSS renders from a lower than native resolution and upscales to native - and Nvidia BS'ized it as TAA replacement, both video processing tasks make the video "double fake", nothing natural about that.

Anyway, have fun.

 

This was the thread I was looking for.... I was searching for Vega7 thread and couldn't find it.

 

Hey, I'll be working the next couple days. If one hasn't been created, could you create a thread for just Vega 7/Radeon VII to discuss the nuances of the card and comparative offerings (meaning comparisons to the 1080 Ti/2080/2080 Ti).

There are certain aspects that would turn here into a flame back and forth, but are worth discussing (just don't need a thread dump here).

In other news, @TANWare new bios released for the TaiChi now supporting spread spectrum option to get it running at flat 100mhz when disabled. Bios 3.35. Wanted to let you know.

Sent from my SM-G900P using Tapatalk

 

Looks like AMD used salvaged MI50 gpu's and used them for consumer space.
They should have disabled about 30-40% of that compute performance in the VBIOS in that case which would probably affect power consumption for the better and free up space for increasing core and VRAM frequencies further up without adversely affecting gaming (which seemingly can't use compute that well).

If they simply kept compute performance as available on V64 for example but ramped up the core clocks and bandwidth, the GPU would probably reach 2080ti levels of performance at similar or better efficiency.

Vega's issues remain the same... too high compute performance which has limited use in gaming.

Oh well, I guess its cheaper and faster for them to just salvage the MI50 discarded dies for prosumers without changing them much at all.

 

https://twitter.com/AdoredTV/status/1083140603712540672?s=20

“"Radeon 7" is so embarrassing I'd rather they hadn't talked about GPU at all.”

 

He's not thinking clearly. He's disappointed his "leak" didn't pay off right now - likely it's a roadmap more than a release map, but that doesn't lessen the blow to his ego / reputation for him right now because it "didn't happen":

Jim‏ @AdoredTV Jan 9
" Navi should be the "real" 4870 that AMD told us Polaris would be."

Jim‏ @AdoredTV Jan 9
" They won't give away that kind of information when it's so far out. Navi *was* planned for demonstration but they had to pull it."

Jim's still stuck on the giant void of "what didn't happen" that he was banking on. It's too bad the leak was likely a marketing target list rather than a production shipping list, he'll just have to be patient and wait till release proves him right, or not.

According to Jim's "leak" all of it was ready to release, with specs and prices, so a "planned to demo" isn't that either.

The 2080 just released, it's also rehashed / rejiggered / rebranded Series 10 hardware with completely useless RTX features - 1 game, really now it's a "must have", Pffft!!

If the Radeon VII matches the 2080 in release benchmarks and gaming tests - +- normal variance due to games favoring one architecture over the other - just like now - then it's a good release.

Nvidiots always feels they can crap on AMD for not matching the top end $XXXX GPU's, but if the AMD GPU matches at each price point Nvidia offerings +- a bit either way, then those are competitive offerings, not to be dismissed, perfectly functional and a great alternative.

If Intel can put out 95w TDP CPU's that crank out 500W under full OC load, and most use is 2x TDP power usage, then AMD can be a little hot on power usage too.

The Radeon VII isn't embarrassing, it's a great alternative to the $100 higher priced ( $799 right now on nvidia.com), and if you want to stop Nvidia's strangle-hold on the GPU market, it's a nice place to start if that is your budget level.

Otherwise the Vega 56 / Vega 64 are cheaper and available new and used, as well as the even lower price point RX 580 / RX 590 / RX 570/ RX 560 / etc.

The only "embarrassing" things I see is Nvidia fanboi's trying to justify spending way too much money on Nvidia RTX BS and same performance different name Series 10 rehashed GPU's.

 

There are only so many "salvaged" MI50 dies available, I doubt that's what's fueled or inspired the release. It's common practice with both AMD and Nvidia to tune features from the high end professional and release them as consumer GPU's.

AMD need's to be releasing new product in the GPU realm, and given Nvidia's poor RTX release and lackluster performance boost generation to generation, AMD found they had a GPU on hand that at least matched the 2nd highest performing consumer gaming GPU from Nvidia's "brand new" 20 series - the RTX 2080.

Even if the AMD GPU was originally designed and built for content creation and professional use at it's heart, it still can compete against the "brand new" RTX 2080.

Nvidia pushing RTX BS doesn't make RTX features a requirement now, in fact RTX has been and will continue to be a huge detriment and embarrassment to Nvidia.

RTX is stinking up otherwise acceptable performance bumps, as long as the new GPU's were shipped at the same price points. That would have been giving more performance for the same money as every other Nvidia release going back 8 generations.

Instead Nvidia employed magical thinking and RTX hand-waving BS to mesmerize their suckers (fans) into spending more money for the same performance as the 10 series. It's embarrassing.

It's also embarrassing how easy it was for AMD to compete against Nvidia's 2nd best consumer GPU, with a process bump from 7nm for Vega architecture.

If production performance matches pre-release benchmarks AMD is releasing on Vega VII, there will be a lot of happy new Radeon Vega VII owners.

Here's a look at AMD's own Radeon VII benchmarks for 25 games
These will have to do until we test the card for ourselves.
By Paul Lilly 17 hours ago
https://www.pcgamer.com/heres-a-look-at-amds-own-radeon-vii-benchmarks-for-25-games/

Spoiler: Article...

"Like everyone else, we're eager to get our hands on AMD's new Radeon VII graphics card and run it through a gamut of games to see how it performs, and especially how it compares to other high-end GPUs. That day will come, probably on or just before the February 7 launch date. In the meantime, it turns out AMD shared more of its own benchmark data than we initially thought.

During the AMD's keynote yesterday, Dr. Lisa Su talked about the Radeon VII while several graphs splashed on the big screen behind her. Those graphs contained only a handful of benchmarks, including a mix of content creation and actual games.

The folks at HardOCP noticed that AMD included a bunch more benchmarks buried in the footnotes section of its press release. It's a jumbled mess, so we cleaned things up for easy viewing. Here's a look:"

Spoiler: Article continued

"Keep in mind that those are all AMD's own figures, and not ours. According the footnotes, AMD benchmarked both the Radeon VII and Radeon RX Vega 64 in a testbed consisting of an Intel Core i7-7700K processor (seems a bit outdated, TBH) and 16GB of DDR4-3000 memory. AMD doesn't mention the motherboard or storage drive, but does say the games were benched at 4K at max settings.

All of the scores are averages of three runs with the same settings. We don't want to extrapolate too much from this since these are not our own numbers, but overall AMD shows the Radeon VII running over 28 percent faster than the Radeon RX Vega 64.

The biggest difference came in Fallout 76, which AMD shows running over 68 percent faster on the Radeon VII. On the opposite end, the smallest performance gap came in Hitman 2, where the difference was less than 8 percent (probably CPU limited based on our own testing of the game).

Take from this what you will. The bigger question in our mind is how it compares to Nvidia's Radeon RX 2080. That will be the Radeon VII's chief competitor at its $699 price point. On stage, AMD showed the Radeon VII running neck and neck with the GeForce RTX 2080 in Battlefield 5 and Far Cry 5, and leaving it behind in Strange Brigade.

That's promising, though the GeForce RTX 2080 has the benefit of real-time ray tracing and DLSS, features that are not available in the Radeon VII. It's also worth pointing out that in our own internal benchmarks using 17 games at 4k 'ultra' settings, the RTX 2080 comes out about 45 percent faster than the RX Vega 64.

We look forward to testing AMD's new GPU in the coming weeks."

 

This Radeon VII seems like an afterthought. It still uses a heap more power to match(ish) a non top tier Nvidia gpu at a similar price point. AMD must have limited supply and not care if it sells that much, or don't have room to move on pricing, simply the reception would be quite different calling it a winner, if it was $150 less than 2080

Not sure why Jensen was that negative about AMD though, I reckon that's more to do with background pressures

There was little need to point out Vega 20's lack of new features, every news article already has, and every reviewer will surely do that for them in a price-performance-feature comparison with 2080. And pointing out a need for extra testing on each freesync monitor imputes that there is a lack of Freesync QC already and brings it back to the existing Nvidia narrative of "it just works". All he's succeeded is stealing a few headlines by sounding like an angry twat *****ing about the competition, which hints to a mix of fear and arrogance

 

The Vega VII $699 is $100 less in MSRP than the 2080 MSRP at $799, you can always "want / ask" for more, but $100 is a nice drop.

If the production released Vega VII competes against the 2080 then it's the contemporary "brand new" competitor against the 2080 - they are both "brand new" GPU releases at the same performance / price point.

Given that everything builds on previous design wins, both are "rehashes" of previous technology.

And, "afterthought" or not, the AMD Radeon Vega II could be a viable alternative to the strangle hold that Nvidia ruthlessly tries to maintain and enforce in GPU's. As an alternative to the RTX 2080, the AMD Radeon Vega II could be the first choice for price / performance.

Who wants to reward behavior like Jensen Huang, or Nvidia?:

If you look at Jensen and Nvidia's "bad attitude" responses and anti-customer attitude across time, they are one and the same. As I've said many times in the past Nvidia's bad boy crap attitude toward customers and competitors is driven from the top, Jensen is proving that once again.

 

2080 FE MSRP is $799
Partner non-A variants are MSRP $699

 

But, check those 2080 AIB prices online, the prices are around $800 - some are less, some are more, it's still about $100+ more than $699.

The AMD Radeon Vega VII comes with a 3-fan design from AMD - AIB's might be different / more or less in cost and features - but given Nvidia was crowing about their 2-fan upgraded FE design, AMD has 1-fan upped Nvidia with 3 fan's.

16GB for 8GB more than the 2080, and more than the 11GB of the 1080ti, plus even the bone stock AMD cooler has better cooling than Nvidia's FE and AIB 2-fan designs.

AMD Radeon Vega VII performs as good as the latest Nvidia 2080 which performs the same as the legacy 1080ti, Vega VII costs less, Vega VII offers more VRAM and expanded 3-fan cooling, and Vega VII comes with far less Nvidia BS - RTX or otherwise.

 

This statement from Jensen reminds me of the statement Steve Ballmer has made after asking him about the iPhone.

We all know how this played out....

Gesendet von meinem CLT-L29 mit Tapatalk

 

Oh, I'm hardly slamming AMD's Radeon VII here (Nvidia seems to do this and its CEO acts like a proverbial 'baby' - which begs the question, if Radeon VII performance is 'lousy' as he says, why the need for offensive tone?).

All I'm saying is that AMD could have 'tuned' it a bit more for the gamers and improved efficiency by dropping compute power by a certain amount... however, on the other end, it IS advertised as a pro-sumer card... so, its more powerful in compute than 2080, and comparable to second best NV gpu in gaming performance (as you said) - so, its a bit of a trade-off... plus as several of us already claimed before, AMD doesn't 'have' to shoot for the very top.

Plus, they offer 16GB of HBM.
What I think that some naysayers are 'irked' about more is that the power consumption is higher on this GPU than even 2080ti and its a 7nm GPU (but then again, we don't know 100% if the TDP will be 300W... who knows, maybe AMD will be able to fine-tune it more... but undervolting is more likely to bring that power consumption down 20-30% - and if it does, it will end up being MORE efficient than RTX 2080 with a compute power of 2080ti).

Putting compute aside, AMD did increase the number of ROP's, etc... so it wasn't just the clocks that saw an increase.
If you noticed AMD Strange Brigade Vulkan performance, the difference there is much higher... I guess we still need to see how the GPU will perform with proper drivers, etc.

 

#CES2019 #AMD #NVIDIA
Who won CES 2019?
Coreteks
Published on Jan 10, 2019
In this video I take a look at the 3 main keynotes from CES 2019, starting with AMD's Lisa Su, then Jensen Huang's RTX 2060 announcement, and finally taking a look at Intel's Foveros Lakefield.

It's funny how disappointment clouds the rational view of even the best tech people, kinda of a lashing out. So far I have seen this happen, and then turn around and retract their statement when they come back down to earth.

That's why I don't like these speculations and leaks that lead up to release, as there's always a bit of future in them that goes past the next event that won't show up. There's always a bit of disappointment built in to these predictions and leaks almost every time.

In this case he should be happy that AMD has released the Radeon Vega VII and that it is competitive with the 2080 and priced it $100 less than the 2080 FE.

Who cares if the card is based on technology already recently released for professional GPU's, Nvidia does this as well too and has done so every generation recently, delivering into the datacenter and professional markets first.

Also, doesn't supporting DX12 automatically support DXR?, and on top of that AMD has explicitly said they support DXR in their drivers, so the Radeon Vega VII also supports DXR. I don't know why he said he didn't know...

There is always disappointment with pricing, even that $249 price for the 1060 was bemoaned because the last generation '60 card went for $199.

It's too bad that Navi didn't release, and same goes for the full Ryzen release, but things take time and patience is something you need to learn in this business and be quick on your feet to continue doing what you need to do with the tools available.

He seems to think he influenced AMD by putting out one of his videos, and maybe he did, but I think he's got it in his head if he complains about the now higher prices Nvidia has brought about - and AMD appears to be following - it's not going to make much difference.

The only difference we can make is to purchase intelligently, each of us on our own. And, to support the actions we agree with by purchasing as needed for our own use, and to not support the actions we don't agree with by not purchasing those.

But I think he is " cutting of his nose to spite his face", as not buying the Radeon Vega VII due to high price will show AMD that making the effort to sell high end GPU's - that cost a lot of money - isn't a viable business, and they will go back to selling low or mid range GPU's only. That's not the outcome I think any of us want.

So we are kinda " stuck between a rock and a hard place", we want AMD high end GPU's, but we don't want to spend that much money? What to do? I guess buy what you can afford, and if you can't afford it, buy what you can afford and hope the next generation from AMD or Nvidia drop price as well as increase performance.

One can only dream...

Radeon VII could be better than we thought?!
not an apple fan
Published on Jan 10, 2019
Radeon VII could be better than we thought?!

Nice, finally someone notices the MI60 has 64 ROPS and the Vega VII has 128 ROPS, which means the die was specially designed for gaming, and so the Vega VII dies aren't MI60 rejects.

 

There are plenty of aftermarket overclocked cards at around $700 on Newegg right now and an overclocked Gigabyte card was $669 on sale just the other day. It also has 3 fans! Speaking of which why is that somehow a feature? The Radeon 7 consumes 75-80w more than the RTX 2080 when looking at both rated TDP. That third fan is very necessary and not a feature to brag about. We also have no idea on the thermals of that reference design, good or bad.

 

Now here is where I'll disagree to a degree.

First, if you can buy a working 1080 Ti, buy it over the Radeon VII or the 2080. Should be $400-600, making it the top pick for this level of performance.

Second, there is one game with RT, which they lowered what it was doing to get better frame rates. DLSS isn't really out and doesn't look as good as a native render from what I've seen. Maybe that will improve with time, maybe it won't. Time will tell.

Third, the Radeon VII had more memory and bandwidth. There is only one game that needs that much memory, the division 2 running at 4K@60, but as more 4K and 8K displays hit the market, along with higher detail texture packets, that will be helpful.

As for power, the 1080 Ti and the Radeon VII seem to have similar TDPs. No one talked much about the 1080 Ti's power draw because it beat the Vega 64 handily. Overclocking, Vega sucked down a butt ton of power. We will need to wait for reviews to see if that is still the case. So you are correct the 2080 uses less power for the performance.

The way I see it, if you want memory heavy high resolution or if you want heavy AA, etc., which can be memory and bandwidth intensive, the Radeon may be for you. If you want DLSS or RT, then the 2080 is for you. If you are concerned about power, the 2080 is for you.

Overall, I recommend waiting for 2020 7nm from AMD or Nvidia for high end cards. With Intel entering the market, likely in the midrange, Nvidia and AMD will want to make statements on the high end.

Further, personal aesthetics, I like a triple fan look over a dual fan, but that is personal opinion and I'd personally put water blocks on the cards so that doesn't matter to me.

What this really does is eat into Nvidia's already dismal sales of the 20 series cards. That really hits Nvidia where it hurts when they are already struggling in regards to the crypto bust and the secondary market resales of their 1080 Ti.

I've already said last month when me and papusan were arguing that the exec leaving that thought up this product wasn't a bad thing because this product shouldn't have been considered being made. But it was made, it just doesn't shake up the market at all, which is the real disappointment, not the cards itself, which does fit into the market. Also, it is great for Linux gamers, but that is an aside.

Meanwhile, it seems my earlier predictions of 2H, likely Q3, launch of Navi was correct. My prediction on launch of 7nm mainstream chips was wrong.

But, this likely also means that the I/O chip can be cut down and binned, as the mainstream I/O die looks to be a quarter cut version of the full Epyc I/O die. It also means Navi may have the I/O removed from it's die, meaning they may use an I/O die to do a multi-die GPU in the next year or two. That is cool news! The I/O die allows for UMA batter of NUMA, which there is no current support for NUMA for games (enterprise workloads don't care, so didn't matter for them).

But that is my take.

Sent from my SM-G900P using Tapatalk

 

So you don't disagree with me at all then. You just need to catch up.

My default #1 recommendation for months has been to buy a used 1080ti and take advantage of the $500 price. I even posted half a dozen posts with links to ebay BuyItNow sales over those months to demonstrate the fact there were plenty 1080ti's available and plenty Sold at around $500.

Finally you caught up. Now it's time to move on.

The 1080ti is no longer my default recommendation now that AMD has a competitive GPU in the Radeon Vega VII . That's what's changed, AMD now competes in the next tier up in performance against the 1080ti and the new 2080.

There's no way I would recommend an Nvidia product against an AMD one that is price competitive at the same price. There are no used Vega VII's yet, so there isn't an option to purchase a cheaper used Vega VII, yet.

If you are going to get a new GPU then support AMD, don't continue to put money into the Nvidia ecosystem, otherwise your next upgrade might be a used RTX GPU - Yuck!!

We can invest in AMD, now that we can get an AMD Radeon Vega VII GPU that is performance level competitive with the 1080ti / 2080, the Vega VII is the best choice.

 

Zen 2 Faster than i9 9900K at Half the Power.
AdoredTV
Published on Jan 11, 2019
Intel gets rekt by lower-midrange Ryzen 5.

This Will Be Intel's Downfall
UFD Tech
Published on Jan 11, 2019

 

AMD might just save us all... - Radeon VII
Linus Tech Tips
Published on Jan 11, 2019


Amazing, Linus gets right on the first try. Thank you AMD.

AMD Radeon VII Hands On!
Joker Productions
Published on Jan 11, 2019


Bad advice from Joker, he's caught up in the Nvidia RTX BS web of promises with zero delivery.

The 2080 vs Radeon Vega VII performance if they are trading blows on average - some games better than others - then get the AMD GPU. We've been asking for a 1080ti competitive (now 2080 competitive) GPU from AMD, and now they delivered at the same price point, what's not to love?

Don't let Nvidia's RTX BS cloud your thinking, any more than "Hairworks" did, sheez, why can't you guys see that RTX is just a bunch of FPS killing needless "eye-candy" that looks like an old barnacle.

RT is a waste of time in real-time until the games are built ground up to support it without rasterization, otherwise it's just an exercise in frustration - building up nice looking scenes and then needing to cut cut cut RT effects until the FPS is minimally acceptable.

And, come on, 1 game after 5 months!?, and it's an FPS game you won't run with RTX ON due to performance, that's why you want an Nvidia GPU over an AMD GPU, Really?

For an $800 price point GPU (or $1300) that Shiny RTX BS isn't enough of a differentiator -> buy a brand new AMD Radeon Vega II 7nm GPU

Sheez, wake up guys.

 

What's embarrassing is his immature behaviour.
What's also embarrassing is that Radeon 7 offers same amount of teraflops as 2080ti and 16gb of hbm (not to mention the massive amount of rops) in a same price range as 2080 with early/preliminary performance apparently matching 2080 in gaming.

Wow... When one looks at it like that, it's Nvidia that seems embarrassing actually.

 

Specs on paper and theoretical performance don’t matter if it doesn’t perform in the real world. All the FLOPS in the world couldn’t save Vega 10 from being a flop. Am look forward to you eating your words when the benchmarks come out and it doesn’t match even 1080 Ti in DX11, let alone 2080.

 

There's a reason I said 'appear to'. Its not definitive.
Also, NV CEO behaviour is embarrassing... to the point of being infantile.

 

The behavior of AMD fanboys like the ones in this forum is just as infantile.

 

It can be, but I said nothing about fanboys. I was commenting on NV CEO's behaviour only.
Do you disagree that his behaviour is infantile, and is there any reason that I shouldn't have mentioned that it was?

 

Wait...
I don't think Vega VII has 128 ROPS.
According to this article, it has 64:
https://www.extremetech.com/gaming/283514-amd-radeon-vii-details-performance-projections

" Updated (1/11/2019, 6 PM EST): An AMD spokesperson has informed us that the widely reported GPU configuration for the Radeon VII is incorrect. The Radeon VII does not have 128 ROPs. Like Vega 64, Radeon VII is a 64 ROP card. The reports of a 128 ROP design were made by multiple well-regarded sites with reporters on-site at CES. It’s not clear why this information was reported incorrectly or how the error was introduced.

ExtremeTech regrets the error. All discussion and speculation of a 128 ROP design in the story below should be considered incorrect. Our story on the correct 64 ROP configuration can be read here."

 

No, I think he was blunt but truthful. Not everyone has humility in victory.

 

I guess that confirms it. It is cannibalized Radeon Instinct MI50, probably with gimped FP64.

 

https://www.eurogamer.net/articles/...-first-radeon-7-benchmark-results-in-25-games

 

Earlier you called AMD fanboys immature.
Does the definition of 'immature' somehow not apply to NV (and by extension its CEO) or their fanboys?

You are correct that not everyone has humility in victory... but isn't it a tad premature to claim NV 'won' anything when Radeon VII hadn't even been released yet?
And Lisa Su did say that Radeon VII will be better in some games, and not as good in others. So, there's no expectation on AMD's part to 'readily beat NV 2080'. On the overall scale, both gpu's will probably be comparable in gaming performance... and as for compute, well, that remains to be seen, but the Teraflops numbers seem to go into Radeon VII favour.

As for Assassin's Creed: Odyssey and Ghost Recon Wildlands... yeah, current extracts and comparison don't do AMD any favours there... nor should they when those games are better optimised for NV in the first place. But they do show 25% and 29% performance increase vs V64.

What I would be interested in seeing is professional software performance comparison between the GPU's in question... not just gaming comparison.
All this emphasis on gaming is way too one-sided for my taste.

 

AMD fanboys are the ones getting their feelings hurt because somebody said something negative (but true) about their favorite company, that's pretty immature.

 

Or put it the other way... AMD still hope for a new miner boom Tuned for *next* wave.

 

You yourself said: "Am look forward to you eating your words when the benchmarks come out and it doesn’t match even 1080 Ti in DX11, let alone 2080."

If anything, your response seems very defensive.... it appears far more 'fanboy-ish' and 'emotional' than mine was (but just because it appears to be, doesn't necessarily mean it is).

Now... what makes you think that feelings of any kind were hurt just because Huang said something negative?
You do realise that humans can make statements about other people behaviours by simply observing them without having an emotional reaction or being 'fanboys' of any kind, right?
Scientists do this all the time - they are required to be as neutral as possible when reporting observations of the environment.

Huang's outburst seemed quite unprofessional for someone in his position (people were fired for less - or are you saying that CEO's shouldn't be held to same [or even higher] standards as everyone else?).

As for that statement being 'true'... you do realise that Radeon VII still hadn't been released and we don't know how it performs in content creation and professional software vs 2080 (nor do we have a final say from independent reviews on its gaming performance either)?

In regards to Teraflops... many people already know they don't tend to matter that much when it comes to games... which is also why I said that AMD probably should have disabled some of the compute performance in the VBIOS and increased core and VRAM clocks further to increase gaming performance - however, since they are advertising the GPU for prosumers as well, its obviously not just a gaming GPU.

 

If I'm not mistaken, I have read something about AMD investing further into mining.
The following article may not indicate that AMD is hoping for a new miner boom, but they seem to be developing a new cloud computing infrastructure based on blockchain technology :

https://www.coindesk.com/ethereum-studio-consensys-teams-up-with-chip-manufacturer-amd

 

Yes! You made it out of the maze!

Exactly, it's a great GPU, it's not a magical $250 1080 level performance GPU, that may never come, but the AMD Radeon Vega VII 16GBis a great GPU that provides real innovation and value for money, and direct competition against the 2080 8GB / 1080ti 11GB.

Would it be nice if the Vega VII were $100 cheaper still than it is at $699? Making it $200 cheaper than the Nvidia 2080 FE at $799? Yes, and I'm still waiting for my "Birthday Pony", but I get on with life without it.

We don't know what's coming after AMD Radeon Vega VII releases from AMD February 7th, maybe AIB versions, maybe Navi down the road, maybe Nvidia 11xx GPU's or RTX price drops, but in any event if AMD provides a price and performance competitive GPU at a price point affordable to those interested, I'm going to recommend staying away from RTX and Nvidia, and getting the AMD offering.

Same goes for the RX 590 / RX 580 / RTX 570 / Ryzen APU's that at their price point are performance and price competitive.

Otherwise get 2 used 1080ti's in SLI rather than a RTX 2080ti or Titan RTX - save the money and your soul from RTX.

 

AMD Radeon VII Could Get Big Performance Boost by a Small Patch

https://www.ultragamerz.com/amd-radeon-vii-could-get-big-performance-boost-by-a-small-patch/

"AMD has already given their GPUs a big performance boost through a simple GPU driver. The things is the AMD Radeon VII is going to feature specs that are far superior to the Nvidia RTX 2080 but, it will perform like the RTX 2080. The benchmarks show that it performs very similarly to the RTX 2080, in some cases a little worse.

With superior specs, you’d expect it to perform at least 15% better than the RTX 2080 but, for some reason, it doesn’t. This is probably due to optimization issues. AMD is trying to get their 7nm GPU series released asap to compete with Nvidia RTX so they didn’t spend a lot of time on optimization. In a few months and a few updates, the AMD Radeon VII may be able to perform up to %20 better than the RTX 2080. By that point, they could start rolling out the lower-end Navi cards which will be much more optimized then the Radeon VII is right now. "

Of course, driver updates will inevitably improve performance, but whether they allow Radeon VII to consistently surpass 2080 in gaming remains to be seen (NV is hardly sitting still with driver updates - but obviously, it depends on whether Radeon VII will have things that AMD could leverage in games to improve performance by that kind of amount over 2080).
I suppose we will have to wait and see if those statements pan out... but more to the point, AMD needs optimisations from developers themselves so games and pro software makes better use of all that compute hardware.
Pro software could leverage it, but much of it still uses NV CUDA preferentially. As for games... most apparently don't use it virtually at all.
Still, I'm surprised by the gains Radeon VII got in Fallout 76 over V64... it does seem very consistent with the Teraflops amount the gpu has... so I'm wondering what could possibly facilitate that game to use that compute hardware (if that's the case)?

 

Well, the previous "big performance boost" was basically a performance fix, since all the games that had big boost before were unusually underperforming against competing Nvidia card and those boost brought up to the level where it was expected to perform.

Since Radeon VII is essentially a Vega 64 with few disabled CU with much higher clockspeed with double memory bandwidth, it really don't have any new features to bring any surprises. Vega 56/64 has been out for almost year and half and lot of optimization work is already in place. So most likely, just like how Vega 64 was against 1080, AMD sponsored titles will work better on Vega while Nvidia sponsored ones will perform better on 2080.

As for Fallout 76, higher than usual performance boost would mean it was more bandwidth limited than performance. Others with modest gains are pretty much expected range within clockspeed difference.

 

Yup, that's typical of both AMD and Nvidia, when new GPU's come out the optimizations follow in software for months and years with new optimizations specific to games and general optimizations for general boost in performance.

AMD calls it the "Fine Wine" approach, as the GPU ages and the software improves the "Fine Wine" gets better.

Reading the article your article quotes gives more detail specifics, and looking up release notes for each AMD software release will give more details on improvements and fixes. This article is 6 months old, there are newer updates, but this is a good example:

AMD giving Radeon GPUs a HUGE performance upgrade, AMD graphics cards
By Mohsen Daemi|July 13th, 2018
https://www.ultragamerz.com/amd-giving-radeon-gpus-a-huge-performance-upgrade-amd-graphics-cards/

"Earthfall launches out of early access on July 13, and RX Vega 56 users will see performance boosts of up to 28% with this driver, while RX 580 8GB owners will receive up to 22% extra frames. Beyond that, up to 27% additional performance is expected for those with the lower-end RX 560 4GB graphics cards. Adrenalin Edition 18.7.1 driver.

Here are the bug fixes and outstanding issues that Adrenalin Edition 18.7.1 driver delivers:

"Fortnite™ Season 5 may experience an application hang on some Radeon graphics products when throwing stars are visible on screen.

Hellblade: Senua’s Sacrifice™ may experience flickering or corruption on some Radeon graphics products.

Display modes may sometimes appear as not available when setting resolution or refresh rates of a display.

CorelDraw™ may experience slower than expected performance.
Memory clocks may remain at higher than expected values on some displays after changing resolution or refresh rates.

Some displays may exhibit black screen flickering when booting to desktop when using DisplayPort.

And here are the known issues that the company is working on resolving:

Virtual Super Resolution settings may not retain after updating to this Radeon Software release.

Wolfenstein II: The New Colossus™ may experience an application crash when disabling Asynchronous Compute via game settings.

Graphics and memory clocks may remain at higher than expected values after watching a video or using Windows DVR while in game. A workaround is to reboot the system.

Cursor or system lag may be observed on some system configurations when two or more displays are connected and one display is powered off.

Radeon WattMan gauges may report as zero on secondary graphics products in multi GPU enabled system configurations."

AMD has released a whole new software package with a lot more features since then and is also up to display driver version 2019-19.1.1, a lot has happened since that article, but in general new GPU's continue to improve in performance by releasing new software updates.

https://www.amd.com/en/support

Radeon Software Adrenalin 2019 Edition 18.12.2 Highlights

Radeon Software Adrenalin 2019 Edition 18.12.3 Highlights

Radeon Software Adrenalin 2019 Edition 19.1.1 Highlights

 

FYI - another rev of the Acer Predator Helios, now the 5000:
AMD Launch 2nd Gen Ryzen 3000 Series Mobile At CES 2019 !
OWNorDisown
Published on Jan 13, 2019
AMD at CES 2019: Ryzen Mobile 3000-Series Launched, 2nd Gen Mobile at 15W and 35W to compete against the 15w and 35w Intel chips with integrated graphics. Combined with Vega graphics expect both better CPU and GPU performance plus also improved battery life.
http://forum.notebookreview.com/thr...s-video-articles.826767/page-18#post-10847987

 

History says that you sshould't expect this to be hold truth yet at the same time you could. These articles and assurances let you think you get some amount of Boost in a half a year - 1 year top after GPU is released yet in real life for that exact amount of Boost you will have to wait 2 or even up to 3 years. So for some people this "Fine-Wine" Technology is advantage of bought GPU but for some people who wants NOW this undermines bought GPU.