Official AMD "Ryzen" CPU Discussion

[jefferz]

Nvidia Drivers Have Over 2 Times Failure Rate of AMD Drivers According to 3rd-Party Analysis

From your link:
In May 2018, AMD commission QA Consultants to independently evaluate the stability of several of the latest graphics drivers from both AMD and NVIDIA.

So they paid them to do “independent” testing and won.

From legitreviews.com
QA Consultants got back to our e-mail this afternoon and confirmed that all the samples were provided by AMD and passed along all the part numbers of the cards used.

So AMD provided the cards for the test, including their competitors.
That should be a massive red flag.
AMD even provided the drivers.

Also look at the driver used for the Quadros, it wasn’t a Quadro driver

Something is stinky here and it isn’t the tofu.

Throw all that posted above out the window. You know, as well as everyone else, that going from 1 driver to the next can drastically affect performance. So AMD bragging about this really doesn’t mean shit unless they’re going to continually test at every driver release.
If they wanted to claim there drivers are better, their testing should have been over a couple of different drivers.

 

[Madmick]

From your link:
In May 2018, AMD commission QA Consultants to independently evaluate the stability of several of the latest graphics drivers from both AMD and NVIDIA.

So they paid them to do “independent” testing and won.

From legitreviews.com
QA Consultants got back to our e-mail this afternoon and confirmed that all the samples were provided by AMD and passed along all the part numbers of the cards used.

So AMD provided the cards for the test, including their competitors.
That should be a massive red flag.
AMD even provided the drivers.

Also look at the driver used for the Quadros, it wasn’t a Quadro driver

Something is stinky here and it isn’t the tofu.

Throw all that posted above out the window. You know, as well as everyone else, that going from 1 driver to the next can drastically affect performance. So AMD bragging about this really doesn’t mean shit unless they’re going to continually test at every driver release.
If they wanted to claim there drivers are better, their testing should have been over a couple of different drivers.

LOL, so you spotted that. It's also not a survey of actual driver stability in games. It's a survey of their performance in a benchmark itself. It's a fucking benchmark. It's like saying that Chrome will be better in webware gaming because it scored better on the Sunspider/Octane/HTML5 benchmarks. It's asinine.

Yes, this is exactly what should incur suspicion, and it's ironic because I saw this just after posting this comment in the HWs yesterday, so it was on my mind when I came to plant this:

While this post holds a great deal of wisdom I feel the need to enlighten the forum with perhaps a bit more.

The USADA is not one of the "bad" institutions even if it ends up failing us in this case (in some cases, there is just too much money flying around). The USADA is the American branch of WADA, under the leadership of Dick Pound, which is the doping body that originally brought Lance Armstrong down, and ushered in the contemporary zeal for anti-doping, and the belief that doping is egregious enough to warrant it. It is WADA that is there alongside the ITF when Maria Sharapova trips over her own lies about taking meldonium for over a decade to treat a "heart condition", but it is the CAS (Court of Arbitration of Sport) that reduces her sanctions, and lends apologism to her lies. It is WADA that is there hip to hip with the IOC when the Russians abuse the Olympics in unprecedented and disgraceful fashion at Sochi, but it is the CAS that orchestrated their road back to competition in PyeonChang 2018.

WADA/USADA = good guy that is trying
CAS = bad guy that is monkey wrench for corrupt Big Men

The USADA has proved exceptionally inconvenient to the powers that be with their testing, and some of their top draws it has taken out since they have taken over anti-doping responsibilities. Things are a million times better off with them in the sport than if it was left solely to the UFC or "third party" testing bodies that are small outfits dependent on their client to exist; neither would the athletic commissions be any better.
Here’s how much control the UFC actually has over drug testing exemptions


https://en.wikipedia.org/wiki/Jon_Jones#USADA_violation_and_second_stripping_of_title

Jones dosen't have a single test in the database in 2018, so if you see Cormier-Jones scheduled immediately after Cormier-Lesnar, I think you can trust the fanbase-- or at least the integrity of the sport-- has been proper fucked in both cases.

I had confidence you would catch it since you're always on the lookout for AMD propaganda, but I wanted other posters to read it without being aided by a skeptical introduction.

 

[PEB]

Madmick looks really stupid (although I also liked NVIDIA as a buy-- I just didn't believe in the Switch).
Madmick looks incredibly smart.

View attachment 405765

In particular, if you'd invested $100k, for example, in May 2017 when the stock cratered, and I expressed bewilderment, noting that AMD was actually building cash reserves for the first time in a decade, and bumped GPU prices manufacturer-side due to the crypto surge, that stock would be worth $149k today; for roughly a 49% ROI in a 14-month period.

The funny part I was talking about AMD when it was 1.75 stock and at one point I believe went as far down as 1.25. The company was nearly delisted from NASDAQ is was then that Microsoft and Sony throw out a lifeline to the company. They had already started to work on the next generation of chips for the game consoles and Sony and Microsoft did not want to see them implode. Microsoft I think even fronted them 1 billion dollars at one point to secure parts for their Xbox One it was an unusual at the time big bet on AMD. But this is not the first time Microsoft did such deals tracing all the way back to the 90's because they knew what would happen if Intel secured complete control of the X86 chip business. Keeping AMD around keeps Intel pushing limits and pressing forward top it off AMD IP was unreal with the Graphics and CPU business under one roof.

 

[jefferz]

LOL, so you spotted that. It's also not a survey of actual driver stability in games. It's a survey of their performance in a benchmark itself. It's a fucking benchmark. It's like saying that Chrome will be better in webware gaming because it scored better on the Sunspider/Octane/HTML5 benchmarks. It's asinine.

Yes, this is exactly what should incur suspicion, and it's ironic because I saw this just after posting this comment in the HWs yesterday, so it was on my mind when I came to plant this:

I had confidence you would catch it since you're always on the lookout for AMD propaganda, but I wanted other posters to read it without being aided by a skeptical introduction.

I should have waited a bit to see if anyone took the bait.
Stuff like this is so frustrating and AMD has a history of this. They win on some odd weird benchmark and over hype like no one else's business.
Then the AMD fan boys run with it. Ugggh.
Don't get me wrong I like their products but I don't need all this other bullshit, it's on the edge of misleading your customers.

 

[Cygnus A]

I should have waited a bit to see if anyone took the bait.
Stuff like this is so frustrating and AMD has a history of this. They win on some odd weird benchmark and over hype like no one else's business.
Then the AMD fan boys run with it. Ugggh.
Don't get me wrong I like their products but I don't need all this other bullshit, it's on the edge of misleading your customers.

Intel is guilty of this too.

 

[Madmick]

Yeah, Intel is doing it, too, lately, but it's irritating when anyone does it, and that websites like Game Debate profit by proxy via clickbait.

 

[PEB]

Looks like AMD about to take the covers off another APU with improvements over the just released one with smaller chip size and improved power and a slight performance bump.

http://www.guru3d.com/news-story/amd-radeon-picasso-surfacesthe-raven-ridge-successor.html

 

[Madmick]

WCCF Tech, but still ridiculously exciting...
Exclusive: AMD’s Threadripper 2990X 32 Core Will Hit 4.0 GHz On Air On New Wraith Coolers

Elsewhere, it appears to have been spotted with an MSRP of $1,800. Currently the most powerful Intel that can hold a 4.0 GHz across all cores is the 10-core Core X series i9-7900X (currently $899).

Frankly, this is a hard to believe. I'm skeptical.

 

[jefferz]

WCCF Tech, but still ridiculously exciting...
Exclusive: AMD’s Threadripper 2990X 32 Core Will Hit 4.0 GHz On Air On New Wraith Coolers

Elsewhere, it appears to have been spotted with an MSRP of $1,800. Currently the most powerful Intel that can hold a 4.0 GHz across all cores is the 10-core Core X series i9-7900X (currently $899).

Frankly, this is a hard to believe. I'm skeptical.

I've seen that cooler talked about somewhere else. It's a dual tower with a fan in the middle and a fancy shroud. 92mm if I remember right.
I'm skeptical as well about the 4ghz on all cores. It would be a massive improvement if true, but the big question is what power consumption would look like.
I'm also skeptical on the 12/24 cores on the desktop chips.

 

[Cygnus A]

Zen + will still use AM4 right?

 

[Madmick]

Zen + will still use AM4 right?

Yes, but the Threadrippers are on the TR4 socket.

I've seen that cooler talked about somewhere else. It's a dual tower with a fan in the middle and a fancy shroud. 92mm if I remember right.
I'm skeptical as well about the 4ghz on all cores. It would be a massive improvement if true, but the big question is what power consumption would look like.
I'm also skeptical on the 12/24 cores on the desktop chips.

Even on the most powerful Noctua coolers I just don't see how this is possible; especially considering we've already seen the R5 and R7 processors, and those effectively added +200MHz to +300MHz (stable) to the first gen of Ryzen. I haven't been able to fill out my Spreadsheet, but here is where it stands:
Google Sheet: Turbo Modes

Assuming they could hold the +325MHz figure they achieved from the R7-1800X to the R7-2700X across all cores, which logically you would expect to be diminished, then the TR-1950X could conceivably jump from 3.7GHz to 4.0GHz across all cores. But that's still just 16 cores, and even that isn't realistic since doubling the cores means more overall heat to dissipate. I feel like we're being fed phase-change cooler figures as if they will exist in the real world. That's assuming there's an iota of reality to these figures at all.

 

[jefferz]

AMD is working on a new Ryzen/Vega APU for a Chinese console. I wouldn't be surprised to see it in the next Playstation or Xbox consoles.

A new semi-custom system-on-chip (SoC) was announced today that packs a quad-core Ryzen CPU and Vega GPU with 24 stream processors, all packed with 8GB of GDDR5 memory. AMD is working with Zhongshan Subor on the new chip, something that will be powering a "new gaming PC and upcoming console" in China. Inside, teh new SoC packs a 4C/8T Ryzen CPU at 3GHz, and 24 Vega CUs at 1.3GHz joined by the 8GB of GDDR5. We should expect Radeon RX 580 level performance, which is fine for 1080p and 1440p gaming at 60FPS depending on the game and detail levels used. The more interesting side here is that the Vega GPU technology is using GDDR5 memory
https://www.tweaktown.com/news/6270...m-chip-ryzen-vega-tech-pc-consoles/index.html

 

[rikwebb]

AMD is working on a new Ryzen/Vega APU for a Chinese console. I wouldn't be surprised to see it in the next Playstation or Xbox consoles.

A new semi-custom system-on-chip (SoC) was announced today that packs a quad-core Ryzen CPU and Vega GPU with 24 stream processors, all packed with 8GB of GDDR5 memory. AMD is working with Zhongshan Subor on the new chip, something that will be powering a "new gaming PC and upcoming console" in China. Inside, teh new SoC packs a 4C/8T Ryzen CPU at 3GHz, and 24 Vega CUs at 1.3GHz joined by the 8GB of GDDR5. We should expect Radeon RX 580 level performance, which is fine for 1080p and 1440p gaming at 60FPS depending on the game and detail levels used. The more interesting side here is that the Vega GPU technology is using GDDR5 memory
https://www.tweaktown.com/news/6270...m-chip-ryzen-vega-tech-pc-consoles/index.html

AMD and Sony are working on Navi for the PS5

 

[Plissken]

AMD and Sony are working on Navi for the PS5

Navi? Nintendo to sue.

 

[ssBaldy]

Navi? Nintendo to sue.

James Cameron too

 

[offshore33]

[URL='https://www.anandtech.com/show/13178/lenovo-posts-specs-of-amds-ryzen-3-2300x-ryzen-5-2500x']Lenovo Posts Specifications of AMD's Ryzen 3 2300X & Ryzen 5 2500X[/url]

 

[PEB]

 

[PEB]

 

[Madmick]

https://www.anandtech.com/show/13124/the-amd-threadripper-2990wx-and-2950x-review

I am certainly impressed. Knowing AMD the MSRP will bleed pretty quickly, too, because everyone still values single thread performance so much, and this places a level of power in editors' hands never before seen, but as I suspected, the all-core 4.0 GHz headlines were a comedy of AMD's usual marketing lies perpetrated through carefully staged leaks.

3.4GHz across all cores. The base frequency is quoted at 3.0GHz, so that's a +400MHz all-core Turbo, but it's not even holding the 3.7GHz all-core Turbo of Zen1's 16-core 1950X. Ian is in the process of writing the "4.0GHz Overclock 500W" section. I look forward to that analysis.

Remarkable achievement, here, but I don't appreciate the exaggerations and lies. Control your hype, AMD.

 

[PEB]

https://www.anandtech.com/show/13124/the-amd-threadripper-2990wx-and-2950x-review

I am certainly impressed. Knowing AMD the MSRP will bleed pretty quickly, too, because everyone still values single thread performance so much, and this places a level of power in editors' hands never before seen, but as I suspected, the all-core 4.0 GHz headlines were a comedy of AMD's usual marketing lies perpetrated through carefully staged leaks.

3.4GHz across all cores. The base frequency is quoted at 3.0GHz, so that's a +400MHz all-core Turbo, but it's not even holding the 3.7GHz all-core Turbo of Zen1's 16-core 1950X. Ian is in the process of writing the "4.0GHz Overclock 500W" section. I look forward to that analysis.

Remarkable achievement, here, but I don't appreciate the exaggerations and lies. Control your hype, AMD.

The second video does a pretty good analysis of the benchmark performance VS Intel's current offering and that their 16 core chip coming at the end of Aug. will have a 3.4 ghz base clock and overclock to 4.2 ghz. That being said Intel supposedly going to be sending out their next generation chip to be tested in a week I guess.

Keeping competitive pressure and price on Intel is forcing them to stay competitive good for everyone. Intel still leads in single core performance and that's likely driving AMD nuts by a small margin but still stings AMD keeps pricing pressure on Intel people will for only a marginal hit will keep AMD going.

I hear early next year AMD starts sampling 7nm EPIC chips for servers putting them at least 8 months to a year ahead of Intel and possibly 3 months ahead of Intel to 10nm. I would be interested in hearing what kind of performance hit they will get.