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

I agree with what you said, but it remains to be seen whether AMD intends on competing in the high end.
Remember that what we do know of Navi is that its supposed to be a 'scalable architecture'.
How that translates in practicality is anyone's guess, but, it could hypothetically translate into Navi coming with MCM design after all like Ryzen did (in the end, MCM is not totally discounted as we don't have official sources saying that Navi, and for that matter Vega Instinct, won't come with it).

So everything is pretty much up in the air... as is speculation about Navi being 1080 level of performance.
I'm sorry, but even that is just a rumour at this point... nothing of the kind was actually confirmed by AMD as far as I know.

But given what we know of 7nm TSMC and how much of a bottleneck 14nm GLOFO process was, there might be reason to be cautiously optimistic.

 

Lisa Su is in post earnings report interviews starting tomorrow at 9am on CNBC - maybe Lisa Su will let some future product info go - the market needs some good news ...

Here's a review of the falling Semiconductor market (parts and stocks) that's affecting AMD as well, but AMD is still on top of the list of performance, so that's good - could be so much better if the market were in a bull run.

Nvidia is dropping even further today:
http://forum.notebookreview.com/threads/nvidia-thread.806608/page-82#post-10812850

Technician gives his instant take on AMD earnings
CNBC Television
Published on Oct 24, 2018
Todd Gordon, TradingAnalysis.com, reacts to AMD earnings. With CNBC's Bob Pisani and Melissa Lee, and the Fast Money traders, Tim Seymour, Karen Finerman, Steve Grasso and Guy Adami.


AMD’s stock craters more than 20% after weak graphics sales
Chip maker reports computing and graphics sales of $938 million while Street expected $1.05 billion
Published: Oct 24, 2018 7:37 p.m. ET
https://www.marketwatch.com/story/a...after-hours-on-weak-graphics-sales-2018-10-24

"Advanced Micro Devices Inc. shares plummeted in the extended session Wednesday after the chip maker’s outlook and revenue fell short of Wall Street estimates owing to weaker-than-expected graphics sales as blockchain-related became a negligible part of the company’s business.

AMD AMD, -9.17% which had been down as much as 25% after hours, were last down 22% in the extended session following the company’s conference call, Shares closed down 9.2% at $22.79 in regular trading.

AMD’s third-quarter revenue rose to $1.65 billion from $1.58 billion in the year-ago period. Wall Street expected revenue of $1.7 billion from AMD, which had forecast revenue of $1.65 billion to $1.75 billion for the third quarter. Estimize, a software platform that uses crowdsourcing from hedge-fund executives, brokerages, buy-side analysts and others, expected revenue of $1.75 billion."

Todays after hours drop post earnings report:

6 month ramp up, price right now is back to July 2018 level...

 

I dug into this earlier today. So, here are the highlights:

I was wrong and AMD didn't time the exit from the mining boom quick enough.

1. AMD has a fair stock of cards, but has plans to sell and absorb the hit over the next year. Because "client" includes mainstream CPUs and HEDT CPUs (Ryzen and Threadripper), and that Client and GPU/compute are grouped together, they saw the $100M hit due to lagging GPU sales that bottomed out more than expected. I think the fall out was compounded by Nvidia doing the long, slow release which literally took an entire quarter and consumers waiting to fully see and understand the new offerings. Either way, regardless of why, the GPU sales were WAY lower than AMD estimated and it hurt (in fact, it is the primary reason they missed market expectations that were about $60-70M higher than delivered)
2. AMD increased margins to 40%.
3. In the Compute/client/gpu segment, 70% came from Client, which is Ryzen and Threadripper. That speaks to growing CPU market (large amounts of growth). It also says that the 4% increase in margin is because AMD gets phenomenal margin at these prices on CPUs, far and away so much higher than the GPU margins. That also means that they wouldn't suffer too much in a price war.
4. On enterprise and embedded side, you have lagging sales as they are in the fifth or sixth year of consoles, so sales are much lower, that they are expected to go lower in 2019, but after that, they look good. That means we should be seeing new consoles around 2020 or 2021. Good news if that is what you are into.
5. AMD's server sales were able to make up for the expected slowdown in consoles. We didn't get near the details we did last quarter, but were told they are still expecting mid single digits in the server market by end of Q4.
   
6. Su directly stated that capacity at TSMC does not seem to be an issue for growth while producing on 7nm.
7. The WSA is still in negotiations, will tell the market when they get farther on the amendments.
8. The 7nm Vega is on track for a launch before the end of the year.
9. Around $97M was spent paying down long term debt, bringing the ratio to 2.2x, so good news.
10. AMD spent around $370M (cannot remember the exact amount, might have been like $367M or in there) of the $511 in operating expenses on R&D. There was a question on whether they plan on trying to reduce that or if 7nm requires it, etc. Su responded they are keeping to more growth in revenue than in operating cost (as a percentage for each).

So, lots of good parts in there even though the mining collapse really hurt.

We will find out more on Nov. 6th on the 7nm server parts at an enterprise segment event.

Either way, tomorrow is Intel's earnings! We might here them have to respond about 10nm!!! YAY!

 

I doubt Intel will say much about 10nm. They will keep any news about it as loose as possible. Both took a big hit on the open market with AMD taking more of the brunt of it. After hours AMD collapsed where Intel held its position.

I do not know if 7nm is too far off to make immediate gains for AMD. Hopefully though some good news on it may arrest the hard plunge that has occurred. It had better be some great news though.

My other concern is AMD let this happen at all. Agreed at around $35 a share the stock was over priced. This was mainly on 7nm hype and lack of it for 10nm. When intel started releasing FUD on 10nm AMD should have countered with 7nm info, although agreed this plays into what Intel is trying to learn. In the end this is part of Intel's game, if you want to compete you have to play.

Now agreed the stock would be down but for AMD to wait for a 50% price drop before doing so? While I love AMD products this is why I hate them also. Lastly I am not sure how 7nm news will help as the company will not realize real from it from it until 2019 and it will have no real effect on the current quarters weak guidance.

 

The company's don't have control over how the investors respond, or the state of the market when real news drops the stocks, AMD couldn't have done any more than Nvidia has done, which is nothing helpful.

Both lost after hours, for the day Nvidia lost more $ wise @ $28.76 and slightly less % wise overall, but the same order of loss - small $/% differences one way or the other are due to timing of the news vs the market, which is already tech and semiconductor shy - AMD is the only semiconductor stock still outperforming the S&P, holding onto more of it's run up in price, but that may not last for long as the whole market keeps reporting earnings drops.


Before the earnings news AMD was faring much better than Nvidia, and only catches up to Nvidia daily loss of $28.76 after hours, AMD day loss of $7.40.

Hopefully new releases from AMD will bolster GPU sales, and AMD continues to reduce RX / Vega stocks with more aggressive sales.

Nvidia needs to do this even more than AMD, lose some margin while dumping product inventory.

Both should bundle cheap memory and storage to offset the cost to the builder to incentivize them - build a bundle sounds like the best way to get it flexible enough to work.

 

Agreed they can not help stop investors reaction but they can dampen it somewhat with keeping the hype up. AMD has been fully tight lipped during the decline, it is not like it happened in one session. Putting them on the fence with no news and then dumping bad, if not at least sketchy, numbers and this is what happens. Again AMD's tight lipped policy is to blame some what.

I should note while all of this is off topic, it seems to be forcing information about the new process in the open, which is on topic. So please keep this on topic as I know it is tempting to hijack the thread.

 

AMD's Big Opportunity
BPS Customs
Premiered 13 hours ago

 

AMD's Lisa Su on quarterly earnings and her forecast for next year
CNBC Television
Published on Oct 25, 2018
AMD CEO Lisa Su sits down with the 'Squawk on the Street' team to discuss AMD's quarterly earnings, what missed and her yearly forecast for the company.

 

I was hoping for some info to start up some more hype, what a let down.

 

It's actually the second and third quarters with the release of second gen Ryzen and TR that things have started to shift on mind share, especially in relation to midrange CPUs. For example, the other day I brought up the 2600 vs 8400 from Intel. I do not believe we would have seen the exact same recommendation in the first or second quarters, especially with the articles or youtubers trying to show the 8400 outperforming on 1080p at medium (which is to alleviate bottleneck of the GPU, but also doesn't paint the picture of expected performance for someone in the $1000 budget category).

So mind share is changing, and better ram support to get the 3200MHz on ram to get the better performance out of IF is a large part of that change, along with platform stability. Me, I know how to overclock and use my hardware, so its always about best value with my expertise. But making the features and overclocking more accessible has helped. Also, people starting to feel comfortable to overclock memory and play with the BIOS has helped, which I've seen increase since the original Ryzen release, which should be seen as a positive on both the AMD and Intel sides. I think Intel's current actions and issues with 10nm have aided in changing mind share as well.

During the conference call, she pushed a couple events in November for 7nm CPU information, and confirmed Rome will happen, meaning the 64-core chip, rather than the 48-core Starship chip that was reported first, meaning they, for sure, will have 64 cores next year (to answer the 64?/48? parts marked on many tech journalists discussions of Zen 2).

We also will have 7nm Vega event before the end of the year, likely going to be in December and will likely give some info on Zen 2 to try to dampen Intel's thunder on their Cascade-X/SP chips released then on the 14nm++ process.

 

I'll look for a longer session if one happens; this one was too short to get into details.

The business news channels are doing a lot of back to back C class interviews due to all of the earnings reports right now, not to mention calming fears of investors with all the stocks dropping in price. You could see the flash alert for a Twitter CFO interview to follow Lisa Su's allotted time.

Maybe AMD will get a longer follow up session that's not so short and rushed to disclose more future goodies. But, by then the stocks should have recovered and it won't be needed.

Besides that AMD has tried to not do the hype train so as to not disappoint the public, that's why the 3rd party hyper's were doing so much damage - pumping up the product before release only to disappoint based on their claims - not on the unhyped claims of AMD.

Truth is it takes time and effort to produce these products, and there is no guarantee they will come out as expected the first go around, sometimes it takes 2 or 3 or more revisions to get things right.

7nm while a new process node might not offer all of the benefits in expected quantities, there may be more or less what is expected, but take a few iterations to improve - like 14nm for Intel, or 10nm for Intel... so there is some uncertainty to account for in the stock as well moving forward until 7nm product releases.

AMD Chief Lisa Su Says the Trade War Is an Opportunity
By Tae Kim, Updated Oct. 25, 2018 10:49 a.m. ET
https://www.barrons.com/articles/amd-ceo-lisa-su-interview-1540478258

"...
AMD on Wednesday posted adjusted earnings per share of 13 cents and sales of $1.65 billion for the third quarter, roughly in line with the analyst consensus of 12 cents and $1.7 billion. But its fourth-quarter revenue outlook was significantly below the Street at a range of $1.4 billion to $1.5 billion, compared with the $1.6 billion average forecast.

The quarterly guidance was the first time this year AMD forecasted sales materially below Wall Street expectations.

Before the results, the chip stock was the No. 1 performer in the entire S&P 500 this year through Wednesday’s close, with a 122% gain. AMD stock had rallied on rising expectations it will gain market share from Intel due to its increasingly competitive chip technology.

Su is still confident about AMD’s product pipelines for both of its CPU and graphics businesses, noting a tendency of some investors to myopically focus on short term quarter by quarter results.

We will be “competitive in high-end graphics,” she said. “We’re making high-performing quality products and building a solid long-term foundation.”"

Advanced Micro Devices (AMD) Q3 2018 Results - Earnings Call Transcript
Oct. 24, 2018 10:27 PM ET
About: Advanced Micro Devices, Inc. (AMD) - Lots of AMD coverage
https://seekingalpha.com/article/42...-amd-q3-2018-results-earnings-call-transcript

 

This was why I had lower CPU clock expectations on Zen+ than many others, because we have to be reasonable.

After thinking about the early clock rate leak, that is why I used a lower 4.5 to 4.6GHz all core clock to get to my estimate of 24% improvement in performance the other day (10% on clocks, 13% IPC). The IPC part seems like a done deal at this part, but the clocks on early silicon can vary greatly from now to final. Going with 5GHz anymore makes no sense as that was the target frequency for GF 7nm, which has been scrapped. So, all of that guesswork related specifically to the GF 7nm node went out the window. Nothing wrong with that, tbh.

I am wondering if AMD would use the same interposers they use for GPUs, though. Since they are likely to have payments for silicon to GF anyways, and since GF makes interposers and assembles packages, there may be an opportunity, since they would be forced to pay anyways, to use GF for packaging and interposers for Zen 2 and/or Zen 3. I'm sure AMD is running the finances on that to see if viable, and seeing if GF is able to get the yields, etc. Those interposers would be made on either the 60nm or 40nm nodes, which GF has a couple interposers in those ranges, so it could be done.

If that is the way they are going, I could also see why Su wouldn't want to discuss the WSA until they put out more information on Rome, along with that agreement tipping the hat to confirm or kill the idea of producing an uncore chip on 14/12nm node. If they give away too much too soon, then it gives Intel an opportunity to tweak their product stack instead of just competing on price, which is what AMD wants Intel to do.

Either way, TSMC has shown to have a working 7nm line, Apple is the driving test monkey for the fab ATM, and AMD will follow after that, while I see both companies participating in the TSMC development process closely. It will likely be better as AMD is now closer to where Nvidia was positioned on nodes back when AMD and Nvidia were extremely competitive (as detailed in AdoredTVs recent video).

But, the crypto bust hit hard, something I do not think either company was ready for.

But, as mentioned, we will see more on the server chips in two weeks on Nov. 6th.

 

Again I was hoping for some further info, but alas. As far as false information, you are correct but this is why I am looking for information right from the official source, no overhyping or guessing.

 

Well, this was pointed out by Laura Graves at the start of the phone conference for investors:

"I would like to highlight some important dates for you. AMD's next Horizon event is scheduled for Tuesday, November 6, 2018, where we will discuss innovation of AMD products and technologies, specifically designed for the datacenter on industry-leading 7-nanometer process technology. Dr. Lisa Su, President and Chief Executive Officer, will present at the Credit Suisse 22nd Annual Technology Media & Telecom Conference on Tuesday, November 27. And our 2018 fourth quarter quiet time will begin at the close of business on Friday, December 14."

So the Horizon event, where we saw the first zen with a boost of 3.4GHz, is what will happen in two weeks. I'm not sure what will be shown at the media and telecom conference. So...

 

AMD - The Bring Up: Coming Soon
AMD
Published on Oct 25, 2018
Want a behind the scenes look at all-things AMD? This new video podcast brings AMD news, expert guests, tech deep dives and more all in one place. Watch our new show, The Bring Up, on YouTube starting Monday October 29th.

 

I saw this the other day. If done well, it will be awesome. If done poorly, not worth anything.

Things I'd like to see:

Honest talk about getting certain frequency memory working on the platform, with going through how to find voltage settings, how to tighten subtimings, etc.

Experts within the company giving detailed talks in IF, interposers, etc.

Discussion of newer standards and patents being worked on.

Bring on guests from Microsoft and the Linux community to discuss the work on schedulers and platform optimizations

Analysis of performance, live benchmarking, etc., with guests from the XOC community to discuss optimizations.

A deep dive into how PBO works.

A look at how graphics cards process data, different types of microarchitecture and instruction sets, discussions of APIs like vulkan and DX12 with the upcoming ray tracing and AI elements of DX12, being introduced this spring with the next windows update.

A deeper look at APUs, at embedded hardware and semi custom hardware. This includes how tech is changing the auto industry, while showing off racing cars and self-driving vehicles, etc.

It needs to give enough general broad appeal to draw people in, but also be deep enough to satisfy those with more knowledge on the topic and not sound like a lecture or an ad. Some of it needs to talk about changes in the tech industry generally, issues being faced, etc., that may not be directly related to AMD or that AMD isn't a leader in at the moment. That can build trust with the audience, which builds some loyalty. Just have to be careful not to overstep and attack yourself in the process.

But, doing this and shaping or controlling the message, more than PR does or the business suits, and making your videos a reference for tech YouTubers and tech press means you can spin doctor more efficiently when needed, call out companies not playing ball once you get your base established (including biased companies in benchmarking or programs favoring Intel in their coding or compiling without reason, etc.). It just makes sense and I haven't seen any tech company go full in on it, instead doing short informational spots on how to use their products. There is an opportunity to make this something people want to watch. AMD can get access to people in the industry tech press cannot. It is using your clout to treat this like building a media outlet, not just a PR place or advertising.

I would point AMD to the show top gear at its prime, or even LTT and what they do. A goofy and fun experience that people want to come back and watch week after week. If AMD does that, this show really can help them accelerate, in my opinion. I have no idea why a company hasn't thought of that until now (granted, there are many issues and expense to do something like that).

Sent from my SM-G900P using Tapatalk

 

PowerColor Radeon RX 590 Red Devil leaked
Published: 26th Oct 2018, 10:49
https://videocardz.com/newz/powercolor-radeon-rx-590-red-devil-leaked

" PowerColor Radeon RX 590 Red Devil: new, but not quite
You’ve guessed it. The new Radeon RX 590 from PowerColor looks identical to Radeon RX 580 Red Devil Golden Sample. The beefy triple-slot cooler was designed to sustain a very high clock of Polaris 20, but is it enough for Polaris “30”?

AMD’s Polaris 10 launched a mid-range wonder, only to be refined twice and end up as a higher-end model two years later.

The RX Vega 56 Red Devil can be bought for around 350 USD, that’s including three games. The RX 590 is therefore expected to cost less. To be honest, I can’t really see another Polaris launching for more than 299 USD at this point.

Question is, would you be interested or would you rather wait for Navi?

PowerColor’s Radeon RX 590 Red Devil listing found
Published: 25th Oct 2018, 12:40
https://videocardz.com/newz/powercolors-radeon-rx-590-red-devil-listing-found

AMD Radeon RX 590 Final Fantasy XV benchmarks results emerge
Published: 26th Oct 2018, 12:16 GMT
https://videocardz.com/78757/amd-radeon-rx-590-final-fantasy-xv-benchmarks-results-emerge

Final Fantasy XV GPU benchmark page lists Radeon Vega 20
Published: 25th Oct 2018, 13:03 GMT
https://videocardz.com/78741/final-fantasy-xv-gpu-benchmark-page-lists-radeon-vega-20

 

The problem is announcement of events in the future will not arrest the decline of stock now or the immediate future. It may signal some stability of the stock coming but does nothing to help todays stock investors. even just a teaser of the 10/29 event may have helped, but no.

 

AMD Ryzen™ Threadripper™ 2970WX Processor
https://www.amd.com/en/products/cpu/amd-ryzen-threadripper-2970wx

AMD Ryzen Threadripper 2970WX (24-Core/48-Thread) Processor 4.2 GHz Max Boost 76MB Cache (YD297XAZAFWOF) by AMD
https://www.amazon.com/gp/product/B07JBQJ1D9

AMD Ryzen™ Threadripper™ 2920X Processor
https://www.amd.com/en/products/cpu/amd-ryzen-threadripper-2920x

2920x not yet listed for sale that I can find...

Threadripper 2970WX & 2920X Review, AMD Effectively Eliminates Skylake-X
Hardware Unboxed
Published on Oct 29, 2018

AMD Threadripper 2920x and 2970wx Full Review & Overclocking
OC3D TV
Premiered 106 minutes ago
https://www.overclock3d.net/reviews/c...

Spoiler: Click to See Video

The Bring Up: Episode 1: AMD Ryzen™ Threadripper™ Processors
AMD
Published on Oct 29, 2018
We Bring Up: 2nd Gen AMD Ryzen™ Threadripper™ Processors, Shock Tests, and Robert Hallock’s “Hot Gos”

Spoiler: Table of Contents:

01:15 Cavin and Bridget talk 2nd Gen AMD Ryzen™ Threadripper™ Processors
01:33 What are the advantages of having such a big chip?
02:25 Upgrade experience
05:07 Cavin visits AMD's “Shock and Vibe” testing room
06:52 Shock Testing Cavin
08:28 Product Showcase: HP® ENVY x360
08:57 2nd Gen AMD Ryzen™ Threadripper™ Processors “AMA-MD” with Robert Hallock
10:28 Simultaneous Multithreading (SMT)
16:38 Linux
19:56 Gaming
21:24 “Hot Gos”: Dynamic Local Mode
Related Links:
Dynamic Local Mode:
https://community.amd.com/community/g...
AMD 2nd Gen AMD Ryzen™ Threadripper™ Tech Talk:
https://www.facebook.com/AMD/videos/2...

AMD Threadripper 2970WX and 2920X Unleashed!
Tech of Tomorrow
Published on Oct 29, 2018
AMD Threadripper 2970WX and 2920X Unleashed!

Spoiler: Click to See Video

Starts @ 00:54

 

Reviewers try and shoot Intel down for not trying to compete on price. They do not have too yet. Their sales are still going well enough from the fan base. There is also not a huge supply of the new chips. So why lower prices to increase demand?

 

I, instead, think they are saying AMD is better value compared to Intel chips, which I tend to agree with. And considering you can get a 12-core 1920X for the price of a 9700K, needless to say the value if you need productivity versus the 9900K, or looking at the 2920X for $650 and 1950X like you and I have for $680-700, is unreasonable when considering the 9900K on price is going between $520-580 in most cases, while having severely limited supplies and likely not having that alleviated until December.

I feel, at this point, they are trying to gently say wait for price drops and make smart decisions with purchases.

Sent from my SM-G900P using Tapatalk

 

Speaking about intel, did that fraud analisys and comparison company which revealed "50% gaming advantage" of intel CPU over AMD published anything regarding their mistake cutting AMD corrs count to half on testings?

 

Tests were redone and while no longer twice as fast Intel's response was 'we are still faster in gaming'.

 

They got roasted (the 9900k roasts so well... ) by so many, check out the video / article links posted here and in the 9900k thread, and trounced in real review comparisons the day of release, I think Intel / PT have slunk away never to be mentioned again.

 

Another leak about Epyc.

 

The numbers are confusing, if it is true that the core counts are doubled being overall 70% faster is an IPC loss per core overall. So it could be true Rome is having a clock issue or something else all together. Without hard data there is no real wau to know if this is good news or fudging the numbers on bad news.

 

Not necessarily. Remember, they are either using IF for MCM or a new interposer or new interconnect. Because of that, you are looking at 8 dies having to communicate, having to wait on each other on scheduling, etc., so 70-80% performance may very well be inline.

Think of it similarly to SLI scaling. When optimized you might get over 90%, but most times you get 70-80%. Here, we are seeing the schedulers maybe not fully utilizing the new uarch and the extra four dies on package, or the uncore chip comms effecting the scaling.

So, two things here are worth noting: 1) he mentioned there is likely a B stepping coming soon, which may increase scaling or resolve certain other issues which may be effecting performance (remember, wide availability isn't expected until close to Computex with the release being around CES, which is two full months away, with the side note of frequencies shown at the Horizon event and CES being below what was seen in final silicon), and 2) there are scheduling and other aspects of the uarch to consider. We saw higher 80 to low 90s on scaling for TR, but I didn't look at the scaling efficiency on Epyc, which I think was in the mid to low 80s. So the scaling doesn't seem too far off the mark for doubling the number of dies present.

 

As I said confusing, we can only speculate and they left it for speculation both ways. Again we need hard numbers as in specs and benchmarks.

 

I hear ya. It is still a leak, so we are seeing or hearing what the leaker had to give or wanted us to see. But, in one week, we will likely have more info with the Horizon event Nov. 6.

 

Threadripper 2970WX & 2920X Gaming Performance, Can You Make Do?
Hardware Unboxed
Published on Oct 30, 2018
Threadripper 1920X, Threadripper 2990WX, Threadripper 2970WX, Threadripper 2950X, Threadripper CPUs

An Epyc Update
AdoredTV
Published on Oct 29, 2018
All roads lead to Rome.

AMD VS Nvidia : Was Polaris a Success?
2kliksphilip
Published on Oct 29, 2018
With the power of hindsight, I look reflect on Polaris and judge whether the Radeon 400 series was a success.

Spoiler: Click to See Video

0:39 - The Story of the Radeon 4800 series
1:58 - AMD falling behind
3:31 - Polaris
4:19 - The previews
5:30 - Was Polaris like the 4800 series all over again?
7:31 - VS Geforce 900 series
8:19 - Efficiency Wars!
9:42 - The Geforce 1060
10:20 - The rest of the lineups
11:58 - How many were bought? And reasons
16:25 - Mining
17:24 - Conclusion

Spoiler: Details...

My Polaris Preview video: https://youtu.be/c7vAN3IVBdY
Adoredtv's undervolting video: https://youtu.be/nWDgrAQ678g

This video started as an afternoon project and became a 4 day slog, going through several different versions. I wanted to make this video easy to understand, fast-paced and somewhat entertaining, though I had to skip over some of the smaller details along the way. Behold, my disclaimers:

* I'm not too interested in getting into an Nvidia VS AMD argument, I haven't deliberately twisted the figures to make one company look better than the other. Treat this as a rough overview of the architectures, and if you want to know more then I suggest you look up detailed reviews yourself.
* There's also an 8 GB 470 and 4 GB 480.
* The 4800 VS Geforce 200 benchmarks were from Anandtech articles, not all games were included in the comparison as not all cards were shown for each. I did my best to feature all of the games that were benchmarked on all of the cards. https://www.anandtech.com/show/2556
* I don't recall the Radeon 4870 being QUITE so fast compared to the Geforce 280. I notice the 94% was in part from a strong Bioshock performance... but why not. It's still a game.
* The 400 VS Geforce 900 benchmarks were from Guru3d reviews, as they had a better range of games and comparisons. Again, I tried to find games and settings that were tested for all graphics cards in this video. https://www.guru3d.com/articles-pages...
* The Geforce 1060 3 GB was problematic as they had changed the list of games reviewed by then. Rather than to cut the games out, I decided to guess the performance from the other results and to make it clear with a big colourful disclaimer in the video itself! https://www.guru3d.com/articles-pages...
* Geforce 970 didn't seem to be benchmarked on Doom Vulkan, so the number you'll see here is from OpenGL, which may have dropped average performance by like, 1%.
* You could probably also undervolt a Geforce 1060 to get better performance.
* Pricing was also difficult. The Radeon 460 was released for more than $99, but because I was comparing it to the Geforce 1050 I used the price it was at upon the Geforce 1050's launch for fairness. I also had trouble finding the Geforce 970 and 980 prices at the time of Polaris's release. It's so hard to do!
* Yes, Steam Hardware is skewed because it represents Chinese LAN cafes as well. But that's because they're all using Nvidia cards
* That being said, I'm proud of what I've found in this video. I've been fascinated by how Polaris fared compared with the older 4800 series and with this video I think I've found some closure. I hope that you enjoy this video.

Radeon™ Vega Mobile: Remarkable Graphics Performance for Premium Notebooks
AMD
Published on Oct 30, 2018
Radeon™ Pro Vega 20 and Radeon™ Pro Vega 16 graphics are at the heart of innovative and high performance next-generation notebooks. These new AMD Radeon™ Vega Mobile graphics processors are available as configuration options on the 15-inch Apple MacBook Pro starting late November. Learn more about Radeon™ Vega Mobile GPUs here: http://bit.ly/2AA9uvy

Spoiler: Click to See Video

How to: Celsius Series AMD Install
Fractal Design
Published on Oct 30, 2018
This brief video walks you through the installation of a Celsius cooler on an AMD-based system.

Spoiler: Click to See Video

 

The 2920x and 2950x should game as well as the 1950x and that is fine so long as you do not need the extreme frame rates of some super high end monitors. The WX series has its issues but that is to be expected.

As far as they have leaked what they want too, of course. It is a totally confusing group of info I am sure is designed to make the CPU's look better than they are but to what degree? TBH those numbers can be skewed from the CPU being a total failure to a honest improvement or anywhere in between.

 

AMD’s Vega Mobile Lives: Vega Pro 20 & 16 in Updated MacBook Pros In November
by Ryan Smith on October 30, 2018 5:00 PM EST
https://www.anandtech.com/show/13532/amds-vega-mobile-lives-vega-pro-20-16-in-november

Buried towards the tail-end of today’s MacBook Air press release from Apple was a second announcement: that they’d be releasing a minor MacBook Pro refresh next month.

Though curious in and of itself – the MacBook Pros were just updated in July – more interesting is what they’d be refreshed with: new AMD GPUs.

Announced by AMD at the same time, we now have confirmation that AMD’s missing mobile GPU is finally shipping. It will be arriving first as the Radeon Pro Vega 20 and Radeon Pro Vega 16, upgradable graphics options for the 15-inch MacBook Pro set to become available next month.

Spoiler: Lots of Details...

As a bit of background, AMD first announced the Vega Mobile GPU – which we believe is codenamed Vega 12 – at the start of the year at AMD’s CES Tech Day. The company stated the GPU would be coming later this year, and even showed it off. While details were scarce at the time, it was announced that it would be a smaller, lower-power GPU specifically for mobile devices. And of particular note, it would have a z-height of only 1.7mm, the same as AMD’s mobile Polaris 11 GPU, making it suitable for installation in relatively thin laptops.

However between CES and today, AMD has been completely silent about the GPU. Even when I reached out to them I couldn’t get a solid answer, and to be perfectly honest, I had given up on seeing the GPU. Sometimes products hit too many problems, or can’t find interested customers; it happens. With everyone (sans Apple) already having their holiday 2018 products on store shelves, it seemed like AMD had all but missed their launch window, and with it, any real chance of adoption.

Instead, as we now know, it will be making it to commercial products after all as an upgrade option for the 15-inch MacBook Pro. And while the lack of a shipping date for the new models is not especially confidence-inspiring given everything else that has been going on, none the less I have to admit that I was premature in giving up on AMD.

The flip side to this however is that because the first Vega Mobile SKUs are going into Apple products, AMD has limited what they’re saying about the products, as is traditional for Apple suppliers. So while the company is proud to announce the part – they even have a short video showing it off – specific details are a bit thinner.

As a result, the following is a mix of details from AMD and some guesses on my part:

Perhaps the most important bit of news here is that the new GPU is definitely an actual Vega GPU. This is no mix-and-match semi-custom GPU like the “Vega M” GPU in Kaby Lake G – which turned out to be Polaris’s graphics core with Vega’s memory controller – but rather it’s a bona fide Vega GPU with all of the features that entails, including Rapid Packed Math. So, if only at long-last, we finally have a second discrete GCN 5 GPU from AMD.

AMD’s announcement and the resulting SKUs also strongly point to this part having a total of 20 CUs (or 1280 SPs). As a mobile part we were already expecting it to be less than half of a Vega 10 (64 CUs), but this is actually smaller than we were expecting. In practice this puts Vega Mobile much closer to Polaris 11 than it does Polaris 10 or Vega 10; in fact even “Vega M” has more CUs. Which again, we’re looking at mobile parts here, so it becomes a matter of what’s viable in terms of chip size and power consumption. But it also means that we shouldn’t expect to see Polaris 10-like performance out of the part, while the performance improvements over Polaris 11 remain to be seen. Apple for their part is touting a “60%” improvement, but they don’t actually specify what they’re comparing the new GPUs to.

Meanwhile AMD was also able to send over clockspeed information for both SKUs. The Pro Vega 20 has an “engine clockspeed” – which I’m interpreting to mean the boost clock given that we’re talking about mobile parts – of 1300MHz. Meanwhile the lower-end Pro Vega 16 tops out at 1185Mhz. The Pro Vega 16 also only features 16 CUs out of 20, so on-paper we’re looking at around 70% of the Pro Vega 20’s theoretical throughput.

Past that, we don’t have any further confirmed information on the GPU itself. Specifically, we don’t know the number of ROPs or the clockspeeds. 32 ROPs is likely since that would be half of a Vega 10, but then “Vega M” included 64 ROPs anyhow. We also don’t have a good estimate on die size at this time, as we don’t have better pictures of the GPU. What we do know however is that the chip is being fabbed on a 14nm FinFET process, which means it’s almost certainly being done by long-time partner GlobalFoundries.

Moving on, feeding the new GPU is a single HBM2 memory stack, which means we’re looking at a 1024-bit wide memory bus. Because these are mobile parts, AMD has clocked them fairly conservatively, with a memory clock of 1.5Gbps. Even then, a single HBM2 stack offers 192GB/sec of memory bandwidth, a more than 70% improvement over the best Polaris 11 product and even more over the various mobile SKUs which Apple uses. So there is far more bandwidth available here to feed the GPU. Meanwhile it’s worth noting that Apple’s SKUs are all using 4GB stacks here, but if AMD wanted to, they should be able to offer 8GB (8-Hi) configurations as well.

Once Apple’s machines do ship, it will be interesting to see what performance is like. The Radeon Pros Apple current uses are no slouches, but they are limited by their relatively small size, not to mention the TDP limits of the notebook itself. Improving performance on the MacBook Pro means not only offering a larger GPU, but also better performance-per-watt than the current GPUs. In the desktop space AMD has struggled mightily here as they’ve aggressively pursued absolute performance, however we’ve never seen what Vega is like in a more conservative configuration. So it will be interesting to see what AMD and Apple can do.

Finally, looking at the broader landscape, I’m curious to see whether Vega Mobile will be adopted anywhere else. AMD is quite late in this generation, and as I’ve mentioned previously they’ve missed the rest of the 2018 laptop updates. However as we still don’t have a good replacement for NVIDIA’s GP106 GPU – the workhorse of many high-end laptops – AMD may not be as late as they seem if they can sign up more laptop manufacturers in the first part of next year.

Source: AMD - Comments

AMD’s Vega Mobile Lives: Vega Pro 20 & 16 in Updated MacBook Pros In November | Anandtech.com
https://www.reddit.com/r/Amd/comments/9sx8iz/amds_vega_mobile_lives_vega_pro_20_16_in_updated/

janzip 4 points 5 hours ago
"Wow I’m pretty tilted if this is true... I just bought a 2018 15 inch MBP a month ago and they will update it already? Feels like I paid for laptop with not the best performance..."

 

I Bought The Cheapest RX 570 In The Country - But What's The Catch? - (Mining GPU)
RandomGaminginHD
Published on Oct 31, 2018
I found a bargain of an AMD Radeon RX 570 8GB online. But why was it so cheap, and did it's past life (as a mining GPU) have a lasting effect on it's performance? Well let's take an in depth look at this thing and find out!

Spoiler: Click to See Video

http://forum.notebookreview.com/threads/nvidia-thread.806608/page-83#post-10815543

Nvidia
http://forum.notebookreview.com/threads/nvidia-thread.806608/page-83#post-10815543

Checking Out the AMD Ryzen Threadripper 2920X + Threadripper 2970WX
Level1Techs
Published on Oct 30, 2018

Spoiler: Click to See Video

https://openbenchmarking.org/result/1810306-FO-1810266SK79
https://openbenchmarking.org/result/1810297-FO-1810266SK73
There's an annoying "clicking" sound throughout this recording, lots of posts in the video comments too, it's not so bad that it's not watchable, so in case you think you are hearing things... you are.

 

He should have done some benchmarks against other RX 570's to confirm it was getting the overall performance of a new card. The only thing I can say about that video is the card was running cool, and the games seem to run 'fine'. Not very scientific that video, and certainly not worth 10 mins of anyone's time. He got a deal on it £70.

 

For those looking for that level of pricing / performance, and would really benefit from that savings from buying a mining GPU I think would benefit from watching it, as would those of us that are looking to recommend such a video to friends we know are hesitant to buy used.

Could the video be done more professionally and more comprehensively? Pretty much every review or video is that way, even the good ones, insight takes time and generous doses of time for reflection, and in hindsight pretty much everything can be improved.

Don't let simple human failures of comprehensive coverage failing to match your critical eye keep you from learning what little bits are new to you, and likely even more helpful others less experienced, exactly the kind of person that would be looking for the savings offered by buying last generation GPU's - especially inexpensive capital depreciated mining GPU's.

For what it is it's more than good enough to help someone in need.

For people thinking about making a move, doing something they are still uncomfortable with doing, seeing someone else go through what they are thinking of doing is enough to focus their thoughts and give them comfort enough to do it themselves.

 

It looks like it could be worth buying a mining GPU, as long as its significantly below the price of a new one, as his seemed to run fine. I just felt cheated after bothering to watch the 10min video, cheated that there was no real content. I wouldn't really want the 100's of people that might read this thread to each waste 10mins of their time watching it. It could well be worth buying a mining GPU, and his seemed to be working fine for him over the months of use he was getting out of it, although he didn't do anything to really make sure it was working fine, he didn't compare the performance to any kind of a standard for that card. Not worth watching that video though in my opinion, but others can make the decision whether they watch it or not.

 

The value of the review is in the longevity of the testing, not in the nit picky performance differences between other brands of 570's - there are plenty of reviews already - the only tests that would have been useful would be before / after testing for that particular card - and find out how much did mining affect performance.

I have seen such reviews, but don't have one to post off hand for the 10 series, but I am sure they exist - I'll leave the research for that to you - it'll give you chance to post a better review on that same topic.

Previous reviews I've seen show no degradation in performance for properly run mining GPU's. Serious miners make money from their GPU's and aren't playing games with them, they undervolt and tune them for longevity, which makes their resale value higher - which is a large % of the payout - more so today with the continually failing concurrency fad.

There are also many reported selling their cards practically new as they bought them after the value was there to put them in service. Maybe many of those are sold out now though, it's worth asking. Given the quick demise of the value of cryptomining in general with maybe a year of service maximum, the used mining cards shouldn't be "worn out" as was the worry months back when people were warning against buying mining GPU's.

The value in mining is getting a return on the electricity investment, and for most GPU's (all?) there is more value in reducing the power draw @ the same resulting performance so most miners will have undervolted and detuned their GPU's to run cheaper - cooler - and therefore might have less peak load usage than gaming cards OC'd to the maximum all the time they were in use.

A visual inspection of the card can tell a lot, whether the metal is pitted, are there are signs of discoloration from high temperatures, and is the general physical condition of the card good. If it's beat up and poorly treated you can get visual clues to ward off getting a dud, or build up a list of reasons to lower the price even further.

It's always a good idea to not waste money, who knows how the future will treat your job / investment market, it's better to buy smart and save.

Living from pay check to pay check and bottoming out the bank account for tech goodies has never been a good way to live long term - and saving $xxx's when buying computer parts can be a big win for the budget.

 

Correct me if I'm wrong, but AMD never said anything about TSMC node being inherently worse than the disbanded 7nm GLOFO node... in fact, both were stated to be quite similar, which is probably why AMD turned to TSMC in the first place for easy porting of their CPU's and GPU's to compensate for GLOFO potential inability to meet demand (as they claimed at the time).

So, I don't think the poor clock increase as you suggest is all we can expect... certainly not basing it all on a rumour and a potential ES (which usually crash anyway and run on lower frequencies when compared to the finished silicon).

AMD originally stated 45% performance increase over Ryzen 1.
I still maintain we are likely to see around 45% clock increase in comparison to Ryzen 1 to base clocks and unknown boosts at same TDP ratios... 60% higher clocks right now seems a bit high, but given how different the node is in comparison to GLOFO 14nm/12nm, who is to say what they could do with it even with early silicon?

To that effect, I personally expect (but won't exactly hold my breadth) that 65W Zen 2 3700 should basically have around 4.3GhZ base, and comparable/greater boosts on single and all cores vs i9 9900K.

Essentially, a 65W 8c/16th Zen 2 should surpass the i9 9900k in performance and efficiency... even if the clocks are same for boosts, the unknown IPC gains would likely equalise at worst Intel;s 9900k or surpass it (depending on how large the IPC gains are over Zen+ - but as of yet, the 13% IPC boost is still a rumour).

 

I can see 45% increase per watt. I would expect then, a 3700x from 95w 1800x, if 8c/16t, will more than likely be a 75w CPU. So instead of a base at 3.6 GHz expect 4.1 GHz. I would not expect to see 5 GHz all core until 7nm+ or even 7nm++, if ever, but on turbo 4.1 could translate to 4.7 GHz. single core @75w.

So taking a 3950x CBR15 of 2950 score should be 3,400 and another 13% could be 3,800 score. a 3990wx could take a 5,000 score to 5,750 and then with 13% up to 6,500 or even higher if it scales better.

 

AMD said 45% performance increase over Ryzen 1 and were quite specific about that.
Ryzen 1 also needed 15% clock increase to get 10% performance.
So, 60% clock increase would be needed to reach that target, but on early silicon seems a tad high, whereas 45% increase in clocks seems plausible, with rest reserved perhaps for very high boosts across single and all cores?

To that effect, a 95W 1800x would go from 3.6GhZ base, to 5.2Gh base.... or 5GhZ base (if you calculate a 40% increase in clocks instead), and the TDP would remain at 95W (because a node jump states you can gain clock increases with same power draw - or less power draw, depending on the efficiency of the node - and TSMC high perf. nodes even those at same nm size simply blow GLOFO nodes out of the water in regards to efficiency).

AMD hadn't said anything as of yet on whether or not they want to drop TDP values as well... and I don't see why they would as of yet.
They will create different TDP part CPU's yes, but what kind of TDP values they decide is up to them and we have no info on that right now, so its not unreasonable to think they'd retain same TDP values.

But, I sincerely doubt they would want to increase the base clocks by a minor bump up... TSMC node is far superior to 14nm/12nm GLOFO... so it doesn't make sense they'd play around with minor increases if it allows much more.

 

Why AMDs Zen2 at 4.5Ghz will win
TheUltimateComputerScientist
Published on Oct 15, 2018
This video is a deep dive into the differences between Intels Skylake architecture and AMD's Zen architecture in order to make estimates at what Zen2 will bring.

Spoiler: Links

Covering rumors on Zen2 clocks, Zen IPC Inner working on the architectures predictions on the future of the Zen architecture.

References:
4.5Ghz Zen2 leak:
https://hardforum.com/threads/the-rad...

HwBot:
Cinebench:
http://hwbot.org/submission/3847918_f...

http://hwbot.org/submission/3896683_l...

Wprime1024:
http://hwbot.org/submission/3847914_f...

http://hwbot.org/submission/3832921_h...

GPUPI:
http://hwbot.org/submission/3847924_f...

http://hwbot.org/submission/3810200_n...

Hardware unboxed video on IPC test:
https://www.youtube.com/watch?v=doGp5...

I9 9900k specs:
https://thepcenthusiast.com/intel-cor...

Anandtech article on Zen+:
https://www.anandtech.com/show/12625/...

Last (but not least), Dr Agner Fog's report:
https://www.agner.org/optimize/microa...

Also Dr Fog's page:
https://www.researchgate.net/profile/...

If you made it this far, thanks for watching and feel free to subscribe for more like this.

Cache bottlenecks in modern CPUs
TheUltimateComputerScientist
Published on Nov 1, 2018
This video will talk about bottlenecks that I've found in the cache of modern CPUs through developing my own real-world Engineering software.

Spoiler: Click to See Video

Thanks to @Erik C. Stubblebine for finding this channel

 

When ever there is a node size decrease, ie 14 to 7nm, TDP has to go down somewhat. This is just the nature of things smaller dies will handle less overall TDP. As I said numbers can mean a lot of different things in the end. This just like going form 14 to 12nm lowered TDP somewhat as well.

Now they can over the naturally lowered TDP required go even further and try for ultra low voltage and passive cooling etc. but this is to be seen. More than likely they will not as with the naturally lowered power draws this will make the 3xxx CPU with additional cores more up to snuff on existing AM4 and TR4 boards and VRMs.

 

Except that there's no guarantee that TDP will go down just because of a die shrink (at least not for Zen 2).
As far as I know, AMD said nothing on this particular matter beyond that the process can offer lower power (just as they hadn't said anything about increasing the number of cores on Zen 2 - it could happen yes, but we have 0 official data on this), and it goes opposite of AMD's expectations for 5+ GhZ on consumer parts (which I think they would prefer to attain).

3xxx series motherboards can sustain 65W and 95W TDP's for consumer CPU's and can also support Ryzen refresh series (top tier CPU's having higher TDP by 10W)... and besides, I think AMD would prefer increasing performance of existing parts whilst preserving existing TDP values, as there's very little benefit at the moment in increasing core counts (unless we're talking about the server parts - but even so, current 32 cores and 64 threads as offered by Ryzen refresh is actually pretty great).

I wouldn't necessarily expect a core count increase until Zen 3 (or 7nm +, as that will be done with EUV) because by then the node will have improved in maturity to possibly not sacrifice too much in regards to frequency whilst increasing cores (because the frequency penalty would likely be larger with first iteration of 7nm).
And you have to bear in mind that software developers need to also get with the times and further optimise for Zen architecture.
Some professional software for example is still not taking full advantage of multi-core CPU's as they should.

 

They will not, at least for now, go above 32c for consumer. Epyc may be a different story. The same may be true of a 3800x CPU, my reason I think they skipped the 2800x to increase core count on the 3rd gen. TDP will lower overall as power itself will lower too. I do not think they want to increase power draw as the VRMs are well loaded up now.

As far as software, do not hold your breath. Even Windows has its issues.

 

AMD ‘Next Horizon’ Event Scheduled For Tuesday Could Mark The Coming Of Zen 2
by Brandon Hill — Sunday, November 04, 2018
https://hothardware.com/news/amd-next-horizon-event-tuesday-zen-2

" AMD has an event scheduled for this coming Tuesday which is entitled "AMD Next Horizon". Heading over to the landing page on AMD's site for the event presents nothing but a blank page at this point, but we expect for things to start filling in November 6th.

If you recall, back in early December 2016 AMD held a " New Horizon" event that revealed the first details on its Zen processors built on 14nm Summit Ridge architecture. At the event, AMD first referenced the Ryzen brand name and championed a 40 percent uplift in instructions per clock (IPC)."

By all accounts, Tuesday's event will be focused squarely on the Zen 2 architecture. And while we were first introduced to Zen and Zen+ architecture with mainstream Ryzen processors, it looks as though the first Zen 2 processors will be from the EPYC server family.

In its most recent earnings release, AMD Corporate VP Laura Graves stated that on Tuesday the company "will discuss innovation of AMD products and technologies, specifically designed for the datacenter on industry-leading 7-nanometer process technology."

The next-generation EPYC processors will be based on AMD's 7nm Rome architecture at a time when Intel is extracting as much life as humanly possibly from its 14nm architecture. In fact, Intel's first 10nm products for the consumer market won't arrive until late 2019, while its 10nm Ice Lake Xeon-SP processors for servers -- which would compete with EPYC -- aren't slated to arrive until 2020.

In recent months, we've heard rumors that Zen 2, and EPYC specifically, will offer a 13 percent average performance boost over Zen+ processors in scientific applications. It's also rumored that 7nm EPYC processors will be available with up to 64 physical cores, with the potential for an astonishing 128 available threads."

AMD Digital Fan Kit
https://www.amd.com/en-us/markets/digital-fan-kit-download

Below are AMD digital assets grouped by theme and readied for use on various social platforms.

Links to Images:
EPYC™ | Ryzen™ | Ryzen™ Threadripper™ | Red Team | Radeon™ | Radeon™ Software | Radeon™ ProRender | Radeon RX™ Vega | Radeon RX™ Vega Logo

 

Anandtech AMD Next Horizon blog
by Ian Cutress on November 6, 2018 8:00 AM EST
https://www.anandtech.com/show/13547/amd-next-horizon-live-blog-starts-9am-pt-5pm-utc

AMD Unveils Zen 2 EPYC 7nm CPU With 2X Performance Per Socket Gain, Zen 3 Set For 2020 (Updated: Benchmarks)
Dave Altavilla, Brandon Hill — Tuesday, November 06, 2018
https://hothardware.com/news/amd-7nm-epyc-server-processors-rome

AMD Next Horizon Event "Full Show"
AMD
Published on Nov 6, 2018
AMD Next Horizon event full run of show Nov. 6, 2018.
For full PowerPoint presentations including endnotes, please visit www.amd.com/nexthorizon


AMD Zen 2 EPYC Rome 64-Core Server Vs Intel Dual Socket Xeon Platium Benchmark!
HotHardware
Published on Nov 6, 2018
More at: https://hothardware.com/news/amd-7nm-epyc-server-processors-rome

Spoiler: Click to See Video

Spoiler: Original Live Keynote Coverage

AMD Next Horizon - San Francisco Nov 6 2018
Alex on Tech
Published on Nov 6, 2018
AMD Next Horizon - San Francisco Nov 6 2018

Spoiler: Click to See Video

AMD Next Horizon - San Francisco Nov 6 2018
Alex on Tech
Published on Nov 6, 2018
AMD Next Horizon - San Francisco Nov 6 2018

Spoiler: Click to See Video

AMD Next Horizon - San Francisco Nov 6 2018 - Demo's
Alex on Tech
Published on Nov 6, 2018
AMD Next Horizon - San Francisco Nov 6 2018

Spoiler: Click to see Video

AMD Enterprise Software Session 1 Nov 6 2018 Next Horizon
Alex on Tech
Published on Nov 6, 2018

Spoiler: Click to See Video

AMD ROCm 2.0 Session 2 Nov 6 2018 Next Horizon
Alex on Tech
Published on Nov 6, 2018

Spoiler: Click to See Video

AMD Epyc Data Centre ROME Session 3 Nov 6 2018
Alex on Tech
Published on Nov 6, 2018

Spoiler: Click to See Video

https://www.youtube.com/watch?v=DaUy880vtRM

AMD Radeon Instinct MI60 Session 5 Nov 6 2018 Next Horizon
Alex on Tech
Published on Nov 6, 2018

Spoiler: Click to See Video

https://www.youtube.com/watch?v=m0h6-VfH3Xo

AMD's youtube channel has 8 brief video's covering todays event

AMD Takes High-Performance Datacenter Computing to the Next Horizon
AMD
Published on Nov 6, 2018
Today, AMD demonstrates its total commitment to datacenter computing innovation, at AMD’s Next Horizon event in San Francisco, by detailing its upcoming 7nm compute and graphics product portfolio designed to extend the capabilities of the modern datacenter.

Spoiler: Details + Links

***
“EPYC: One year on and AMD has transformed the server landscape” - ZDNet: https://www.zdnet.com/article/epyc-on...

“It’s the enterprise market in general that AMD is after, and it’s committed to a broad range of open standards to get there.” - Techgage: https://techgage.com/news/amd-announc...

“[Radeon Instinct] GPU accelerators will be going toe-to-toe with NVIDIA’s Tesla family of products...” - HotHardware: https://hothardware.com/news/amd-rade...

“AMD's CEO Lisa Su More Bullish Than Ever On Epyc In The Datacenter” - Patrick Moorhead, Moor Insights & Strategy, on Forbes: https://www.forbes.com/sites/patrickm...

“[AMD] took a major step forward by demonstrating the world’s first 7nm GPU.” - The Verge: https://www.theverge.com/2018/6/6/174...

“AMD Zen 2 Update: 7nm EPYC in labs now…” - Anandtech: https://www.anandtech.com/show/12912/...

 

I would have like to see a few more demo benchmarks but it does show that not only is Rome real but a force to be dealt with.

 

I would have liked this event to have shored up AMD stocks better by getting rid of the FUD, but until silicon makes it to market I guess it will not. Hopefully this will show not matter what Intel dreams up on paper it will have a ways to go to compete with AMD 7nm. It would be nice if Vega could knock Nvidia off but I do not see that happening anytime soon.

 

If one chip that is early silicon can beat a 2Px28 core system, what chance does the 48 core, with 8 fewer cores than Intel's 2P have?

Also, changes to cache, monster uncore, improved IF, double floating point, improved instruction fetch and prefetchers, etc.

Sent from my SM-G900P using Tapatalk

 

I'll agree that Rome may topple a few crowns. No matter what, with current silicon, Intel has no chance of running head to head with AMD at 7nm. I do not think it so much Intel plans on being as fast so much as they need something out there to be comparable to Rome. They need to stay in the game.

As far as GPU, the 2080TI prices may lend to a fairly high end Vega solution eventually. At 7nm this could bring hope to the fronts for AMD and a fairly large HBM card. We will see where this all plays out.