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

Here are potential news about upcoming Raven Ridge APU's (Zen cores with Vega IGP) :
http://wccftech.com/amd-pinnacle-ridge-cpu-zen-2-core/

If this is accurate, then upcoming APU's will essentially top out at 4c/8t with Vega IGP.
Not bad, but I think AMD COULD squeeze a 6c/12t in there with Vega IGP - that and HBM (now THAT would be great because IGP would in that case have all the bandwidth it needs with large amounts of video memory - and hey, perhaps AMD could use HBM for other purposes when its directly on die with the cpu.
HSA/HUMA would certainly make use of that.

Ok, so if these APU's are due to come out in H2 this year... then will we be seeing laptops at the same time, or at the end of the year (early 2018)?

 

No, my 'tone' did no such thing; it's all in your incorrect assumptions.

Yeah; it still kicked quite some @$$. Just not enough in the aspects that matter to me.

My workloads have been described here many times; I'm tired/bored to even look them up. And I've specifically answered your requests previously too (have you forgotten?).

Put me on your ignore list - what a shame. That is not learning nor communicating at all. All you're proving is that you think you're the only one that is right. Sorry, but that's simply not the case. You have your valid points as I do, but ignoring input from others just because you assume things about them only hurts you in the end.

 

HSA is kinda not a thing anymore but I want a 4c/8t part under 45w on a 13inch laptop for sure.

 

To build on this, Intel was not expected to take a hit this quarter. They released the Kaby-s series. This was usable on all Z170 platforms, with new features requiring Z270. They expected some hold-off for Ryzen, but not enough to actually change the projections.

Second, Analysts projections ARE NOT Intel's projections. They are people that crunch the numbers in public releases and study the industry, along with general patterns for tech companies, to derive the approximate amounts expected for dividends, earnings, etc. If you do not understand what GAAP is, EBITDA, adjusted EBITDA, 8-Ks, 10-Ks, 10-Qs, Form 14/A, etc., are, then you really are not understanding corporate reporting standards under the Securities Act of 33 or the Exchange Act of 34, the SEC's promulgated rules, as amended, and why information matters to the market.

Third, it also says you do not follow the debt related to each company, including default dates, assets owned by each company, equity offerings, different classes of stock, etc.

Fourth, what truly needs to be analyzed is how AMD's Ryzen effects Intel's plan to move some consumers from performance "S" chips to the HEDT platform through using Kaby-X and how it will effect the coffeelake chip (14nm cannonlake 6-core released on the performance platform). That is where the majority of head to head will come, as AMD beat Intel to that gap area where people's needs are not currently being met.

Then comes analyzing the impact of Naples on the server market (the primary reason @tgipier commented that Intel was in trouble, which @Mr. Fox promptly asked why). These chips, to solve one issue on cache, would require seeing it as two processors. Server based software is already designed for this, meaning if the driver is coded right, that hit goes away. Next, they are placing 4 of these 8-cores on a single die while maintaining the system on a chip design. This gives 32-cores and 64-threads. As the 8-core pretty much had 10-20% benefit in the highly mutlithreaded workloads related to commercial purchasers (some tests showed 3-6% to be fair), the 32-core chips may outperform even the next year's top offering, depending on the speed of the chip. This has scared Intel into moving up production of its 3600 pin field programmable chip, which will be a pain to dial in optimizations at first adoption (although the back-end benefit has significant promise). But, due to that looming issue, you can go with a faster platform with less optimization needed and easier for deployment, or Intel's new platform. Now, in mission critical and high-speed trading, Intel is not going to lose much. But in STEM servers (one area AMD really knocked some dust off), you are going to see declining market share for Intel. Other server areas as well, including gaming servers (dependent upon a good Vega execution, while noting that Nvidia still enjoys the majority of this segment and how well hardware plays together will be a factor that may, in some instances, benefit Intel).

Now, this isn't going to put Intel in the red, nor rip them limb from limb, but it is an all but foregone conclusion that AMD will take server share in the billions, which is a win the company so desperately needed. Some call it a relative win. Others call it a qualified win. But, here, I'll have to step up and bolster @tgipier .

 

You can post a wall of text, but you still seem to miss the point.

I gave a rejection and a write off for me or anyone else in my position (productivity first...). Please don't assume things and go off on needless tangents. I fully stated the reasons why an AMD solution - even one as great as Ryzen - isn't something to get excited about when the computer is there to make you money (and you depend on the computer...).

Nobody has full control of what others do. AMD has control of when it releases it's new chip. By all accounts; it was rushed. Don't sugar coat it; it was rushed.

I don't have to recognize the efforts of what a game producer has achieved with AMD hardware - I don't game (period). Is it a good effort? Sure. Doesn't sway me one way or another when deciding to choose a platform.

If the industry truly doesn't want to fully back AMD/Ryzen - maybe there's a reason, no? But I truly don't believe the industry would do that (increased AMD sales = more $$$$ for them...) so it's back to AMD not taking the time that is obviously needed to give a better first impression of what we can expect from them.

Another thing about the 'it'll come sometime later' mentality in the latter part of this thread so far - I hope it does, because it needs to... but I don't recommend to anyone to part with their money now and expect something in the next quarter year or so (it usually ends up bad for the buyer). Instead; I buy what is in front of me today, that I can use now. This is where AMD has failed for me. I don't want 'it should work, eventually'... - I NEED it DOES work, today... (better than what I have now).

As stated, this isn't my first platform purchase. And it would certainly not be a single example either (multiple dozens, after passing my testing phases).

A new user looking for their first desktop may be served well enough. I need more than 'well enough'.

If you're on the ball, you'll notice that what I'm stating here is no different than the post you're responding to. So I'll stop and let you re-read what I've already stated.

Don't lose the forest for the trees - some of the info/details you have posted previously were very informative. But as others before me on this thread have stated; the great 'theoretical' strides and technical, financial and manpower obstacles that were overcome to produce a product as good as Ryzen do not overshadow the fact that it is not a great (overall!) CPU/Platform at this point in time.

At least, not for anyone with a system that was already at the 'top' before Ryzen came out.

 

Intel supplies the overwhelming majority of US military microprocessors (with Qualcomm NXP and Nvidia distant seconds). US military budget was 600 Billion in 2016.

This is where Intel is "winning"

 

That's the US economy, not just Intel

 

Let me get to the point, then. If you have already said something, and others have made points similar or the same as yours, WHY SPEAK?

In your last post, I had to address your lack of nuance. That is your problem. You lack foundation of what you are saying and speak in a conclusory manner. That is something I will not stand for!

As to what you say here, you give more that adds to the dialogue. I'll agree that the final revision was in January with most vendors having the third revision at CES, not the final silicon. The motherboard companies are relying on that as the reason for having 5-6 weeks on the firmware (I find that to be a cop-out as you can work with nearly finished silicon to design firmware, then rely on the final version to tweak for performance, something they did not do). Also, that left that same amount of time to finalize the driver, of which the MB manufacturers relied heavily on M$ stock drivers, making getting a good working product out difficult, as well why the driver still needs work. But, many of the firmware issues could have been dealt with while not having that driver nor the final revision, as the third revision was roughly the finalized silicon with a 3.6/3.9 clock. Now, it can be argued that was not given until January because, in December, AMD used a 3.4 variant for the Horizon show. To that, I respond that they used the 3.4 base clock revision at CES in review machines. This could mean they had too few, or it means they wanted to build to a grand release with a higher clock. That is an unknown UNLESS you know someone in the know, which by all means, please share.

The example of the crossfire scaling was just that, an example. It is meant to show that AMD is working on developing the relationships necessary to accomplish what you are asking down the road. That is something to put in your roledex for later and should not NOW influence a purchase decision. Instead, it is addressing the critique that they aren't working with software developers. Before now, they did not spend the money to develop these relationships and it is an honest critique. But to write off them starting to do what everyone has said to do to improve support is ludicrous.

As to the reason, part of it is the past ten year history. AMD boards cost less, are cheaper to produce, but do not sell volume like Intel, nor do they allow for the margin Intel boards do. If you look at pricing this round, they now cost as much as Intel (although they do not have the full feature set in many cases, including thunderbolt). These hardware partners were unsure of sells, margin with the higher price, and did not want to get stuck with aging inventory. I'm not saying that decision was wrong. It's a business decision to which the management receives immunity from review as it is not grossly negligent to operate in such a fashion to assure that they will be able to move inventory. But, addressing the lack of design feature sets and the magnitude of under-developed firmware is where my anger is pointed. So it is not as simple as you suggest, there are many more nuances to the situation. I mentioned the inventory concern before (especially with the highest end boards and AMD being viewed as a budget company, meaning higher risk for over-producing the highest end models and being left with aging inventory).

As to the "it will come someday" mentality, maybe you misread part of the intentions. I, as well as others, have recommended waiting for the issues to be ironed out before making the purchase decision and to not necessarily purchase based on the platform alone, as what will be done by both Intel and AMD on releases in the interim is in flux. Intel will certainly release the 10nm around 4Q 2018 or 1H 2019 for HEDT (estimating between 15 and 18 months before cutting off Skylake-E/X). GloFo is transitioning Fab 8, the current primary 14nm Fab, to 7nm in 2018, saying capacity will be ready for risk chips by the end of the year. This suggests AMD may move to 7nm on the interim. But, GloFo just as easily could transition another fab to 14nm and produce at the same node until 2020-21. Also, GloFo refused to renew the licensing agreement on finFET IP from Samsung. Now, this could be due to getting the design team from IBM, which has been playing with sub-10nm for awhile now, including looking to the sub-5nm nuances. But, I would say do not spend money on an unsure bet. Bet on known information.

So, no, you are not reiterating your last statement in a new way. You have provided me, this time, statements that I can use to build a dialogue. The difference may be subtle to you, but it is huge to someone that analyzes statements for a living.

Finally, I agree with the decision not to buy the new platform if you already possess BW-E. I've said that numerous times. There is not even a guarantee that SL-E/X will offer enough benefit to give a purchase recommendation (unless the absolute top available is needed, which amounts to few purchasers who already possess BW-E and are not purchasing Xeons). But, that information will not be available until August. Now, if you are holding off for the fixes on AMDs platform for 2 months, roughly, that brings you to May/June. Many trade shows happen then. So holding off another two months so a direct analysis on productivity for specific uses is not unreasonable, especially considering these platforms can serve 2-4 years, updating boards as needed on other considerations. But, once again, I already stated that in a prior post.

So I find myself coming full circle and asking, if this has all been said before, WHY ARE YOU STILL SPEAKING?

 

Wow, I voiced my opinion and conclusion of what I had been accused of 'not understanding' very early on in this thread. With some points that just happen to have also been echoed by others too.

What? you are the only one who can repeat themselves? And in your case, ad nauseam (over 60 pages worth)?

You have a voice here as I do. Get used to it. If you can't see the nuances I bring to the discussion, it's your failing not mine.

 

When I said intel is losing, I am talking about marketshare increases. Markshare wise, Intel is not going to improve. They more or less hit a stone wall and multiple companies are chipping at them.

 

I'm asking you to base it on facts. You state opinions without a base. If you notice, I have become much more lax as information is available and widely disseminated. That is the difference. You still keep your needs amorphous as to productivity, nor do you point to facts that are widely known, rather speaking in generalities. When you make a point, and there is publicly available information, I do try to fill the holes. I made this critique and explanation long ago. Please, either provide the bases of your statements, or go. If you had not done the same thing before release, I would let it go as I do with others here. Also, I have a report with others that have done so, which means I have a better idea of what they are basing the statement on and the rationale of their arguments. No such relationship exists here. Do you understand?

 

Sorry, I stopped the analysis there. You are correct on Nvidia taking out their Xeon Phi, ARM in relation to IoT, where they are bleeding money, etc. I got lazy and reached for the low hanging fruit for analysis. Did not mean to give slight there.

 

All good, I was more talking about the other people who are taking it wrong then you.

I noticed you talked about FPGA. Yeah, IF intel get that working and optimized, its gonna be a interesting market for sure.

You always have a clear analysis on things. I enjoy reading your analysis most of the time.

Clear logic and clear argument, even if I dont agree, it makes for good debate

 

Definitely, but it requires specialized software, which significantly increases the price of adoption/optimization. That takes time and money. It will be worthwhile after companies do so, while also locking them into Intel for the foreseeable future due to the software costs, etc. But, that also means risk related to adoption and rebuilding platforms and software, some of which will be ground up. With AMD coming up with a competitive server product, it will make the decision harder for some companies to justify the additional expense. So there may be a purchase of a small number while designing for deployment 3+ years down the road for larger infrastructures (you could do it quicker, but the closer you get to release, the more costs increase exponentially; some will be done in the first year). This makes the future server platforms much more interesting, as you choose Intel and need more optimization, or you can get AMD and continue running what you have with less optimizations needed and better performance (edit: than existing platforms, most likely). Intel's pays off on the long haul, but front loads some costs....

 

https://www.kitguru.net/tech-news/f...-is-going-on-with-am4-x370-motherboard-stock/
Good read, giving credence to my analysis on zero faith from MB manufacturers, which also shows why they gave no concern to designing the firmware.

 

I'd be pretty irritated if I was building Ryzen right now...

Meanwhile in Korea

 

Wow, I was gone for six hours, and missed three pages of rants. I have come to one conclusion, well three. Most people in this particular forum own laptops (I assume). Most people on this particular forum do not own desktops (another assumption). People that do not have a desktop, and might be considering building and or buying one, could not go wrong with the Ryzen platform.

Sent from my SM-G935T using Tapatalk

 

@TheReciever ......lmao
Incredible video!

Sent from my SM-G935T using Tapatalk

 

ASRock X370 TaiChi Motherboard Review- Ryzen

 

Im wondering if I can drive games according to which monitor is connected to it? Like Monitor 1 > GPU0 and Monitor 2 > GPU1

Then have the games be driven by separate GPU's on their respective monitors? I think you used to be able to do something like that in the old days but I am not hopeful about things right now since Windows isnt really concerned with anything outside the norm.

I wont be buying Ryzen anytime soon though, maybe Zen+. Its much cheaper for me to get the adapter for my laptop to accept desktop GPU's than to buy a desktop (regardless of brand).

 

Just as I said already.............

 

A little more on Intel's acceleration of the field programmable chips and move to 7nm. Notice how the article ignores GloFo's process and when Intel's competitor will release products, instead focusing on the timeline of TSMC.
https://www.fool.com/investing/2017/02/25/intel-corporation-reportedly-plans-to-mass-produce.aspx

Here is another one discussing the server market, although less analysis went into the adoption hurdles of the field programmable chips....

https://www.thestreet.com/story/140...alcomm-just-made-things-more-interesting.html

And another questioning delivery of Cannonlake 4Q this year.

https://www.fool.com/investing/2017/03/06/intel-corporation-expresses-doubt-about-10-nano-ra.aspx

 

Delays are the name of the game any more.

 

Edit: Sorry, this popped into my head as soon as you said that...

Edit 2: In all seriousness, the main reason for delays are due to the delayed EUV lithography, which was expected for 2015, but now won't be deployed widely until 2018. Because Intel planned on that for 10nm, it created a wall, allowing all others to catch up with its dash on miniaturization.

 

TSMC timeline is not exactly accurate. I would add an additional 6 months or 1 year.

 

Well, in the context of x86, TSMC isn't exactly as relevant (except if GloFo can't meat the wafer agreement for AMD). They ignore the move to 10nm upcoming by them and also do not discuss Samsung's move to 10nm and beyond either.

Sent from my SM-G900P using Tapatalk

 

Speaking of GF, its always the weak point for AMD. What happened to that wafer agreement? I remember AMD had to pay them a lot of money.

 

It was amended, both increasing the number of wafers and giving larger concessions to allow AMD to use TSMC if GloFo has a shortage, etc. I actually posted an article discussing the WSA earlier in this thread. With that said, them licensing the IP from Samsung on 14nm is what brought us to here, but the IBM team is what is bringing the 7nm moving forward. So, I wouldn't write them off as a weak point quite yet, as that seems to be a turning point. In the past, yes. But, GF is doing the highest end graphics cards and the CPUs currently, with the rest of the GPU lineup on TSMC 16nm. . . .

 

I have a sudden desire to rewatch Resident Evil. I don't know if I should thank you for that or not.

 

The Altera deal is precisely why I said earlier in this thread (a month or so ago) while arguing with certain user, that GrIntel has no clue where the future is or even if they can bring it. As @ajc9988 and @tgipier stated it would require A LOT of money and time to make that FPGA a thing. It would've been best if GrIntel focused on making this a thing, rather than sinking a crap ton of money into trying to penetrate the mobile market. I don't call them GrIntel for no particular reason after all. Greed ruins, we'll see where it would take this particular company. Maybe that's the reason behind the AMD deal - they realized that the FPGA thingy wont happen over night and that they literary flushed in the toilet pretty nice amount of money which would've been better suited elsewhere, so they have to stay afloat somehow. Hopefully it would mean better optimizations for APUs. Decisions, decisions, right?

 

TBH though, at Intel's position, they had to expand into other markets. By obligations to the shareholders, a public company's job is to make money. A public company's number one job is to make as much money as possible, so you cant fault it for being greedy. AMD is the same way. Intel tried to compete in a market that it utterly failed in.

 

I agree, hence the "decisions" thing. This side-step would make things much more interesting though. AMD obviously pushes for hUMA and GrIntel might actually help them get there while doing their FPGA. Haven't commented on the HBM yet, but here it is now - it's not coincidence that recently they called it cache. That's how I see things.

 

If the next AMD processor have a "L4"HBM cache, I would be interested for sure. Zen+/Zen 2 is gonna hopefully fix the features lacking from Ryzen and be the CPU for my next build.

 

What should be noted, the DDR5 standard was supposed to be filed this past fall. It still has not been filed, but HBM future standards have been filed. In addition, we are still waiting to learn the impact of the high bandwidth cache controller, which allows for hUMA. So, AMD may have helped to shape the future (plus Intel's playing with their version of phase change memory)...

Sent from my SM-G900P using Tapatalk

 

http://wccftech.com/amd-pinnacle-ridge-cpu-zen-2-core/

Edit: @Raiderman - Here is what was used to check cache for that example awhile back-
https://technet.microsoft.com/en-us/sysinternals/cc835722.aspx

 

I am basing it on facts (mine; direct experience gathered over the last few decades). The issue is that you refuse to acknowledge the basic truths I am stating.

Just because I base my conclusions on my productivity; that is not amorphous. Others can and should do the same based on their workloads. That is the only real 'score'/'number' that matters (or; should matter).

I do provide the basis of my information/statements; you don't accept it. That is not a conversation. All the stat's and schematics you can provide do not make the Ryzen platform attractive to me in the least - the proof is on the internet (but yet, you conveniently gloss over those 'reports').

The fact that I did do the 'same thing before release' should be a sign for you to consider deeper what I am stating; rather than being a reason to want to dismiss offhand what I am offering (food for thought...).

There is no relationship here, period. We are exchanging ideas. It seems to me; you don't want to consider anything other than the 'stats' you believe so much in.


See:
http://www.tomshardware.com/reviews/amd-ryzen-7-1800x-cpu,4951-12.html

The above quotes are representative of the unbiased reports I like to read on the www... And they fully support my conclusions, which is not the case with your analysis every time***.


*** Please see the next post in this thread from me for an example.

 

Here is an example of your 'analysis' that is the exact opposite of how I read the same article (link in your quote).

MB manufacturers had zero faith based on what AMD's early samples showed them. Why on earth would they roll up their sleeves and do a deep dive into Ryzen (or any other less than promising platform from any manufacturers, for that matter) with their expensive R&D staff when AMD showed them less than stellar samples?

When AMD finally delivered on most of what they promised (this is a first in a decade for them, remember...) they didn't leave their partners time to actually produce their ends of the platform AMD wants in every DT build.

You see the difference between what you read and think you understand and what I read and comprehend?

I'll repeat it again: AMD rushed this launch. Nobody is to blame expect them for the state of the Ryzen platform at this point in time.

 

I will explain this once more before I report you for thread crapping again. When you spoke BEFORE any information was known, you were making assumptions on a historical bias that was no longer relevant. Some of what you said was proven false with later numbers, such as performance in STEM and modeling. You then say the same thing after WITHOUT nuance or qualification. That means it has no relevance. Then, you use your own uses as a basis, but you do not share what area you are in, meaning we couldn't point you to relevant information if we had that information. You then repeat the tired line that you've said elsewhere in the forum, as if you are so god damn important that people should look you up. This is narcissistic. You have not established yourself as an insider in one of the companies producing these products, nor as someone whose opinion matters. As such, you add nothing and everything you say remains amorphous to the community. Hence my black box statement. To me, you add nothing to the conversation.

As to the MB manufacturers, they directly stated, from insiders off the record, what I assumed. As soon as that proof came forward, I gave it to allow people to weigh my assertions in light of facts later made public. That is what proper discourse is, pointing to information to explain and allow others to weigh the merit of your information. I've been shown wrong on a couple things, such as the OC utility. That is fine. People provided facts made public after my statement that showed it to be false. I then change my analysis in light of new information. That is the purpose of forums and discourse.

You, on the other hand, provide none of this. That makes your opinion USELESS! I give two ****s as to your opinion is concerned, as you do not address new facts. You regurgitate your opinion without addressing ANY FACTS.

Meanwhile, you are saying companies should **** on consumers just because they are not interested in a product or platform. I agree with producing less. But to put out a sub-standard product, you just said yourself, is to say **** YOU BUY ME to consumers. This alone should give consumers pause on these companies! Yet, you don't see it like that (when they all derive from the same underlying facts)!

Further, you are saying that AMD should have had finalized silicon MONTHS before standard (usually, silicon is finalized 2 months before, 3 months tops). You send early silicon so they can get ready for the release. Considering how long they had the early silicon, **** them and **** you for the insinuation that it justifies sub-standard work product. For legal work, you would be disbarred for sub-standard work because of having to provide pro-bono service or forego work product for a higher paying client. THAT IS NO JUSTIFICATION.

What you are saying does not make it a rushed launch! YOU ASSUME THAT A LAUNCH SHOULD BE PUSHED BECAUSE ADJACENT COMPANIES DID NOT SEE VALUE IN IT UNTIL IT WAS PAST THE TIME THAT THEY SHOULD HAVE AND NORMALLY WOULD HAVE DONE THAT WORK! That makes no ****ing sense, because then no smaller company would ever put any product out. So, please, just ****ing leave!

 

@tilleroftheearth

Go to 0:45 and watch to 0:58 and follow what is asked of you!

 

Okay...

Very mature response. Please report yourself while you're at it. I am participating in a conversation. You just want one opinion expressed: yours.

Your rant below shows you don't read and comprehend one thing of what I write. Again: not my issue.

I showed very plainly how your conclusions are opposite mine. You could have politely said to agree to disagree with me; but no. It's your way, or the highway...

Btw, I'm not Tomshardware... they stated it was a rushed launch. I'm simply agreeing with them.

Take care.

 

You already told me in much more colorful language to leave. But thanks for the video.

 

Microsoft Windows Bug Is Holding Back AMD Ryzen

"Microsoft now kind of confirms the issue that Windows 10 does not detect the simulated SMT threads properly, and the simulated SMT threads are weaker compared to real CPU cores"

@tilleroftheearth Why blame everything on AMD? We all know that "The Redmond Morons" is a master in screw us all with their New Modern Pastel Colored Baby Boy Tiled OS - X <Trash>.

Don't expect everything is shiny and golden when we talk about Micro@Screwers. Everything they touch will be dirty!!

 

I don't.

But first; I don't excuse the primary person/company that should be more directly responsible/interested in making sure bugs like this aren't addressed in a timely fashion either.

And by timely fashion? I mean before releasing their highly modified (from ES samples...) on their launch day.

The 'Redmond Morons' that you refer them by may not do everything properly. But it is not them I blame if I move to a new version of their offerings without fully testing out my workflows and I find a bug or worse. I'm the one that is held accountable at that point.

Just like AMD should do everything it can to change it's own baby's diapers when it is but a wee young thing and needs the actual parent company (and not distant uncles or aunts) to care for it so it can grow as quickly and as strong as possible.

 

"May"? Give me a breeak. Of course can we blame AMD as well. But the main "Ostrich" is still M$. Don't defend the M$ Morons!! Not everything is shiny down there in Redmond.

 

Maybe I should have said 'It is not them (MS) I would blame first...'.

MS isn't the main ostrich (?) and I'm not blindly defending them.

The facts are that AMD made drastic changes to their shipping CPU's up until launch day (or at least; that is how it seems from ES models supplied to select users to what was actually released).

How is anyone able to account for that unless AMD directs them (and gives them the time) to fix/patch/change things so they work properly for the new tech that they (AMD) is unleashing onto the world?

I don't defend anyone/anything without reason. Least of all the person/company that is primarily responsible. Which in this case is AMD, without a doubt.

To put it another way; AMD - just like any other company, or, any other processor from Intel that I am considering to purchase - needs to ensure it is working/compatible/optimized (at some basic level...) for the O/S, software and workloads for the users they are targeting to sell to.

Nobody will hand hold them. This is busine$$ with time & money involved for all.

In the final analysis, it seems that there is some hope that MS may have it ready (already) in a few days on MS Tuesday this 14th day of March... Either way, AMD is the one sitting on their hands when they should have seen this in their internal testing on their shipping CPU weeks before, no?

 

Hint... Google push out warning about security problems to the Redmond Hq. What happens then? The Old sucky lady only complaining and maybe push out fixes when they have time. M$ have more than enough with pushing out unwanted trash!! Not fixing their screwed up OS. And I'm sure same happens with their work with AMD. They will always come second.

Always said. Don't be a Guinea Pig. Not for hardware or software. Always something screwed up with lunch.

 

You can't force other companies to do their ****ing job! Don't you think some of this **** was told to those companies. Some of it is growing pains that new platforms get. But to think these other companies didn't have any part in NOT DOING THEIR ****ING JOBS is ludicrous and ignorant! A business decision is deciding which components to use, pricing, number of production, etc. The choice to not waste money on firmware development is to directly say **** YOU to both AMD and TO THE CONSUMER! Now, M$ is used to saying that, as is APPLE! But, if we examine the lineups, only a few made the effort (although there are caveats on all MB manufacturers). The first and most notable is Biostar. This company hasn't really come up at the higher end of boards recently, but they swung for the fences and put out the only board on release day supporting up to 3600MHz. When you go out of your way to "punch above your weight class," (not meaning their products are inferior, rather exalting entering into the ring for the top boards) you deserve praise. Many journalists were surprised at this, and consumers, which is why they deserve a shout-out! (Disclaimer - I used their board for my last AMD CPU, from 2002, and lasted through 2008-09, when it no longer supported the OS with my move to windows 7 and I needed more mobility, so purchased an HP pavilion laptop (a bane, but another story)). The second is MSI, which at least gave decent build quality and slapped their Xpower Titanium name on it. This board is not without issues, but, worth noting. Asus chose to cut down the offerings to the ****ing Hero series. They were not even going to give a formula series as they had in the past. It was not just a **** you to AMD, it was a **** you to anyone not buying Intel, because of margins,etc. They didn't care to try, then changed the tune when the numbers started rolling in, both on pre-orders and stock sell-outs. Then, the light came on, "maybe this is what the consumers wanted." But, this already helped to lose some steam, which should properly be blamed for not doing the proper analysis and polling online to figure out demand during this time period. Instead, they focused on the Z270, whose sells were likely underwhelming because of support on the Z170 platform. Eggs in the wrong basket. Not to mention the later release of the M9E and M9Apex, which seriously likely didn't do as well because of waiting for Ryzen (coincidentally, I think there may have been gamesmanship here to show games are better on the 7700K and there is no stock of AM4 boards, so jump on this other, more expensive platform with a larger margin for us). But the parenthetical is pure conjecture at the moment. I could say similar about Asrock, to a degree, but even the OC formula came out much later for Z170, so, I won't delve too deep at the moment.

As to your Tom's Hardware quote, way to pick a journal dumping on gaming performance and not productivity, which you said you were focused on productivity, not gaming. This shows you are a joke!

But, the blame on adjacent companies should be there as well (I'm not saying don't blame AMD as some of this should properly be put on the company). But, it is obvious to anyone that these other companies deserve blame as well!!!

ALSO, sorry, but the changes at the end are optimizations, microcode fixes and driver fixes, etc. NO MAJOR ARCHITECTURE CHANGES OCCUR IN THE LAST COUPLE MONTHS! So this ******** that they were changing a lot of big things on the platform is horseshit. They have a problem with the IMC that should have been dealt with and has been part of the focus for the past 6 months (or at least from what I've read to date). That does not mean you cannot work on your firmware in the interim! Also, M$ would have had the silicon as well and had the back and forth. The fact they did not even try to diagnose issues and assumed it is just AMD's silicon is part of why they did not look or do the work. AMD deserves blame, too, as they should have pointed that out (which this is the third or fourth time in the past couple pages I've said that, at least). But, it is M$hafts software, meaning they should have more idea on how to terse this out and find the issues. They didn't!

All of this points to market share and protecting higher margin products, while not giving a **** about AMD consumers due to smaller user base. There is no excusing putting out ****ty products in the way you suggest. Just because you don't realize the potential success of a product does not mean you can **** in a bag then tell the customer to be happy with the **** they were given. IT STILL IS ****! That is the apple way, the alienware way, etc., and consumers should demand more from these companies.

As I said, you should expect growing pains. Because the MBs and hardware will vary so much once sold, bugs will inevitably come rolling in and the first couple weeks to a month will hold many updates to fix those. You cannot test every use scenario. I concede that. But there is a difference between that and what these adjacent companies did. So once again, why do you keep trying to shift all blame (or the majority) to AMD when it is obvious the industry, as a whole, gave a big **** You to AMD and consumers?

@Papusan - Here are links confirming the cache issue I mentioned as the second one (not the round tripping, CCX issue) (further addressing the same issue your link above discusses):
http://segmentnext.com/2017/03/09/amd-ryzen-windows-10/
http://wccftech.com/amd-ryzen-performance-negatively-affected-windows-10-scheduler-bug/

I cannot find the link talking about using a different thread scheduler that Intel had to use years ago because of the similar thread scheduling problem. That would answer it directly for the time being until the fix truly comes in.

 

There is no such thing as a free launch

 

Asus Crosshair VI AM4 Ryzen Overclocking Motherboard Review
"Perspective review as 3rd AM4 build"
Published on Mar 10, 2017
Is this the best of the launched Ryzen Motherboards so far?
Full Review:
https://www.overclock3d.net/reviews/cpu_mainboard/asus_crosshair_vi_hero_am4_motherboard_review/1

 

So, I agree the memory is all AMD. In fact, AMD limited what options are allowed on tweaking ram and estimated two months from release to resolve the issue. That, I will blame on AMD. For the specialized tweaks in the bios (the presets for overclocking enthusiasts which do require finalized silicon and testing), I will blame AMD (this is a new caveat which I thought was obvious, but I guess some readers of my comments may not have realized I do except this from firmware functionality). But other quirks should have been addressed before hand (unless it can be shown that those quirks were introduced due to early silicon, which changed to the final version, which, mostly, I find dubious). At this point, this is PR handling and blame shifting (standard practice as this can effect goodwill; see comment above on AMD taking it on the chin).

Still skimming the video. Shame it did not compare the flagship of each company...

Sent from my SM-G900P using Tapatalk