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

Look real close at Papusan's Avatar.............That pretty much sums up how I feel about Microsoft, and there so called patches. I am pretty sure AMD confided in MS about this issue, knowing how Microsoft could screw up a wet dream, shows how much confidence I have in them. I have WAY more faith in AMD than I do Microshit.

 

When the fact that Windows (as a whole) has 90% or so of the market - on the most variable hardware possible - it isn't hard to understand why they 'push out fixes when they have time'.

Also with that ~90% market share; 9 out of 10 users do not view it as unwanted trash on their computers either. All the other options available are not as good (obviously).

Another thing to consider here is that an O/S is not a stale and stagnant part of computing anymore - security fixes are not something that will go away - as the O/S & Software become more capable and at the same time more complicated (under the hood...) there will be the hackers there ready to break them to their advantage.

This is why I like the future of Windows 10 as an 'idea'; it is always the newest O/S you can have. With the most stability, performance and security (all three - not just one or two...) at any given moment in time.

And finally; an on-going company first takes care of it's shareholders and it's customers - everyone else will always come second to those two (business 101).

100% agree! Don't be a Guinea Pig.

 

I am focused on productivity. Mine; not their version of it.

The 'joke' here may be your incorrect conclusions based on merely superficial analysis?

Can't force other companies to do their job? You mean MS or the MB manufacturers being forced to do AMD's job?

Let's state this again, succinctly; AMD launches a new product - has access to all possible O/S's it would be used on - gives a substandard ES sample to 'everyone' and then delivers a much improved version weeks/days before launch... but seemingly without giving key players/partners the much improved versions in time to properly support their new product.

Continuing; even if the above wasn't true... in their (AMD's) internal testing - did they not see the scheduling/core/cache issues? What did they do about it? Nada.

What should they have done about it? Obviously delay the launch until that little issue was fixed.


Anyway, why are you so angrily defending AMD? I haven't said any untruths about them. And some/most of what you accuse me of you've said yourself in the preceding pages.

And the point that you shout about 'no major architectural changes in the last two months'? Uhm, that is neither here nor there. What was initially shown to their partners was not worth their partners efforts (obviously). Or, the partners simply didn't feel like spending/wasting $$$$$$$ on a moving target. NOW that the target has stopped moving; they'll get to it soon enough.

The point being that this was a business decision based on AMD's actions (not to mention their past track record up to this point). Great that Biostar took a chance! Their efforts paid off.

Most others with bigger production capacity (hence more to lose) didn't want to take that chance.

In an ideal world - one with unicorns and tooth fairies - this wouldn't happen (and I agree it shouldn't in an ideal world). But this is the world we live in. Big business rules - no friends or free handouts to be found unless all sides make the $$$$$$ they think they deserve. Corporate Executives and Shareholders are who is catered to first and mostly 'only'. If consumers could be taken out of the equation (they're trying with ongoing, automatically renewed monthly subscriptions to everything...), they would be.

Again; agreed, other companies didn't step up for AMD as much as AMD needed them to. But that doesn't excuse AMD from proceeding with the party when the booze and food haven't been paid, arranged for and verified first.

 

So, you don't run Windows?

See my previous posts immediately above for my take on AMD's issues with MS & M/B manufacturers.

Having used Windows since v1.0 and right now on Windows 10 - my faith is firmly in MS to give me the tools I require to put food on the table and a roof over my head.

Is MS Perfect? No.

Are they Proven? Yes, many times over. No matter which O/S I've compared it to over the decades.

 

Lol this I can't agree with this.
MS is ****. Specially when it comes to professional versions of it's OS for actual "production" houses.
That is why most are either sticking to Windows or switching to UNIX/Linux.

 

*sarcasm* I assume is not your strong suit, since reading my avatar or my signature eludes you, I will explain. I do not use Windows 10, I use Windows 7 with the update service disabled. Microsoft as of late, cannot make a patch to save their own arse. That has been well documented. The Kool-aid oozing from the posts you have written only show..... A. You like to troll, B. you are a 15 year old... C. You work for Intel, or Microsoft.

 

My Avatar is created just for... BGA and Windoze!! Pict says more than words.

Yeah, You hit the nail on the head. I ain't stupid I have seen too much!!

 

It could be!! If they would listening more to their customers. No one ask for adds and much of the other ****y they put in their OS. But it's too much to ask for!! The hate will continue!!

 

Sarcasm? No, I'm asking a simple question. Btw, thanks for your answer.

But Win7 is still MS no matter what you've disabled.

No oozing happening here - like I've said; Windows 10 makes all previous O/S from MS and any other O/S I've tried too look like a bad movie about 2025 from 1957.

I also don't work for anyone except myself. Nor am I 15 years old (I wish!).

I am just maybe more vocal than others about the products that actually work, day after day and year after year for my needs, is all.

 

I agree that Microsoft did release fine products at one time, and I used to support them. To think that they still have products that work year after year is up for speculation. Windows 10 works "sometimes", and then they release an update that screws millions of people over. Along with their bullish tactics, and unwillingness to listen to the user, as of late, has left a bad taste in peoples mouths. Today, I think AMD has a better name than they do, so I am willing to trust AMD more than MS, that changes something as simple as the behavior of the "close" button.
For me, and many others, Microsoft has soiled their name, and would most likely put the blame on them, and rightfully so.

 

M$ have jumped down in reputation the last 2 years. No doubt. Just look on the web. They can thank themselves for all this!! A working OS isn't the same as a good OS. I would expect with competition in OS software, we would see another attitude from them. Monopoly isn't good for none. I'm still shocked how ugly Windoze X is. The GUI is designed for children who play with colors. With children... I mean in age 5 or 6. Maybe the main target group Not for the users who is grown up with Windoze from the beginning!!

All the talk about Windoze... Because AMD Ryzen doesn't work correctly yet!!

 

Fair enough (on your comments below).

But ~90% market share leaves little to the imagination; they do have products that work for most/all and have been doing so for many, many decades. Far longer than any other class leading O/S can claim. And for more varied hardware/software/drivers/workload combinations and permutations than all other O/S makers combined.

I don't put much stock in a 'name' as there is nothing there to trust once $$$ have exchanged hands. I put my $$$ and trust in platforms that push my productivity higher than what I am replacing. The names of those current platforms happen to be MS & Intel - specifically; Win10x64Pro + i7 QC or better hardware with maxed out ancillary components (except maybe/optionally GPU's for 99% of my workstations).

Updates and hiccups is the new norm (has been for a long time now...) - turning off Updates is not doing oneself any favors for any system that is online, even infrequently.

AMD's progress these last few months is great for the industry as a whole. But there is much room for improvement. This, I don't think is up for debate.

In any event, it is far from the "The Pinnacle of Modern Multi-Core Processing Power" as advertised by AMD. Even when TDP and $$$ are taken into account and a full/complete platform is assessed (DT only at this time).

See:
https://www.amd.com/en/ryzen

 

Turning off updates, because I can, and will continue to if I wish. That is the difference. I have a choice to do whatever I want, whenever I want. and am not forced to do or have something on an OS I paid to use. I did not say my system was not up to date, rather updating is turned off, so I may choose which of them I prefer to install. I have that control, as you do not. You are force fed whatever, MS thinks your system needs, and that means advertising based on your keystrokes, and if you think you are locked down, you need to think twice.

Ryzen is NOT far from the "Pinnacle of modern Multi-core processing power, as you suggest, hence many benchmarks have shown. If you are as much into the productive side of computing as you say, then you would already know this. Suggesting otherwise would only prove/show fanboism (not a word), on your part, thus making your debate pointless, and meaningless.

 

So, let's look at the facts of what the difference between the chips were during that time period. First, the IMC. All early testing used either 2133 or 2400 DDR4. They didn't have it working as well as it does now until around January samples (the 36/39 and 36/40 final silicon). This is an ongoing problem and one I have conceded lies with AMD. Second, let's look at the speed stepping. Early samples sent out were from 3.1-3.4GHz. Now, comparing the numbers we now know, that would have barely broken even with HW-E at stock. Considering these were locked (in some instances at least, if not all), I can see the underwhelming nature of the chips, even though it is a huge jump from prior offerings. Third, AMD even said internal testing took place on Windows 7 machines, but they were going with Windows 10 as the officially supported OS. This is because both chip companies made a deal with M$ to do so. Considering this, the scheduler issue may not have been known until a couple months ago. This is where I agree with @Raiderman , they likely reported it and M$ likely did nothing about it. Does that fully absolve them on this third factor, NO! But, M$ has taken 6+ months to fix issues reported to them in the past for conflicts with hardware. This is especially true when the user-base is smaller and fewer are effected. So I will dump plenty of hate its way!

But, considering the first two factors heavily (considering the MB manufacturers, in their defense, stepped up with drivers for their MBs for Win 7 while resolving the scheduler issue), there is no good reason not to make a firmware that works! Not being impressed is reasons to make fewer boards, use less expensive components, etc., based on expected sales and prior platforms. IT DOES NOT MEAN TO PUT OUT AN IMPROPERLY WORKING PRODUCT DUE TO THEIR NEGLIGENCE ON DESIGNING A PROPERLY FUNCTIONING FIRMWARE!!! It's like you don't understand where is and is not acceptable to cut the corners. I deal with these kinds of decisions regularly. Lesser components = still properly functioning for the components and features selected. Not designing the firmware to put that off after the release because you'll have to work on it anyways after launch is egregious! THAT is my problem. Even that video earlier said the difference between the review and release firmware and the new ones are night and day! WHY? BECAUSE THEY SAID **** THE CONSUMER!

As I said, my main beef is that they didn't do their ****ing jobs. I will complain about the offerings on the highest end, but, as both of us have said, selection on MB components and design are a business decision. That, I will concede. But you don't understand the line between the two, which is very disturbing for someone claiming decades in the industry!

 

What website shows reputation gained/lost?

Huh? A working O/S is a good O/S. One that works better with new hardware with each update is even better. Sure a working O/S can be lousy (see the fruity companies efforts...) - but what real world workflow, overall, doesn't benefit from the latest Win 10 Version 1607 OS build 14393.693 version (seriously?) - given that the software chosen requires Windows to run?

Yeah; monopolies are not good for anyone. But they're effectively a monopoly because of their strengths and benefits to the users - not because of their weaknesses.

Considering that there is over 400 million Windows 10 installations today, I'm sure that a show stopper like a misguided GUI would have turned more than a few of those users away by now.

Yet, the numbers keep increasing for Win10, while decreasing for all other Windows versions, albeit not as quickly as MS would wish for.

Yeah; 'Windoze' gets blamed for many things it shouldn't be blamed for; this time for AMD's latest and greatest.

 

I don't need to turn off updates: I control my network(s) completely otherwise. So, no difference.

Your updates for Win7 though will stop very soon - that is the difference.

I don't need to think twice (again); I already have (at the evaluation part - almost two years ago).

Yeah I do know. See link below. But many benchmarks by definition is not all benchmarks. "Pinnacle' doesn't mean almost - it means above all else. This debate is not pointless or meaningless to me. My business decisions are based on such observations (and more).

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

I think you can see I'm not force fed anything I don't want or cannot use to my advantage?

And anytime you hear 'bang for the buck' or 'value' is exactly when you know there is something better out there if you can afford to spend more $$$$.

 

I can see now, that you are only a fanboy, as you cannot seem to grasp the fact that what you say, does not make any sense. Your quote not mine... "They are far from the Pinnacle".....your reverse stepping and double talk makes for a confusing discussion, and thus pointless, and meaningless. You say one thing, then mix words to come out as if you had said that in the first place, or as if that is what you meant to say.
Windows 10 has been stuck at 400 million for almost 3 months, and as of February of 2017 the user base for Windows 10 has dropped, and Windows 7 has gone up.

 

The facts are that unless you work at a very high level within AMD, you don't know any of what you wrote to be 'facts'.

Have you written microcode for processors before? I haven't - but I can well imagine that the hardware would look like something completely different to the O/S - depending on the microcode changes introduced.

Like already stated; AMD was a moving and elusive target for O/S and M/B manufacturers for mere weeks prior to their official launch.

I don't see how it becomes anyone else's responsibility? Even with your possible sequence of events quoted below.

 

http://news.softpedia.com/news/wind...as-windows-10-loses-market-share-513431.shtml

 

Really, no fanboy.

But you quote one of my posts and comment as if to another? You're confusing me.

I haven't changed nor double talked in this thread so far.

Sept 2016 was the 400 million mark for Win10. There is no way it is less now?

Where do you see the user base dropping in Feb 2017? Link?

And Win7 going up? Nah... now you're just playing with me.

 

Uhm, that is a blog post of a 'report'.

Good clickbait. No data. No links to source. Not too reliable?

 

Click bait? Netmarketshare?? Would you like me to post the Net market share link, I just pulled one up out of thin air. You are really kind of......well....I wont say it. Suffice it to say you have truly been awarded the ignore button. Only because I appreciate a good debate. Not one that is filled with an unwillingness to be wrong at any point.

 

For my newly awarded ignore list user!

http://www.netmarketshare.com/

 

Ah, another one to put me on their ignore list? Ignore whoever you want at your whim.

But just so you know; I love to be proved wrong - it means I've learned something.

But comparing the 400M installed base to netmarketshare shows your misunderstanding of the facts.

Those two are not comparable to each other. They're measuring different things.

So, are you willing to be possibly wrong and continue debating?

 

Thanks for semi-ignoring me.

But I know the link - see my post above.

 

Let me restate: the F3 revision was available BEFORE CES! That contained the current fix on the IMC (still waiting on the microcode update in two months) and operated at 100MHz less boost. The companies, to one reporter, said 5-6 weeks, while the one above said 2-3 weeks. Considering that means they may have had the F3 revision in December, which a reporter at CES said it was practically finished silicon, and the December 3.4 variant (which I think had a 3.7 boost, which seems awfully like the silicon of the 1700 and 1700X on timiings), I'd say it was at minimum 8 weeks they had with the F3 variant. Because of this, I'm going to call ********. Timeline and known issues say the silicon that was practically done and had the ram fix to that point was in their hands by the start of the year. So, as I said, getting the firmware to a good base before hand, then finalizing with the final silicon after that point, which we have established was at least 5-6 weeks for some vendors, suggests the MB companies are to blame for the none functional base firmware. I've already detailed above on what exceptions I give.

 

Those 'facts' even taken at face value still do not 100% support your argument.

Is it a possible version of the truth. Yeah - the assumed 'facts' do hint to that.

Is it probable (especially considering the massive undertaking needed from all sides to bring a new platform to life)? I do not believe so.

I guess I am more humbled than you by the processes needed to release a BIOS/firmware (or in MS' case; a new scheduler/cache 'fix') for thousands if not millions of different possible hardware components and have it validated for public consumption in as little as 5-6 weeks. Even 8 weeks is pushing it for a brand new processor with brand new capabilities and internal workings. And that is assuming that the microcode updates didn't require another fresh new round of compatibility testing.

If there was a total of 1 million test setups available and everyone in the industry worked night and day in harmony - ~6 weeks still wouldn't be enough time to validate and replace BIOS/firmware and/or O/S patches with what had been already created and tweaked for months and months beforehand.

And as I've already stated over and over; in business; nothing happens in harmony (at least not for the primary benefit of one (company) and/or just for the benefit of the customer).

 

@ajc9988 - I salute you, you lone super hero warrior samurai. You are fighting an uphill battle, so my thoughts are with you. Sorry that I don't have any more nerves to help you and it seems @Raiderman drew a line there as well

It is. As I said earlier - AMD are in no position to point fingers, because they would easily end-up with a larger FU (totally dropping them), than they are getting now (crappy support). I definitely do not support this behaviour, but it's not like we can do anything about it, so I just hope that firmwares would be fixed and most people would be happy in the end. The damage is done however, but it's just part of the "beautiful" world we live into recently

 

Yeah, Im not bothering with that, general lack of citations and refutation of citations presented means there is no discussion to be had.

 

Here's a Bulgarian contribution (Russian article).

 

Interesting Vega development if true: http://www.tweaktown.com/news/56645/amd-radeon-rx-vega-beats-gtx-1080-leaked-tests/index.html

Sent from my SM-G900P using Tapatalk

 

Are you referring to me, as this is based on prior articles posted in this thread to give details on the timeline. If you need them consolidated, I can search later today to provide the sources (there are some articles I didn't post because the content was not as relevant at the time read or at the time I decided not to post)...

Sent from my SM-G900P using Tapatalk

 

No, not referring to you.

 

On videocardz there is one more leak (the one above including), with a (687F: )C1 code (revision?) chip showing performance in between 1080 and 1080Ti. With these two leaks, the logic says that this (687F) is the small Vega and the C-something is the revision. What it would turn-out in the end remains to be seen, and whatever it is, I really hope that it would be soonish. April's 1st seems like a nice date for release like this. Let's see who is going to be fooled

On topic - while browsing around, here's what I've found. Make sure to read the comments as well. Basically more blame shifting, so again whatever the truth, I just hope that we'll get solid products sooner, rather than later.

 

Out of the comments, the ones I found most pertinent and germane:

I doubt this is true on all fronts, some of motherboards had Intel (R) specific bios options (which, obviously, would not exist on Ryzen), some tech reviewers on YouTube said during their chat, so removing a feature (at least hiding it) is as simple as it gets, really doubt whole guilt is on AMD's behalf.~George Janiashvili

Wow, that is pretty bad... We've seen companies copy others UEFI/BIOS builds before, so that doesn't seem too far off. Also, keep in mind that most all AMD AM4 boards are using Intel NICs, so maybe some of the code that was copied over was due to that. ~ Nathan Kirsch

Oh, please! The MB makers are hardly left "high and dry." They have just grown soft under Intel support levels. There is limited support because there is limited money. The resources simply are not there for what they want. If they cannot accept that, they may need to re-think their business. Maybe it would be better for AMD to make motherboards, like they do GPU's.~ RandSec

ya i'm not sure where he got that... it's been Q1 release for months already. so the most it could possibly have been moved up was 4 weeks or so (from end of march to start of march). ~ The Countess

Am I the only one that remembers that they launched the AM4 platform LAST SUMMER for their APUs? Granted, it was for a Piledriver-based CPU, but still...it's not like it was just dropped in their lap in February. ~ Ammaross Danan

If taken together, it says the MB manufacturers said FU to AMD and purchasers in favor of the 200 series Intel motherboards, as I previously said, which is a HARD SELL because of compatibility with the 100 series chipsets. They even outright said they focused on Intel. The rest is excuses!

Edit: Also, at CES, AMD made clear the release would be in the 1Q. Then, they hinted at late February to early March. This is why the speculation of release at the Capsaicin and Cream event circulated for awhile.

Edit 2: MB manufacturers are also trying to conflate Chinese New Year effecting manufacturing with the firmware development team. Firmware development can happen without the manufacturing team being involved. Also, it comes every year and any business heavily involved in China plans around the time off for that celebration (unless you are a complete idiot (see Diodes's management trying to cancel leave for Chinese New Year a couple years ago)).

 

With your narrowly focused view point, I would agree with your conclusions.

But you are failing to take the other side's perspective into account. These are business dealings. Nothing happens unless $$$$$$ change hands.

The shareholders of any of the companies would feel very different than you if they started giving AMD freebies to get their new platform off the ground.

I understand your side of this 'debate'. I don't understand why this aspect is any less important to the conversation?

A one sided conversation that is supposedly built on 'facts' but doesn't at least explore other explanations/possibilities of the past events is not a conversation at all.

Worse; you accept other random noise from other sites without 'citations' - yet to have an intelligent conversation with me is too much for you without you reverting to yelling, swearing and otherwise acting like anything but an adult. Too bad.

With an adult conversation; what could happen is you teach me something. And who knows; you may learn something from me too.

 

chew* @xtremesystems, Gigabyte X370 board died, so he took a quick trip to Microcenter, and picked up an Asus B350, already has it at 4.1ghz....meanwhile...

Fugger, the owner of @xtremesystems, and someone who has not tested AMD in over a decade decided to dive into Ryzen, with a little LN2

Also notice the evaluation build of Windows 10...lol,

.

 

Nice! I might setup an Ent. Lean for others to try. It's not perfect, but it gets me what I get in non-optimized win 7. It has almost all the remote stuff removed, the apps are gone, M$ store is gone, Windows Update smashed to pieces, etc. For time spy, you cannot be on the LTSB like he is here (I have modified versions of both that and anniversary edition, but you have to break the Windows Update on AE after install). Took a lot of time to get some benches to compare to win 7 (which I also can strip down, but all my keys are pre-sp1 and will not activate unless... So I tear it down on live install)....

Edit: 3000th post! WOOHOO!


Sent from my SM-G900P using Tapatalk

 

Followup video: Ryzen 7: Good CPUs, Crippled by Expectations

 

Have any of you with a Ryzen CPU gaming PC benchmarked a game with SMT ON (for a game that gets a boost from SMT) but with cores 8-15 disabled?

That would keep the communications on the same "half" of the CPU, not invoking that +100ns penalty for threads communicating across the halve's of the CPU (CCX - core complex).

40ns communication latency between thread communications when on the same CCX, and 140ns latency (infinity fabric between CCX's) between threads across the CCX's.

You could also keep all 15 cores running, but set the game / game engine / game manager core affinities all to 0-7 or 8-15 to force threads to stay on a single CCX, but allow other non-communicating threads to run on the other cores.

Watch the first video, they take forever spitting this out

AMD Ryzen and the Windows 10 Scheduler - No Silver Bullet

https://www.pcper.com/reviews/Processors/AMD-Ryzen-and-Windows-10-Scheduler-No-Silver-Bullet

Ryzen R7 1700 GAMING BENCHMARKS (7 games tested vs. 7700K!)

 

Interesting, and thanks for the summary you wrote too. Do they have written articles as well as the youtube videos, because I hate watching videos for information as opposed to quickly reading parts of articles?
EDIT: oh yeah, I missed the link you gave, yes it's a written article plus the video, thanks.
EDIT 2: after quickly reading/skimming the article it seems that the conclusion is that the somewhat dissapointing gaming performance is due to latency in communication between the the 2 halves of the Ryzen CPU, and that it's not a Windows scheduling problem - this is not a particularly rosy conclusion for the future of Ryzen as a gaming CPU when it comes to the 8 core models. I'm assuming that the 4 core Ryzen CPUs won't be split in half, and therefore won't suffer from this issue?

 

First Ryzen Quad Cores Will not Pass 3.2 GHz?

 

Well that would suck for gamers (& others), assuming that the lower stock clocks were indicative that these CPUs won't overclock as high as the 8 core models that have just been released - but only if you assume that. I think in reality they'll be able to overclock just as far the 8 core models.

 

If it turns out that first Ryzen Quad Cores don't Pass 3.2 GHz... Then I will say that this CPU is a BIG Flop before it is launched!! Unlikely, but you never know

 

As explained, the NUMA + feature support needs to be added by developers, maybe through MS development tools it could be semi-automatic - forcing all apps to recognize "multi-processor" + "multi-core" would do it, as mentioned some apps only fill the first processor cores, and ignore the 2nd or additional processors, but not all do.

It's like a technology timing issue - as new apps get developed on new technology builds they incorporate the features automatically.

If your app doesn't need more than 4 cores + 4 threads, then setting affinity should help avoid the cross CCX penalty, maybe setting to Cores 8-15 so the OS can assign other apps 0-7 - does Windows 10 still favor low cores first over high cores?

I wonder if AMD could come out with a service that would run and check performance of applications and automatically distribute thread affinity for apps so users don't have to do it manually?

 

Although in relation to the bit I put in bold in your message above, latency is still latency in communication between the two processor cores/halves, so that latency is still gonna be there regardless (it's part of the hardware design), so that's not ideal if your game is needing to use more than 4 cores. You're saying that maybe that won't be a problem when developers make games with this in mind, but I'm not sure, we'll have to see.

 

There are already plenty of apps not affected and showing full use and no penalty for Processor to Processor separation, some games show little penalty, with others showing a 10% drop in the cores vs high ghz 4 core performance.

I can live with 10% less performance in somethings, to make up for it in others.

Even if 100% of the things you run now are 10% slower for $329 with an OC'd 1700, for many it's nice to be able to give AMD the money instead of Intel for a change

 

Fair enough, I can understand that view point, the point I was getting at is that it's not a 'perfect' design and that latency between the 2 halves will always be there (according to that article you linked) - so even though they are good value & offer good performance it seems that Windows 10 scheduler is not to blame for the somewhat disappointing gaming performance & that future updates from AMD or anyone else look like it might not fix the issue.

 

I red that they have issues with the Quads, that's why they are pushed for H2. Don't know what the problem is exactly, but my guess is if they can't fix the design and the yields of the 8c are good, they could lock an entire CCX and push it for 4c. If none of this happens, well the 4c would be a flop. The 6c would be the giveaway - if there are batches that can be unlocked to 8c - good yields; if not - you get the picture.

Yet microsoft said that they'll be working on a fix.

 

The end result right now, how it performs in worst case and best case applications, still provides a great alternative to Intel.

It's up to you to decide how the new AMD architecture, 8 cores / 16 threads, fit into your workload.

For me a top end loss in some games, especially when the FPS is still far higher than I need in 1080p, doesn't reduce my interest in having all those cores available for other uses.

Noone has mentioned it here yet that I have seen, but many report much smoother lag free game play as compared to their Intel 4 core CPU's.

Consistent smooth game play is more useful than higher FPS I won't see.

I'm still very interested in the Ryzen 7 CPU's even in their current state.