Why does Ryzen 7 1800X performs so poorly in games?

[lolfail9001]

How about you directly reference owners of the boards and their performance issues... Oh wait they don't seem to be having any.

Curious how AMD talks about issues that do not exist, then, is not it?

 

[Presbytier]

How about you directly reference owners of the boards and their performance issues... Oh wait they don't seem to be having any. You make it sound as if they don't work at all.

Why don't you go to your local PC store and buy a 7700K and spend the next few years talking about how great it is in the Intel forum section. Because so far with your lack of ownership of a 1700/1700x/1800x, you lack any real knowledge that you can impart here.

You are absolutely right we can just ignore all the people reporting these issues including tech sites, owners and reviewers alike. Obviously none of these issues are real because i personally have not experienced it. Oh and space is not a vacuum you can breathe just fine out there because I have never been out there and neither have you it must be true.

P.S. me thinking a product is a bit disappointing is not a personal attack, so maybe stop taking this so personally. I also don't like fruit loops and i have never had a fruit loops fan react so negative to me not liking it.

 

[Gideon]

Look the SMT issue is similar to Intel's HT issue, I'm sure that will be fixed, the CCX issue, well that is something quite different.

Razor1 your the only talking about a supposed CCX issue. There are reports about windows 10 scheduler issue and the fact that Windows 7 does not suffer from it and the chip performs better. So unless you can point to at least two articles about it, it's just in your head man.

 

[razor1]

Razor1 your the only talking about a supposed CCX issue. There are reports about windows 10 scheduler issue and the fact that Windows 7 does not suffer from it and the chip performs better. So unless you can point to at least two articles about it, it's just in your head man.

That is not a just in Windows 10, that problem is not scheduling problem, CCX is a latency issue accessing L3 Cache. When disabling specific cores if that problem wasn't there, we wouldn't see the latency change, but we do see it change. So again.......

This is why I stated to another on this very forum, the Windows scheduler if you look at what it does, has nothing to do with the way the cache works in the CPU, that is handled by the CPU instruction scheduler. The windows scheduler is specific for Windows needs and to stop conflicts of its processes and application processes, those issues can be fixed. The CCX latency problem, windows scheduler has no means of controlling that. Its impossible, programmers have little or no control over it either. They can code around it by making sure critical path code stays on one thread, but that is it.

 

[JustReason]

Curious how AMD talks about issues that do not exist, then, is not it?

That isn't the point. Yes there are issues but in the context of ownership and real world... ? He is making it sound like they cant beat a FX 4300. So what if it doesn't top a 7700K.

Go back 6 months or even a year:

1: We KNOW it wont clock over 3.3Ghz because the node based on Samsung is tailored to 3.0Ghz. Now I am not degrading this line of thought, it was good and reasonable for the information that was available. But bear with me a little longer.

2: AMD told us 40% IPC over Excavator. Most thought AMD would not make that as they tend to give inflated specific situation type projections... Maybe hitting Sandy but no way getting Haswell.

3: AMD would use so much more Power like they have done for a while, Intel was far ahead in efficiency.

Now given what we thought and what we expected, here is what we got:

1: Clocks up to 4.0Ghz stock and so far 4.1Ghz OC. Well above 3.3Ghz, but again this is not a bash at those that originally expected lower as the information pointed to that outcome. (In other words Bias was not at play).

2: We got >50% IPC putting AMD between Haswell and Broadwell.

3: Thus far AMD has been within the realm of Power efficiency with Intel. And some preliminary reviews put this new AMD Zen at better than Intel with lower TDP (35W) (not sure on the authenticity of this as a whole in regards to Intel).

As of now there are some issues ranging from minor to severe ( that one line of boards that has been bricking/not posting). Seems quite a few are enjoying them just fine and many speaking to their equaling their own Intel rigs in gaming. Even Kyle spoke to playing one of his VR games and having no CPU related issues. Doesn't mean the issues don't have merit but we must consider that this is a new arch From AMD and it will take time as most software and hardware companies don't generally put an exhaustive amount of time into their AMD lines.

But afore most we must be rational and reasonable here. It isn't the best of all choices but it is a damn good if one so chooses.

And if one were to have a choice between 4 cores or 8 being both nearly identical in price and performance difference was slight as in this case, why not get the 8 core. It is better to have them and not need them than to need them and not have them.

 

[Presbytier]

That isn't the point. Yes there are issues but in the context of ownership and real world... ? He is making it sound like they cant beat a FX 4300. So what if it doesn't top a 7700K.

Go back 6 months or even a year:

1: We KNOW it wont clock over 3.3Ghz because the node based on Samsung is tailored to 3.0Ghz. Now I am not degrading this line of thought, it was good and reasonable for the information that was available. But bear with me a little longer.

2: AMD told us 40% IPC over Excavator. Most thought AMD would not make that as they tend to give inflated specific situation type projections... Maybe hitting Sandy but no way getting Haswell.

3: AMD would use so much more Power like they have done for a while, Intel was far ahead in efficiency.

Now given what we thought and what we expected, here is what we got:

1: Clocks up to 4.0Ghz stock and so far 4.1Ghz OC. Well above 3.3Ghz, but again this is not a bash at those that originally expected lower as the information pointed to that outcome. (In other words Bias was not at play).

2: We got >50% IPC putting AMD between Haswell and Broadwell.

3: Thus far AMD has been within the realm of Power efficiency with Intel. And some preliminary reviews put this new AMD Zen at better than Intel with lower TDP (35W) (not sure on the authenticity of this as a whole in regards to Intel).

As of now there are some issues ranging from minor to severe ( that one line of boards that has been bricking/not posting). Seems quite a few are enjoying them just fine and many speaking to their equaling their own Intel rigs in gaming. Even Kyle spoke to playing one of his VR games and having no CPU related issues. Doesn't mean the issues don't have merit but we must consider that this is a new arch From AMD and it will take time as most software and hardware companies don't generally put an exhaustive amount of time into their AMD lines.

But afore most we must be rational and reasonable here. It isn't the best of all choices but it is a damn good if one so chooses.

And if one were to have a choice between 4 cores or 8 being both nearly identical in price and performance difference was slight as in this case, why not get the 8 core. It is better to have them and not need them than to need them and not have them.

I never touched on any of those things you responded to me lamenting the state of Ryzens launch which is plagued by MOBO issues amongst all the other issues. This is not a smooth launch by no means.

 

[tangoseal]

Hopefully a new patch for 10 is out soon.

 

[JustReason]

You are absolutely right we can just ignore all the people reporting these issues including tech sites, owners and reviewers alike. Obviously none of these issues are real because i personally have not experienced it. Oh and space is not a vacuum you can breathe just fine out there because I have never been out there and neither have you it must be true.

P.S. me thinking a product is a bit disappointing is not a personal attack, so maybe stop taking this so personally. I also don't like fruit loops and i have never had a fruit loops fan react so negative to me not liking it.

read my post to lolfail9001.

 

[Gideon]

That is not a just in Windows 10, that problem is not scheduling problem, CCX is a latency issue accessing L3 Cache. When disabling specific cores if that problem wasn't there, we wouldn't see the latency change, but we do see it change. So again.......

This is why I stated to another on this very forum, the Windows scheduler if you look at what it does, has nothing to do with the way the cache works in the CPU, that is handled by the CPU instruction scheduler. The windows scheduler is specific for Windows needs and to stop conflicts of its processes and application processes, those issues can be fixed. The CCX latency problem, windows scheduler has no means of controlling that. Its impossible, programmers have little or no control over it either. They can code around it by making sure critical path code stays on one thread, but that is it.

Aida64 still has not been fixed yet as far as I know, so how do we know if were getting the correct numbers? Is there another program out there that shows that kind of info and works with Ryzen?

 

[Presbytier]

read my post to lolfail9001.

I did and even replied to it

 

[JustReason]

I never touched on any of those things you responded to me lamenting the state of Ryzens launch which is plagued by MOBO issues amongst all the other issues. This is not a smooth launch by no means.

Seriously having to explain these posts is tiresome.

The performance of these processors is far above what most were expecting, in most cases 6 months ago: BY A COUNTRY MILE.

Here you are complaining about the performance because it isn't beating the 7700K.

As far as the issues with MoBos, it isn't as horrendous as you infer, being only one seems catastrophic. The rest are operating fine and giving their owners great performance when you read what they tell you in these posts. The only Issue in the MoBo at the moment is memory speeds and stability, which is quite common in new runs more so on new architectures. Those issues will take some time, based on past updates, about 6 months for quality speeds and just abit more for higher OC speeds.

The Windows and software patches will come and some of these architecture related issues will be sorted out, to what degree is still unknown. Seems the CCX issue could be a handful but thread load related issues, be it core parking and populating will be quick.

Your doom and gloom posting is what gets you so much trouble and friction with other members in these threads. You are being far too extreme and in much the same way as you accuse others on the other end. I see there are issues but it doesn't detract from the performance enough to be too concerned. At the present state it performs quite well, so if it doesn't receive any benefit from updates or such it isn't a loss at all. But if it does then well all the better.

 

[razor1]

Sis Sandra should be able to do it too. haven't seen those tests yet though.

There is a problem here too.

That is two different software suites to test cache performance showing similar results. Even if it was a problem in one of the suites, is going to happen in both?

 

[lolfail9001]

1: We KNOW it wont clock over 3.3Ghz because the node based on Samsung is tailored to 3.0Ghz. Now I am not degrading this line of thought, it was good and reasonable for the information that was available. But bear with me a little longer.

Well, "we" were not even that off, turns out that optimal operating range ends precisely at 3.3Ghz. Rest is basically factory overclocking. Should have guessed from looking at Polaris, i admit.

2: AMD told us 40% IPC over Excavator. Most thought AMD would not make that as they tend to give inflated specific situation type projections... Maybe hitting Sandy but no way getting Haswell.

Most also did not realize that the Excavator was slower than Piledriver in their estimations. Now that's a plot twist i do not think anyone has expected. But yes, i believe no one within their reason expected getting Haswell, that's why i was so quick to hop on hype train when Passmark leak happened. Also because it corroborated Geekbench results from early August, but i digress.

3: AMD would use so much more Power like they have done for a while, Intel was far ahead in efficiency.

Well, AMD instead misled about TDP once again, what's not to love. Granted, it should be pretty darn efficient at low clocks.

But afore most we must be rational and reasonable here. It isn't the best of all choices but it is a damn good if one so chooses.

Yes, and that's the sort of conclusion most of reasonable people made. Drama, if we are brutally honest, stems from few people refusing to accept that Ryzen may actually have some flaws beyond teething issues (and it does, let's not kid ourselves here). It still makes it a perfectly acceptable CPU, that is great in certain circumstances (dev box or poor man's workstation), but is not perfect. Or if it does, that these are totally non-issues and are all fake and superstitious. And few people pretending that just because Ryzen has some flaws and teething issues, it is the worst thing since Bulldozer.

 

[JustReason]

Sis Sandra should be able to do it too. haven't seen those tests yet though.

There is a problem here too.

That is two different software suites to test cache performance showing similar results. Even if it was a problem in one of the suites, is going to happen in both?

Has anyone tried doing any testing with that memory latency, like other means to see if it is accurate? Most of the benches seem reasonable except that one and the Multimedia. Have not seen any real indepth discussions just mentions of those benches.

 

[Gideon]

Sis Sandra should be able to do it too. haven't seen those tests yet though.

There is a problem here too.

That is two different software suites to test cache performance showing similar results. Even if it was a problem in one of the suites, is going to happen in both?

Perhaps if they use the same method. When Aida64 fixes their software and shows high latency still then it's a issue, but I have a hard time with believing that since it seems to have no issue with the cache in Windows 7. Something just does not make sense there.

 

[razor1]

Has anyone tried doing any testing with that memory latency, like other means to see if it is accurate? Most of the benches seem reasonable except that one and the Multimedia. Have not seen any real indepth discussions just mentions of those benches.

nope haven't seen any more don't think it was really on top of reviewers priorities to do it :/

 

[razor1]

Perhaps if they use the same method. When Aida64 fixes their software and shows high latency still then it's a issue, but I have a hard time with believing that since it seems to have no issue with the cache in Windows 7. Something just does not make sense there.

Two separate companies, two or more different programmers using the same algorithms? Highly unlikely, that is like a million toothpicks and with a million ways to order them and picking out the exact same order.

 

[JustReason]

Well, "we" were not even that off, turns out that optimal operating range ends precisely at 3.3Ghz. Rest is basically factory overclocking. Should have guessed from looking at Polaris, i admit.

Most also did not realize that the Excavator was slower than Piledriver in their estimations. Now that's a plot twist i do not think anyone has expected. But yes, i believe no one within their reason expected getting Haswell, that's why i was so quick to hop on hype train when Passmark leak happened. Also because it corroborated Geekbench results from early August, but i digress.

Well, AMD instead misled about TDP once again, what's not to love. Granted, it should be pretty darn efficient at low clocks.


Yes, and that's the sort of conclusion most of reasonable people made. Drama, if we are brutally honest, stems from few people refusing to accept that Ryzen may actually have some flaws beyond teething issues (and it does, let's not kid ourselves here). It still makes it a perfectly acceptable CPU, that is great in certain circumstances (dev box or poor man's workstation), but is not perfect. Or if it does, that these are totally non-issues and are all fake and superstitious. And few people pretending that just because Ryzen has some flaws and teething issues, it is the worst thing since Bulldozer.

Just want to clarify AMD has always done that with their TDP, which is why over years people keep posting TDP does not equal W. Watt usage thru a CPU does not all turn in to heat. Like I said about the FX 8350: it originally listed as 125W but then after boards were released got changed to 140W (TDPs). The actual W usage at stock on these can hit 200W during benches. Like I said TDP does not equal W, at least with AMD. Even their GPUs have always exceeded TDP figures with W usage figures.

 

[Gideon]

Two separate companies, two or more different programmers using the same algorithms? Highly unlikely, that is like a million toothpicks and with a million ways to order them and picking out the exact same order.

Oh come now, no programmer ever copied another program before But still does not explain why it speeds up so much in Windows 7 unless that Windows 10 scheduler is really causing havoc.

 

[lolfail9001]

Perhaps if they use the same method. When Aida64 fixes their software and shows high latency still then it's a issue, but I have a hard time with believing that since it seems to have no issue with the cache in Windows 7. Something just does not make sense there.

There are tests of cache performance in Win 7?

 

[razor1]

Oh come now, no programmer ever copied another program before But still does not explain why it speeds up so much in Windows 7 unless that Windows 10 scheduler is really causing havoc.

ah no they don't man, in business they don't do that. its one thing to take a math problem, and make your own algorithm out it, but to copy it outright, not that hard to catch.

Coding is like writing book, you can have two authors having the same plot and conclusion, but getting from point A to Z will be totally different.

Windows 10 schedular is screwing some things up, not all of it though.

 

[Presbytier]

Your doom and gloom posting is what gets you so much trouble and friction with other members in these threads. You are being far too extreme and in much the same way as you accuse others on the other end. I see there are issues but it doesn't detract from the performance enough to be too concerned. At the present state it performs quite well, so if it doesn't receive any benefit from updates or such it isn't a loss at all. But if it does then well all the better.

I don't think I was being Doom and gloom nor does I think i have problems with really any poster. I also have not denied that it works for some people, but the launch has not been smooth nearly all the motherboards are having issues and the fact is most of these could have and should have been addressed pre-launch and AMD has not been as forthright post launch about this issues either. You also have to admit AMD knew about all these issues ahead of time and decided put their product up for pre-order and decided their customers did not need to know about potential launch bugs and issues ahead of time.

 

[Gideon]

There are tests of cache performance in Win 7?

Not that I have seen, just bench runs of games that showed about a 10% gain or so.


https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-8#post-38775732

 

[razor1]

well not all 10% some but average is closer to 5%, that is what I expect to see once the windows 10 scheduler is corrected. Probably another 5% form the memory issues coming from the lower speed memory.

 

[Gideon]

One thing I will say is it would have been nice if this was all smoothed over before hand on the launch, but I think Ryzen outperformed our expectations. I just think AMD tried too hard to keep this out of anyone hands and have info leak out and they hampered their own launch doing that. Either way I am happy to have something new to muck around with and rough edges just dont scare me as I bought a AMD Slot A processor and a VIA motherboard. Oh fun times with drivers, but man that chip screamed once I got it sorted out.

 

[lolfail9001]

just think AMD tried too hard to keep this out of anyone hands and have info leak out and they hampered their own launch doing that.

And here i am thinking that Intel's Conroe review policy have set a baseline for how you do a disrupting launch.

 

[NKD]

I don't care about the resolution if its GPU limited, its GPU limited, you will not know what the CPU puts out PERIOD.

Might as well have AMD rub gamer's faces with workstation loads and tell us its gaming performance is the same as Intel's CPU's by showing us GPU limited scenarios, end results the same thing, when you have a GPU limited scenario you don't know shit about how the CPU is doing. This is what they did and now they are getting the results from what they sowed.

And there is no fuckin excuse for threatening reviewers for their tests.

This is not a political thing, but its the same crap that happened with Trump won, these people should get a life, AMD lost again, nothing is going to change that! Doesn't matter who they threaten, whom ever they beat up, whom ever they wish to point and shoot, stupid people do stupid things and should be punished for doing those things.

Death threats are against the law, those people should be strung up in jail and then let the courts decide their future.

amd is behind in gaming. They will sell shit load of ryzens as well. Both are facts I am willing to stand behind. AMD has not lost as long as ryzen sell, which it will. Gamers will by 7700k, that is still pretty high end system. When the 4/6 core ryzens come out they will probably sell out as well due to price/performance. I wont recommend anyone buying anything other than ryzen 1700, but those people will cuz they like the x behind the number, and don't see it doesn't do anything.

 

[razor1]

amd is behind in gaming. They will sell shit load of ryzens as well. Both are facts I am willing to stand behind. AMD has not lost as long as ryzen sell, which it will.

I don't think they will sell a shit load of ryzens, lets see Newegg preorder, reviews, don't see a shit load of reviews up there.

Amazon Ryzen was in the top three sales when launched,

Look at them now

https://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189

Not selling as well as their old AMD counter parts.....

Granted its a snap shot of sales with a time slice, but still, its going back to where it should be selling over all. Since Amazon's sales go by numbers of sales based on time, gives us an idea the initial launch well preorder sales came in heavy, after that its gone down to normal or average levels again.

Lets rephrase that, AMD is not lost as long as Ryzen is there, even at current volume sales of AMD products, it will help by rising margins.

 

[cybereality]

I think Ryzen will sell, but probably moreso the cheaper chips coming later this year. If you look on Amazon, most of those top sellers are the $200 - 300 chips, and even the FX is beating Ryzen (again because they are so cheap).

 

[NKD]

I don't think they will sell a shit load of ryzens, lets see Newegg preorder, reviews, don't see a shit load of reviews up there.

Amazon Ryzen was in the top three sales when launched,

Look at them now

https://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189

Not selling as well as their old AMD counter parts.....

Granted its a snap shot of sales with a time slice, but still, its going back to where it should be selling over all. Since Amazon's sales go by numbers of sales based on time, gives us an idea the initial launch well preorder sales came in heavy, after that its gone down to normal or average levels again.

Lets rephrase that, AMD is not lost as long as Ryzen is there, even at current volume sales of AMD products, it will help by rising margins.

whats wrong with the picture? I see both 1700 and 1800x in top ten still, no? Market for 8 core CPUs will always be less than 4 core and 6 core. You and me both know 4-6 core ryzen chips are going to be really good bang for buck. They will certainly be replacing few of those intel processor.

Look no denying its slower in games. but common now, I am not ignorant to the fact that these are going to give amd a jump start in CPU that they needed. People aren't stupid with their money. Most people that need multi core performance for cheap will go that route. High end gaming and if thats all you do. You will get 7700k, and call it a day.


Cheaper chips will always sell. I think 4-6 core ryzen chips will sell really good to be honest for gamers even.

 

[IdiotInCharge]

Cheaper chips will always sell. I think 4-6 core ryzen chips will sell really good to be honest for gamers even.

This is the kicker: and while motherboard makers are having difficulties, when AMD releases these chips at super-competitive prices (i7 configurations for i3 prices), that difficulty should be solved and they'll start yanking back marketshare until Intel is prepared to respond.

 

[Gideon]

I don't think they will sell a shit load of ryzens, lets see Newegg preorder, reviews, don't see a shit load of reviews up there.

Amazon Ryzen was in the top three sales when launched,

Look at them now

https://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189

Not selling as well as their old AMD counter parts.....

Granted its a snap shot of sales with a time slice, but still, its going back to where it should be selling over all. Since Amazon's sales go by numbers of sales based on time, gives us an idea the initial launch well preorder sales came in heavy, after that its gone down to normal or average levels again.

Lets rephrase that, AMD is not lost as long as Ryzen is there, even at current volume sales of AMD products, it will help by rising margins.

Well how many people want to buy a processor and not know when they can get the motherboard they want. I think orders will pick up once motherboard availability and issues are resolved.

 

[Presbytier]

This is the kicker: and while motherboard makers are having difficulties, when AMD releases these chips at super-competitive prices (i7 configurations for i3 prices), that difficulty should be solved and they'll start yanking back marketshare until Intel is prepared to respond.

Maybe, but maybe not. Those 4/8 chips won't be performing anywhere near i7 Kaby lake, so if anything they will eat into i5 and i3 sales but i7 sales will probably remain steady

 

[Gideon]

Man can't believe how well that old 8350 still sells.

 

[Presbytier]

Man can't believe how well that old 8350 still sells.

yeah considering a i3 or pentium is probably a better buy price to performance.

 

[IdiotInCharge]

Maybe, but maybe not. Those 4/8 chips won't be performing anywhere near i7 Kaby lake, so if anything they will eat into i5 and i3 sales but i7 sales will probably remain steady

'Anywhere near'

If the performance of the eight-core parts is consistent with the six- and four-core parts per core, then they will actually be quite near the i7 CPUs- behind a little bit, of course, but they will be priced like Intel's two- and four-core parts which they will exceed very nearly across the board.

(the top i5 part will likely still be faster for certain lightly threaded but demanding workloads, just as they are over the current eight-core R7 parts)

 

[OutOfPhase]

Just want to clarify AMD has always done that with their TDP, which is why over years people keep posting TDP does not equal W. Watt usage thru a CPU does not all turn in to heat. Like I said about the FX 8350: it originally listed as 125W but then after boards were released got changed to 140W (TDPs). The actual W usage at stock on these can hit 200W during benches. Like I said TDP does not equal W, at least with AMD. Even their GPUs have always exceeded TDP figures with W usage figures.

Actually, all the power drawn into a CPU does actually get emitted as heat. A CPU using 100W of power is generating 100W of heat.

The companies simply define what TDP means differently. Intel uses it as a max envelope spec, and AMD (as I understand it) uses it as a nominal/typical steady-state spec. In other words, yes, AMD's chips will frequently cross that number and that's expected. As you say, during benches is a good example.

 

[Shintai]

Just want to clarify AMD has always done that with their TDP, which is why over years people keep posting TDP does not equal W. Watt usage thru a CPU does not all turn in to heat. Like I said about the FX 8350: it originally listed as 125W but then after boards were released got changed to 140W (TDPs). The actual W usage at stock on these can hit 200W during benches. Like I said TDP does not equal W, at least with AMD. Even their GPUs have always exceeded TDP figures with W usage figures.

It never got changed to 140W, despite the chips drawing 140W. When you cant compete, TDP is the first thing going out the window. What AMD does is to list the TDP at some unrealistic condition either temperature wise or voltage that wont be seen in real world.

 

[legcramp]

It never got changed to 140W, despite the chips drawing 140W. When you cant compete, TDP is the first thing going out the window. What AMD does is to list the TDP at some unrealistic condition either temperature wise or voltage that wont be seen in real world.

It draws less power than the 6900K.

 

[Shintai]

I don't think they will sell a shit load of ryzens, lets see Newegg preorder, reviews, don't see a shit load of reviews up there.

Amazon Ryzen was in the top three sales when launched,

Look at them now

https://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189

Not selling as well as their old AMD counter parts.....

Granted its a snap shot of sales with a time slice, but still, its going back to where it should be selling over all. Since Amazon's sales go by numbers of sales based on time, gives us an idea the initial launch well preorder sales came in heavy, after that its gone down to normal or average levels again.

Lets rephrase that, AMD is not lost as long as Ryzen is there, even at current volume sales of AMD products, it will help by rising margins.

The DIY segment is tiny tho.

Mindfactory.de that was one of the 9 preorder sites sold ~1785 Ryzen chips so far.
http://www.mindfactory.de/Hardware/Prozessoren+(CPU)/AMD+Desktop/Sockel+AM4.html