The Radeon Technology Group (RTG) has received its first Zen 2 sample!

I haven't had anywhere near as many issues with memory compatibility recently as I used to with my launch 1700. Even the pretty generic 3000 mhz cas 16 stuff runs rated timings and speeds with zen+ for the most part. If they get to the point where it is plug-and-play like Intel with the next-gen that's a win for them. From my experience it seems like they are almost there.

Yup, it's definitely within reach; however, because of the current design relying on uncore speeds and those speeds being tied to RAM speeds, they definitely need to get plug-and-play support rolling for faster memory or figure out how to minimize the impact of slower memory. I'd prefer the latter, of course, fast memory in speed or latency is expensive, and prohibitively if you want both :D.
 
Yup, it's definitely within reach; however, because of the current design relying on uncore speeds and those speeds being tied to RAM speeds, they definitely need to get plug-and-play support rolling for faster memory or figure out how to minimize the impact of slower memory. I'd prefer the latter, of course, fast memory in speed or latency is expensive, and prohibitively if you want both :D.

As we know from The_Stilt's deep dive on the original Zen ES chip, it ran the Infinity Fabric (uncore) at 1:1, and was said that it'd likely be the case for the EPYC chips, and TR by association (whether or not that came to be the case I don't know and I don't know how we'd even be able to determine what speed it runs at).

I don't think, or at least recall, it ever being explained why they ended up going with a 1:2 speed. I mean, as far as I know, Stilt's ES chip was perfectly stable and I think it ran at 3.2GHz even. So perhaps it had something to do with being able to clock the CPU higher as being the reasoning? Yet, if EPYC does run 1:1, that'd seem to imply that TR does as well, unless the changes between a Zen's CCX configuration/features are switchable in the AGESA; thus, wouldn't require it to be adjusted in silicon. But I digress, if it's 1:1 in TR, well they have high-clocking variants, so it wouldn't seem to be hurting their CPU speeds.

The same holds true for if it was hindering the RAM speeds, since it had been an issue in early Zen ES (or at least rumored to be). Sure we've been able to achieve higher speeds now, but even Zen 1 have benefited from that with the newer AGESA updates and are able to run their memory faster. Again, if it were the case though and if TR are using 1:1, they once more turn the theory on its head since they too do well with memory clocks...

As a result, I've always wondered (felt) that it was done that way in order to provide an easy (see: free) way to uncork performance in a later generation of Zen. I had kinda hoped it'd be switched back on in Zen+, but doesn't appear to be the case. Maybe Zen 2 will get it? *shrug*

Either way, I've still been plenty happy with my 1700X! Though, I was blessed with DDR4-3200 capability day one, having intentionally purchased single-ranked Samsung B-Die RAM for my Ryzen before any of the issues came to light, as I bought it solely on the reviews that it was capable of overclocking really well in Intel systems. Whether that would've changed my feelings today, I really can't say, but I expect it'd have bittered my impression slightly. Nevertheless, I've been very happy with it and that's been with it running at only a 3.7GHz OC, as I never bothered to go higher since I didn't want to increase core voltage.

Fingers crossed that Zen2 is a drop in on my board, not that I currently even need to upgrade. :D
 
Yup.

I don't expect that they'll fail to deliver improvements, but their target isn't standing still either. You're talking about 8C/16T at 3.7GHz, whereas the 9900k is coming in at 4.7GHz on all cores. Certainly the 2700(X) will do better, and hopefully there's a 2800(X) that's binned for that purpose, but it's a tall order for AMD. I know that I'd have a hard time choosing AMD for a pure gaming build these days still; Intel has pushed that edge of the envelope just a bit more over time since the first Ryzen release.

Now, if I needed brute processing power, I'd go grab a TR2... :D
 
Yup.

I don't expect that they'll fail to deliver improvements, but their target isn't standing still either. You're talking about 8C/16T at 3.7GHz, whereas the 9900k is coming in at 4.7GHz on all cores. Certainly the 2700(X) will do better, and hopefully there's a 2800(X) that's binned for that purpose, but it's a tall order for AMD. I know that I'd have a hard time choosing AMD for a pure gaming build these days still; Intel has pushed that edge of the envelope just a bit more over time since the first Ryzen release.

Now, if I needed brute processing power, I'd go grab a TR2... :D

Of course, how many people build "pure gaming builds" ?

I mean the 2700 non-x went for $220 the other day on Amazon. The closest priced Intel chip is the i5-8400 at $225 on NE. All things considered, I'd take the 2700 even for a gaming build ;).
 
Of course, how many people build "pure gaming builds" ?

I mean the 2700 non-x went for $220 the other day on Amazon. The closest priced Intel chip is the i5-8400 at $225 on NE. All things considered, I'd take the 2700 even for a gaming build ;).

Assuming an absolutely inflexible budget, I would too- but generally speaking I'd recommend stepping up.

I'll also note that it is easy to pick a specific dollar amount to make a 'closest to' argument. It's why I don't really engage there, because that's almost never reality.
 
Assuming an absolutely inflexible budget, I would too- but generally speaking I'd recommend stepping up.

I'll also note that it is easy to pick a specific dollar amount to make a 'closest to' argument. It's why I don't really engage there, because that's almost never reality.

Assuming a "flexible" budget everyone would end up with TR2s or 9900ks and SLI'd 2080Ti's. Obviously chips are marketed to certain segments based on price, and price/performance for your needs is king. Either that or I just don't understand the argument you're making about it almost never being reality. I get it that you can just randomly pick any dollar amount to suit your argument, but it seems like AMD and Intel have certain price points for their processors (and have generally been similar for several generations), and it's just a matter of matching up where the price/performance sweet spot lines up for your budget and needs. AMD has been reactionary in price and has been for 10+ years and generally undercuts Intel slightly in price/performance.
 
Assuming a "flexible" budget everyone would end up with TR2s or 9900ks and SLI'd 2080Ti's.

In some rare cases, sure; what we see though is not a flat price/performance line, but performance 'ceilings'. Today, no matter how much you spend, you can't get faster than an 8700k/8086k for gaming, especially with overclocking considered but even without. The same goes true as you step down in price bands, and as well, keeping in mind stuff like memory pricing where the Intel parts are currently less sensitive to the use of more affordable mainstream memory over the boutique memory that makes available Ryzen CPUs shine.

And that's why the application is important. For most enthusiasts, gaming is the only truly performance sensitive application. That's why I tend to recommend Intel first, because AMD hasn't gotten the per-core performance up to par yet.

So, if performance for stuff ranging from compile times to photo or video editing is important enough where time is money- those are important to me too, but not that important!- then the base recommendation changes, and I'd actually point toward a 2700x for fastest stock speeds, and perhaps even grabbing ECC RAM as well for the compiling case and/or if an onboard storage array is desired.

AMD has been reactionary in price and has been for 10+ years and generally undercuts Intel slightly in price/performance.

This is basically a result of marketing their products from an inferior position, and likely something that they learned when they had the stronger products and tried to price them appropriately with respect to performance: they found out that they can't gain marketshare and mindshare that way, and that marketshare and mindshare are important!

Now let's bring it back to the point above, and that's that the application matters. If the application is limited to per-core performance while also say wanting five or six threads available, that being many games and most that are CPU limited, then AMD's per-core limitation is a detriment regardless of how many cores they put in a socket.

And that is why I'm skeptical on Ryzen as a default for gaming. Even the Intel 8400 can hit clockspeeds that Ryzen can't while having enough cores for games, which means that it can be as fast or faster for the majority of games. Now, I wouldn't recommend the 8400 over a 2700- and if that is a hard price ceiling, AMD it is!- but I also don't consider the price different to step up to be very significant relative to the overall price of the build. And if the price is that inflexible, I'd even consider dropping down to a 2600(X) instead.
 
Also, to add, GamersNexus did a pretty good video on the different impacts of video editing across various high-end CPUs. The summary is that there really isn't a reason to choose anything other than the 8700k for their use case which to me looks fairly typical of what a consumer would produce, though I'd argue that the conclusion also applies to the 2700(X).

And that is why I specify gaming. If the 8700k (and many higher-end Intel consumer CPUs) are faster for gaming due to higher per-core performance, there's little reason to compromise gaming potential for inconsequentially faster video renders or compile times unless those workloads have a meaningful effect elsewhere like income or perhaps family time, for example.

Hopefully AMD steps up the per-core performance with Zen2 and makes this a bit more interesting!

Video from GN:
 
  • Like
Reactions: Nobu
like this
To be honest I have yet to see a problem with per-core performance on my system. Certainly not enough of a problem to pay Intel's price.
 
To be honest I have yet to see a problem with per-core performance on my system. Certainly not enough of a problem to pay Intel's price.

Same here, I haven't seen a problem with multi-threading performance to pay AMD's price of more expensive RAM and lower per-core performance ;).

[just pointing out that this does go both ways]
 
Same here, I haven't seen a problem with multi-threading performance to pay AMD's price of more expensive RAM and lower per-core performance ;).

[just pointing out that this does go both ways]
Meh, the cheap-ass ram I bought runs fine at 2933mhz, which is good enough for me.
 
Same here, I haven't seen a problem with multi-threading performance to pay AMD's price of more expensive RAM and lower per-core performance ;).

[just pointing out that this does go both ways]

I think you are overblowing the RAM issue, especially with Zen+ and newer bios releases. I haven't had any problem running 3000mhz Cas 16 RAM at rated speeds no matter the brand (even with the POS Asus X370 Prime Pro board which was notoriously finicky as a first release board).

I've run a 1700, 1600x, 7700k, 7700, 8400, 8700k, 2600, and a 2700x in the past year. I haven't really noticed anything detrimental to gaming at 4k 60hz with my 1080Ti with any of them. The only issue I had was the 1700 had problems running memory at anything above 2933mhz, but that was last year. I do like the drop in CPU replacements with AM4 though.

Admittedly not a twitch gamer though... And if I can find a deal on a 9900k, I might pick one up to play with. I'm not going out of my way to pick one up though at $480 when I have a perfectly good 2700x. I paid $325 for the 8700k and $250 for the 2700x FWIW.
 
Last edited:
I think you are overblowing the RAM issue, especially with Zen+ and newer bios releases. I haven't had any problem running 3000mhz Cas 16 RAM at rated speeds no matter the brand (even with the POS Asus X370 Prime Pro board which was notoriously finicky as a first release board).

Let me say that that's fair; what I'm getting at is the 3200C14 stuff is expensive, and as you've no doubt seen from Kyle's benchmarks, even 3600 is desirable for pushing Ryzen to be more competitive with Intel on a per-core basis, and the cost here does go up here significantly. Obviously I'm not as concerned about this myself as I'm running 3000C15, but even in this thread you can see signatures with some very expensive memory.

Meh, the cheap-ass ram I bought runs fine at 2933mhz, which is good enough for me.

My 3000C15 stuff is fine for me too, for my 8700k- if I were to assemble a Ryzen build, though, I'd certainly explore using something faster. Might not find it worth the price increase given current pricing, but I'd look :).
 
Let me say that that's fair; what I'm getting at is the 3200C14 stuff is expensive
I blame this on AMD personally, at least I'm pretty sure they're to blame, but it could potentially be the mobo makers at fault...
Reasoning: They have it in place that you are locked out of Odd CAS timings, and it has curiously gone back-and-forth on whether you've been able to use them or they auto-switch to the next-highest Even timing.

For example, with my G.Skill TridentZ 3200 CL15 kit on my MSI X370 Titanium --keeping in mind my knowledge of advanced DRAM options (not subtimings per-se) was virtually nill-- for the first few BIOS releases you were unable to use CL15, or CL13, or CL17 etc etc. Then I had found out that if you Disabled the GearDownMode it would allow for 1/2 latency timings, which translates into Odd numbered; therefore, you could utilize CL13, CL15, CL17, etc. I believe this occurred with AGESA SummitPI 1.0.0.4b. Later on in the BIOS released it came to be that you didn't need to use GearDownMode to use Odd CAS (using same AGESA 1.0.0.4b). I stuck with an early 1.0.0.4b BIOS version for an incredibly long time, as later BIOS/AGESA updates didn't work out as well for me in other regards. At some point I went to a newer BIOS with SummitPI 1.0.0.6 and lost Odd CAS again, which I chalked up to it being the introduction Raven Ridge APUs so reverting back to their 'better safe than sorry' "Evens Only!" approach. It gave be other issues with getting 3200 working as well so... I once more rolled back.

Cue the PinnaclePI 1.0.0.2a AGESA and their improvements to memory speeds, which even Summit Ridge Ryzens could benefit from, so I gave it a go. Once again, locked to Even CAS, but this time, I couldn't even use GearDownMode to gain back my Odd CAS ability :( I could enter it in the options, it saves as Odd and shows that after a reboot, but the motherboard is applying 16. Thankfully my system has been ~98% stable at CL14 so I've been doing alright and kept with that AGESA-equipped BIOS. (I haven't looked for a newer one lately, but I may go check right now.)

Reason I don't know specifically if it's AMD or Mobo maker (MSI in my case) to blame is due to the flip-flop of that ability. I'm quite certain that initially it was in fact AMD having it locked out, but after that I don't recall. There's also my ignorance and memory (brain) that leaves me with uncertainties, as I can't specifically recall if I was running 1.0.0.4b for the longest time, or the older 1.0.0.2, as I seem to recall skipping quite a number of AGESAs due to complications getting my system stable again at the same settings.


TL;DR - If a person could be guaranteed the ability to run their Ryzen with RAM using Odd CAS latencies, it would definitely make buying RAM a far easier and cheaper endeavor! The viable kits would expand as now you wouldn't have to rely on an AMD-tailored DDR4-3200 14-14-14 kit for the best performance, but could opt for one like I did at DDR4-3200 15-15-15 and be at virtually the same performance (in my tests, between CL14 and CL15, there was no performance/latency difference in AIDA tests).

Which IIRC it was the RAM frequency that offered the greater benefit for gaming on Ryzen, and that only when you would get into the higher timings did it start to impact framerates a bit, with CL14 to 16 yielding roughly the same performance at DDR4-3200 but where CL18 at 3400 would suffer. Basically I understood it as a situation where the amount of throughput the increased Infinity Fabric speed offered outweighed the higher timing's overall latency... up to a certain point of course.
 
Ive got my 2950x so I will be excited to get a 3950x or whatever they are called.

Zen 2 cores on my x399 board .... oh its so lovely to not have to change my motherboard next generation.

Only reason I chose 2700x over getting 9900k. Motherboard! I will just pop in zen 2 chip to my x470 board and not have to worry about shit! Sell the 2700x for 220 or so and be done with it. I decided to put my money on AMD after like 8 years. I think I am going to be in red camp for a while for CPU side. I got my ryzen running at 4.2ghz at 1.375v rock solid on my existing water cooler. Cant complain. More than enough for everything I want to do. Only thing you have to be careful with ryzen is high speed memory. I accidently made a mistake of ordering 3600mhz but it wasn't off the qvl. MSI board just didn't want to play along, ordered Hyper X 3466mhz since 3600 wasn't in stock anywhere that was on QVL. Ran AIDA and bamn been running rock solid at 3466mhz for 2 hours already. G-skill just didn't want to be stable. So as long as you order memory of the QVL list it should be gold! I was a dumb ass I thought I ordered 3600 model of the QVL list but it was few letters off lol. I saw results and after 3466 there wasn't much difference, so decided to let go of my bragging rights and settle with that lol!


As long as you buy ram off the QVL list (referring to higher speed ram than officially supported) no one really should have any issues with ram speed.
 
Also, to add, GamersNexus did a pretty good video on the different impacts of video editing across various high-end CPUs. The summary is that there really isn't a reason to choose anything other than the 8700k for their use case which to me looks fairly typical of what a consumer would produce, though I'd argue that the conclusion also applies to the 2700(X).

And that is why I specify gaming. If the 8700k (and many higher-end Intel consumer CPUs) are faster for gaming due to higher per-core performance, there's little reason to compromise gaming potential for inconsequentially faster video renders or compile times unless those workloads have a meaningful effect elsewhere like income or perhaps family time, for example.

Hopefully AMD steps up the per-core performance with Zen2 and makes this a bit more interesting!

Video from GN:


I am not sure if its really related to per core performance because there is not a crazy difference there. I think it really has to do with higher boost on intel CPUs and them not having to deal with CCX for 8 core chips. Also How Ryzen really makes a big difference when using higher speed ram. I think that is the area they might focus on improving. Pretty much improving interlink speed between CCX. However if they go with 8 core modules with 7nm, that might fix their issue. Right now infinity fabric scales well with higher speed ram but after like 3400ish it maxes out it seems. So communication speed between the CCX will be the key, but AMD will be ahead in this game though. I think Intel will be going the same route to multi chip modules from what it seems for monolithic chips. So AMD is already a step ahead when intel takes that turn. I bet AMD will keep improving the link speed between CCX as they keep going to bigger chips.
 
They do (cannon lake), and fwiu it's even shipping in low volume, they just don't have any desktop chips out yet and are having trouble with yields.

The only available product of that line is very underwhelming. Lacks the iGPU of the chip it replaced and still uses more power at lower clock speeds. Their only indication of 10nm is nothing to worry about for AMD. In fact it's a point of concern for Intel. The i3 8121U is a regression from it's 14nm predecessor.
 
Same here, I haven't seen a problem with multi-threading performance to pay AMD's price of more expensive RAM and lower per-core performance ;).

[just pointing out that this does go both ways]
Not sure why you thing ram is more costly in an AMD system rather than an Intel one. Ram is Ram and you pay the going price regardless of where you plug it. So no, that only goes one way a little ipc vs a lot more $$$.

Just as an aside I have a pair of 32GB dual channel kits of Corsair vengeance 3200 CL16. Couldn't run that at full XMP settings on my older Intel I7-6700K based system. Wouldn't run at XMP settings on my current AMD system either, but at least I could get the AMD system closer with a drop just of a speed notch. It wasn't so happy with that in the Intel system that almost forced me right back to jedec settings. If I had bought as a 1st owner I could have RMAed them. And yes they cost the same regardless of which system I wanted to run them on.
 
Last edited:
Not sure why you thing ram is more costly in an AMD system rather than an Intel one. Ram is Ram and you pay the going price regardless of where you plug it. So no, that only goes one way a little ipc vs a lot more $$$.

This lol! I got hyper x 3466 for 187 and I have that running at 3600 and cpu at 4.2ghz at 1.375v set in bios but VID is actually 1.344v under load. It was simply plug and play for me. As long as you get the memory off QVL it should work just fine.
 
bahahahahahaha! Wccftech: "gotta keep them articles coming''! I don't think we needed this attention. OP isn't posting shit now ROFL!

I know i was laughing a lot when I saw it. Like really......They reek of desperation
 
I know i was laughing a lot when I saw it. Like really......They reek of desperation

They make an article about anything and then after the short rumor they will copy paste two pages of information about what ryzen 2 is suppose to be to make it longer. Its like every article repeats ten articles before it, to make it longer. ROFL! They live on hype. Like the 9900k, omfg it is so fast 11.5% faster in ashes then 8700k, but then they won't mention the part hey it has 2 more cores and is boosting much higher all core than 8700k. So when you look at that, that 11% isn't so mighty impressive. But hey gotta get them hits!

Also I don't think they read down the thread where the OP said they took the kit away from RTG already lol!
 
Wait, if the zen 2 ES was being used by RTG, then it maybe:

1) Has an integrated Vega and is possibly a single functioning ccx, just like current Raven Ridge. I doubt AMD has a Zen 2 updated Raven Ridge sample, yet. Maybe.

Or,

2) doesn't have an integrated Vega at this time, but is still a single ccx, just like Raven Ridge, and is intended for RTG to pair with Vega and run tests/develop drivers.

"Different interconnect" is probably referring to updated infinity fabric design. I.e. ccx to Vega bridge.

My point is, maybe, just maybe, it's a single ccx of 8 cores, and not a full Zeppelin. Otherwise, why would RTG have their hands on a normal Zeppelin (2 x ccx), when they are doing design work regarding an updated "interconnect" i.e. infinity fabric?
 
They make an article about anything and then after the short rumor they will copy paste two pages of information about what ryzen 2 is suppose to be to make it longer. Its like every article repeats ten articles before it, to make it longer. ROFL! They live on hype. Like the 9900k, omfg it is so fast 11.5% faster in ashes then 8700k, but then they won't mention the part hey it has 2 more cores and is boosting much higher all core than 8700k. So when you look at that, that 11% isn't so mighty impressive. But hey gotta get them hits!

Also I don't think they read down the thread where the OP said they took the kit away from RTG already lol!
Personally, I just like stopping by WCCF because of it being a good place to find... well to accurately put it, they are the Rumor Mill. It's a one-stop-shop for me to see what tech rumors are floating around, and from there I take it with either a grain of salt, or if my current tech knowledge invokes a raised eyebrow, I'll consider it BS.

While they DO report a lot of rumor as news, and furthermore do exactly as you said by fluffing up said rumor-reports with recycled tables and known-details... In fairness, they do generally end up showing us rumors that pan out. Whether or not they've correctly worded/interpreted the rumor or leak, that's another story... It's as I said though, if you use your own common sense then you aren't going to fall for some of the nonsense. Case in point is what you've pointed out with the 9900K. They've spun it as being an architectural improvement, whereas to anyone that's followed the rumors around the 9900K will also be aware of it having a core count advantage as you did, so can conclude that it also makes up for that gain over the 8700K.

In the end, the data is still useful, even if their regurgitation of it isn't so much.


Wait, if the zen 2 ES was being used by RTG, then it maybe:

1) Has an integrated Vega and is possibly a single functioning ccx, just like current Raven Ridge. I doubt AMD has a Zen 2 updated Raven Ridge sample, yet. Maybe.

Or,

2) doesn't have an integrated Vega at this time, but is still a single ccx, just like Raven Ridge, and is intended for RTG to pair with Vega and run tests/develop drivers.

"Different interconnect" is probably referring to updated infinity fabric design. I.e. ccx to Vega bridge.

My point is, maybe, just maybe, it's a single ccx of 8 cores, and not a full Zeppelin. Otherwise, why would RTG have their hands on a normal Zeppelin (2 x ccx), when they are doing design work regarding an updated "interconnect" i.e. infinity fabric?
Or it's just like OP said once or twice, and they were using the ES to hammer out some driver issues/tweaks. :p
AMD has said that their cards interact with the Infinity Fabric even when they're dedicated cards, since the PCIe bus is tightly woven into the IF.

Your ending point could still ring true to a point though, in that there has potentially been a large change to the IF and therefore to properly leverage and work with the new IF it's needing driver tweaks.
I'll be pleasantly surprised if Zen2 has 8 cores per CCX, but personally I'm not expecting any increase in core count per-CCX.
Now, could they add a third CCX per-die, for 12 total cores? That feels way more plausible; however, my gut says that we won't see core count increase until Zen3, with PCIe 4.0, maybe DDR5 (massive maybe), and as such a new socket with higher pin count.

We've had the Tick-Tock going. Zen Ticked, Zen+ Tocked, so Zen2 would Tick and I'd anticipate a Zen2+ Tock, meaning Zen3 would be another Tick. They promised us our sockets will last us for a number of years (2020, or was it 2021 for AM4?). If we get Zen2 again in the May timeframe of 2019, and Zen2+ around the same time 2020, we're on track for that promise and "AM5" for Zen3 in 2021 isn't too far fetched.


NOTE: I have no insider connections, I don't know anything more than most others here, nor do I even have any schooling on the subject. I'm basing my hypothesis purely on what AMD has said and currently been doing. All that said, I'll be tickled if any of this happens to play out how I've said. :D
 
  • Like
Reactions: NKD
like this
I cant remember if mockingbird has been wrong, but thank you for providing this info to us.
 
Personally, I just like stopping by WCCF because of it being a good place to find... well to accurately put it, they are the Rumor Mill. It's a one-stop-shop for me to see what tech rumors are floating around, and from there I take it with either a grain of salt, or if my current tech knowledge invokes a raised eyebrow, I'll consider it BS.

While they DO report a lot of rumor as news, and furthermore do exactly as you said by fluffing up said rumor-reports with recycled tables and known-details... In fairness, they do generally end up showing us rumors that pan out. Whether or not they've correctly worded/interpreted the rumor or leak, that's another story... It's as I said though, if you use your own common sense then you aren't going to fall for some of the nonsense. Case in point is what you've pointed out with the 9900K. They've spun it as being an architectural improvement, whereas to anyone that's followed the rumors around the 9900K will also be aware of it having a core count advantage as you did, so can conclude that it also makes up for that gain over the 8700K.

In the end, the data is still useful, even if their regurgitation of it isn't so much.



Or it's just like OP said once or twice, and they were using the ES to hammer out some driver issues/tweaks. :p
AMD has said that their cards interact with the Infinity Fabric even when they're dedicated cards, since the PCIe bus is tightly woven into the IF.

Your ending point could still ring true to a point though, in that there has potentially been a large change to the IF and therefore to properly leverage and work with the new IF it's needing driver tweaks.
I'll be pleasantly surprised if Zen2 has 8 cores per CCX, but personally I'm not expecting any increase in core count per-CCX.
Now, could they add a third CCX per-die, for 12 total cores? That feels way more plausible; however, my gut says that we won't see core count increase until Zen3, with PCIe 4.0, maybe DDR5 (massive maybe), and as such a new socket with higher pin count.

We've had the Tick-Tock going. Zen Ticked, Zen+ Tocked, so Zen2 would Tick and I'd anticipate a Zen2+ Tock, meaning Zen3 would be another Tick. They promised us our sockets will last us for a number of years (2020, or was it 2021 for AM4?). If we get Zen2 again in the May timeframe of 2019, and Zen2+ around the same time 2020, we're on track for that promise and "AM5" for Zen3 in 2021 isn't too far fetched.


NOTE: I have no insider connections, I don't know anything more than most others here, nor do I even have any schooling on the subject. I'm basing my hypothesis purely on what AMD has said and currently been doing. All that said, I'll be tickled if any of this happens to play out how I've said. :D

Yep I go there for the same reasons you mentioned lol! There comment section though is fanboy wars lol that can't use their own brain it seems lol!
 
Yep I go there for the same reasons you mentioned lol! There comment section though is fanboy wars lol that can't use their own brain it seems lol!
Oh yea, I forgot about the god awful comments there... Reason? I have Ghostery (Chrome extension) block the Discus on WCCF, and then an Element Hider remove it as well just in case.
I have never in my 23 years being online, seen such a TOXIC commentary environment. Even YouTube isn't half as bad as WCCF's comments section...

Generally I like to read the comments on websites because often times there are users who will either know something that is correcting an article or is able to just offer some other relevant insight on the topic. While there will always be those fanatics who butt heads, it's usually limited enough that it's easy to ignore and continue reading comments.

THAT is NOT how their comments section is. It's all fanatics and trolls, vomiting their unfounded bias on another commenter who exhibits the opposite opinion, and what they say in return is not done in effort to have a civil debate... no, what they say is at the extreme level they say it only to get under the skin of the other people, and it's everyone that posts.

TL;DR
- WCCFTech comments are wall-to-wall, non-stop flamewars, and it would be in your best interest to disable it on their page if you do enjoy their site. Because the stupidity contained within those comments is truly enough to ruin your day.

[/off-topic]
 
The magic fairy told you?

Edit: and this is not good.

Engineering samples are meant to crash a lot. but ONLY coming close to matching an 8700K? Who freaking cares? It needs to BEAT a 9800K. Intel is going to have its own 8/16 CPU that will be the definitive gaming champion, AMD needs to beat that.
 
I'll take a 8c/16t Zen 2 vs 8c/8t 9800k even if the performance in games is less. I'm not going to pay that much for 8 core no HT Intel. Zen2 8 core will probably be cheaper too and I don't have to get a new mobo.
 
I'll take a 8c/16t Zen 2 vs 8c/8t 9800k even if the performance in games is less. I'm not going to pay that much for 8 core no HT Intel. Zen2 8 core will probably be cheaper too and I don't have to get a new mobo.
WE hope lol
 
If folks have money for top-end Zen 2 or 9th gen i7s, they already have the money for top-end GPUs and probably also already have 4K displays... meaning... *gasp* ...

... they'll be gaming at 4K.

Guess what!

Your top-end Zen 2 or 9th gen i9 won't matter cause you're gonna be GPU limited either way.
 
Last edited:
If folks have money for top-end Zen 2 or 9th gen i7s, they already have the money for top-end GPUs and probably also already have 4K displays... meaning... *gasp* ...

... they'll be gaming at 4K.

Guess what!

Your top-end Zen 2 or 9th gen i9 won't matter cause you're gonna be GPU limited either way.

Talk about assuming everyone thinks like you? LOL! No not everyone will buy a top end GPU with zen 2. No its not the same price and hell no to everyone having 4k displays. People who game go for higher hz like 1440p 144hz or so vs 4k 60hz. So everyone buying zen 2 going for 4k is just flat out wrong. There is more to gaming then just 4k. lol
 
If folks have money for top-end Zen 2 or 9th gen i7s, they already have the money for top-end GPUs and probably also already have 4K displays... meaning... *gasp* ...

... they'll be gaming at 4K.

Guess what!

Your top-end Zen 2 or 9th gen i9 won't matter cause you're gonna be GPU limited either way.

Yeah not true. Most gamers with cash go for 1440p, and there are people who go for 240Hz 1080p displays as well.

But that's all white noise. The truth is that benchmarks in reviews and elitist culture matter more than real world application to a person's buying habits. It's a MUCH better feeling knowing you have "the best" gaming CPU versus the feeling of having "a decent" gaming CPU.
 
Yeah not true. Most gamers with cash go for 1440p, and there are people who go for 240Hz 1080p displays as well.

But that's all white noise. The truth is that benchmarks in reviews and elitist culture matter more than real world application to a person's buying habits. It's a MUCH better feeling knowing you have "the best" gaming CPU versus the feeling of having "a decent" gaming CPU.

Something both sides fail to take into consideration is that you can use your computer for more than just playing games. Sacrilege bringing up that point I know.

The biggest problem I have with the Intel chip is the price. I didn't pay $480 for my 7820x. They are pricing it out of the realm of mainstream.
 
Keep in mind that Zen 2 will hopefully have an IPC advantage (not just be on par) over Intel since it is rumored to be 15% faster than Zen 1/Zen+ in IPC. I ran the calculations and this means that at 4.5 GHz with a 15% IPC advantage, Zen 2 is projected to score roughly 213 points for single-threaded testing in Cinebench R15. For reference, here is the work required for finding the projected Cinebench R15 score using the Ryzen 7 2700X as a baseline-> 177 pts / (4.3 GHz) * (4.5 GHz) * 115%. The source of the 2700X baseline score was AnandTech which I found to be a happy medium in the pantheon of benchmarking configurations. Other review sites like Guru3D had even higher scores such as 180 for single-threaded so this is not a far-reaching estimate.

In comparison, Intel's Lake family (meaning Skylake through Coffee Lake as well as likely Whiskey Lake) score approximately 215 points for single-threaded testing at 5 GHz. That means should Zen 2's rumored 15% IPC improvement come through and given margin of error compared to Intel, AMD is already sitting on an effective tie in single-threaded performance even with a 500 MHz slower stock boost speed. Finally, this is only an early engineering sample meaning we could very well see boost clocks easily 100 MHz higher or even surpassing this.

So actually backpedaling a bit, if Zen 2 manages anything better than 4.5 GHz along with the rumored IPC advantage, AMD could in fact hold the single-threaded performance crown. Mark my words: historically, AMD has already underpromised and overdelivered with the first generation of Zen where they had promised a 40% IPC improvement which ended up being in excess of 50%. I expect them to carry on this tradition of dazzling audiences with Ryzen. I foresee a jaw-dropping demonstration of all the raw untapped potential there is of low hanging fruit, just ripe for the picking, in this their second major architectural release in the Zen family.

---

Also, to add, GamersNexus did a pretty good video on the different impacts of video editing across various high-end CPUs. The summary is that there really isn't a reason to choose anything other than the 8700k for their use case which to me looks fairly typical of what a consumer would produce, though I'd argue that the conclusion also applies to the 2700(X).

And that is why I specify gaming. If the 8700k (and many higher-end Intel consumer CPUs) are faster for gaming due to higher per-core performance, there's little reason to compromise gaming potential for inconsequentially faster video renders or compile times unless those workloads have a meaningful effect elsewhere like income or perhaps family time, for example.

Hopefully AMD steps up the per-core performance with Zen2 and makes this a bit more interesting!

Video from GN:

Full disclosure: They are using Quick Sync Video (QSV) acceleration (introduced in one of Premiere's April 2018 updates) whose image quality is not acceptable for some users like myself though admittedly QSV is not the total pixel haze of death it once was. For video transcoding for home watching, I would still lean very heavily towards a more universally pleasing software encoding solution like x264 or x265. But for streaming where bandwidth starved compression is guaranteed anyway to result in artifacts somewhere down the broadcast chain, depending on your tastes and visual acuity, QSV may be an acceptable compromise.

 
Last edited:
Keep in mind that Zen 2 will hopefully have an IPC advantage (not just be on par) over Intel since it is rumored to be 15% faster than Zen 1/Zen+ in IPC. I ran the calculations and this means that at 4.5 GHz with a 15% IPC advantage, Zen 2 is projected to score roughly 213 points for single-threaded testing in Cinebench R15. For reference, here is the work required for finding the projected Cinebench R15 score using the Ryzen 7 2700X as a baseline-> 177 pts / (4.3 GHz) * (4.5 GHz) * 115%.

Color me skeptical on a 15% IPC increase. Maybe a 15% *overall* performance increase is possible. Meaning combination of clockspeed and IPC. AMD shot for a 10% performance uplift with Zen+, which they... more or less delivered on. A combination of minor 3% IPC bump, minor ~5% max boost clock boost, and precision boost rework (good for a nice practical performance boost) got them over the finish line, if barely. I expect Zen2 will deliver more than this, because of a much better process. So a 15% bump is not out of the question. More, even, is possible if there is a core count increase to go along with it.

One possibility which crossed my mind is that Zen2 may dispense with the CCX. I wondered the other day if 2x 4 core CCXs was a stopgap measure from AMD. Perhaps when designing Zen, they realized that while it was impressive - and ran right over bulldozer, it wasn't quite going to reach parity with Intel in clockspeed or IPC. So AMD used IF, and expanded a 4 core design to be an 8 core design so that it would beat out the 7700k in at least some tasks, and create a market for Zen. This might have happened early in the design process. Anyway... the IF would still be useful for gluing together Epyc - and so they just added it on die as well as a way to glue together multiple dies.

Then Zen 2 moves to an 8 core "CCX" that's the whole die. Full ring bus internally. But, of course, you can still glue together multiple dies for Epyc and TR, for rapid, cheap core count scaling. A 4 core version could continue to be produced for APU design, with the interconnect still used.

If true Zen2's IPC improvements may be tied partly to the reduction in latency from eliminating cross-CCX traffic and, perhaps, a 2x 256 bit AVX setup (instead of 2x 128). Combine with what we assume will be much better clock scaling... and you have something that should compete well against a 9900k, while still preserving the ability to glue multiple dies for TR and Epyc at relatively low cost.

Of course, this is all supposition. I have no idea. We could still see a 6 core CCX or something - and this ES (presuming the rumor is true) could have just had some cores disabled because partly broken ES.
 
Back
Top