Ryzen 3000 Eng. Samples @ 4.5 GHz+? / +15% IPC? & AMD X570 chipset gets PCIE 4.0 and USB 3.1 gen2

Joined
Jan 18, 2008
Messages
1,019
Some rumor and some info?

"AMD Ryzen samples have been sent to motherboard manufacturers for product validation, these engineering samples already run 4.5 GHz...

...Typically, and I do say this with a bit of hesitation and care, engineering samples run slower opposed to the final SKUs. Apparently, the manufacturers are also preparing for CPUs with 12 and 16 cores (well yeah .. doh!).

While these engineering samples reportedly are running 4.5 GHz, IPC improvements also are mentioned to be 15% clock for clock."

Sounding pretty good no? Now give us 5 GHz and 16 cores!


Maybe a new build is finally coming. Goodbye my sweet Haswell?...


index.png


index.png
 
Engineering samples this close to launch are probably identical to launch CPUs. I wouldn't expect any changes between now and launch unless there is a drastic issue.

I hope the specs are true. I need to upgrade soon to get another m.2 slot and have been waiting for Ryzen 3 benches before making any decisions.
 
Last edited:
Something something ipc something ...

Sweet jesus look at that rock on her finger.

... anyway,

Yeah, i would agree with Kardonxt and not expect any significant difference in speeds at this point. In fact if something changes NOW it would be most likely downward, as more statistical analysis comes in and they build in safety margins for yields.
Still - promising and fun. I'd really like a very highly clocked 8 core with a little bump in IPC as well.

I was about to say I was looking forward to the [H] review, and now I'm sad again.
 
Well you can quote me from most of the other Zen rumor threads saying I'd be happy as hell to see an 8 core with a modest IPC gain that I can run ~4.6-4.7ghz 24/7 (overclocked).. So this rumor is almost as if somebody has read my mind. As long as the price isn't much more than a launch 2700x, I'm finally upgrading from my aging 4790k and into 9900k performance territory.

The only reason I never bought a 2700x is because the single core performance is practically the same as my current setup from 2014. This would give it the edge.
 
This is certainly possible.

I'd go a step further and it's almost certain that it will not be a graph for optimized power. They will almost certainly tilt towards performance, but the numbers remain to be seen.

If they burn some extra watts, this forum will point that out. If they miss the mark on performance completely, everyone will point that out.

And yes, we'll all hope for a golden chip for our own purchases, but it would be wise so assume AMD will burn close to the limit for even stock settings.
 
I'd go a step further and it's almost certain that it will not be a graph for optimized power. They will almost certainly tilt towards performance, but the numbers remain to be seen.

If they burn some extra watts, this forum will point that out. If they miss the mark on performance completely, everyone will point that out.

And yes, we'll all hope for a golden chip for our own purchases, but it would be wise so assume AMD will burn close to the limit for even stock settings.

Well, Intel pretty much redefined what TDP means when it released a 180W processor with a "95W" TDP.

You have to wonder if AMD is going to follow suit.
 
Well, Intel pretty much redefined what TDP means when it released a 180W processor with a "95W" TDP.

You have to wonder if AMD is going to follow suit.

I assume they will, and I'm honestly not sure what to think of it. For "most people", the usage is a browser and maybe a word processor. Both have very low utilization numbers overall, so the claimed duty cycles are probably not completely inaccurate.

But is that fair to claim that? I guess I would personally prefer if the TDP represented something closer to a max envelope rather than coming up with increasingly-creative time based metrics of bursty workloads. But at the same time, most people ARE using simple apps which sit idle and have bursts (browsers, most notably).

Glad I'm not in marketing. "He is a product of your profession. Not mine."
 
Sounds like the 12 and 16 core versions won't be available at the same time as the 6 and 8, or am I interpereting wording wrong?
 
I'd go a step further and it's almost certain that it will not be a graph for optimized power. They will almost certainly tilt towards performance, but the numbers remain to be seen.

If they burn some extra watts, this forum will point that out. If they miss the mark on performance completely, everyone will point that out.

And yes, we'll all hope for a golden chip for our own purchases, but it would be wise so assume AMD will burn close to the limit for even stock settings.

I'm totally ok with that.
They already gave us 220w cpu's, do I see 250?
 
Keep in mind that AMD must consider current motherboard VRM designs and limitations (X370, B450, X470, etc.) when deciding TDP and power draw in general for these new CPUs.
The current high-end 2700X is 105W TDP. Thus the new CPUs will also max at 105W TDP in order to not push current MB VRMs beyond their design specs. Whatever speed that allows with whatever number of cores there are is what it will be.

Now, this is pure speculation on my part, but it is possible for AMD to have two series of Zen2 CPUs: Series 1 is the "backward compatible series" with 105W max TDP for current-gen and previous gen MBs. Series 2 (possibly released later in the year) is the "next gen higher power series Zen2" designed for newer generation MBs with higher power VRM designs (including new X570-based and maybe some old and brand new 2nd gen high-end enthusiast X470 boards too if the VRMs are sufficient to handle > 105W TDP).
This would allow AMD to keep their promise of allowing current AM4 systems to be upgraded to Zen2 while also providing enthusiast CPUs with higher performance in certain cases where the MB can handle the higher current load and TDP greater than 105W.
But I'm just guessing and maybe this is all too complicated. If it is too easy for Johnny to buy a 3850X and pop it into a A320 MB and pop the VRM, that could be problematic.
 
Keep in mind that AMD must consider current motherboard VRM designs and limitations (X370, B450, X470, etc.) when deciding TDP and power draw in general for these new CPUs.
The current high-end 2700X is 105W TDP. Thus the new CPUs will also max at 105W TDP in order to not push current MB VRMs beyond their design specs. Whatever speed that allows with whatever number of cores there are is what it will be.

Now, this is pure speculation on my part, but it is possible for AMD to have two series of Zen2 CPUs: Series 1 is the "backward compatible series" with 105W max TDP for current-gen and previous gen MBs. Series 2 (possibly released later in the year) is the "next gen higher power series Zen2" designed for newer generation MBs with higher power VRM designs (including new X570-based and maybe some old and brand new 2nd gen high-end enthusiast X470 boards too if the VRMs are sufficient to handle > 105W TDP).
This would allow AMD to keep their promise of allowing current AM4 systems to be upgraded to Zen2 while also providing enthusiast CPUs with higher performance in certain cases where the MB can handle the higher current load and TDP greater than 105W.
But I'm just guessing and maybe this is all too complicated. If it is too easy for Johnny to buy a 3850X and pop it into a A320 MB and pop the VRM, that could be problematic.

That's easy.

Allow the new 4-core Ryzen 3, 6-core Ryzen 5, and 8-core Ryzen 7 on older motherboards while restricting 12-core Ryzen 9 (and possibly 16-core Ryzen 9) to new motherboards.
 
Keep in mind that AMD must consider current motherboard VRM designs and limitations (X370, B450, X470, etc.) when deciding TDP and power draw in general for these new CPUs.
The current high-end 2700X is 105W TDP. Thus the new CPUs will also max at 105W TDP in order to not push current MB VRMs beyond their design specs. Whatever speed that allows with whatever number of cores there are is what it will be.

Now, this is pure speculation on my part, but it is possible for AMD to have two series of Zen2 CPUs: Series 1 is the "backward compatible series" with 105W max TDP for current-gen and previous gen MBs. Series 2 (possibly released later in the year) is the "next gen higher power series Zen2" designed for newer generation MBs with higher power VRM designs (including new X570-based and maybe some old and brand new 2nd gen high-end enthusiast X470 boards too if the VRMs are sufficient to handle > 105W TDP).
This would allow AMD to keep their promise of allowing current AM4 systems to be upgraded to Zen2 while also providing enthusiast CPUs with higher performance in certain cases where the MB can handle the higher current load and TDP greater than 105W.
But I'm just guessing and maybe this is all too complicated. If it is too easy for Johnny to buy a 3850X and pop it into a A320 MB and pop the VRM, that could be problematic.


Anyone that ran a first gen at 4.1Ghz or then 2nd gen @ 4.35Ghz AC turbo is well past that 105W TDP. I've done both on an Asus x370/470 Prime/P. Pro. This is without sctivr cooling on the VRMs since I have a massive custom loop and I am running 130% current with a very high Vcore.


There is no issue. AmD is going to lock out the A320 crowd and possibly the low low end B350 boards very easily. Via BIOS support.


AMD can simply say they cannot force the Mobo ODMs to release bios support.

The next easy way for this is to also only guarantee a base clock speed on older but good 3xx/4xx boards and allow the Mobo makers to make their own version of PBO/2/3 like Asus did for their x370 boards despite it being a "x470” only feature.

I'm certainly ready. The extra boost in .01% and 1% Frame Times coupled with a low input lag display and FreeSync is gaming on a new level.
 
Even the shittiest of shitty, diarrhea, defecation, fecal matter a320/b350 board will be able to run perfectly good 7nm CPU's ...so the $50 motherboard folks (and believe me I understand the desire to run a $50 board) should be happy with the choices they get, I mean you paid $50 and get access to THREE different process shrinks.... that should make you an AMD fan for a very, very long time to come.
 
MY guess is JUST like with the Phenom 2 to FX and beyond: board manufacturers will update their QVL on boards that they believe will support the new CPUs.
 
Even the shittiest of shitty, diarrhea, defecation, fecal matter a320/b350 board will be able to run perfectly good 7nm CPU's ...so the $50 motherboard folks (and believe me I understand the desire to run a $50 board) should be happy with the choices they get, I mean you paid $50 and get access to THREE different process shrinks.... that should make you an AMD fan for a very, very long time to come.
I keep hearing that the a320's will not be supported and I have no problem with that.
 
Any bettors out there? The most exciting part in before a launch :p

I'm predicting:

Ryzen 9 3900X: 16C/32T 4.8GHz (single core) 4.4~4.5GHz (all core) boost - $499 (MSRP) (9900K competitor, ~5% behind in avg in single thread/poor thread synthetic benchmarks, multithreaded waaaaaay ahead, better bang-for-buck advantage)
Ryzen 9 3800X: 16C/32T 4.6GHz (single core) 4.2~4.3GHz (all core) boost - $449
Ryzen 7 3700X: 12C/24T 4.7GHz (single core) 4.3~4.4GHz (all core) boost - $349
Ryzen 5 3600X: 8C/16T 4.5GHz (single core) 4.1~4.2GHz (all core) boost - $269
Ryzen 3 3300X: 6C/12T 4.2GHz (single core) 3.8~4.0GHz (all core) boost - $179

I expect it to clock manually max 200-300MHz with a quite significant ~1.25% increase in voltage.
 
Lets be honest, do any of us here use A320 boards???
forgot they even existed, lol.. but given they were bare minimum boards doubt the vrm's are even close to being able to support zen 2. realistically the only ones that were probably using a320 in mass was oem's which has little effect on them in the grand scheme of things.
 
15% IPC sounds nice and about the best anyone could have realistically expected. I just hope that is 15% clock for clock.
 
Back
Top