AMD launches Zen+ 12nm Ryzen and X470 motherboards

To get that CB score on an R7, you would need an amazing overclock. On an 8700k, you would need exotic cooling. This does it stock.
 
Still taking caution on it, the source is from the crowd that brought you the 4C/8T i3 8100 running at 4.4ghz that happened to actually be a 4790K. I don't think the speculation is far off but some stuff looks fishy.
 
KazeoHin said:
Output IPC can change with something like reorganised caches, different uncore clocks, things like that. BUT the change in IPC will be very minor.
Click to expand...

So an architectural change, then, or do you mean a programming change that would still leave identical hardware with identical performance per clock?
 
Nightfire said:
Fake as hell. Look at the embargo date. Always wait for at least 2 sources.
Click to expand...
Looks like they shopped some old slides... Are there any similar looking slides from that date?
Maybe it's a typo on the slide and they meant March 2018.
 
It's a fake, which is why I always say be careful at this time, fake slides do their rounds, AMD will never be lazy so as to make that error on a slide.
 
Hopefully there's not a huge reason to upgrade to X470 from X370. If I'm already running 3200Mhz memory, I can live with a less efficient chipset (~2W difference).
 
If that's correct, I'll probably be picking up a 2700x board/chip/ram; a 2400g to put in my current setup, and sell my 1700.

This slide would indicate we're going to get around 4ghz all core speed using Precision Boost Overdrive (ie x470) on the top parts. If that pans out, I really see the benefit in grabbing the new chipset as I think you'll have a tough time going more than 100mhz higher than that all core with traditional OC unless you have exotic cooling. In most stuff that uses 3-4 cores, it looks like we can expect 4.1-4.2ghz, which would be a pretty big shot in the arm in a lot of games that right now will peak at 3.7 on an 1800x with good cooling. We could be looking at 15-20% improvement in that sweet spot for gaming.

AMD-Ryzen-2000-tecnologias-3.jpg
 
sover said:
IPC is instructions per cycle, it doesn't change without an architectural change. Increasing clock increases performance, but not IPC.
Click to expand...

Total IPC of a chip is a combination of the IPC of the core, the IPC of the last level cache, and the IPC of memory.

The IPC of the core cannot change without a change in the microarchitecture, but the total IPC of the chip can change if the LLC and/or memory run faster.

Precisely, people overclocks memory in RyZen chips to increase the IPC. A 3.7GHz Zen chip with 3200MHz RAM is faster than 3.7GHz Zen chip with 2400MHz because the IPC of the chip has changed, although the cores are the same and the IPC of the cores is the same.
 
Nightfire said:
To get that CB score on an R7, you would need an amazing overclock. On an 8700k, you would need exotic cooling. This does it stock.
Click to expand...

R7 1700 does 164 at 4GHz. It would do 178 showed in this leak if it was clocked at 4.3GHz. This leaked score is 15% behind the score of a 8700k stock.
 
jenosmaverick said:
No 2800x. Hmmmm am really considering that r7 2700 for my Dan case paired with the upcoming C7 copper.

https://videocardz.com/75194/amd-ryzen-2000-series-exposed-pricing-performance-leaked
Click to expand...

So confirmation that 400-series mobos will be needed to extract all the performance from the new chips. The new chips will be capped in older 300-series mobos.

2933MHz confirmed. Just as I said it has the same memory clocks than Raven Ridge.

Very interesting the higher clocked model has now an official TDP of 105W and requires a new cooler. So those 350MHz extra aren't coming only from the updated 14nm+ process (aka '12nm'), but from pushing thermals 10W above. This could explain why there is no 2800X.

Average gaming increase only 5% at 1080p.
 
Last edited:
kirbyrj said:
Hopefully there's not a huge reason to upgrade to X470 from X370. If I'm already running 3200Mhz memory, I can live with a less efficient chipset (~2W difference).
Click to expand...

I would say IF you can get 3600 or even 4000mhz speeds it might be worth it. But that might be wishful thinking. Only have 1 more month to wait!
 
juanrga said:
So confirmation that 400-series mobos will be needed to extract all the performance from the new chips. The new chips will be capped in older 300-series mobos.

2933MHz confirmed. Just as I said it has the same memory clocks than Raven Ridge.

Very interesting the higher clocked model has now an official TDP of 105W and requires a new cooler. So those 400MHz extra aren't coming only from the updated 14nm+ process (aka '12nm'), buy from pushing thermals 10W extra. This could explain why there is no 2800X.

Average gaming increase only 5% at 1080p.
Click to expand...

did you ignore all the "its fake" post about that link?
 
I appreciate the graphs, but I don't know exactly what that's comparing. Stock to stock? Is MCE enabled on 8700k? Why does it say Ryzen 5 Launch 4/11/17 in the corner?
 
So, looking at these two images, what I see is: Ryzen 7 2700X (compared to Ryzen 7 1800X) made huge gains with games that are lightly threaded, but made (almost) no gains with games that are heavily threaded.

That's probably because base clock only went from 3.6 GHz to 3.7 GHz while the boost clock went all the way from 3.7 GHz to 4.35 GHz.

What's important is that the games that Ryzen 7 2700X made the most improvements in are the games that Ryzen 7 1800X fare the most poorly.


Ryzen-7-2700x-vs-Ryzen-7-1800X.png

Ryzen-7-2700x-vs-Core-i7-8700K.png
 
These slides look pretty legit. Real AMD slides have had mistakes/typos before. Nothing about them screams fake, no crazy performance claims, no crazy "boost up to 4.6GHZ!!!~" claims either. People have been wondering what the justification for the 400 series chipsets would be and the slides make sense too, slightly better boost performance (I doubt there will be a significant difference in scenarios where you're overclocking). But to me it looks more like AMD was throwing a bone to the mobo manufacturers while appeasing the users who already bought motherboards by not locking them out. I would say that this refresh looks pretty decent despite what the usual suspects are saying.. look at how many product releases/refreshes Intel had between Sandybridge and Coffee Lake that had next to no performance difference, at least we're seeing some decent clockspeed gains here to bridge the gap - will be interesting to see how far they get on 7nm.
 
Last edited:
Glad to see AMD is progressing. Intel needs a little heat under their feet. Personally, I am waiting for Zen2. Sceptre fix in silicon is not optional imo.
 
SimianRob said:
These slides look pretty legit. Real AMD slides have had mistakes/typos before. Nothing about them screams fake, no crazy performance claims, no crazy "boost up to 4.6GHZ!!!~" claims either. People have been wondering what the justification for the 400 series chipsets would be and the slides make sense too, slightly better boost performance (I doubt there will be a significant difference in scenarios where you're overclocking). But to me it looks more like AMD was throwing a bone to the mobo manufacturers while appeasing the users who already bought motherboards by not locking them out. I would say that this refresh looks pretty decent despite what the usual suspects are saying.. look at how many product releases/refreshes Intel had between Sandybridge and Coffee Lake that had next to no performance difference, at least we're seeing some decent clockspeed gains here to bridge the gap - will be interesting to see how far they get on 7nm.
Click to expand...

The "leak" is probably a trial balloon so that AMD can get feedback on the prices.
 
Brackle said:
I would say IF you can get 3600 or even 4000mhz speeds it might be worth it. But that might be wishful thinking. Only have 1 more month to wait!
Click to expand...

Memory speed is reliant on the imc not motherboard.. I don't think we'll see any major changes on that front til zen 2. Maybe we'll see a better chance at hitting 3200mhz on hynix chips but I won't hold my breath.. none the less if the clocks are accurate I'll probably replace my 1600 with a 2600x anyways.
 
SimianRob said:
These slides look pretty legit. Real AMD slides have had mistakes/typos before. Nothing about them screams fake, no crazy performance claims, no crazy "boost up to 4.6GHZ!!!~" claims either. People have been wondering what the justification for the 400 series chipsets would be and the slides make sense too, slightly better boost performance (I doubt there will be a significant difference in scenarios where you're overclocking).
Click to expand...

In another forum I have given three examples of AMD official slides about Zen with glaring typos, inconsistent typefaces/designs, and incorrect datetimes. I recall perfectly how some people then pretended the slides were fake, but they were official and latter available at AMD website.

Those leaked slides contain information looks totally legit from technical viewpoint. Clocks agree with prediction of "200--400MHz extra". IMC agrees with prediction of "2933MHz". Top R7 model price agrees with prediction of "above $350". Boost overdrive only available on 400-series mobos agrees with predictions that full performance on the new chips would require "new mobos", with performance crippled in older 300-series mobos.

The gaming slide implies minimal IPC bump, in agreement with predictions made around the technical changes on the new chips.

Moreover, the info in those leaked 2000-series slides agrees well with other info has been leaked. E.g those leaks about the qualification sample ZD2600BBM68AF_38/34_Y or the slides with 2700X performance.

So the information in the slides looks correct for both technical and economical reasons.
 
Last edited:
sirmonkey1985 said:
Memory speed is reliant on the imc not motherboard
Click to expand...

Not true. The memory slots are connected to the IMC through the mobo. If the mobo cannot support certain speeds or signals are unstable, then it doesn't matter if the IMC can handle those speeds.

Moreover, we have word from AMD's James Prior that "The 400 Series motherboards are designed to improve memory stability and overclocking".
 
juanrga said:
So confirmation that 400-series mobos will be needed to extract all the performance from the new chips. The new chips will be capped in older 300-series mobos.

2933MHz confirmed. Just as I said it has the same memory clocks than Raven Ridge.

Very interesting the higher clocked model has now an official TDP of 105W and requires a new cooler. So those 350MHz extra aren't coming only from the updated 14nm+ process (aka '12nm'), but from pushing thermals 10W above. This could explain why there is no 2800X.

Average gaming increase only 5% at 1080p.
Click to expand...

Man you get so much wrong. One can configure p-states on a 300 series board to do the same thing the 400 will just do it automatically, also if someone is overclocking they are overclocking for all cores most times so then it becomes a moot point. It's a small bump in the TDP and it comes with a cooler now that the old X chips did not, so you now get a cooler for it instead of a empty box. They could make a 2800X but they learned there is no point since most people will pick the slightly slower chip and just overclock it, a smart move on their part. Single threaded games will get the biggest boost, not really a shocker with the increased frequency and ones that use all the cores see very little difference, go figure. Your spin is always obvious.
 
Gideon said:
Man you get so much wrong. One can configure p-states on a 300 series board to do the same thing the 400 will just do it automatically, also if someone is overclocking they are overclocking for all cores most times so then it becomes a moot point. It's a small bump in the TDP and it comes with a cooler now that the old X chips did not, so you now get a cooler for it instead of a empty box. They could make a 2800X but they learned there is no point since most people will pick the slightly slower chip and just overclock it, a smart move on their part. Single threaded games will get the biggest boost, not really a shocker with the increased frequency and ones that use all the cores see very little difference, go figure. Your spin is always obvious.
Click to expand...

One of the slides says that 400-series motherboards are "required" for overdrive boost.

You missed my point about the higher TDP. My comment was aimed to remark how that 350MHz bump has required 10% higher TDP, because 12LP alone couldn't provide the clock bump.

And why the games that use all cores see little difference? Because there is no 10--15% higher IPC, neither 10% higher base clocks as others believed. The IPC gain is practically zero and the base clock is less than 3% higher.

Funny that you accuse me of "spin", when Pinnacle Ridge is very very close to what I said it would be. I got ridiculous accusations of trolling and shilling when I said that Pinnacle Ridge would have 200--400MHz higher. :)
 
Last edited:
juanrga said:
One of the slides says that 400-series motherboards are "required" for overdrive boost.

You missed my point about the higher TDP. My comment was aimed to remark how that 350MHz bump has required 10% higher TDP, because 12LP alone couldn't provide the clock bump.

And why the games that use all cores see little difference? Because there is no 10--15% higher IPC, neither 10% higher base clocks as others believed. The IPC gain is practically zero and the base clock is less than 3% higher.

Funny that you accuse me of "spin", when Pinnacle Ridge is very very close to what I said it would be. I got ridiculous accusations of trolling and shilling when I said that Pinnacle Ridge would have 200--400MHz higher. :)
Click to expand...

Ryzen 1st gen has great multi-thread performance, but single thread-performance wasn't great (compare to Intel's offerings).

Now, Ryzen 2nd gen is going to come out with better single-thread performance (thanks to higher boost clock).

So, what are you complaining about, exactly?
 
I'm really curious as to how binned the processors will be across the lineup. Will users be able to overclock the Ryzen 5 2600 or 2600X to similar 4.3/4.4 levels that the 2700X offers? It seems like they're really trying to segment the processors by cooler as well, which is sort of an artificial barrier with thermal headroom. I think it's going to be a tough battle for the midrange. The 2600X at $250 might be a tough sell if the 2600 can hit the same clockspeeds and offer a better value compared to the Intel counterparts. But the Ryzen 5 1600 was a huge seller for AMD, I can see why they might want to push more people towards the "X" model and get a little more cash out of the midrange/value chips. B450 + Ryzen 5 2600 could be a good value proposition since Intel have struggled to get h370/b360 boards out.
 
Last edited:
SimianRob said:
I'm really curious as to how binned the processors will be across the lineup. Will users be able to overclock the Ryzen 5 2600 or 2600X to similar 4.3/4.4 levels that the 2700X offers? It seems like they're really trying to segment the processors by cooler as well, which is sort of an artificial barrier with thermal headroom. I think it's going to be a tough battle for the midrange. The 2600X at $250 might be a tough sell if the 2600 can hit the same clockspeeds and offer a better value compared to the Intel counterparts. But the Ryzen 5 1600 was a huge seller for AMD, I can see why they might want to push more people towards the "X" model and get a little more cash out of the midrange/value chips.
Click to expand...

probably something like this:

Top 15% --> Ryzen 7 2700X

Next 20% --> Ryzen 5 2600X

Next 25% --> Ryzen 7 2700

Bottom 40% --> Ryzen 5 2600

I would guess that binning would make a bigger difference this time around.

Last time, the worst binned would go to low end processors (Ryzen 3 1200, Ryzen 3 1300X, Ryzen 5 1400, Ryzen 5 1500X), but this time, the low end (Ryzen 3 2200G and Ryzen 5 2400G) would be served with a different die.
 
Last edited:
I just think the prices are a bit screwy on these slides. Why would they keep the 1500x and 1300x at higher prices then the RR counter parts? Also, I have my doubts that the 2600 will release at $200. Especially with Intel dragging their feet on B-360.
 
SimianRob said:
I'm really curious as to how binned the processors will be across the lineup. Will users be able to overclock the Ryzen 5 2600 or 2600X to similar 4.3/4.4 levels that the 2700X offers? It seems like they're really trying to segment the processors by cooler as well, which is sort of an artificial barrier with thermal headroom. I think it's going to be a tough battle for the midrange. The 2600X at $250 might be a tough sell if the 2600 can hit the same clockspeeds and offer a better value compared to the Intel counterparts. But the Ryzen 5 1600 was a huge seller for AMD, I can see why they might want to push more people towards the "X" model and get a little more cash out of the midrange/value chips. B450 + Ryzen 5 2600 could be a good value proposition since Intel have struggled to get h370/b360 boards out.
Click to expand...

Figure the best chips can go to 4.5 maybe 4.6 since previous gen could hit 4.2. But to be honest it's a bit early yet to know if all the info we have seen is actually true. I will be waiting for the 7nm version to come out, they seem to expect it to hit 5GHz. Should be interesting as Intel will no longer have the process advantage they used to rely on.
 
Nightfire said:
I just think the prices are a bit screwy on these slides. Why would they keep the 1500x and 1300x at higher prices then the RR counter parts? Also, I have my doubts that the 2600 will release at $200. Especially with Intel dragging their feet on B-360.
Click to expand...

They're not making them anymore, the RR chips are what they're being phased out for. They also may launch lower end non igp parts 2-3 months later like they did with the 1xxx chips (after they've built up enough with defective cores).
 
I tried a contact in AMD and the origin of the slides was denied to be true, on the information they couldn't say anything due to the probing nature of the data would violate NDA terms. I could not get any confirmation on whether there is a tier higher SKU and similarly details on clocks, TDP and price are non disclosures, The slides have overlaps and things that seem like they are photo-shopped some really badly and the 2017 repeated is something in house editing will not repeatedly overlook and that many mistakes will result in disciplinary. This is a professional corporate and constant mistakes are just not really seen, I work in the Legal field and my documents are checked and fact checked and checked again before submission, I have not had a typo on a document in 7 years.
 
You must log in or register to reply here.
Back
Top