Pieter3dnow
Supreme [H]ardness
- Joined
- Jul 29, 2009
- Messages
- 6,784
This time around Jim has some new bits of information and shows us that some of the old bits seem to be confirmed by other sources...
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
AdoredTV pre Computex roundup Zen 2 leaked information
- Status
Will be interesting, to be sure. If these hold true, the kids will be getting the 1700 and 2700 cpus and my office and VR machines will be getting upgrades. Will finally be able to rest on their machine and I'll have the home lab I can use to mimic my work environment almost exactly.
- Joined
- May 18, 1997
- Messages
- 54,458
Address the message, not the messager. If you try to turn this into an "anti-AdoredTV" thread, like we have seen over and over again, derailing the topic of conversation, you will lose access to your account.
BrotherMichigan
Limp Gawd
- Joined
- Apr 8, 2016
- Messages
- 376
There are some pretty confident claims made near the end of the video, so it will be interesting to see how those hold up over the next week or so. All I can say is that I've got a custom water cooling loop just waiting for a 5GHz part.
Pieter3dnow
Supreme [H]ardness
- Joined
- Jul 29, 2009
- Messages
- 6,784
There plenty of people that if they have anything contradicting from sources they are happy to share it with us. The frequency is not earth shattering base is just above 4 ghz and the turbo can be up to 5ghz we don't know how many cores but the number sounds enticing. We don't know if the turbo will push it outside tdp.
Cooling is a whole different segment if you take a look what Kyle did here at [H] when cooling Threadripper. And the spacing between the chiplets on X399 was somewhat larger between modules.
Pieter3dnow
Supreme [H]ardness
- Joined
- Jul 29, 2009
- Messages
- 6,784
Well if you goto https://www.youtube.com/channel/UCrwObTfqv8u1KO7Fgk-FXHQ/videos
Buildzoid explains it why there is a fan needed but he could not describe under which scenario...
Gideon
2[H]4U
- Joined
- Apr 13, 2006
- Messages
- 3,215
Rumors and a few leaked specs on the server side have been looking good, so I think there is a good chance these new Ryzen chips will be a hot item.
Thatguybil
Limp Gawd
- Joined
- Jan 21, 2017
- Messages
- 164
Don’t like the whole secret information stick, but base at 4-4.2 range on an 8 or 12 core chip would be huge. All core turbo of 5 ghz would be nice.
I seriously wonder if gen4 will improve nvme SSD performance.
I seriously wonder if gen4 will improve nvme SSD performance.
Brackle
Old Timer
- Joined
- Jun 19, 2003
- Messages
- 8,437
Not sure if it will increase performance, as much as giving the nvme drives more bandwidth from my understanding?
Will be interesting to see some benchmarks
IdiotInCharge
NVIDIA SHILL
- Joined
- Jun 13, 2003
- Messages
- 14,675
It should only improve performance with respect to using multiple devices at capacity simultaneously. Which is... mostly pointless, given that there's almost no use case for that much bandwidth on a desktop. The point would most likely be that any of multiple devices could be used at capacity as opposed to being limited as on previous CPUs and chipsets.
And I'm still almost entirely skeptical. Not that I want to be, but the likelihood of AMD making detrimental architectural changes like splitting the memory controller from all CPU cores while also improving per-core performance is extremely low. They might get the clockspeeds up and that will certainly help, but for most desktop / workstation usecases that aren't pure compute, which is mostly video work in the consumer space, they're still going to be nipping at Skylake's heals. And Skylake is old.
I agree FrgMstr! I get tired of all the wccftech sucks crap too. News is just news. As individuals it's just information. You must bring to light for yourself if it is credible or not. But the biggest mistake Leonidas did was kicking the messengers of the god king into the death pit.
If the 16 core 3850x (or whatever it'll be called) is announced and released, I'm all in. This is exactly the CPU I need to do what I need to do, without breaking the bank or otherwise compromise, with a Threadripper.
If it boosts to 4.4 Ghz, 5 Ghz or more or less doesn't really matter much. People put way, waay too much emphasis on this arbitrary number. Makes no sense.
16 cores, smallish footprint and case volume, and I'll be happy for years.
If there is a decent midrange GPU released with great thermals, I'm in butt first on that too.
And, NVME with 5000 MB/s transfer speed. Yes, please.
If it boosts to 4.4 Ghz, 5 Ghz or more or less doesn't really matter much. People put way, waay too much emphasis on this arbitrary number. Makes no sense.
16 cores, smallish footprint and case volume, and I'll be happy for years.
If there is a decent midrange GPU released with great thermals, I'm in butt first on that too.
And, NVME with 5000 MB/s transfer speed. Yes, please.
Boil
[H]ard|Gawd
- Joined
- Sep 19, 2015
- Messages
- 1,439
I guess if one was an Intel fanboy, one would pray these rumors were not true...
5ghz does make sense for gamers because that is the biggest reason why Intel is better than AMD in gaming. Cores help, but clockspeed is still the king.
- Joined
- May 18, 1997
- Messages
- 54,458
Battlefield V NVIDIA Ray Tracing - CPU Testing
We have been doing some deep dives into playing Battlefield V 64-person multiplayer lately and testing what exactly the cost of using NVIDIA Ray Tracing is in terms of framerate performance using new NVIDIA RTX 2070 and RTX 2080 cards. We did get questioned on using a 5GHz overclocked 7700K instead of the suggested CPU that EA recommends.
With the active cooling and extra 8 pin cpu power pins showing up. I wonder where the limit will be for x470 to x570? Even seen some rumors that it may pull between 300-400W for the CPU. I still have faith I can use my CH VII, but if I have to swap than I’ll upgrade next year.
Pieter3dnow
Supreme [H]ardness
- Joined
- Jul 29, 2009
- Messages
- 6,784
We won't know for a good while since release will be a prolonged one over the few months after launch we will get to see the 16 core versions.
Even the X370 version of CH VI hero had enough leads to supposedly run 16 core (4 and 8 pin).
If the results weren't so close I may have resorted to calling you an Intel shill!
(and before anyone gets theirs in a twist, it is a joke. Removing CPU bottlenecks as much as possible is a good idea(tm) when testing GPUs)
Thatguybil
Limp Gawd
- Joined
- Jan 21, 2017
- Messages
- 164
I wonder when when / if we will see the heterogeneous core concept that we see in phones and tablets. Big fast core + low power cores.
Big fast wide cores performance intensive task, slower simpler lower power cores perform lower priority threads / tasks.
Big fast wide cores performance intensive task, slower simpler lower power cores perform lower priority threads / tasks.
Pieter3dnow
Supreme [H]ardness
- Joined
- Jul 29, 2009
- Messages
- 6,784
Those are battery operated devices which "need" these things. If you look at laptops they tend sometimes not use a real desktop cpu to begin with and if they do they tend to have a special (vendor) TDP setting you won't find on the desktop ...
IdiotInCharge
NVIDIA SHILL
- Joined
- Jun 13, 2003
- Messages
- 14,675
The 8550U in my ultrabook is a 15w part, but isn't materially different from the 77w 6700k I was running in my desktop a few years back.
Except in terms of clock behavior. Huge difference there, and that results in power usage and themals appropriate for the form-factor.
What Intel has shown is that within a given envelope, they can produce a single CPU that can be used both for higher end desktops and for ultrabooks; if the number of cores is varied too, they can go in both directions.
I expect that ARM devices are amenable to other configurations due to some combination of the instruction set being simple enough that the 'low power' cores can be effective while using very little die space, thus being inexpensive to implement, and the software being able to 'survive' on said cores when not loaded by the user. Might also be some inflexibility in their 'big' cores that prevents them from functioning at extremely low power as Intel has been able to do with refinements to Skylake.
Here's hoping that AMD has been able to make similar refinements to Zen 2. I'd love to see an AMD APU in my next 2-in-1 ultrabook- I don't need much, certainly no more CPU than this 8550U, but I'd certainly like something with a 120Hz VRR panel with just enough grunt to push 120FPS in simpler games (League of Legends in my case) with the settings more or less buried at 1080p. The 8550U does just about 60FPS for reference.
Thatguybil
Limp Gawd
- Joined
- Jan 21, 2017
- Messages
- 164
They need it because of battery limitations and thermal budget.
While desktops don’t have power limitations we are running into thermal limitations. Air cooling is hitting a wall with watts/mm^2 and I don’t think AIO water cooling will ever truly becomes mainstream.
Really depends, but those tin foil heatsinks with 40mm screamers can't possibly be any better at cooling than a nice copper finned heatpipe setup.
Thatguybil
Limp Gawd
- Joined
- Jan 21, 2017
- Messages
- 164
He was pretty far off.
Not in terms of clocks he wasn't the specs are the same just the branding and pricing is different.
So AdoredTV didn't lie nor are his sources fake it's just that these things can change at anytime.
Though I'm wondering what on earth is the 16-Core going to cost 699USD? on an AM4?
I mean it's quite possible it will just be 100 dollars more like 599USD or they will shift the product stack down a pricing tier introduce a few 50 models but I think I'll wait a bit before buying a new CPU like October and see what AMD has in store then.
He also got the performance of Navi correct a RTX 2070 level GPU is there but what I find interesting is the RTX 2080 Ti comparison.
No, the clocks are wrong.
He was on Twitter all week promoting 5GHz boost frequency.
IdiotInCharge
NVIDIA SHILL
- Joined
- Jun 13, 2003
- Messages
- 14,675
The specs are what was possible, so...
...I could have made this up. But I've been following tech for over twenty years, so the patterns are quite obvious.
Hell, I could have been his source.
Then he said he was told no 5ghz boost at launch.
He was just backpedaling
IdiotInCharge
NVIDIA SHILL
- Joined
- Jun 13, 2003
- Messages
- 14,675
Let me throw this out here- the linked information in the OP was wrong. That part of this thread has run its course.
I highly recommend dropping the youtuber chearleading. We don't have much patience for it, especially for ones that peddle rumors.
Pieter3dnow
Supreme [H]ardness
- Joined
- Jul 29, 2009
- Messages
- 6,784
How many times can you get it wrong. Find out when you don't get out of your house nothing can happen to you until your house collapses on you
.Find people with sources that want to post stuff on the internet since everything turns into a flame war if you get a single letter wrong. What is worrying is that people like you don't want to listen to what he says:
TAKE EVERYTHING WITH A GRAIN OF SALT
That is what he starts his announcements on what his sources tell him!
There is difference between this and what you and many others blame AdoredTV for.
I highly recommend dropping blaming youtube people with speculation tech videos for wrong information.
Just for you IdiotInCharge here is the definition of speculation:
https://www.dictionary.com/browse/speculation
OnceSetThisCannotChange
Limp Gawd
- Joined
- Sep 15, 2017
- Messages
- 238
Surprisingly enough, he undersold IPC increase and oversold the clocks, in the end it's balanced.
I expected 10% increase in total, as we have been getting 5% increases (if that) for a decade now, ie going from 4.3 with my 2700x to something like 4.7 ghz equivalent, being good enough, especially as with B-die RAM, 2700x is quite fast as is.
Based on the keynote, we are getting 15% IPC increase and another 5% boost in top clocks (from 4.35 to 4.6), which is about 20% total, and would equal to about 5.2ghz 2700x from single threaded performance perspective, while MT being even better?!?
I am still sceptical that it will be that good, especially when compared to a well tuned 2700x (perhaps this is the gain comparing stock to stock Zen+ to Zen2), but combined with that leak of 16core Zen2 being 30% faster in cinibench MT at 4.2ghz CPU speed, it could actually be true.
While he missed naming/clocks, the most important part - total performance - could be about right, educated guess or not, I am positively surprised with this launch. This was beyond my expectations, if it turns out to be as good as advertized right now.
I expected 10% increase in total, as we have been getting 5% increases (if that) for a decade now, ie going from 4.3 with my 2700x to something like 4.7 ghz equivalent, being good enough, especially as with B-die RAM, 2700x is quite fast as is.
Based on the keynote, we are getting 15% IPC increase and another 5% boost in top clocks (from 4.35 to 4.6), which is about 20% total, and would equal to about 5.2ghz 2700x from single threaded performance perspective, while MT being even better?!?
I am still sceptical that it will be that good, especially when compared to a well tuned 2700x (perhaps this is the gain comparing stock to stock Zen+ to Zen2), but combined with that leak of 16core Zen2 being 30% faster in cinibench MT at 4.2ghz CPU speed, it could actually be true.
While he missed naming/clocks, the most important part - total performance - could be about right, educated guess or not, I am positively surprised with this launch. This was beyond my expectations, if it turns out to be as good as advertized right now.
The new cpus are pretty consistent with around $20 per thread, at least the x models. That would put the 16 core part at around $650.
Being on AM4 is a disadvantage, but remember, x570 has alot going for it. If memory frequency is around 4000 mhz, bandwidth should not be much of an issue using dual channel. There are only 20 pcie lanes, but these are also pcie4, which is like 40 pcie3, so not too bad.
The biggest downside is the lack of surface area for cooling.
The biggest advantage is that you will now have a lot more uATX and ITX options to pack 24 or 32 relatively affordable threads into a sff powerhouse.
- Status