Whats the point of a 16c TR part if a 16c Ryzen exists?
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***
- Thread starter Jamin280672
- Start date
Whats the point of a 16c TR part if a 16c Ryzen exists?
AMD: “No Chiplet APU Variant on Matisse, CPU TDP Range same as Ryzen-2000”
My hopes for an 8 core 16, thread APU have sunk for this year............
My hopes for an 8 core 16, thread APU have sunk for this year............
Caporegime
- Joined
- 13 May 2003
- Posts
- 34,515
- Location
- Warwickshire
Presumably they'll have different power consumptions, cache, clock speeds to cater for the server market vs. desktop / gaming.
Quad channel and ECC memory I guess. Same as 4/6/8 core Xeons.
I expect high end of 8C will be £350. Plenty of profit but still undercuts Intel by a decent marain. I think 16C AM4 will be more like 9900K pricing. I could be wrong of course
Caporegime
- Joined
- 8 Jan 2004
- Posts
- 32,588
- Location
- Rutland
Cheap entry to TR4 platform if all you need is the PCI-E lanes for cards or multi NVME setups. Same happened with 1900X.
Whilst the first two points are quite likely, based on leaks and deductions from the demo, they are not confirmed.
Cinebench is more sensitive to memory timings than it is to speed. So the slow speed isn't that much of an issue for this test. They probably picked that because it is a baseline standard. However, it could also be that the memory controller needs more refinement to work at faster speeds / tighter timings.
Last edited by a moderator:
Exactly the same reasons why there is an 8c ryzen and 8c threadripper now...
Memory at 2666MHz was picked as historically Ryzen was bad on slow memory and it's the stock Mem controller speed for the Intel part. Makes sense to beat the Intel part on it's home turf than to bias it in Ryzen's favour and garner negative press as a result. The 9900K release was a very good example of why that's a terrible thing to do. Espcially as the media is now the Youtubers rather than the press.
Soldato
- Joined
- 19 Feb 2011
- Posts
- 5,849
+1 exactly this... AMD presented the rigs to actually leave no excuse of gimping, stock speeds, supported ram speeds, decent cooler etc. They intentionally put out a comparison with no room to question the build. Very well done and should be applauded.
Obviously they only showed enough to peak interest and spark dialect and debate on what we was actually seeing. Which in its own way will create hype.
As it stands now, post relief I've seen quite a few posts of people asking "what CPU shall I buy?" To be answered with "wait for Zen2, we just seen their midrange destroy Intel's best CPU"
Now while we know the above is not 100% true, as the 9900k was not destroyed, it was matched by presumably a mid range part, we can deduce this was midrange from the system power figures we were shown and using common sense, and by looking at AMDs own info, such as their slide showing same TDP etc, well we know that AMDs 2700x is not as low TDP in the demo... So therefore I think it's safe to say it wasn't a chip higher than 8c 16t.
On top of this we were told it was an 8c chip, and show the actual CPU, and AMD baited us here, they intentionally showed us a CPU they knew straight away everyone would know had room to double up. They didn't make any mention of more cores etc during the keynote, for once their presentation was actually on point. The old less is more is relevant here. They showed just enough to make people draw their own conclusions, all while keeping as much as possible unsaid and playing on an even field.
Anyone on these forums, who can come into this thread, claiming to be a hardware enthuisiast and nay say, ridicule, or down play that presentation is either trolling, a moron, or actually has no idea what they are talking about, or more likely all of the above.
6 months is going to drag by, which will suck, but the truth is, in 6 months I firmly believe we will have a new CPU king, whether it be midrange or high end, AMD will be taking the single thread and multithread crown, you can save this post and ridicule me with it in 6months if this is not true.
Associate
- Joined
- 7 Apr 2017
- Posts
- 1,762
Exactly, it's almost as if you have to be on one side of the fence or the other, PC enthusiasts... More like extremists!
Like you I think these look GREAT on paper, but paper burns! I want to know the single core strength, all core boost frequencies and even then, until I see gaming benchmarks I'm not going to be fully sold. Having said that they look incredibly promising and see no reason why they won't be the standard in performance... But hype aside I want the cold hard facts. I don't see any issue with some scepticism, it's not like your mentioning your own Intel cpu at every given opportunity.
I have a ryzen, I want them to do well, but I'm also realistic about the chip I currently ownand it's limitations. Seems as though some folk here have a hard on for a couple of posters in this thread. One is clearly trolling but not the person I've quoted.
but I truly believe this is AMD's cure2duo 
Just to clarify, I'm not on any fence...I'll buy a Zen 3XXX chip if the performance is there. I have said countless times I want Zen to do well and destroy 9900k.
Fact is if you want to spout speculations and create false data to reinforce an ideology then prepare to be challenged on it. This is not trolling....This is discourse
I mean how can anyone be called a troll when:
A: They will be buying Ryzen if the performance is there and ditching 9900k
B: They wanted Ryzen to destroy intels Flagship 9900k
C: They wanted Radeon 7 to beat 2080
D: They have more AMD chips than Intel chips
The word Troll is to easily used in this forum when in fact people just have a different take on what is real.
Then you have that old chestnut: "well you're just trying to justify your recent purchase as new tech is coming and your 9900k will be obsolete."
Why would I be justifying my purchase when:
A:I could buy a 9900k for every PC in the house if I wanted and not even notice
B: Saving 200 quid over a 2700x buyng a 9900k is literally peanuts to me.
C:The cooler bro...you missed off the cost cooler for your 9900k....I can reuse this for Ryzen if I want and I want silence.So again its not an issue.
D:You hate AMD bro....Erm no...Three servers in my house are all AMD....I built my nephew a 2700x system. I have used AMD chips for over 25 years...
I mean being this clear is not enough for some...Which to me is very odd behaviour.
Last edited:
Man of Honour
- Joined
- 17 Feb 2003
- Posts
- 29,640
- Location
- Chelmsford
Ok No more accusing others of trolling.. If you think people are genuinely trolling in an attempt to take a topic off subject, then report , let us deal with it. Then ignore it.
Anyone calling others trolls is trolling and will be dealt with.
Anyone calling others trolls is trolling and will be dealt with.
Last edited:
Associate
- Joined
- 27 Apr 2007
- Posts
- 966
They will at some point but they will probably focus on areas of higher volume and profit before that niche.
Their job is to make money, not score points.
It also depends on their supply of chips and demand for Epyc etc.
They will need to bin heavily for a 64C TR I imagine even at 7nm.
Everyone knew that the upcoming 3000 APU are Zen+ not Zen2.
Ryzen 3/5/7 support ECC
PCI-e Lanes, Quad channel ram, upgrade to 64 core.
Same way Intel has octa-core on the HEDT platform, or had quad core also. (X79 etc).
Caporegime
- Joined
- 8 Jan 2004
- Posts
- 32,588
- Location
- Rutland
Why would they need to bin heavily?
The multichip design means the binning process is far easier. Yes they will be binning for lowest power draw for the high core count chips, but it's nothing like what you need to do for a monolithic chip.
Scoring points is how you make money. Density is everything in the server market, an untouchable 64c chip will make them money, that's where the margins are highest.
Soldato
- Joined
- 26 Oct 2013
- Posts
- 4,132
- Location
- Scotland
Yeah AMD have struck absolute gold with the chiplet design. Using it from top to bottom will save mountains of money in the long run and is definitely a fantastic way forward. Binning will be so easy and even non-functioning chips will take up a tiny amount of the wafer compared to a monolithic design.
I really don't know what you're talking about. Obviously everything we are discussing is in the context of Cinebench since that's what was demoed and is pretty much the only piece of evidence we have (aside from the die shot). Obviously the 16c/32t Ryzen 3 part will not be twice as fast compared to the i9-9900K at everything, especially since not that many applications are capable of using that many threads in the first place. I know you like to think that gaming is all that matters but I never mentioned that once.
As for "whole lot of ifs", I had one "if", which was "if the 8c/16t part matches the Intel flagship"...which it does in the one demo we've seen. That doesn't mean it'll match it at everything but that's not what we're talking about here. Cinebench is a good indicator of performance in highly threaded apps, we already know this. We also know that the current performance delta in Cinebench between Ryzen 2 and Coffee Lake is more than the average performance delta in 1080p high refresh rate gaming, so the fact that they've caught up in Cinebench is at least a good indicator that it will perform better in gaming too. I've already been through this once in this thread.
I'm sure you'll make another incoherent reply not addressing anything I've said but hopefully this makes things clearer for others.
HEDT has far more PCIe lanes and quad channel memory. Ryzen 1 had an overlap between desktop and HEDT at 8 cores, so it wouldn't be strange for Ryzen 3 to have an overlap between the two at 16 cores.
Matisse is the codename for the CPUs though, so I wouldn't expect any variants with iGPUs. We'll probably have to wait until Ryzen 4 for a Zen 2, chiplet based APU.
Ryzen 1 CPU: Summit Ridge (14nm Zen)
AM4 1st gen APU: Bristol Ridge (28nm Excavator)
Ryzen 2 CPU: Pinnacle Ridge (12nm Zen+)
Ryzen 2 APU: Raven Ridge (14nm Zen+*)
Ryzen 3 CPU: Matisse (7nm Zen 2)
Ryzen 3 APU: Picasso (12nm Zen+)
Ryzen 4 CPU: Vermeer (7nm+ Zen 2?)
Ryzen 4 APU: Renoir (7nm Zen 2?)
*Kinda debatable since it seems to have most improvements from Zen+ but is still on 14nm.
Last edited: