Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Aug 6, 2020.
Ditto ... Hopefully my Asus CH VII x470 will run no problem I paid enough for it.
However it does mean TSMC's process has matured so much so that there is confidence in doing an ES chip like this one. A little hope that every core will be capable of hitting boost clocks unlike it is today where only 1-2 cores can actually hit boost frequencies hence why almost nobody does all core overclocks with AMD.
This new ZEN3 line will crash Intel!... it has a higher probability that this new CPUs will hit 5.0Ghz and with the improve of IPC will be a BEAST!
The blue side will must put this 10XXX series at maximum price of USD 400 if they want no lost the entire market...
I don't care about the boost speed. I just want it to beat 10900k with fast memory in ANY games in 1080p/1440p
Then I will upgrade my 3900x
I m on the same boat.
Regardless of how much it will perform worse/better than intel or in games blah blah.
2700X to the serie 4000 equivalent seems a nice upgrade if i do not have to change anything else.
Gotta be honest, but your original post doesnt make any sense, and the reply to it is valid.
Since its only a rumor it will be 5000 series, there's no reason to compound rumors into other rumors. Otherwise we will end up with a 20th down the line rumors with paragraphs stating all said rumors for days before you get to the new rumor..
Wow so soon eh? Man I love my 3900X & you can complement it with some fast RAM depending on the motherboard I had seen some B550s staring upwards of DDR4-5000 no?
So I've had an 1800X, 2700X, and now a 3900X.
I am sticking with the 3900X for now, I would want at least a 25% IPC increase and a good chunk of multicore IPC before I upgrade. Will probably skip 4000 series and wait for the 5000 series hopefully with DDR5 memory. 12/24 core cpu for me is more than enough and I don't see this chip being saturated by any game in the coming years and it blasts through video rendering using Handbrake more than fast enough for me.
I hope the 4000 series is good, I'm holding onto my money for a GPU and RAM upgrade at the end of the year.
8370E new - $119.99 to 2700X new - $129.99 will be interesting next year, what i decided to drop in my Asus X470 Prime
In that picture Gigabyte "says" only x570, B550 and A520, while Asus has better uefi skills.. And it was later confirmed that all 4xx and 5xx will support it, I think after that uefi MB limitation rumors.
Cool, ya it would more than likely be my last upgrade. I'm not going new board DDR5 ect. in next gen. So hopefully ...
Zen2 already runs read/write bandwidth full speed in the 3900X and up. It's only in the lower-core count processors where write is 1/2 the speed of read--interestingly enough, however, it doesn't affect performance *at all*...
Might want to consider the fact that if the 10900k didn't boost it wouldn't beat the AMD CPUs *now* at any speed...! And in productivity applications the 10900k won't win regardless...! I really like my 3900X, too... Still want to see Zen 3, though. I think it will comfortably trounce anything Intel has--especially in gaming-specific comparisons. I don't really expect Intel to offer anything competitive with Zen2/3 until early 2022, earliest--based on what they've said about reaching 7nm. But then--where will AMD be at that time? Intel has its work cut out, imo.
Does boost mean anything? Doesn't it usually last for a few seconds and fall back to normal clocks.
Give me 24/7 stable high clock and I shall call dicks out for AMD.
Sometime down the road if AMD is continuing to make Zen2 CPUs in 2 years or so I may upgrade my 3700x to a 3900x Or at that time Zen3 CPUs are more reasonably priced then I may upgrade to 12c/24t Zen3 equivalent. So I have even more cores and threads and that would match my x58 based dual CPU Workstation/Future File server.
Can't compare XT versions to various Zen architectures-- XT indicates minor changes to existing architectures made possible by improvements in the 7nm production nodes. XT is still 100% Zen 2. Something like Zen 3 is much different--just like Zen2 was much different from Zen1/+. AMD has stated that Zen1/+ (same architecture, different nodes), Zen 2 and Zen 3 and Zen 4 are all new architectures as differentiated from the earlier architectures in the Zen architecture family nomenclature--(now, that was a mouthful...).
As such, I am inclined to believe that a Zen 3 Engineering Sample boosting to 4.8GHz would be slower than a theoretical Zen 3 maximum boost--although maybe no more than 100-200Mhz slower than max shipping boost.. But this is assuming the rumor has legs and is substantial and is based on actual ES Zen3 silicon. Much depends, as well, on improvements in Z3 IPC, more than mere improvements in boost maximums, of course.
Still, it is fun to speculate!
Yep, lots of news now showing AMD stating ZEN 3 for B450 X470 up.
Good news ASUS began aggressively releasing UEFI updates again for almost their entire product line far back as many X370s. You have many current products with the latest AGESA updates. I’d say you’re gravy’
Well yes they are short bursts. However certain OEMs have long released UEFI features to complement Intel/AMD turbo, boost. Let’s take ASUS and their “ASUS Performance Enhancement” for example. With Ryzen the feature increases the duration of the boost clocks when enabled. It’s actually not that bad. Works well with my 3900X & B550-E.
Why would my original post not make any sense to you?
I agree that it's all based on rumour and not likely will happen.
But even Videocardz mentioned it in their article of Zen3 ES.
Until that moment Zen3 doesn't get officially released all you can hear or read about it only rumour, so which one of the leaks you believe in that's only up to you.
1) Each individual die does write a half of read, but because there's two of them, the total measurement appears as the same
2) dual-channel DDR4 up to 3800 is already at the limit with the full IF read bandwidth of a single die, when having two it doesn't increase bandwith
3) but it DOES when using a 3900X/3950X with even faster ram (DDR4-5000) for example, with 1:2 link ratio. Write however will stay slower
4) They chose this half width outgoing (write) links to save on die space/power, and while it doesn't affect performance when a thread that stays in it's core...
5) It does affect it when threads move between dies ! Which happens quite a lot in software that is unaware of this strange architecture !
Not having this deficiency in the new Zen3 could reduce die-to-die latency by quite a lot.