Read more.Quote:
A high boost clock for what is thought to be an AMD Zen 3 Engineering Sample chip.
Printable View
Read more.Quote:
A high boost clock for what is thought to be an AMD Zen 3 Engineering Sample chip.
The jump to 5000 skus would make sense considering how messy the 4000 tier would become. I'm still holding out with my 1700X, looking forward to this set of chips, but I don't think I want to spend $750 on a CPU. I'd like to shift to 12 cores, but 16 is overkill for most DIYers.
looking VERY promising.
Does anyone know if these purport to retain the socket compatibility with X570?
AMD desktop 4000 series 'should' be on am4 and supported on x570.Quote:
Originally Posted by Tattysnuc
I sure hope so. I def bought one earlier this year with the notion X570 would have compatibility with 4000 series. Given the previous backlash I think they learned not to go back on their word.Quote:
Originally Posted by Tattysnuc
4.8GhZ boost would be only 100 MhZ higher than 3950X boost.
Still, if this is indeed just an engineering sample, then its possible clock speeds would improve later on.
But, I wouldn't hold my breadth on this end.
The Enhanced 7nm node Zen 3 will be made on is still 7nm (albeit refined).
We don't know how much of a clock increase will there be (if any).
Zen 3 will primarily focus on this being a new uArch which has a more unified design (with big reduction in latencies) and of course IPC improvements (to say the least).
I don't think we should expect large clock improvements on this node.
Given its already early August, and Zen 3 is slated for release late this year, its possible that these might be it in regards to frequencies.
I think for the average consumer it would be more confusing. I can just see the conversations I will end up having:Quote:
Originally Posted by globalwarning
I want the latest laptop, but I can only find 4000 series. Where are the 5000 series?
Oh, they are just on the desktop.
So when do I get 5000 series in a laptop?
You have to wait a year, the same time as the 6000 series comes out.
So why are the numbers different?
Because the internal architecture (and at this point I can see the person glaze over with no interest in CPU internals)
Now this is news, forget the phone garbage
Clock speed alone doesn't tell the whole story so that may not matter if the IPC uplift is significant.Quote:
Originally Posted by deksman2
Apparently just the unified massive cache is a massive perf netter.Quote:
Originally Posted by watercooled
I'm almost regretting not hanging on for this with my upgrade last year - the CCX behaviour together with non-passive 570 chipset were just niggly enough on Zen 2 to push me to Intel (on offer). But if they pair Zen 3 with a passive high end chipset then that's certainly what I'd be wishing for and recommending to others who are upgrading later.Quote:
Originally Posted by Tabbykatze
I get where you're coming from but if you take the AMD GPUs into consideration we're on course for a huge car crash once they release their mid-term XT variants.Quote:
Originally Posted by globalwarning
Igor provides an update concerning an Engineering Sample with a 4.9GHz boost
https://www.igorslab.de/en/cracks-on...ns-from-intel/
It is real. AMD says it will launch the first Zen 3 CPUs later this year, as well as its highly anticipated RDNA 2 GPUs.
RDNA2 wont have HMB2 though apparently which should keep cost down if its GDDR6 no?
I'm always holding my breadth. One of these days I'll find a diet which isn't totally ruined by habitual drinking.Quote:
Originally Posted by deksman2
To play Devil's Advocate (although I think the last Pope got rid of the office so he could make a tonne of saints), I'd say 7nm doesn't need advancing on if Intel isn't capable of getting anything more than 14nm to work properly with decent yields. I'd say they might as well refine the current node if they can provide meaningful performance increases with architectural advances rather than process advances.
Clock speed is only a useful metric if all other things are equal. IPC obviously makes a difference as well as how many cores can beat along at a given clock speed.
Here's hoping the latency issues are eradicated. They are certainly the low hanging fruit and once they've got their Zen architecture to where it should be, then they can start really pressing on with the node. Myself I like to make one big change to things at once, then I can be sure of anticipating the feedback and I know where and feedback has arisen.
You may well be right with the frequency comment but I'd suggest:
- This may be a "leak" purely to get discussion and excitement going. I'd never considered moving from my current CPU (it's more than sufficient) but this started that little voice in the back of my head going "maybe if it's THAT good..."
- Engineering samples will be messed with all over the place. They'll change the configured clock speeds, etc and run tests multiple times on the one chip. It wouldn't surprise me if they change the number of cores and many other things. I dunno how these things work but with motorbikes they make a "mule" with loads of variables that can be configured for testing.
Honestly, I'm just being argumentative. Ignore me.