If this is true for most common workloads and if this is coupled with significantly higher clocks, we could be talking game over for Intel’s i7 as their 14nm++++ process hits the limit and as their 10nm process struggles.
AMD may have on its hands the chip that will take the crown in both single and multi-core performance (whether the actual core count increases is not known). There is also speculation of whether AMD knows they will have the industry leading product and will start raising price points as a result.
Overall, 2019 could be the best year in ages for the CPU, just when people started thinking the GPU was the future of computing
And it might be more overclockable, which would mean more IPC. Im just wondering if it will run in my motherboard, ASUS Prime X370-PRO. If i do the latest BIOS firmware update.
I will admit that since AMD messed up their numbering scheme and have them way to close to Intel’s, when I read that i though it was an Intel board
Overall AMD made a statement that Zen 2 CPU’s will drop in replacements.
only thing is that if all Zen2 will support PCIE 4th gen (vs current 3rd gen). ON the Epyc line that will be the case, and there if you want PCE 4th gen you’ll need new boards, but it will drop into existing TR4/SR4 socket jsut run at PCIE 3rd gen speeds.
Well GPU still has its place, and no CPU will rivial that (at least not for the next 10 years, and NVIDIA will ensure that is the case by releaseing very Compute capable cards)
Now CPU side, AMD has it covered. Now here is to hoping that they will do the same to the GPU market, especially if the tackle the Chiplet approach to GPU’s that would be VERY VERY good.
i’ve not done the calculations yet, but price performance wise, surely there’s not much of a performance difference between upgrading my Ryzen 1700X upto a 16Core Zen2 @4.0GHz + vs adding another GTX1070(at their current high prices)…
Doing a full upgrade fully agree. However the issue sis that not all aspects of blender are GPU only. Unfortunately not even part is even multithreaded.
Definitely adding more GPU’s is more price concious as long as you mainly do a lot of renderings.
My first PC was K6 200 and after that I had an Athlon 700, so have a soft spot for AMD and am glad to see them doing well. I have a 4790K right now, which is quite OK for everything I need to do, but all the latest rumours about Ryzen 2 and Navi are getting me all worked up to build a new system (which is fun to do in itself…) that’s all AMD. However, some of the rumours sound too good to be true, and Ryzen 2 isn’t due until April or thereabouts, so going to have to wait a while to see what kind of performance improvements will actually be available. Having said that, the Youtube channel AdoredTV has a lot of detailed info on the architecture of AMDs chips, company strategy and discussion of rumours and their reliability, so I’ll be following that channel for my info. AMD may also be making some announcements in January at CES, so the next few months are going to be an interesting time to watch the CPU market.
By the way, I thought one of the reasons for rewriting the depsgraph in Blender 2.80 was to make it more multithreaded, so perhaps AMD chips will start seeing better performance in Blender itself, not just in rendering performance?
there’s always the memory matter. I understand that 8gb is more than enough for the big majority of people. If on the other hand the kind of work you do is memory hungry, things change quickly. To have more than 12gb of vram you need to pay A LOT.
In general speed is super important obviously, but when I run out of vram is trouble. So I am starting to look at these CPUs with a lot of interest. Comparable speed and no more ram problems sounds very good.
The rumor mill is now in full swing with speculation about the Zen2 processors.
There is a debate going. On one side you have AdoredTV gushing over just how much of a leap the new CPU’s are, but other sites disagree.
The rumor in a nutshell is that among AMD’s new lineup are Ryzen 9 models with 16 cores and a boost up to 5.1 Ghz. While the clockspeed can indeed be increased a bit due to the new fab process, the fact that AMD has not released any official information means that all of this should be taken with a grain of salt.
I wouldn’t get too hyped over the 7nm process and whatnot, afaik it is in reality the same as Intel’s 10nm, if not 14nm process. There are different ways to measure stuff and AMD chose to measure it differently to get an edge in numbers game for PR purposes.
Global Foundries and Samsung also use this kind of numbering to describe their processes as well. What is true is that the number itself is just one variable among others such as transistor density.
I’m skeptical about this, Zen 1 had half the AVX2 throughput than the Intel chips, if I recall correctly. It never really showed up in a lot of tests, because few applications (or even benchmarks) use AVX, but that would be an easy workload to show such gains on.
Still, the 7nm process should put AMD in the ballpark of Intels non-delivering 10nm process. My guess would be an 8-core to match the 9700Ks performance at an attractive price.
I don’t know who might think that, but from a developer’s perspective I can almost guarantee you that isn’t going to happen. By my estimation, 99% of programmers can’t program GPUs and the other 1% would prefer not to, if they had the choice, that is if CPUs had the same parallel throughput. With Larrabee (and its successors) dead in the water, that’s probably not in the cards, but one can still hope.
One of the impressive bits is that AMD managed to match the performance of the Intel i9 9900K with an 8-core engineering sample. There is a good chance that the performance will be even higher come launch day as we’re still at least several months away from its release.
In addition, the demo is being done with the Ryzen 5 model containing a design that leaves room for a second 8-core chiplet, so the idea of a Ryzen 7 or Ryzen 9 chip with 12/16 cores is not that farfetched.
In addition, the demo is being done with the Ryzen 5 model
This is rumored, but more than likely true, and very strategic. What they are showing is that the IPC advantage that intel has, is now gone, at a lower power consumption rate.
