What is more interesting to me is the platform longevity.
Chances are it'll be 2020 before AMD change sockets (DDR5). I've also read they expect a couple of CPU revisions on this socket.
Personally, I'm not convinced X370 will cut it as a prosumer platform a few years down the line, but X399 will. My usage scenarios have changed drastically in the past few months, I no longer need the heavy lifting I had envisaged - but I'm still likely to get the 1900X in a few months along with the best mobo (review dependent) and get onto the platform. Then, if, in a year or two I do need to run large simulations, update the CPU to Zen2 or Zen3, if I need the NVMe or PCIe drives, get them too then, RAID0 and go for it.
If they don't care about gaming, then why do we have an 8700K listed hereWell they obviously don't care otherwise they would not have changed the cache organisation and spent all that effort on improving AVX performance - maybe you are getting confused between marketing and the actual engineering side,when their own rebadged Core i7 7700K,destroys the Core i7 7900K and almost any other CPU in gaming. Even the Broadwell E CPUs are faster in games.
Look at the Hardware.fr review:
http://www.hardware.fr/articles/967-18/indices-performance.html
They use games which have been tough on AMD in the past,and look at where Skylake-X is located relative to the other Intel CPUs.
Every change with Skylake X is to improve non-gaming performance.
There shouldn't be a chipset change until pcie4/5 or DDR5 goes mainstream. They have said as much.This.
The longevity is what interests me. Gaming may be a large use case for me, but it isn't the only one and I fully intend to move from 3 1200p monitors to 3 4K monitors in the near future. If TR2 or TR3 slot into one of these boards in the future, it is a huge deal.
It's like saying because it's says it has a gaming mode that the Vega FE is therefore targeted at gaming and sucks as a gaming card because it comes in third.
There are parts of these systems that are good at gaming and for the most part including with the clock speeds the 1950x, these perform well in games. Just because they aren't the fastest doesn't meant they can't be used for gaming.
We aren't Ricky Bobby. Not being first doesn't mean they are last.
There shouldn't be a chipset change until pcie4/5 or DDR5 goes mainstream. They have said as much.
Ending up with a chip that's less than stellar at gaming on release does not mean that the designers didn't care about gaming. Especially when the chip in question was probably not designed for the HEDT market in the first place, but for the server market.
I'm not seeing any excuses here. Who in their right mind would buy either a 1950X or a 7900K for 1080 gaming first? It would be more than a bit stupid, regardless of Team Red or Team Blue.So many AMD fanboys that made snarky comments about x299's gaming performance are now busy making and excuses for TR.
Never one to make any comments about the x299's gaming. I think the point was that there was some major regression in IPC as far as gaming was concerned. Which was a valid comment. Though it doesn't really matter as Intel was never going to make a HEDT specific die. Me personally don't see what the downside is in gaming on the 1950x considering you can isolate it to get 1800x-ish performance.So many AMD fanboys that made snarky comments about x299's gaming performance are now busy making and excuses for TR.
Yeah, Threadripper hasn't had a chance to have any regressions yet, so hang on, Intel fans.Never one to make any comments about the x299's gaming. I think the point was that there was some major regression in IPC as far as gaming was concerned. Which was a valid comment. Though it doesn't really matter as Intel was never going to make a HEDT specific die. Me personally don't see what the downside is in gaming on the 1950x considering you can isolate it to get 1800x-ish performance.
My recollection about at least one of the claims for more cores on the desktop was that we could all have some task going on in the background and still game decently, since we had lots of cores to handle lots of threads. In fact, I recall one poster in particular who liked to post about how 8 core Piledriver was supposedly so good with something significant running as a background task.Yeah that kind of my point. We say that gaming shouldn't be a priority on the Chip and won't be for those smart enough to understand the product that they are choosing. But as far as marketing goes. It's not like these are terrible gaming chips. But when someone looks at the UMA and NUMA gaming and talks about how bad it is affecting games (barely on the Numa one). It's like the best of both worlds. There an almost insignificant drop off related probably more to the slightly reduced drop off in clocks. But in the end it nor any of the i9's (I'll leave the door open for the 7900) is going to be a world beating gaming CPU. A great gaming platform sure. Good gaming CPU's sure. But the fact is that if gaming was realistically the only reason to get a CPU, none of the CPU's in the benchmarks including the 7740x are great CPU's. The only one to get is the 7700k. After that it becomes about tradeoffs and compromises (which is what Zen was designed around). How important is 16 cores, 32 threads, 4 channel memory, with 8 slots, and endless PCIe. I don't know why recognizing that it's not the best in gaming ever means it can't be advertised as a gaming system. Subtract HEDT from the equation. I could go on amazon and look at a dozen "gaming systems" if most of them have different configurations does that mean only one system can be called a "gaming system"?
I guess I am having a disconnect on why stating that the primary use not being gaming disconnects them from offering them as gaming systems. That's like saying that a 7700 can't be advertised as a decent Virtualization system just because we understand that 90% of the use case of the 7700 isn't virtualization.
So many AMD fanboys that made snarky comments about x299's gaming performance are now busy making and excuses for TR.
Well, take your machine for instance, its a quad core CPU on the X299 platform.
Its not really a good solution for prosumer applications given the rest of the market or the best at gaming. What is it supposed to be?
[You may point out that the 1900X on X399 will not be fast, I'll reply with I/O driven prosumer scenarios.]
Well, what do you guys think? I don't know how I feel about it. If your software can actually use the cores, then its a beast and way better than a 7900X. If you can't use the cores, then clearly its not the best option. I'm just still in shock that I have to wrap my head around a 16 core CPU being on the market. Good god.
So many AMD fanboys that made snarky comments about x299's gaming performance are now busy making and excuses for TR.
Trolling is not allowed
Not to mention baiting and calling others fanboys.
Markfw
Anandtech Moderator
Stop considering official $ prices and think about local prices too. Where I live 7900X is $150 more expensive and 1950X is $300 more expensive. Price wise 7900X competitor is 1920X in local prices which isn't really such a good deal at all.
Prices fluctuate depending on time and location. The only good even metric we have to compare the different CPUs is msrp. Comparing it to local prices doesn't make sense. Whose local prices are we going to use? Those at your location? Mine? Someone in Botswana? China? Uruguay?
Stop considering official $ prices and think about local prices too. Where I live 7900X is $150 more expensive and 1950X is $300 more expensive. Price wise 7900X competitor is 1920X in local prices which isn't really such a good deal at all.
If Intel dropped the 7900x to 699 it would be a clear winner and would take some potential TR buyers. At 999 it makes no sense to me.
If you don't consider local prices at all price based arguments are invalid as 1950X simply doesnt cost $999 at all.