Now, somebody desperate to justify their pet theory that Vulkan is not the way forward for 3D rendering might point to some cases where Vulkan (or DX12) actually slows the game down. This is because the initial implementation of Vulkan is sometimes just a translation layer for OpenGL, which is single-threaded, so the translation is single-threaded too. On top of that you have the translation overhead. Result: no improvement, or actually a regression. However, that was then, this is now. (...we continue to see much lower CPU utilization when using Vulkan rather than OpenGL.)
Vulkan improves performance in several ways: 1) distributes rendering across multiple cores 2) lifts state out of the inner render loops 3) shortens the path for submitting render data 4) supports fine grained control over caching of render assets. (Not an exhaustive list.) None of this is a secret; all of it is known to anybody who could call themselves a game developer. You are very obviously not one of those. But do us all a favor and please stop shitting on the internet with your ignorance. If you are confused then do a little research.
Your numbers say that AMD share on Steam increased from 11% in March to 15% in July, the most recent numbers. Look, disability with math does not mean you are a complete imbecile. But other signs do point in that direction.
So start with 2x8GB and add 2x8GB later, that's what I did. Now the remaining issue is, if I want to go higher my slots are already fully populated. Waste! Can't stand it. My solution is to fantasy-build a TR, partially populated with 64GB, and hand this rather nice machine down.
you *are* correct that the margin is very much reduced or inverted with the 2 DX12/Vulkan games out there.
Thanks for that, now try to connect the dots. I really don't care if an obsolete game engine runs 10% slower, I care about what happens with the new architecture that everybody is moving to. And you can hardly call any of those old crap engines unplayable on Ryzen, at least not without losing whatever cred you have, which is looking a bit thin at the moment to be honest.
If memory price is too painful just at the moment then you could pick up 2x4GB for $90 to get started and add another 2x8GB later when the price comes down.
You also provide essentially zero evidence that the camera on an S9+ or iPhone X or Pixel 2 XL is inadequate, while other posters link to hundreds of examples of rather impressive photos produced by those cameras
They look so great in a jpeg in a browser, don't they? This subject has been hashed to death, it's not for me to go through all the points again, and I'd miss a bunch. Just one: let's see you shoot raw on a phone at 6-10 frames/second like any passable DSLR can do. Without that, you're going to miss a lot of shots.
"Which was the last Hollywood blockbuser filmed in portrait? Maybe they are stupid?"
I'm not Hollywood. I have gigantic, high res monitors (dirt cheap these days). So I shoot video in the orientation that best suits the subject, simple. I'm also not hidebound by tradition.
When changing orientation I don't "fill" with anything, that's the job of the player if necessary. The world does not consistent solely of movie screens.
AMD is not hurting Intel significantly now, but the writing is on the wall.
AMD has something like a one year window of great looking margins and income, because of Intel's laziness/greed/blunders and because of AMDs great engineering and product sense. Also, the decision to go fabless is looking like pure genius as of today, whereas for a number of years it was widely regarded as a harbinger of doom. It's that fabless thing that has the best chance of carrying AMD's current advantage beyond the current product cycle.
As I see it, Intel has only two realistic paths to getting back out in front: 1) stop competing with TSMC and friends and join them instead 2) engage in massive illegal trustmaking activity again. Obviously, there are issues with the second one.
AMD parts can do a lot of stuff that Intel can't, such an encrypted RAM and offering a huge number of PCIe lanes. And of course, they offer more threads at a better price.
On the other side, ECC in consumer parts really ups the ante.
There is no way that the 8 threaded i7 competes with the 16 threaded 2700X unless it is half the price, which isn't going to happen.
It will continue to compete, because enough people want the fastest, and that is simply the i7, period.
That's a fool's game. There is always a faster one next quarter. But I agree, there is a great supply of fools. And inertia is a thing too, it is sometimes wise. But I still see Intel coming out on the wrong side of the decision tree here.
My 8700K kills a 2700X in every game I've ever seen benchmarked.
You mean crap games, not yet using Vulkan/D12. And "kills" is a wild exaggeration. 2700X is widely recognized as a perfectly good gaming part, built for the future instead of the past, and also the best workstation in class. Cores to spare, aggregate throughput without peer.
I fully understand that aggregate core performance of the 2700X per dollar is superior- but you need to accept that that metric just doesn't fucking matter to a whole lot of people.
I recognize that a lot of people are fucking stupid, I will give you that. Nutrition may not matter to them either.
I see your point. Maybe this part? solves the problem. Not sure about the socket or number of PCI lanes. Anyway, it's hard to complain about the 2400G at $150 as a throwaway placeholder, maybe even a keeper.
Haha, same with me. My 1700 build is a budget workstation, about $1k of parts. Now I intend to drop more than twice that on a 16 core TR2 and I don't bother justifying. The build itself is the point.
Well, I do max out all my cores on a regular basis, so it's not just for fun. But dammit, it is fun. I'm going for 16 cores instead of 32 because it's clocked higher, so on balance, better for my mix of serial and parallel loads. But I can certainly see the value of the 32 core part, it will be the go to part for video editing, for example.
If budget was still the main thing, I would go for the 16 core Ryzen. But the TR has twice the memory channels and TR motherboards have twice the dimm slots, so that is enough justification to go up the food chain. Still trying to figure out what I'm going to do with all those PCI lanes. Obviously, the crytocurrency miners already figured that one out.
Yes, it's a no brainer for a budget build, even if you plan to stick in a GPU at some point. Give your wife or PC a highly respectable sit-down PC that costs maybe $500 to build, or $600 for the deluxe version. For myself it is equally a no brainer that I want 8 Zen cores, and now being thoroughly addicted I will go on to 16 (above that the core clock starts to drop) which is not something I need, it is something I want.
For Intel, ARM is the invisible bite that doesn't show up directly because it is related to the PC market decline, which is entirely explained by people doing without PCs in favor of mobile ARM devices. If it wasn't for ARM, Intel would be sellilng a billion more processors a year than it now does, think about it. Intel badly wanted that mobile market and were utterly defeated. They could easily start peddling ARMs themselves, but not for the margin they got used to and are now dependent on.
Meanwhile. AMD is nice enough to not undercut too much. Not because they don't want to, but because they can't afford it. The last thing Intel should do now is go kill AMD with antitrust thuggery again, that would be extremely unwise, it would just push AMD into the arms of somebody much richer, and much more of a threat. A curious kind of detent we have going on now, with AMD enthusiasts the big winners.
These 8 core chips coming soon from intel better be very very competitive priced too
Let's speculate... I can't see the 8 core i9 selling for $329, which is where AMD has the 2700X right now. Mind you, the i9 does have a (lame but functional) graphics core, while you must install a GPU for the 8-core Ryzen, so that's a slight point on Intel's side. Very slight. It would be great to see AMD will respond with some minimal GPU on their 8 core Zen 2 parts next year, but AMD called it right anyway: everybody who plugs in that 16 thread beast also happily plugs in a GPU. On the lighter side, I do recall my moment of shock horror when my shiny new 1700 bench build came up with no video. Thought it was dead until I noticed the fan speed changing. Then duh.
There is no way that the 8 threaded i7 competes with the 16 threaded 2700X unless it is half the price, which isn't going to happen. Even the 2400G looks better than the new I7, at least it has a credible GPU. Is Intel banking purely on brand loyalty this cycle?
You cherry picked the most recent sequential month to month data. Words do not describe the depth of your intellectual dishonesty.
Here, watch this: 10 EPIC games with VULKAN support (PC 2018). There is a good reason why it is easier to name the game engines that do not support Vulkan than those that do. What reason? Better performance. Those engines that currently do not support Vulkan, do support DX12 or Metal, basically the same thing. So 100% of major engines are already moved to Vulkan or similar. For most of those, their Vulkan render path gives the best performance.
Now, somebody desperate to justify their pet theory that Vulkan is not the way forward for 3D rendering might point to some cases where Vulkan (or DX12) actually slows the game down. This is because the initial implementation of Vulkan is sometimes just a translation layer for OpenGL, which is single-threaded, so the translation is single-threaded too. On top of that you have the translation overhead. Result: no improvement, or actually a regression. However, that was then, this is now. (...we continue to see much lower CPU utilization when using Vulkan rather than OpenGL.)
Vulkan improves performance in several ways: 1) distributes rendering across multiple cores 2) lifts state out of the inner render loops 3) shortens the path for submitting render data 4) supports fine grained control over caching of render assets. (Not an exhaustive list.) None of this is a secret; all of it is known to anybody who could call themselves a game developer. You are very obviously not one of those. But do us all a favor and please stop shitting on the internet with your ignorance. If you are confused then do a little research.
Your numbers say that AMD share on Steam increased from 11% in March to 15% in July, the most recent numbers. Look, disability with math does not mean you are a complete imbecile. But other signs do point in that direction.
Do you understand what is meant by the phrase "dogs watching television?" No? Down boy.
AMD has declined in every segment except Linux users
You need to lay off that crack pipe.
you seem to think that engines that don't use Vulkan are obsolete
Yah, I do, for sound technical reasons that everybody seems to understand except you.
Bye Felicia
Reverse Relativity, sounds catchy.
So start with 2x8GB and add 2x8GB later, that's what I did. Now the remaining issue is, if I want to go higher my slots are already fully populated. Waste! Can't stand it. My solution is to fantasy-build a TR, partially populated with 64GB, and hand this rather nice machine down.
2700X is widely recognized as a perfectly good gaming part,
Steam CPU Statistics
Look at all that recognition!
What, AMD Steam share up from 11% in March to 15% in July? Undone by your own link.
By the way, you have a crap posting style, you come across as a pimply teen taking revenge on the internet for being bullied.
you *are* correct that the margin is very much reduced or inverted with the 2 DX12/Vulkan games out there.
Thanks for that, now try to connect the dots. I really don't care if an obsolete game engine runs 10% slower, I care about what happens with the new architecture that everybody is moving to. And you can hardly call any of those old crap engines unplayable on Ryzen, at least not without losing whatever cred you have, which is looking a bit thin at the moment to be honest.
If memory price is too painful just at the moment then you could pick up 2x4GB for $90 to get started and add another 2x8GB later when the price comes down.
You also provide essentially zero evidence that the camera on an S9+ or iPhone X or Pixel 2 XL is inadequate, while other posters link to hundreds of examples of rather impressive photos produced by those cameras
They look so great in a jpeg in a browser, don't they? This subject has been hashed to death, it's not for me to go through all the points again, and I'd miss a bunch. Just one: let's see you shoot raw on a phone at 6-10 frames/second like any passable DSLR can do. Without that, you're going to miss a lot of shots.
"Which was the last Hollywood blockbuser filmed in portrait? Maybe they are stupid?"
I'm not Hollywood. I have gigantic, high res monitors (dirt cheap these days). So I shoot video in the orientation that best suits the subject, simple. I'm also not hidebound by tradition.
When changing orientation I don't "fill" with anything, that's the job of the player if necessary. The world does not consistent solely of movie screens.
AMD is not hurting Intel significantly now, but the writing is on the wall.
AMD has something like a one year window of great looking margins and income, because of Intel's laziness/greed/blunders and because of AMDs great engineering and product sense. Also, the decision to go fabless is looking like pure genius as of today, whereas for a number of years it was widely regarded as a harbinger of doom. It's that fabless thing that has the best chance of carrying AMD's current advantage beyond the current product cycle.
As I see it, Intel has only two realistic paths to getting back out in front: 1) stop competing with TSMC and friends and join them instead 2) engage in massive illegal trustmaking activity again. Obviously, there are issues with the second one.
AMD parts can do a lot of stuff that Intel can't, such an encrypted RAM and offering a huge number of PCIe lanes. And of course, they offer more threads at a better price.
On the other side, ECC in consumer parts really ups the ante.
There is no way that the 8 threaded i7 competes with the 16 threaded 2700X unless it is half the price, which isn't going to happen.
It will continue to compete, because enough people want the fastest, and that is simply the i7, period.
That's a fool's game. There is always a faster one next quarter. But I agree, there is a great supply of fools. And inertia is a thing too, it is sometimes wise. But I still see Intel coming out on the wrong side of the decision tree here.
My 8700K kills a 2700X in every game I've ever seen benchmarked.
You mean crap games, not yet using Vulkan/D12. And "kills" is a wild exaggeration. 2700X is widely recognized as a perfectly good gaming part, built for the future instead of the past, and also the best workstation in class. Cores to spare, aggregate throughput without peer.
I fully understand that aggregate core performance of the 2700X per dollar is superior- but you need to accept that that metric just doesn't fucking matter to a whole lot of people.
I recognize that a lot of people are fucking stupid, I will give you that. Nutrition may not matter to them either.
I see your point. Maybe this part? solves the problem. Not sure about the socket or number of PCI lanes. Anyway, it's hard to complain about the 2400G at $150 as a throwaway placeholder, maybe even a keeper.
Haha, same with me. My 1700 build is a budget workstation, about $1k of parts. Now I intend to drop more than twice that on a 16 core TR2 and I don't bother justifying. The build itself is the point.
Well, I do max out all my cores on a regular basis, so it's not just for fun. But dammit, it is fun. I'm going for 16 cores instead of 32 because it's clocked higher, so on balance, better for my mix of serial and parallel loads. But I can certainly see the value of the 32 core part, it will be the go to part for video editing, for example.
If budget was still the main thing, I would go for the 16 core Ryzen. But the TR has twice the memory channels and TR motherboards have twice the dimm slots, so that is enough justification to go up the food chain. Still trying to figure out what I'm going to do with all those PCI lanes. Obviously, the crytocurrency miners already figured that one out.
Yes, it's a no brainer for a budget build, even if you plan to stick in a GPU at some point. Give your wife or PC a highly respectable sit-down PC that costs maybe $500 to build, or $600 for the deluxe version. For myself it is equally a no brainer that I want 8 Zen cores, and now being thoroughly addicted I will go on to 16 (above that the core clock starts to drop) which is not something I need, it is something I want.
For Intel, ARM is the invisible bite that doesn't show up directly because it is related to the PC market decline, which is entirely explained by people doing without PCs in favor of mobile ARM devices. If it wasn't for ARM, Intel would be sellilng a billion more processors a year than it now does, think about it. Intel badly wanted that mobile market and were utterly defeated. They could easily start peddling ARMs themselves, but not for the margin they got used to and are now dependent on.
Meanwhile. AMD is nice enough to not undercut too much. Not because they don't want to, but because they can't afford it. The last thing Intel should do now is go kill AMD with antitrust thuggery again, that would be extremely unwise, it would just push AMD into the arms of somebody much richer, and much more of a threat. A curious kind of detent we have going on now, with AMD enthusiasts the big winners.
These 8 core chips coming soon from intel better be very very competitive priced too
Let's speculate... I can't see the 8 core i9 selling for $329, which is where AMD has the 2700X right now. Mind you, the i9 does have a (lame but functional) graphics core, while you must install a GPU for the 8-core Ryzen, so that's a slight point on Intel's side. Very slight. It would be great to see AMD will respond with some minimal GPU on their 8 core Zen 2 parts next year, but AMD called it right anyway: everybody who plugs in that 16 thread beast also happily plugs in a GPU. On the lighter side, I do recall my moment of shock horror when my shiny new 1700 bench build came up with no video. Thought it was dead until I noticed the fan speed changing. Then duh.
There is no way that the 8 threaded i7 competes with the 16 threaded 2700X unless it is half the price, which isn't going to happen. Even the 2400G looks better than the new I7, at least it has a credible GPU. Is Intel banking purely on brand loyalty this cycle?
A blow to the ego, certainly, but real damage? I doubt it. ARM bites Intel a whole lot harder.
Meltdown and Spectre don't depend on hyperthreading, they just run faster with it.
Have you forgotten about Spectre and Meltdown already? That's why they're not hyper-threaded.
Brilliant, so that's why i9 is not hyper-threaded? Oh wait.