If the gaming performance is true it's a very bad sign for AMD as a very large chunk of people who buy those CPU's are not looking for professional application performance.
It's also a very bad sign that it's doesn't gets double the performance of a non hyperthreaded Intel 4 core parts while Intel 8 core parts do.
Considering it still loses to a 6900K (yes it's a very expensive part but it doesn't matter) in professional benchmarks and you lose native support for USB 3.1, vastly superior PCIE and storage performance, more PCIE lanes, and Thunderbolt it also doesn't bode well for AMD as the professional market is still going to go with Intel CPU's almost regardless of the price point.
And all it seems that Intel needs to do to beat AMD at their own game is simply to release 6 and 8 core non-HT parts at a 400-500 and 600-700$ price points since it would match the performance of a 6900K in virtually every professional application if Zen does become at any point a threat to Intel's sale.
I can only speak for myself, but I never expected these chips to compete with intels top offering in any setting or use case. Certainly not for gaming.
I own an AMD cpu because it was affordable, not because it was great, but good enough. If this is "good enough" the chips will do fine IMO.
All I'll say about competitive AMD is don't hold your breath.
The problem is that it's the 2nd time that AMD is supposed to offer something "competitive" the early Bulldozer benchmarks were promising but then the actual results were garbage.
This is going to be a 500-600$+ part, it loses to an Intel's 230$ parts in gaming and doesn't really beat their current top offering that while does cost around 1000$ when you factor the final price of such system and the fact that you'll lose PCIE lanes, storage performance native USB 3.1 and better networking and Thunderbolt when it comes to systems that cost 5000-6000$ or more it wouldn't really matter.
There is very little reason to buy AMD cpu's atm they don't really offer good and affordable platforms if you bought one recently you made a mistake because you bought a platform without native USB 3.1 with previous gen PCIE and with very poor storage performance, sure it might not matter to you but if you only browse the internet then a Core i7 system or a NUC would work just as well.
AMD needs a competitive part and more than one and not only in price; price is irrelevant NVIDIA and Intel smoke AMD sales on both fronts when the performance lead is large enough regardless of the price time and time again.
AMD needs to win benchmarks and to offer better prices to gain a substantial market share.
From the translations/discussion this sounds like it's a 3.15/3.3 GHz 8C/16t engineering sample (ES7?). Current 8C clocks are up to 3.4/3.8 and the 4c/8t (the 6700k competitor) is expected to have a 3.8 or 4.0 GHz base. IPC-wise, the Zen chips appear competitive and in line with AMDs claims so personally I don't see this as a bad sign at all.
The pictures on the graph show the AMD part to be a 3.33 / 3.4 boost clock one.
If they can't push it to beat a 6700K in gaming it won't matter, especially at it's price point, you'll be giving up way too much on the platform itself.
ATM it's slower than a 220-230$ Core i5 in games and those get better PCIE performance and native NVME and USB 3.1 support and Thunderbolt which does matter to people.
As far as professional applications goes, can't beat a 6900K has only 24 PCIE lanes which means no multi-GPU support if you want to use PCIE based storage and even with a 16/8 GPU setup you are still going to get throttled heavily because USB 3.1, network, sound and every other peripheral is running over PCIE these days (if you have even a single NVME SSD you'll have to lock it to 8/8 or 8/4 which means you can actually bottleneck the GPU somewhat on the X4).
Overall no sorry doesn't look good, especially considering that in benchmarks that AMD showed with the 8c/16t part they were beating a 6900K at 3.4ghz it doesn't seem to be the case with this evidence.
As far as the IPC part it wasn't really in question at this time, the clocks were and still are, AMD is going to be at very big disadvantage as far as the platform goes ("chipset" if you like) because Intel does PCIE, storage and LAN considerably better than AMD and also has native USB 3.0/3.1 support.
AMD needs to be able to provide better clocks than Intel to be really valuable in this, and at least as many PCIE lanes as Intel does; 24 is simply not enough.
The aggregated benchmark numbers are total bullshit, a complete rip off & insultingly anti-useful. This is serious propaganda, total malarkey. Who knows which of those games can use how many cores? Who knows how the results from various games get aggregated together to produce the "98%" result. I struggle to think of a less informative way to leak results.
The only game in the aggregated results that I strongly recognize as being built for multi-core- what Ryzen is really firing shots across the bow at Intel at- is BF4. And I'm not sure how scale-out it goes- 4 cores, 6 cores? I have no idea how these other games scale- Far Cry 4, GRID, AutoSport, X3:TC, Witcher 3, Anno 2070. I'm pretty confident at least two of them don't scale well across cores at all. It may not be how games are built now, but in the future, we need to have the expectation that good games scale out, and part of the problem with setting this expectation has been the huge premium Intel charges- the exotic status it sets- for many-core parts, and AMD seems set to put a huge amount of performance on the table for very little cash, if game makers are willing to keep up.
Being a consumer magazine, a large part of the readership are probably gamers. It makes perfect sense to include a test of what users should expect if they buy the hardware for playing games.
Like it or not, game performance is a useful real-life metric for many consumers.
You're a moron. That's not in any way what I was trying to refute.
Benchmarks are great! These are however a crime against all respectable benchmarks ever.
By aggregating all the results together, we have no idea how to tell whether it's just some games that the processor doesn't do well with, & we have no idea how to find out which types of games that may be. Your willful disregard for all the points I'm making are an insult to the inquiry that any decent benchmarkers would do- we have to know what games do ok and what games fail to begin to make a pattern.
If only old, ancient single core games suffer, & suffer badly, so be it. If all games are a little bit under par, ok. But we need something other than the terrible aggregated synthetic mess of a "benchmark" that is presented here which gives us no idea where on the spectrum this processor lies. This benchmark is a sham and bullshit & could not work harder to prevent any kind of meaningful conclusion from being drawn.
9 comments
[ 4.6 ms ] story [ 29.0 ms ] threadIt's also a very bad sign that it's doesn't gets double the performance of a non hyperthreaded Intel 4 core parts while Intel 8 core parts do.
Considering it still loses to a 6900K (yes it's a very expensive part but it doesn't matter) in professional benchmarks and you lose native support for USB 3.1, vastly superior PCIE and storage performance, more PCIE lanes, and Thunderbolt it also doesn't bode well for AMD as the professional market is still going to go with Intel CPU's almost regardless of the price point.
And all it seems that Intel needs to do to beat AMD at their own game is simply to release 6 and 8 core non-HT parts at a 400-500 and 600-700$ price points since it would match the performance of a 6900K in virtually every professional application if Zen does become at any point a threat to Intel's sale.
I own an AMD cpu because it was affordable, not because it was great, but good enough. If this is "good enough" the chips will do fine IMO.
All I'll say about competitive AMD is don't hold your breath.
This is going to be a 500-600$+ part, it loses to an Intel's 230$ parts in gaming and doesn't really beat their current top offering that while does cost around 1000$ when you factor the final price of such system and the fact that you'll lose PCIE lanes, storage performance native USB 3.1 and better networking and Thunderbolt when it comes to systems that cost 5000-6000$ or more it wouldn't really matter.
There is very little reason to buy AMD cpu's atm they don't really offer good and affordable platforms if you bought one recently you made a mistake because you bought a platform without native USB 3.1 with previous gen PCIE and with very poor storage performance, sure it might not matter to you but if you only browse the internet then a Core i7 system or a NUC would work just as well.
AMD needs a competitive part and more than one and not only in price; price is irrelevant NVIDIA and Intel smoke AMD sales on both fronts when the performance lead is large enough regardless of the price time and time again.
AMD needs to win benchmarks and to offer better prices to gain a substantial market share.
As far as professional applications goes, can't beat a 6900K has only 24 PCIE lanes which means no multi-GPU support if you want to use PCIE based storage and even with a 16/8 GPU setup you are still going to get throttled heavily because USB 3.1, network, sound and every other peripheral is running over PCIE these days (if you have even a single NVME SSD you'll have to lock it to 8/8 or 8/4 which means you can actually bottleneck the GPU somewhat on the X4).
Overall no sorry doesn't look good, especially considering that in benchmarks that AMD showed with the 8c/16t part they were beating a 6900K at 3.4ghz it doesn't seem to be the case with this evidence.
As far as the IPC part it wasn't really in question at this time, the clocks were and still are, AMD is going to be at very big disadvantage as far as the platform goes ("chipset" if you like) because Intel does PCIE, storage and LAN considerably better than AMD and also has native USB 3.0/3.1 support. AMD needs to be able to provide better clocks than Intel to be really valuable in this, and at least as many PCIE lanes as Intel does; 24 is simply not enough.
The only game in the aggregated results that I strongly recognize as being built for multi-core- what Ryzen is really firing shots across the bow at Intel at- is BF4. And I'm not sure how scale-out it goes- 4 cores, 6 cores? I have no idea how these other games scale- Far Cry 4, GRID, AutoSport, X3:TC, Witcher 3, Anno 2070. I'm pretty confident at least two of them don't scale well across cores at all. It may not be how games are built now, but in the future, we need to have the expectation that good games scale out, and part of the problem with setting this expectation has been the huge premium Intel charges- the exotic status it sets- for many-core parts, and AMD seems set to put a huge amount of performance on the table for very little cash, if game makers are willing to keep up.
Like it or not, game performance is a useful real-life metric for many consumers.
Benchmarks are great! These are however a crime against all respectable benchmarks ever.
By aggregating all the results together, we have no idea how to tell whether it's just some games that the processor doesn't do well with, & we have no idea how to find out which types of games that may be. Your willful disregard for all the points I'm making are an insult to the inquiry that any decent benchmarkers would do- we have to know what games do ok and what games fail to begin to make a pattern.
If only old, ancient single core games suffer, & suffer badly, so be it. If all games are a little bit under par, ok. But we need something other than the terrible aggregated synthetic mess of a "benchmark" that is presented here which gives us no idea where on the spectrum this processor lies. This benchmark is a sham and bullshit & could not work harder to prevent any kind of meaningful conclusion from being drawn.
Personal attacks are not allowed on HN and will get your account banned, so please don't do this again.
Also, the site guidelines ask you to drop this sort of name calling from your comments here:
> total bullshit, a complete rip off & insultingly anti-useful. This is serious propaganda, total malarkey
(Though I admit that is an impressive run.)
https://news.ycombinator.com/newsguidelines.html
https://news.ycombinator.com/newswelcome.html