Who's planning to buy Zen?

[Trimlock]

Should be shitloads cheaper platform than Intel's X99 and the likes.

Considering how low to low-mid range that board looks I'd hope (for AMD's sake) it would be ALOT cheaper than x99.

 

[drescherjm]

My main hope is that these boards with support 128GB of RAM.

32GB unbuffered dimms will be expensive if produced. I doubt consumer class would support REG-ECC.

Now the server socket will have a much higher limit and support REG ECC.

 

[jardows]

I've got an office computer that is ending its service in the indefinite future (motherboard appears to be going out). It's currently a SB Pentium B850. Does what it needs to do, but some new components are in soon order. If it hold on long enough, I'll have to consider the Zen. Right now I'm looking at an i3-6100, so whatever AMD puts out in that price range could be the ticket! My main rig is a Xeon E3 v2, so I'll be riding it a bit longer.

 

[Zedicus]

Zen (Ryzen) is the first time i have been excited about anything in the computer industry in YEARS.
hopefully this time it actually pans out, if for no other reason than to get intel moving again it would be acceptable.
I will buy it just because it says AMD rather its good or terrible, just to support the competition. here is hoping i do not have to always explain my purchase again for the next 10 years. even the current AMD stuff is very good in a very specific area of computing. having it be an actual competitor instead of a 'how do i make it work for me' platform would be great though.

 

[Child of Wonder]

32GB unbuffered dimms will be expensive if produced. I doubt consumer class would support REG-ECC.

Now the server socket will have a much higher limit and support REG ECC.

Well aware of that. Holding out hope.

If not, I'll nab a 16 core Naples. MOAR RAM.

 

[jeremyshaw]

I think I will, even though I have serious doubts it will be competitive with Intel on anything but price/perf. Whatever decent Mini-ITX AM4 board comes out will likely take the place of my Pentium G4400 setup.

If it straight up blows away my i7 2600k in every single metric, I will get the AMD chip on principle alone.

 

[razor1]

well I sure hope it blows away your i7 2600k lol.

 

[jeremyshaw]

well I sure hope it blows away your i7 2600k lol.

Same, here. I have 32GB of DDR4 RAM waiting for a good CPU and MB (and being wasted on a G4400).

 

[tangoseal]

I think I will, even though I have serious doubts it will be competitive with Intel on anything but price/perf. Whatever decent Mini-ITX AM4 board comes out will likely take the place of my Pentium G4400 setup.

If it straight up blows away my i7 2600k in every single metric, I will get the AMD chip on principle alone.

well I sure hope it blows away your i7 2600k lol.

Same, here. I have 32GB of DDR4 RAM waiting for a good CPU and MB (and being wasted on a G4400).

I'm not sure if you guys are paying attention but the Ryzen (AMD ZEN) yesterday through their special event just shown the world that the chip is faster than the I7-6900K which is Intel's 8 core 16 thread desktop top of the line.

Go and read more in the threads here for the details. The Ryzen was clock locked at 3.4ghz with Boost disabled while the Intel was running full steam with turbo running etc... and the ZEN was as fast, I think it was even faster in the tests.

Unlock the chips boosts, which will be done in the retail parts, and it will certainly be faster than Intel's fastest chip. And cost significantly less.

And JeremyShaw your comment about the I7-2600K .... LMAO, (Not at you) but at what unfolded yesterday. It completely and utterly destroys that old Sandy - it absolutely smokes my Sandy 3930K. The official event proved that yesterday.

 

[razor1]

yes we have been paying attention, hence why it should blow away a 2600k, if it doesn't then that is just AMD being AMD.

I could care less what AMD shows us, it could just be cherry picked, till those reviews come out, there is nothing certain about what they showed us, please look at my posts about blender rendering and how setting changes within the render can have different affects on different CPU's.

 

[tangoseal]

yes we have been paying attention, hence why it should blow away a 2600k, if it doesn't then that is just AMD being AMD.

I could care less what AMD shows us, it could just be cherry picked, till those reviews come out, there is nothing certain about what they showed us, please look at my posts about blender rendering and how setting changes within the render can have different affects on different CPU's.

I just did and got 33 seconds on my 3930K at 4.6ghz (Goes to show just how slow Sandy Bridge is by todays standards). But what settings did AMD use for their presentation? And yes you are right about marketing etc... lets really see what the real reviews come out to show.

There is no way AMD went head to head with Intel's baddest boy on the block at 100 samples and both chips too that long if my dilapidated 5 year old multi generations old chip is 33 seconds. There is something very off about the samples being set to 100. I think AMD used 200 to be honest in which event my chip came in at over 01:20'something seconds.

 

[razor1]

I just did and got 33 seconds on my 3930K at 4.6ghz (Goes to show just how slow Sandy Bridge is by todays standards). But what settings did AMD use for their presentation? And yes you are right about marketing etc... lets really see what the real reviews come out to show.

There is no way AMD went head to head with Intel's baddest boy on the block at 100 samples and both chips too that long if my dilapidated 5 year old multi generations old chip is 33 seconds. There is something very off about the samples being set to 100. I think AMD used 200 to be honest in which event my chip came in at over 01:20'something seconds.

Blender has many more settings that can be influencing instructions sets one of the reasons why Cinbench is so much faster on Intel chips is because of instruction set support, so it might not be just sample sizes that are making a difference, we don't know what they did, which is why can't really make an conclusions yet, but I'm thinking its going to do well, its a good showing.

 

[tangoseal]

Blender has many more settings that can be influencing instructions sets one of the reasons why Cinbench is so much faster on Intel chips is because of instruction set support, so it might not be just sample sizes that are making a difference, we don't know what they did, which is why can't really make an conclusions yet, but I'm thinking its going to do well, its a good showing.

Also keep in mind the Handbrake test was very impressive. 54 vs. 59 seconds.

What I see, is not the numbers necessarily, but it is the fact that there is no boost and the AMD part was running at, either it was 3.2 or 3.4, but the Intel was running at full steam ahead power with Turbo, I think it is 3.7 or 3.8 full speed. That was impressive because imagine the number differences when boost is enabled or when you overclock the piss out of the Zen. Not defending assumptions - just pointing out what I found to be very impressive. With a grain of salt of course.

 

[Dan_D]

Also keep in mind the Handbrake test was very impressive. 54 vs. 59 seconds.

What I see, is not the numbers necessarily, but it is the fact that there is no boost and the AMD part was running at, either it was 3.2 or 3.4, but the Intel was running at full steam ahead power with Turbo, I think it is 3.7 or 3.8 full speed. That was impressive because imagine the number differences when boost is enabled or when you overclock the piss out of the Zen. Not defending assumptions - just pointing out what I found to be very impressive. With a grain of salt of course.

The 6900K has a clock speed of 3.2GHz with a turbo clock of 3.7GHz. Again, I'm suspicious of this because AMD's own statements on performance put it around Ivy Bridge or Haswell performance levels not Broadwell-E. It could very well indicate that this benchmark was cherry picked or something was off with the 6900K's configuration. AMD could have been understating Ryzen's performance but that wouldn't be typical for them based on the companies previous behavior with processor launches.

I can't wait to get ahold of one of these things and see for myself what's going on with it.

 

[pendragon1]

Su did say its beating expectations...

 

[Zion Halcyon]

Su did say its beating expectations...

Precisely, and the original intent of Zen was indeed at Haswell levels, however early last year, there was a LOT of talk of Zen OUTPERFORMING AMD's expectations internally. It was steady enough and coming from enough different outlets to warrant some credibility.

Given that, I don't think this is more AMD fluff - I think this is something that really can hang with Intel's big bad, and I think that even took AMD by surprise.

Given that, it shocks me that they let Jim Keller go AGAIN. Keep that man on the payroll, because it seems when he partners with AMD, all he does is win.

 

[Pieter3dnow]

Precisely, and the original intent of Zen was indeed at Haswell levels, however early last year, there was a LOT of talk of Zen OUTPERFORMING AMD's expectations internally. It was steady enough and coming from enough different outlets to warrant some credibility.

Given that, I don't think this is more AMD fluff - I think this is something that really can hang with Intel's big bad, and I think that even took AMD by surprise.

Given that, it shocks me that they let Jim Keller go AGAIN. Keep that man on the payroll, because it seems when he partners with AMD, all he does is win.

Look at the old youtube Jim Keller/AMD presentation when he was still with AMD Jim would stress the word his (or the) team , not me. For AMD to keep Jim Keller AMD has to have money and interesting things to do. Look at the diversity when he changed jobs it is would suggest that if it is not something of a challenge he would rather work somewhere else. If AMD can rack up enough money again and have something interesting maybe there is a change of a return to AMD.

When there is a Ryzen retail sample then wake me up please . If it can only get better then what is shown now sign me up for a 8C16T Ryzen .

 

[jeremyshaw]

I'm not sure if you guys are paying attention but the Ryzen (AMD ZEN) yesterday through their special event just shown the world that the chip is faster than the I7-6900K which is Intel's 8 core 16 thread desktop top of the line.

Go and read more in the threads here for the details. The Ryzen was clock locked at 3.4ghz with Boost disabled while the Intel was running full steam with turbo running etc... and the ZEN was as fast, I think it was even faster in the tests.

Unlock the chips boosts, which will be done in the retail parts, and it will certainly be faster than Intel's fastest chip. And cost significantly less.

And JeremyShaw your comment about the I7-2600K .... LMAO, (Not at you) but at what unfolded yesterday. It completely and utterly destroys that old Sandy - it absolutely smokes my Sandy 3930K. The official event proved that yesterday.

I was among the forefront of unfolding that shitstorm last night through this morning. That "benchmark" is at best laughable, and at worst, fraudulent. At any rate, if I was using Blender to render small scenes, I would use my GPU which wallops all CPUs (in the Blender Cycles renderer) in perf/watt, perf/dollar, perf/time.

I am certain in pure, outright performance, just about any modern CPU can crush even the best Sandy Bridge CPU. As noted already, there is more to a compute device than just a single metric of performance (it is why I haven't "upgraded" my CPU with any of Intel's latest offerings, either; they may be faster, but not for the cost; I am already a bit biased in favor of Ryzen, simply because Kaby Lake is just insulting, from a desktop perspective).

 

[Spirit_Retro]

I'm not sold yet... need independent benchmarks.

 

[tangoseal]

I was among the forefront of unfolding that shitstorm last night through this morning. That "benchmark" is at best laughable, and at worst, fraudulent. At any rate, if I was using Blender to render small scenes, I would use my GPU which wallops all CPUs (in the Blender Cycles renderer) in perf/watt, perf/dollar, perf/time.

I am certain in pure, outright performance, just about any modern CPU can crush even the best Sandy Bridge CPU. As noted already, there is more to a compute device than just a single metric of performance (it is why I haven't "upgraded" my CPU with any of Intel's latest offerings, either; they may be faster, but not for the cost; I am already a bit biased in favor of Ryzen, simply because Kaby Lake is just insulting, from a desktop perspective).

Gotcha ... we just have to wait on official reviews that really test the things I guess.

 

[Shintai]

Precisely, and the original intent of Zen was indeed at Haswell levels, however early last year, there was a LOT of talk of Zen OUTPERFORMING AMD's expectations internally. It was steady enough and coming from enough different outlets to warrant some credibility.

Given that, I don't think this is more AMD fluff - I think this is something that really can hang with Intel's big bad, and I think that even took AMD by surprise.

Given that, it shocks me that they let Jim Keller go AGAIN. Keep that man on the payroll, because it seems when he partners with AMD, all he does is win.

Zen have never had any intent vs an Intel product. only vs AMD products. The Haswell etc part is just from forums.

 

[limitedaccess]

There's been no mention on the actual marketing positioning of the products. So whether or not Zen is Intel performance comparable isn't even relevant from a buying decision stand point.

If you want an 8 core CPU with Intel performance that's already available from Intel. It just isn't popular or hyped because it is positioned much higher than people want to spend.

Sure maybe they hit those Intel comparable performance targets and they also do want to massively disrupt prices, I'd buy that since the deal makes sense. But a buying decision just based on performance comparable to Intel is silly without factoring in pricing.

 

[Insula Gilliganis]

Not falling for AMD hype again!! When AthlonXP changes his name to Ryzen, then I will know it is time to buy AMD again!!

 

[Dreamerbydesign]

Zen have never had any intent vs an Intel product. only vs AMD products. The Haswell etc part is just from forums.

Actually AMD targeted and was competitive with Intel. But that was 10-15 years ago when Intel dropped the ball with netburst.

 

[Zion Halcyon]

Zen have never had any intent vs an Intel product. only vs AMD products. The Haswell etc part is just from forums.

Not true. Then again, that sort of thing came from the various rumor sites, which places like this have a love-hate relationship with - people love to hate them, until one rumor they like ends up being true, and then they love them until its time to hate them again.

Thing of it is, the initial Haswell hopes were reported quite a bit from a bit ago, and it wasn't the case of one place making a report and everyone else citing the one place either.

Even if you took all but the most reported rumors with a grain of salt, it still gives a pretty clear picture of the progression of Zen. That is, unless you're one of the above types who hates rumor sites unless they agree with their pre-conceived notions.

 

[Shintai]

Not true. Then again, that sort of thing came from the various rumor sites, which places like this have a love-hate relationship with - people love to hate them, until one rumor they like ends up being true, and then they love them until its time to hate them again.

Thing of it is, the initial Haswell hopes were reported quite a bit from a bit ago, and it wasn't the case of one place making a report and everyone else citing the one place either.

Even if you took all but the most reported rumors with a grain of salt, it still gives a pretty clear picture of the progression of Zen. That is, unless you're one of the above types who hates rumor sites unless they agree with their pre-conceived notions.

Not true? Why dont you show me AMDs comments about Zen vs Intel. And I am not talking about the demos.

Everything about Zen have been in relation to construction cores from AMD. 40% IPC over XV etc.

 

[Spoonie_G]

Sadly it just wouldn't make sense for me to buy a ZEN now. My Z170 build is little over a year old. But I still might upgrade next year just for the sake of the industry.

 

[ManofGod]

I may have to buy it before release just so I can actually be able to get it right off the bat. (Pre release purchase or whatever it is called.) I will probably go with the Asus mainboard just to be on the safe side for the home computer. At work, since I will probably not overclock, an MSI one might do as well. (I would do Asrock because they work well at stock clocks but, overclocking is out of the question on their AMD stuff, they cheaped out in the AM3+ boards.)

 

[ManofGod]

Not falling for AMD hype again!! When AthlonXP changes his name to Ryzen, then I will know it is time to buy AMD again!!

That's cool, you can do what you want with your money. Fact is, it will be a fantastic upgrade path for those on FX processors such as myself but, for those with recent Intel systems, probably not so much. However, it would still make a great second system. Personally, I do not prefer the stupid high prices of Intel but then again, some do I guess.

 

[Dan_D]

Precisely, and the original intent of Zen was indeed at Haswell levels, however early last year, there was a LOT of talk of Zen OUTPERFORMING AMD's expectations internally. It was steady enough and coming from enough different outlets to warrant some credibility.

Given that, I don't think this is more AMD fluff - I think this is something that really can hang with Intel's big bad, and I think that even took AMD by surprise.

Given that, it shocks me that they let Jim Keller go AGAIN. Keep that man on the payroll, because it seems when he partners with AMD, all he does is win.

I think AMD may have targeted Ivy Bridge / Haswell's performance levels because they were so far behind Intel at the time the project started and by the time they had a basic design already figured out. Their estimates were either low on purpose or they were just being realistic about how much of the technological gap between the two companies that they could potentially close. As for Jim Keller, I get the impression he goes where he wants to based on the challenge or interest he has in a project. I don't think it's a problem with AMD not giving him the money to keep him.

When there is a Ryzen retail sample then wake me up please . If it can only get better then what is shown now sign me up for a 8C16T Ryzen .

I admire your positive outlook but the fact remains it could get much worse for AMD. I don't think most people realize just how much of a mountain AMD has to climb to compete well with Intel here. They are effectively six generations behind Intel in terms of performance. The only saving grace is that two of Intel's processor generations were extremely minor tweaks which help those CPUs sell in certain specific markets. Devil's Canyon was a tweak of Haswell for overclockers and Kaby Lake's improved iGPU and lower power consumption make it ideal for mobile applications. That's still four generations of performance that AMD is roughly having to compete against. By the time sales get rolling they'll soon be under fire by the next generation of Intel CPUs.

Also you have to consider that AMD may be cherry picking the best applications to showcase Ryzen. AMD has a history of being technically accurate with its statements and still disappointing people. Remember when AMD said Bulldozer was 50% faster than Phenom II and the Core i7 950? It turned out to be true in one or two benchmarks and that was it. They technically told the truth but forgot to mention that the processor was weak everywhere else. I'll give you another example of how this could go wrong. Ryzen could be 41% faster than Excavator which is ahead of their target expectations. It could still be slower than Broadwell-E in all but a handful of benchmarks. Ryzen could clock like shit and 3.4GHz could damn near be best they can do with the thing. We might see a boost clock of 25-200MHz and that's it with overclocking headroom that doesn't go much beyond that. At 4.0GHz it would be a loser compared to Haswell-E or Broadwell-E in all but stock vs/ stock performance. That's not necessarily the end of the world as those CPUs will probably be much more expensive than Ryzen is likely to be.

I have no idea what to expect all said and done. Even if Ryzen doesn't clock that well it might still be successful if it competes well enough against Intel's offerings at stock speeds provided the price is right. Unfortunately, I don't know if that will drive sales enough to keep the lights on at AMD. Ryzen needs to be a success or AMD as we know it is likely done for. AMD needs a CPU with performance good enough to charge a fat markup on it. That's not to say the CPU needs to be $600 or $1,000, but they need a healthy margin and volume sales.

Essentially, it breaks down like this: There qualities in any performance product. Three to be exact. Cheap, Fast, and Good. You can have any combination of the three but not all three at once. You can have fast and cheap but not good or good and fast but not cheap. This is generally a car analogy but I find it applies with almost everything related to performance be it firearms or CPUs. No processor that was good and fast was ever cheap. The super cheap CPUs were sometimes good but they were never fast. We've also had fast CPUs that were cheap but they weren't good. Excessive errata, high temperatures, high failure rates, bad motherboard platforms, stability issues, or a major weakness for certain software.

I was among the forefront of unfolding that shitstorm last night through this morning. That "benchmark" is at best laughable, and at worst, fraudulent. At any rate, if I was using Blender to render small scenes, I would use my GPU which wallops all CPUs (in the Blender Cycles renderer) in perf/watt, perf/dollar, perf/time.

I am certain in pure, outright performance, just about any modern CPU can crush even the best Sandy Bridge CPU. As noted already, there is more to a compute device than just a single metric of performance (it is why I haven't "upgraded" my CPU with any of Intel's latest offerings, either; they may be faster, but not for the cost; I am already a bit biased in favor of Ryzen, simply because Kaby Lake is just insulting, from a desktop perspective).

I don't totally understand the enthusiasts who are still holding onto their Sandy Bridge rigs. Yes that platform and processor had some really long legs. There is no doubt that Sandy Bridge is one of the all time greatest CPUs ever made. For gaming alone, Sandy Bidge is fine with a modern GPU but there are still advantages to going with a newer platform. Skylake actually shows improvement with DDR4 RAM clocked high enough.

M.2 is really the main thing that would drive me to upgrade. The NVMe storage devices make the old shit look like its standing still. Now that SSD's are reaching the 1TB+ range with fast performance I now put my games on SSDs for faster loading. This saved me from a lot of the issues in Arkham Knight as it wasn't designed to use the conventional hard drive and streaming textures from it caused a ton of issues with it. For fuck's sake a lot of those older generation motherboards are still stuck on SATA 3G. When they incorporate aftermarket controllers much of the time they aren't much if any better than the Intel 3G ports and are less flexible. For storage alone I'd make the switch. I get that USB 3.1 and Thunderbolt / Thunderbolt 3 aren't huge crowd pleasers either. There just isn't enough stuff out there for those interfaces.

In non-gaming applications it makes little sense to be on Sandy Bridge anymore. Not counting the HEDT segment, you are 5 generations behind now and the performance each generation may have been incremental, but now its added up to something significant. You are between 20% and 55% behind the newer systems as far as IPC goes. Granted you Sandy Bridge guys at 5GHz may have a case to stay where you are but the bulk of you running under 4.5GHz are a bit behind. We've had several CPU generations now that can clock to 4.7GHz or close to it. Kaby Lake may not excite a lot of you but the rumor is that they'll clock better than that. I just don't know how many CPUs can do that or how difficult such an overclock will be to achieve as of yet.

Zen have never had any intent vs an Intel product. only vs AMD products. The Haswell etc part is just from forums.

AMD probably aimed as high as they could, only intending to reach Ivy Bridge or Haswell level's of performance. No one wants to build a product that's second best. No one sets out to do that.

Actually AMD targeted and was competitive with Intel. But that was 10-15 years ago when Intel dropped the ball with netburst.

People often forget those days. There was a lot of dominoes that had to fall in order to make that happen. It was smart business acquisitions combined with timing and luck that make AMD competitive at the time.

 

[razor1]

Dan_D

Exactly what I've been saying all along! Even if Zen is up to the task to keep up with Broadwell, Intel's response could come after Kaby Lake, that is 1.5 years. If Intel didn't plan for Zen to be so good, 3 years then we will see Intel's response. All in the meantime, Zen + should keep the status quo with Intel's current chips (if Zen matches Boardwell), so by the time we see Intel's response, they will again have a node advantage, and IPC advantage granted not as high as before, but still there.

Price wars only hurt these companies guys, so if AMD prices lower like at 500 bucks a pop Intel will match up to this, margins go down for them a bit but still volume sales will remain fairly the same on either side.

Because AMD can't differentiate their product from Intel is the biggest reason why they can't gain. And with Intel's mind share for the past 10 years, consumers are not going to look at AMD, if prices are the same and performance are the same, whats the need to switch their buying habits?

 

[Zion Halcyon]

I think AMD may have targeted Ivy Bridge / Haswell's performance levels because they were so far behind Intel at the time the project started and by the time they had a basic design already figured out. Their estimates were either low on purpose or they were just being realistic about how much of the technological gap between the two companies that they could potentially close. As for Jim Keller, I get the impression he goes where he wants to based on the challenge or interest he has in a project. I don't think it's a problem with AMD not giving him the money to keep him.



People often forget those days. There was a lot of dominoes that had to fall in order to make that happen. It was smart business acquisitions combined with timing and luck that make AMD competitive at the time.

Ha, I got ya. So Jim Keller, Tech Ronin. Kinda cool, actually.

As for the old days, I remember them and agree. Pre Athlon, AMD was just another budget processor with severe overheating issues.

I will say, I do see history repeating itself here, as AMD has been making a lot of killer deals - they have a very smart business team now, it seems like.

 

[jeremyshaw]

[...]
I don't totally understand the enthusiasts who are still holding onto their Sandy Bridge rigs. Yes that platform and processor had some really long legs. There is no doubt that Sandy Bridge is one of the all time greatest CPUs ever made. For gaming alone, Sandy Bidge is fine with a modern GPU but there are still advantages to going with a newer platform. Skylake actually shows improvement with DDR4 RAM clocked high enough.

M.2 is really the main thing that would drive me to upgrade. The NVMe storage devices make the old shit look like its standing still. Now that SSD's are reaching the 1TB+ range with fast performance I now put my games on SSDs for faster loading. This saved me from a lot of the issues in Arkham Knight as it wasn't designed to use the conventional hard drive and streaming textures from it caused a ton of issues with it. For fuck's sake a lot of those older generation motherboards are still stuck on SATA 3G. When they incorporate aftermarket controllers much of the time they aren't much if any better than the Intel 3G ports and are less flexible. For storage alone I'd make the switch. I get that USB 3.1 and Thunderbolt / Thunderbolt 3 aren't huge crowd pleasers either. There just isn't enough stuff out there for those interfaces.

In non-gaming applications it makes little sense to be on Sandy Bridge anymore. Not counting the HEDT segment, you are 5 generations behind now and the performance each generation may have been incremental, but now its added up to something significant. You are between 20% and 55% behind the newer systems as far as IPC goes. Granted you Sandy Bridge guys at 5GHz may have a case to stay where you are but the bulk of you running under 4.5GHz are a bit behind. We've had several CPU generations now that can clock to 4.7GHz or close to it. Kaby Lake may not excite a lot of you but the rumor is that they'll clock better than that. I just don't know how many CPUs can do that or how difficult such an overclock will be to achieve as of yet.


[...]

I think you nailed it for me. I am not really an enthusiast. Maybe I was, once, but all I ask of my current home desktop is to be a good host for whatever GPU I currently have (which by enthusiast standards, I should have updated when the 1080 came out - but my GTX780Ti was a relative bargain, at the time, compared to current GPUs, so I hesitate to replace it). When I feel the desktop experience has really moved on, I will get more capable components (the last one for me was a, that was a 24" 144Mhz 3D display, now complementing my older U2711).

My laptop has M.2. I agree it is blazing fast. My underdesk ubuntu server has a G4400, 32GB of DDR4, and a GTX1050Ti (and M.2, with its slot in the least convenient place ever for a mITX board). It and my laptop boot up much more quickly than my desktop, and embarrass my desktop just about everywhere outside of heavy CPU/GPU loads.

It is not so much that Sandy Bridge (and by extension, my desktop) cannot handle tasks adequately. It is that newer platforms do not do it appreciably better - at least not enough to justify the cost (to me). That being said, constant upgrades to the home network (new NAS, replacement domain server, HTPC, etc) have kept newer generations of HW in the household (mostly Intel/nVidia - I did force an AMD card in to keep myself up to date with them), just not performance enthusiast hardware.


TL,DR; Perf/cost is not up to snuff with newer stuff. Experience is better on newer HW, but not egregiously so. SB (at least later ones like mine) have SATA 6G.

 

[Pieter3dnow]

I admire your positive outlook but the fact remains it could get much worse for AMD. I don't think most people realize just how much of a mountain AMD has to climb to compete well with Intel here. They are effectively six generations behind Intel in terms of performance. The only saving grace is that two of Intel's processor generations were extremely minor tweaks which help those CPUs sell in certain specific markets. Devil's Canyon was a tweak of Haswell for overclockers and Kaby Lake's improved iGPU and lower power consumption make it ideal for mobile applications. That's still four generations of performance that AMD is roughly having to compete against. By the time sales get rolling they'll soon be under fire by the next generation of Intel CPUs.

Also you have to consider that AMD may be cherry picking the best applications to showcase Ryzen. AMD has a history of being technically accurate with its statements and still disappointing people. Remember when AMD said Bulldozer was 50% faster than Phenom II and the Core i7 950? It turned out to be true in one or two benchmarks and that was it. They technically told the truth but forgot to mention that the processor was weak everywhere else. I'll give you another example of how this could go wrong. Ryzen could be 41% faster than Excavator which is ahead of their target expectations. It could still be slower than Broadwell-E in all but a handful of benchmarks. Ryzen could clock like shit and 3.4GHz could damn near be best they can do with the thing. We might see a boost clock of 25-200MHz and that's it with overclocking headroom that doesn't go much beyond that. At 4.0GHz it would be a loser compared to Haswell-E or Broadwell-E in all but stock vs/ stock performance. That's not necessarily the end of the world as those CPUs will probably be much more expensive than Ryzen is likely to be.

It is still simple , it has to do well, not beat Intel , it prolly won't beat Intel as it seems now. But I can not imagine why you would pull Bulldozer into this the outset for Bulldozer was so different from what is done with Ryzen. I know that Kyle gave the story about his PR person telling those things. But now it is down to AMD to prove things they can not hide. Most people know that when AMD shows these benchmarks it is because that is where they do better then Intel it does not tell you all about the overall performance.

There 2 things which is still somewhat of a unknown that is the definitive clock speed 3.4 now and 3.6 at launch ? Then the turbo if what they are saying is true that would promise some better performance.

I'm not convinced it will beat Intel (in all benchmarks) but that is not important for me.

 

[Dan_D]

Dan_D

Exactly what I've been saying all along! Even if Zen is up to the task to keep up with Broadwell, Intel's response could come after Kaby Lake, that is 1.5 years. If Intel didn't plan for Zen to be so good, 3 years then we will see Intel's response. All in the meantime, Zen + should keep the status quo with Intel's current chips (if Zen matches Boardwell), so by the time we see Intel's response, they will again have a node advantage, and IPC advantage granted not as high as before, but still there.

Price wars only hurt these companies guys, so if AMD prices lower like at 500 bucks a pop Intel will match up to this, margins go down for them a bit but still volume sales will remain fairly the same on either side.

Because AMD can't differentiate their product from Intel is the biggest reason why they can't gain. And with Intel's mind share for the past 10 years, consumers are not going to look at AMD, if prices are the same and performance are the same, whats the need to switch their buying habits?

Intel's response could be much faster than you think. Intel has a process node advantage. Intel doesn't need a new processor architecture to respond to AMD either. If I had to hedge a bet today, I'd bet that Intel CPUs will out clock AMD's Ryzen by a large margin. So even if Ryzen can match Broadwell or Broadwell-E stock for stock, I doubt it can overclock as well. Given how well Intel CPUs overclock (even the bad ones) I think Intel could drop higher clocked parts on us within weeks to retake the performance crown. Those might not gain the same leads while overclocking but overclocking doesn't concern the masses. It can add to a companies reputation either way, but it doesn't effect the buying habits of any group of people besides enthusiasts.

As for your latter comment, you do make a point. Its the same one I've made many times. Intel's "Intel Inside" marketing campaign was hugely successful. It successfully leveraged Intel's strength's and their competition's short comings to create instant brand recognition. Your average consumer probably thinks of Intel as Honda and AMD as the KIA or Hyundai of the processing world. In IT circles its even worse. For two decades non-Intel x86 servers have been demonized as cheap junk in many circles regardless of performance. That thinking is what kept the Opteron from doing as well as it should have.

 

[Dan_D]

It is still simple , it has to do well, not beat Intel , it prolly won't beat Intel as it seems now. But I can not imagine why you would pull Bulldozer into this the outset for Bulldozer was so different from what is done with Ryzen. I know that Kyle gave the story about his PR person telling those things. But now it is down to AMD to prove things they can not hide. Most people know that when AMD shows these benchmarks it is because that is where they do better then Intel it does not tell you all about the overall performance.

There 2 things which is still somewhat of a unknown that is the definitive clock speed 3.4 now and 3.6 at launch ? Then the turbo if what they are saying is true that would promise some better performance.

I'm not convinced it will beat Intel (in all benchmarks) but that is not important for me.

I only bring Bulldozer into the mix to illustrate how AMD likes to make comments that are technically true, but still quite misleading. It has been very adept at this over the last decade or so.

 

[razor1]

Intel's response could be much faster than you think. Intel has a process node advantage. Intel doesn't need a new processor architecture to respond to AMD either. If I had to hedge a bet today, I'd bet that Intel CPUs will out clock AMD's Ryzen by a large margin. So even if Ryzen can match Broadwell or Broadwell-E stock for stock, I doubt it can overclock as well. Given how well Intel CPUs overclock (even the bad ones) I think Intel could drop higher clocked parts on us within weeks to retake the performance crown. Those might not gain the same leads while overclocking but overclocking doesn't concern the masses. It can add to a companies reputation either way, but it doesn't effect the buying habits of any group of people besides enthusiasts.

As for your latter comment, you do make a point. Its the same one I've made many times. Intel's "Intel Inside" marketing campaign was hugely successful. It successfully leveraged Intel's strength's and their competition's short comings to create instant brand recognition. Your average consumer probably thinks of Intel as Honda and AMD as the KIA or Hyundai of the processing world. In IT circles its even worse. For two decades non-Intel x86 servers have been demonized as cheap junk in many circles regardless of performance. That thinking is what kept the Opteron from doing as well as it should have.

Of course Intel can have a response sooner, I was taking the angle of Intel underestimating Zen or not prepared for it

 

[Vader1975]

I am planning on buying one, but I may wait for their 2nd generation of it.

 

[Dan_D]

Right now I'm adopting a wait and see attitude with it. I'm cautiously optimistic that it will be good. It may not be faster than a lot of what Intel has to offer, but again we'll see. If its a solid performer for virtualization and server work at the right price I may definitely be interested in it.

 

[tangoseal]

Even if the benchmark was done on a slightly manupulated and even on a cherry picked CPU the numbers look good. Lets say it is 10% slower than pictured at the conference. Well then that still puts it above about everything else Intel offers except its most top tier chips. If this is the case then this chip is still a win and in the right direction.

Opinion nothing more. Doing what others said and not going to hold my breath but am highly optimistic at the benchmark results.