The real reason there is so much hunger for Ryzen

[Mockingbird]

Intel has completely separate the mainstream platform (i.e. Intel Z270) from the enthusiast HEDT platform (i.e. Intel X99) in order to keep prices of the latter high.

Intel has limited the mainstream to quad-core while the enthusiast platform can have up to ten-cores.

The barrier to entry for the enthusiast platform as well as for switching platforms are very high.

For example, enthusiast platform motherboard (i.e. Intel X99) is also significantly more expensive the mainstream platform motherboard

Switching from the mainstream platform to enthusiast platform requires buying new motherboards and quad-channel memory as well as processors.

Had Intel released hex-core and octo-core processors for the mainstream platform, they could have become popular, but Intel didn't do this specifically to keep prices of hex-core and onto-core processors high.

As a result, there is been a lot of pending demand for hex-core and octo-core processors for the mainstream platform and this is precisely what AMD is offering.

Conversely there may not be as much demand for quad-core Ryzen processors because many people with quad-core Intel Core processors (Sandy Bridge and later) wouldn't consider quad-core Ryzen to be much of an upgrade or even a downgrade for those with quad-core (Intel Skylake or later)

 

[tamz_msc]

Yeah, I wish things hadn't changed since the days of X58/P55.

 

[Topweasel]

I wouldn't assume demand for a 4c8t or 4c4t Ryzan to be low. It's just at best we will have to wait out its release. It could be 6 months before those offerings are available. But remember how the price slotting should be on these chips. You are getting to near i3 pricing on chip that will be competing with 6700k or 6500k in performance.

Few people looking at building a Ryzen now are looking at the absolute value, though that might be the 1600x. They are looking for MT monster for cheap. But when it comes down to it I expect Ryzen to do less damage to the 7700k sales and more to the HEDT, i5, i3, sales. The 7700 will still have some value as relatively cheap max IPC game performer. But the value disappears on both sides of it.

 

[LTC8K6]

AMD has been offering hex-core and octo-core mainstream cpus for a very long time, though. Demand for more cores in the DT arena apparently was not that high.

It certainly may be higher these days, though.

Intel isn't pining for the fjords yet.

 

[IntelUser2000]

Intel has completely separate the mainstream platform (i.e. Intel Z270) from the enthusiast HEDT platform (i.e. Intel X99) in order to keep prices of the latter high.

There's another issue with HEDT. It's based off on server so on PC it's largely a waste. A Chinese site comparing BDW with Ryzen showed the latter having significant power use advantage. The difference will largely have been made up if BDW was on mainstream LGA115X.

SKL-X and KBL-X aren't the answer to Ryzen, at the most Coffelake is. KBL-X should have been on the mainstream platform.

I am extremely surprised at reports of Ryzen IMC having significant efficiency advantage over Intel's. Intel's traditional area of strength is the memory controller. Again, this is a repeat of Athlon in 1999. Doing something up until that point thought to be not possible to do. Some reports had Athlon having advantage in I/O operations.

But when it comes down to it I expect Ryzen to do less damage to the 7700k sales and more to the HEDT, i5, i3, sales.

The real damage is changing perception. From now they have a chance to turn around their brand. Brand recognition is largely based on trust. It'll take a few years to gain, and few to lose. Then sales follow.

 

[LTC8K6]

I don't recall the 5775C showing any significant power advantages, though. Clock for clock it was not much different than Haswell, iirc.

 

[CatMerc]

AMD has been offering hex-core and octo-core mainstream cpus for a very long time, though. Demand for more cores in the DT arena apparently was not that high.

It certainly may be higher these days, though.

Intel isn't pining for the fjords yet.

Of course not, those 8 cores were weak and no better than Intel 4 cores even in the most multithreaded tasks.

 

[BigDaveX]

I am extremely surprised at reports of Ryzen IMC having significant efficiency advantage over Intel's. Intel's traditional area of strength is the memory controller. Again, this is a repeat of Athlon in 1999. Doing something up until that point thought to be not possible to do. Some reports had Athlon having advantage in I/O operations.

A better comparison might be the original DDR1-equipped Athlon 64, which had absolutely incredible memory efficiency, and is probably still the single most efficient memory controller ever featured on an x86 platform. For whatever reason none of their products from the AM2 Athlon 64s onwards were ever as impressive in that regard, so I guess that when Keller came back he helped them rediscover their memory controlling mojo.

 

[raghu78]

I think AMD R3 and R5 are likely to be as disruptive as R7 as they bring more cores to the respective price points at which they are launched. Ryzen R5 1600X is launching in Q2 with 3.6 Ghz base and 4.0 Ghz boost clocks at USD 259. This chip will run very close to 7600k in most single thread benchmarks while absolutely murdering it in multi thread benchmarks. Similarly the R3 4C/4T is going to be very close to core i3 in single thread while destroying it in multithread benchmarks. Moreover the R3 is going to be the best gaming CPU below USD 150 as most games today run faster on 4 real cores rather than 2 cores with SMT.

 

[JimmiG]

AMD has been offering hex-core and octo-core mainstream cpus for a very long time, though. Demand for more cores in the DT arena apparently was not that high.

It certainly may be higher these days, though.

Intel isn't pining for the fjords yet.

The problem was AMD *needed* 8 cores to be competitive with Intel's 4 cores. People weren't willing to give up that much single-threaded performance to get more cores. Even in benchmarks that effectively use 8 cores, AMD barely won against Intel's 4C/8T CPUs.

 

[scannall]

For me anyway, I need more cores. And I have wanted to upgrade my OC'd Xeon 5660 for a while. But, not at Intel's prices. AMD has the product I want, at the price I am willing to pay. Intel doesn't. It's just that simple.

 

[BigDaveX]

The problem was AMD *needed* 8 cores to be competitive with Intel's 4 cores. People weren't willing to give up that much single-threaded performance to get more cores. Even in benchmarks that effectively use 8 cores, AMD barely won against Intel's 4C/8T CPUs.

I think AMD's intention for Bulldozer was more along the lines of what Zen (apparently) ended up being; competitive with Intel's mainstream desktop chips in single-thread, while completely obliterating them in multi-thread. Between the underwhelming IPC and not being able to clock high enough, however, it ended up a massive disappointment.

 

[Flapdrol1337]

Imo the reason is there hasn't been a significant price/performance upgrade since sandy bridge.

Every time intel released a new chip it was slightly faster, but also slightly more expensive. Today you can get a little bit better performance per dollar than with a 2500K, but not much.

By the looks of it Ryzen will undercut intel significantly, finally a worthwhile upgrade that isn't that much more expensive. And they can easily do this since they're not wasting space on an igp.

 

[LTC8K6]

I've decided to wait until the initial rush calms down with RyZen and also see what develops with Intel in a couple months.

I'm going to give Intel a chance to respond.

 

[Mockingbird]

I think AMD R3 and R5 are likely to be as disruptive as R7 as they bring more cores to the respective price points at which they are launched. Ryzen R5 1600X is launching in Q2 with 3.6 Ghz base and 4.0 Ghz boost clocks at USD 259. This chip will run very close to 7600k in most single thread benchmarks while absolutely murdering it in multi thread benchmarks. Similarly the R3 4C/4T is going to be very close to core i3 in single thread while destroying it in multithread benchmarks. Moreover the R3 is going to be the best gaming CPU below USD 150 as most games today run faster on 4 real cores rather than 2 cores with SMT.

Call me skeptical, but people who already have Intel Core i3 are probably not enthusiasts (obviously I define "enthusiasts" differently from Intel) and probably don't know what AMD Ryzen 3 is or even shopping for new computers.

Additionally, they might not willing to pay for an extra a GPU (there won't be integrated GPUs until Raven Ridge).

 

[PingSpike]

I think AMD R3 and R5 are likely to be as disruptive as R7 as they bring more cores to the respective price points at which they are launched. Ryzen R5 1600X is launching in Q2 with 3.6 Ghz base and 4.0 Ghz boost clocks at USD 259. This chip will run very close to 7600k in most single thread benchmarks while absolutely murdering it in multi thread benchmarks. Similarly the R3 4C/4T is going to be very close to core i3 in single thread while destroying it in multithread benchmarks. Moreover the R3 is going to be the best gaming CPU below USD 150 as most games today run faster on 4 real cores rather than 2 cores with SMT.

I think you're right about the R5 1600x. I gotta wait for benchmarks and things anyway so I might end up getting that one. That's a product that really makes Intel's current stuff look like ass. I'm not sure about the i3 though. Intel already seppuku'd their i3s with the g4560, they are a pointless product now. But in a way maybe that makes a $120-130 quad core ready to sweep in I guess.

 

[kaesden]

I think AMD's intention for Bulldozer was more along the lines of what Zen (apparently) ended up being; competitive with Intel's mainstream desktop chips in single-thread, while completely obliterating them in multi-thread. Between the underwhelming IPC and not being able to clock high enough, however, it ended up a massive disappointment.

The real problem with Bulldozer was that although you could get an "8-core" chip, it practice it was only a 4-core chip due to only having 4 floating point units, with 8 integer units so it was about as accurate to call Bulldozer an 8 core as it is to call an i7 an 8-core chip because of hyper threading. Ryzen fixes all of this and now an 8-core AMD chip, is really a full physical 8-core chip.

 

[Markfw]

I will respectfully disagree. X99 motherboard can be had for under $150, I just got one for $130 (open box, but new) And CPU's ? I got 14 cores and 28 threads, and it was NOT an ES, for $349. (see sig) So for $20 more than the cheapest Ryzen in this release, I get 28 threads instead of 16.

HOWEVER, that 28 threads@2.5 ghz, not 16 at 4.0 if you get the $500 1800x. So, Ryzen may win this one too.

Now you can get socket 1151 motherboard for less than that, but not a lot. And CPU's ? I am looking to get a quad-core with hyperthreading to upgrade my G4400, but can't find one for under $200 (maybe Ryzen will help that)

However, I really do think there is a good chance we have a winner in Ryzen, tomorrow will tell for sure. For the first time in 10 years, I want to see a winner from AMD !

 

[richaron]

Call me skeptical, but people who already have Intel Core i3 are probably not enthusiasts (obviously I define "enthusiasts" differently from Intel) and probably don't know what AMD Ryzen 3 is or even shopping for new computers.

Call me skeptical, but this sound like the usual misunderstanding by people who think spending (wasting?) money on hardware makes someone an "enthusiast". I think a real enthusiast would understand many people have vastly overpowered rigs, and a real enthusiast would happily make do with an i3 if they knew it was all the performance they need (as opposed to just throwing money at a build).

As an enthusiast I'm interested in Ryzen not because I'm out to get the top of the line CPU (I'm currently GPU bottlenecked and will be fine with current CPU performance for the near future).

I am interested in the possibilities which Ryzen will open up. Including more competitive choices with modern platforms, 'disruptive' prices, and pressure on future performance gains.

 

[Dulanic]

I haven't bought since my i7 930 a long time ago. In finally going to update to a ryzen. I felt like Intel was over pricing with no true competition.

Now that there is a great CPU for a good price I felt it was time.

 

[jihe]

Intel has been treating their customers like redacted. How long have people asked for soldered heat spreaders instead of the terrible TIM on their K-series?




No profanity allowed on the tech forums.


esquared
Anandtech Forum Director

 

[IntelUser2000]

I don't recall the 5775C showing any significant power advantages, though. Clock for clock it was not much different than Haswell, iirc.

I've never mentioned the 5775C.

Ok, my problem is that I assume people will know all the acronyms used in computers. Here's what I said:

There's another issue with HEDT. It's based off on server so on PC it's largely a waste. A Chinese site comparing BDW with Ryzen showed the latter having significant power use advantage. The difference will largely have been made up if BDW was on mainstream LGA115X.

HEDT: High End Desktop
BDW: Broadwell, but in conjunction with HEDT, it means BDW-E, or Broadwell-E or way-too-expensive enthusiast oriented Broadwell
LGA115x: Socket for all mainstream Intel platforms.

The mainstream Intel platforms have significant power use reduction over HEDT. One benchmark showed Ryzen having significant advantage in power use over HEDT BDW. The reason is Ryzen puts 8 cores in a optimized consumer platform and HEDT Intel platforms are just derivatives of server ones.

A better comparison might be the original DDR1-equipped Athlon 64, which had absolutely incredible memory efficiency, and is probably still the single most efficient memory controller ever featured on an x86 platform.

Nehalem was far better than anything IMC AMD had. Intel never had a IMC for DDR technology so we'll never know but considering the advantage Nehalem and post chips had I'd have bet on that lead continuing. The way of doubling bandwidth with DDRx technologies sacrifice bit of latency and efficiency to do so.

Anyway that was not my point. Athlon was the original comparison because that chip did not have an absolute lead and/or maintained it for a significant period of time. Athlon 64 few years later did. Athlon came at a time when Intel was stagnating with cores. They added multimedia extensions and called it a third generation Pentium.

Another thing was Athlon broke the mentality people had back then of AMD being the copycat of Intel. Athlon was genuinely different. True, the details are different but many things have parallels. In a few years with Zen derivatives and with Intel really being stuck with process advancements going nowhere we might see Athlon 64-like performance and marketshare lead.

 

[jihe]

I will respectfully disagree. X99 motherboard can be had for under $150, I just got one for $130 (open box, but new) And CPU's ? I got 14 cores and 28 threads, and it was NOT an ES, for $349. (see sig) So for $20 more than the cheapest Ryzen in this release, I get 28 threads instead of 16.

HOWEVER, that 28 threads@2.5 ghz, not 16 at 4.0 if you get the $500 1800x. So, Ryzen may win this one too.

Now you can get socket 1151 motherboard for less than that, but not a lot. And CPU's ? I am looking to get a quad-core with hyperthreading to upgrade my G4400, but can't find one for under $200 (maybe Ryzen will help that)

However, I really do think there is a good chance we have a winner in Ryzen, tomorrow will tell for sure. For the first time in 10 years, I want to see a winner from AMD !

Nice deal on the 2683v3, were you able to downgrade microcode to make it run at 3Ghz all cores?

 

[Markfw]

Nice deal on the 2683v3, were you able to downgrade microcode to make it run at 3Ghz all cores?

Nope. 2.5

 

[WhoBeDaPlaya]

Depending on how high the IPC is, a decently clocked (say 4.5GHz) Ryzen should just about beat my dual X5690 Z800 in encode jobs, which is nice