UPDATED AGAIN: New Ryzen prices and clock speeds rumoured: lowest-end 8c/16t for $322, lowest-end 6c/12t for $223

[MageTank]

2 minutes ago, Ramaddil said:

https://www.pcper.com/news/Processors/AMD-Details-Zen-ISSCC

 

From that link

In some of the basic measurements of the different processes we see that Intel has advantages throughout.  This is not surprising as Intel has been well known to push process technology beyond what others are able to do.  In theory their products will have denser logic throughout, including the SRAM cells.  When looking at this information we wonder how AMD has been able to make their cores and caches smaller.  Part of that is due to the likely setup of cache control and access.

 

One of the most likely culprits of this smaller size is that the less advanced FPU/SSE/AVX units that AMD has in Zen.  They support AVX-256, but it has to be done in double the cycles.  They can do single cycle AVX-128, but Intel’s throughput is much higher than what AMD can achieve.  AVX is not the end-all, be-all but it is gaining in importance in high performance computing and editing applications.  David Kanter in his article covering the architecture explicitly said that AMD made this decision to lower the die size and power constraints for this product.

 

Basically it seems like its a educated guess given the die size of the CPU. 

I see. People were speaking here as if the information was absolute. So it's another "let's wait and see" situation.

[Ramaddil]

That's what I am getting from this.. I think there is some healthy skepticism given AMD's lack of competition to intel to think that this processor will be more of the same.  For me I am going to be optimistic and hope that this processor is great for the money.  I have been saving for awhile and cant wait to build my first 8 core 16 thread cpu. 

[79wjd]

9 minutes ago, MageTank said:

I see. People were speaking here as if the information was absolute. So it's another "let's wait and see" situation.

Considering the 8c/16t is priced so much lower than Intel's, there's no way there isn't some aspect that is noticeably worse and based on some of the benchmarks, FPUs would be a reasonable assumption as to where that deficit might lie. 

[Ramaddil]

That is kind of false logic.. If AMD came up with a better and cheaper way to manufacture something, wouldn't it stand to reason that they could sell it cheaper than intel's $1000 CPU?

[MageTank]

1 minute ago, djdwosk97 said:

Considering the 8c/16t is priced so much lower than Intel's, there's no way there isn't some aspect that is noticeably worse. 

I agree. However, I personally think that aspect will be overclocking. Let's be real. 8c/16t at 95w is one thing, but to pretend a 65w SKU exists with the same thread count? The only Intel SKU's that match those numbers, are sub-2ghz on the clock speed. My fear is the IMC will be vastly crippled compared to Intels, as we saw with the FX series. 

 

I just can't imagine it would be FPU, as the benches we've seen (even if they were only 4k) seemed to be quite good, unless the FPU performance is simply "good enough" for gaming. The truth is, we won't really know until reviewers get their hands on them and test this for us. 

[Ramaddil]

I am not trying to say it will be better or that AMD is some how better than Intel, I think people are jumping the gun without giving it a fair shake.

[MageTank]

1 minute ago, Ramaddil said:

I am not trying to say it will be better or that AMD is some how better than Intel, I think people are jumping the gun without giving it a fair shake.

Nobody is accusing you of that. I think you are assuming our comments are directed towards you, when they are just broad general statements. We are simply sharing our opinions as to why we believe AMD's product is cheaper than Intel's. Is it possible AMD is simply trying to undercut Intel by a large margin with a highly competitive product? Sure, but AMD has already told us they are tired of being referred to as the "cheaper option". It stands to reason that if their product is as competitive as they claim it is, then they should seek the biggest profit margin that they can with it. However, directly cutting the cost in half compared to Intel's comparable product, places a fair bit of doubt in my mind. Either they are comfortable with a lower margin (which they can't really afford to be) for the sake of trying to win consumer market share back, or their product is inferior and they know it.

 

I personally hope it's the first option, and that AMD is releasing cheap for now, just to win their customers back for this first generation of Zen products. If we get another round of noncompetitive CPU's, then we all lose as consumers. Intel and AMD fans alike. 

[79wjd]

14 minutes ago, MageTank said:

I agree. However, I personally think that aspect will be overclocking. Let's be real. 8c/16t at 95w is one thing, but to pretend a 65w SKU exists with the same thread count? The only Intel SKU's that match those numbers, are sub-2ghz on the clock speed. My fear is the IMC will be vastly crippled compared to Intels, as we saw with the FX series. 

 

I just can't imagine it would be FPU, as the benches we've seen (even if they were only 4k) seemed to be quite good, unless the FPU performance is simply "good enough" for gaming. The truth is, we won't really know until reviewers get their hands on them and test this for us. 

Would gaming even show weak FPU performance? They showed BF1 at 4k, but at 1080p the FX8370 isn't far behind, and if there's any IPC gains at all, then that performance delta would be completely negated especially once you consider they did their runs at 4k further minimizing the bottleneck shown here. 

[Ramaddil]

Well to be honest I am looking at this CPU in particular for my server, Gaming doesn't matter to me at all for this CPU.  My server does encoding, backups, security cameras, and trans-coding in plex.  More cores and threads for me is better for a lot of reasons.  While it is likely given what people are saying it probably wont beat intel's $1000 CPU it will surely beat the AMX FX 8320 that is in my server now.  I used that CPU because my main was upgraded not because I bought it for it. 

 

So given individual use cases it wont be good for everyone it will be good for some no matter what, and cheaper than Intels 8c/16th option.

[MageTank]

7 minutes ago, djdwosk97 said:

Would gaming even show weak FPU performance? They showed BF1 at 4k, but at 1080p the FX8370 isn't far behind, and if there's any IPC gains at all, then that performance delta would be completely negated especially once you consider they did their runs at 4k further minimizing the bottleneck. 

I can't speak for battlefield or that star wars game they benched (I don't have either one) but yes, gaming can show weak FPU performance, especially MMO's. I used to play an older MMO called Perfect World, and it was heavily dependent on FPU performance, and I believe GW2 is as well (if memory serves me correctly). 

 

Speaking of IPC, has anything came out recently to solidify exactly where Zen's IPC is at? Last I checked, it was still somewhere around Haswell/Broadwell, but I wouldn't mind something more concrete. 

[79wjd]

4 minutes ago, MageTank said:

I can't speak for battlefield or that star wars game they benched (I don't have either one) but yes, gaming can show weak FPU performance, especially MMO's. I used to play an older MMO called Perfect World, and it was heavily dependent on FPU performance, and I believe GW2 is as well (if memory serves me correctly). 

 

Speaking of IPC, has anything came out recently to solidify exactly where Zen's IPC is at? Last I checked, it was still somewhere around Haswell/Broadwell, but I wouldn't mind something more concrete. 

AMD showed a Cinebench run against a 6900k iirc....that should give a decent idea, no? 

 

10 minutes ago, Ramaddil said:

Well to be honest I am looking at this CPU in particular for my server, Gaming doesn't matter to me at all for this CPU.  My server does encoding, backups, security cameras, and trans-coding in plex.  More cores and threads for me is better for a lot of reasons.  While it is likely given what people are saying it probably wont beat intel's $1000 CPU it will surely beat the AMX FX 8320 that is in my server now.  I used that CPU because my main was upgraded not because I bought it for it. 

 

So given individual use cases it wont be good for everyone it will be good for some no matter what, and cheaper than Intels 8c/16th option.

I agree 100%. Even the FX chips still had their niche use cases. The question is, in what cases will Zen be good and in what scenarios will it be lackluster (if at all). Those aren't questions that can be answered until after some review samples have gone  out.  

[Ramaddil]

I also think in the long and short of it, some people don't need a crazy expensive cpu to play most games at 1080 anyway.  I honestly have no desire to play in 4k at this time.  I think a lot of us are now just doing more than gaming anyways or will soon be doing it more in the future.  I personally try to future proof as much as I can given the amount of money I have.  Instead of building a machine that can just do gaming i try to build a machine that is good in multiple areas. 

 

I think some competition is a good thing for all PC builders, options breed competition which in turn gives way to further advancement that benefits the consumer and the manufacturer.  

[MageTank]

34 minutes ago, Ramaddil said:

I also think in the long and short of it, some people don't need a crazy expensive cpu to play most games at 1080 anyway.  I honestly have no desire to play in 4k at this time.  I think a lot of us are now just doing more than gaming anyways or will soon be doing it more in the future.  I personally try to future proof as much as I can given the amount of money I have.  Instead of building a machine that can just do gaming i try to build a machine that is good in multiple areas. 

 

I think some competition is a good thing for all PC builders, options breed competition which in turn gives way to further advancement that benefits the consumer and the manufacturer.  

It would honestly be more difficult on the CPU at 1080p than what it would at 4k. A slower CPU runs better at 4k because you end up with less frames to have to process for the GPU. Running at 1080p, especially with a high end GPU, you shift the bottleneck towards the CPU. This is why we still don't know Ryzen's relative gaming performance, because we've only seen it at 4k (which most CPU's are capable of handling just fine). 

[vanished]

2 hours ago, djdwosk97 said:

No idea, but pcper was saying something along the lines of AVX being doable in 1 cycle on Kaby Lake/modern Intel CPUs while it would take several on Zen.

 

2 hours ago, Dabombinable said:

Games, CAD and similar. AKA, if Ryzen has a pathetic FPU design like their CMT architectures, and their older ones-they'll probably have to bring back the 3DNow! instruction set if they want the CPU to be used in more than basic office machines.

 

rip  

[79wjd]

4 minutes ago, Ryan_Vickers said:

 

 

rip  

 

7 hours ago, djdwosk97 said:

Yay Bulldozer 2.0!

Intel4lyfe

[PianoPlayer88Key]

6 hours ago, MageTank said:

It would honestly be more difficult on the CPU at 1080p than what it would at 4k. A slower CPU runs better at 4k because you end up with less frames to have to process for the GPU. Running at 1080p, especially with a high end GPU, you shift the bottleneck towards the CPU. This is why we still don't know Ryzen's relative gaming performance, because we've only seen it at 4k (which most CPU's are capable of handling just fine). 

That's what I've thought, a better GPU helps with better graphics quality at a given resolution & frame rate, whereas a better CPU helps with high frame rates when your resolution & details are turned down.  (Correct me if I'm wrong.)

 

Something I've been wondering about ...

 

Let's say someone wants to go crazy high on their FPS, and they're willing to turn down game settings to achieve it.  For example, run everything on lowest settings, and lowest resolution.  (For example, in CS:GO it's 640x480, and in Rise of the Tomb Raider it's 800x600.)  They have something like SLI Titan XPs (or 4x SLI Titan XMs), or at least SLI GTX 1080 Hybrids.  (Maybe even SLI Quadro GP100s.)

 

In the above mentioned games (CS:GO, or RotTR), what CPU would be necessary such that it would not bottleneck those GPUs at lowest resolution/settings?  Like, say you want to run an older / less-demanding game at the 343k fps that slow mo guys filmed a video at, with a sufficiently powerful GPU (like SLI GV110 volta cards) in an older game like CS 1.6.   Or any newer but less demanding one, as long as it's not engine-capped - I tried Rocket League and it wouldn't go over 250 fps.

 

I ran some brief tests on my laptop (i7-6700K & GTX 970M) and the abridged results are: (open the spoiler)

Spoiler

 

CS:GO, map Dust II, vs bots, spectating:

 

1920x1080, lowest: ~540fps looking @ sky, ~220fps looking over map

1920x1080, highest: ~475fps sky, 175fps overview

640x480, highest: 410fps sky, 260fps overview

640x480, lowest: 480fps sky, 300fps overview

 

RotTR, built-in benchmark:

 

1920x1080, highest: overall 17.84, max 47.27 in Geothermal Valley

800x600, highest: overall 46.95, max 99.27 in Mountain Peak

800x600, lowest: overall 162.68, max 329.21 in Mountain Peak

1920x1080, lowest: overall 102.33, max 187.51 in Mountain Peak

 

 

 

It looks like I'm running into a limit in CS:GO (either an engine, or cpu limit).  Also there's an issue where in fullscreen mode, it caps my FPS at 60, even though vsync is off and xbox game dvr is disabled in registry.

 

[porina]

13 hours ago, MageTank said:

I agree. However, I personally think that aspect will be overclocking. Let's be real. 8c/16t at 95w is one thing, but to pretend a 65w SKU exists with the same thread count? The only Intel SKU's that match those numbers, are sub-2ghz on the clock speed. My fear is the IMC will be vastly crippled compared to Intels, as we saw with the FX series. 

 

I just can't imagine it would be FPU, as the benches we've seen (even if they were only 4k) seemed to be quite good, unless the FPU performance is simply "good enough" for gaming. The truth is, we won't really know until reviewers get their hands on them and test this for us. 

To me the pieces are falling together, even if still not confirmed.

 

When I think of heavy lifting FP, I'm thinking AVX and FMA more than old x87. Earlier slides on Zen architecture had mostly showed quite simply they weren't implementing as wide a FP unit as Intel. My reading of this was that the execution potential of Zen FPU would be about half an Intel one. This was reconfirmed in the more recent ISSCC slides and the PCPer discussion. This is still an update from current as it will be one per core, and they may still hold in this area if they can offer roughly double the cores vs. Intel at a price level.

 

This also ties in to the power consumption. Intensive use of those instructions is one of the use cases to get Intel CPUs really hot. If the FPU isn't so strong, they wouldn't require as much power and could specify a lower TDP for a comparable core configuration.

 

My prediction is: for general use it will offer more bang per buck than Intel. For FP intensive uses that wont be the case, but wont be as bad as current. The difference between use cases will lead to internet arguments for years to come.

[captain cactus]

13 hours ago, MageTank said:

I agree. However, I personally think that aspect will be overclocking. Let's be real. 8c/16t at 95w is one thing, but to pretend a 65w SKU exists with the same thread count? The only Intel SKU's that match those numbers, are sub-2ghz on the clock speed. My fear is the IMC will be vastly crippled compared to Intels, as we saw with the FX series. 

 

I just can't imagine it would be FPU, as the benches we've seen (even if they were only 4k) seemed to be quite good, unless the FPU performance is simply "good enough" for gaming. The truth is, we won't really know until reviewers get their hands on them and test this for us. 

This might be the reason for a 65W 8-core CPU:

 

 

If its starting voltage is only 0.9V then yes, a 65W 8c16t CPU isn't entirely impossible. 65W at 0.9v is just 72A for a stock clock of 3.6GHz.

[JoeCoke]

18 hours ago, Mr.Meerkat said:

But with the pricing of the pentium G4560, dual cores aren't that likely to die with it almost costing half as much...

A man can dream.

[MageTank]

6 hours ago, PianoPlayer88Key said:

-snip-

 

 

52 minutes ago, porina said:

-snap-

 

37 minutes ago, lots of unexplainable lag said:

-snoople-

Nothing I like more, than to go to sleep and wake up with a ton of great information to digest. You guys are great, lol. This does explain a lot about these rumored Zen SKU's, and actually eases my mind quite a bit. Perhaps we might see some real competition after all. 

[captain cactus]

Updated op with a new price sheet with all the Ryzen CPUs on it as well as the Intel pricing equivalent.

[Sakkura]

Those 6-core R5s are looking good. Heck the whole lineup looks good if per core performance is decent. 

[leadeater]

On 2/11/2017 at 4:49 AM, porina said:

To me the pieces are falling together, even if still not confirmed.

 

When I think of heavy lifting FP, I'm thinking AVX and FMA more than old x87. Earlier slides on Zen architecture had mostly showed quite simply they weren't implementing as wide a FP unit as Intel. My reading of this was that the execution potential of Zen FPU would be about half an Intel one. This was reconfirmed in the more recent ISSCC slides and the PCPer discussion. This is still an update from current as it will be one per core, and they may still hold in this area if they can offer roughly double the cores vs. Intel at a price level.

 

This also ties in to the power consumption. Intensive use of those instructions is one of the use cases to get Intel CPUs really hot. If the FPU isn't so strong, they wouldn't require as much power and could specify a lower TDP for a comparable core configuration.

 

My prediction is: for general use it will offer more bang per buck than Intel. For FP intensive uses that wont be the case, but wont be as bad as current. The difference between use cases will lead to internet arguments for years to come.

 

If this is to be believed FP performance is much better than it would look simply going by an architecture diagram.

1700X:

 

Then there is this which is FANTASTIC if you run a PS2 emulator on your PC .

Source almost has zero credibility so you know, better to wait and see: http://wccftech.com/amd-ryzen-7-1700x-389-8-core-cpu-benchmarks-leaked/

 

Also AMD is pretty hot on pushing a lot of FP tasks to GPUs using OpenCL meaning if AMD can get some really helpful tools out to developers FP tasks that would be weak on Ryzen could be pushed to GPU.

 

Edit:

OpenCL 2.1[edit]

The ratification and release of the OpenCL 2.1 provisional specification was announced on March 3, 2015 at the Game Developer Conference in San Francisco. It was released on November 16, 2015.^[34] It replaces the OpenCL C kernel language with OpenCL C++, a subset of C++14. Vulkan and OpenCL 2.1 share SPIR-V as an intermediate representation allowing high-level language front-ends to share a common compilation target. Updates to the OpenCL API include:

• Additional subgroup functionality
• Copying of kernel objects and states
• Low-latency device timer queries
• Ingestion of SPIR-V code by runtime
• Execution priority hints for queues
• Zero-sized dispatches from host

AMD, ARM, Intel, HPC, and YetiWare have declared support for OpenCL 2.1.^[35][36]

https://en.wikipedia.org/wiki/OpenCL#OpenCL_2.1

 

Looking at this old discussion, while we may have addressed most of the issues raised it might still not happen yet, OpenCL usage in games. The problem I see is actually Nvidia and poor async compute and HSA. Even with help from AMD the effort to do it vs the target audience isn't great.

https://www.reddit.com/r/gamedev/comments/2aabpc/why_dont_more_games_use_gpgpu_opencl/

 

[captain cactus]

New pricing in the OP showing converted prices to $ without a 15% import tax.

 

AMD's going balls deep with this. It's brilliant.

[vitor_cut]

I just saw this at OC3D. it should shake things in cpu market

 

https://www.overclock3d.net/news/cpu_mainboard/clock_speeds_and_pricing_have_been_leaked_for_amd_s_entire_ryzen_lineup/1

 

Quote

Clock speeds and pricing have been leaked for AMD's entire Ryzen lineup

 

Clock speeds and pricing have been leaked for AMD's entire Ryzen lineup, from high-end 8-core processors all the way low-end quad-cores. 

 

These new Chinese prices and CPU clock speeds are included below, though it is worth noting that the previously leaked R7 1800 model was missing from the list.  

 

We have converted the prices included in this leak to USD, taking into account the current exchange rate and China's 15% sales tax. Wh have compared these prices to previously leaked US prices, with end prices being very similar. 

 

 

 

AMD Ryzen CPU	Core Count	Thread Count	Base Freq	Boost Clock Speed	Leaked Chinese prices (Yuan)	Converted US Prices	Previous US Price leaks
R7 1800X	8	16	3.6GHz	4.0GHz	4399	$555.83	$490.29
R7 1800	8	16	-	-	-		-
R7 1700X	8	16	3.4GHz	3.8GHz	3199	$404.21	$381.72
R7 1700	8	16	3.0GHz	3.7GHz	2599	$328.39	$316.59
R5 1600X	6	12	3.3GHz	3.7GHz	1999	$252.58	-
R5 1500	6	12	3.2GHz	3.4GHz	1799	$227.31	-
R5 1400X	4	8	3.5GHz	3.9GHz	1599	$202.04	-
R5 1300	4	8	3.3GHz	3.6GHz	1399	$176.76	-
R3 1200X	4	4	3.4GHz	3.8GHz	1199	$151.50	-
R3 1100	4	4	3.2GHz	3.5GHz	999	$126.22	-

 

 While these prices are not the same as previously leaked Ryzen pricing, though these prices are within the same ball park. It is especially interesting to see these new prices for AMD's lower-end Ryzen CPUs, with AMD's R3 1200X and R3 1100 costing than Intel's i3 7350K?

 

One thing that also must be remembered is that all of AMD's Ryzen SKUs are overclockable, giving enthusiasts a great opportunity to achieve great performance/$ out of their Ryzen CPUs, especially when using lower-end CPUs. 

These priced will likely force Intel to reconsider the pricing of their Kaby Lake CPU lineup, as it seems like AMD will greatly upset the status-Quo.