AMD rumoured to be creating consumer oriented 16 Core/32 Thread Ryzen CPU

A 200w TDP would indicate 16 cores at ~4.0 Ghz, bring that shit ON. 16 cores at 4.0 would destroy just about any HEDT product on the market.
 
to say that the HEDT has no market outside enthusiasts is short sighted
No. Its just true is all. Outside of the professional market how much software can make use of 16C/32T? Pretty much none. Even 8C/16T is a bit ahead of the curve right now.

that keep potential buyers out of some of the more costly Xeon markets
That is what Opterons/whatever AMD decides to call server Zen is for. Its already a given that AMD will be competing on price there so pro looking to build a home workstation would be easily enticed to buy AMD this time around. LED RGB lighting certainly won't do it for them.

One of the reasons AMD would want in on this market is because it's quite big
LOL no. The workstation market is tiny and has been for a long time. Its more of a prestige/marketing thing + easy money for AMD or Intel to pursue since its just a rebrand/rebinned server part and server platform with overclocking enabled. They probably won't sell more than a few tens of thousands (vs millions of PC's) per year.
 
I think AMD is anticipating Intel's new up coming HEDT for late this year or early next, Purley, here. You're right that their current high end CPU are targeted at Intel's current HEDT of course.

Agree. This seems to be HEDT competitor for Skylake-X
 
Except that CanardPC claims the clocks are 3.1GHz.

Canard also claims that the chips have a 150w TDP. So that puts them at about 75w per CCX without taking into ccount efficiency loss in MCM fabric, which there surely will be. Pretty amazing stuff! An unlocked 16 core chip with broadwell IPC is a power-user's dream come true!
 
Canard also claims that the chips have a 150w TDP. So that puts them at about 75w per CCX without taking into ccount efficiency loss in MCM fabric, which there surely will be. Pretty amazing stuff! An unlocked 16 core chip with broadwell IPC is a power-user's dream come true!

No. 150W was the marketing TDP for the first gen sample with 2.4GHz base clocks. 180W is the marketing TDP for the second gen sample with 3.1GHz base clocks. And CanardPC claims the real TDP is higher than 200W.
 
Given the target frequency for a system like this it will be competitive with Intel... power is a number one priority for AMD right now.

Wait till it launches and then bitch.
You misunderstand.
Not bitching as I agree with lolfail that the CPC figure may be a bit unusual.
Also their CPU power figures for R7 did not match some other sites and was higher, hence why I mentioned maybe they also adjusted the TDP figure for the 16C to correlate with their own R7 benchmarks.

Cheers
 
Last edited:
two 90W dies plus 20W for the package explain the ">200W" claimed by CanardPC.
Except 1700 is not 90W in practice, simple as that. 20W for the package is fair enough (well, overshooting maybe, but w/e), so we land with about 160-170W TDP.
 
No. Its just true is all. Outside of the professional market how much software can make use of 16C/32T? Pretty much none. Even 8C/16T is a bit ahead of the curve right now.

Content creation, encoding, many simultaneous VMs, rendering, etc can make use of that many cores easily. There are plenty of non-professionals who do that stuff. Few will want to shell out $1000 for a CPU but for those who do it would certainly be worth it.
 
200w ? And just think...people use to be happy with 220w 9590.....This should be leaps & bounds faster for the same or less TDP.
 
AMD needs to quit fucking around and release a price-conscious industry standard form factor SMP dual socket Ryzen platform into the consumer and enterprise segments a la Athlon MP.

They would see some record CPU sales while putting an actual measurable dent in Intel's market share grip.

A cheap dual 1700 setup would be totally killer for enthusiasts and businesses alike...and even better with the capability to go dual 16c/32t.

IMO, AMD dropped the ball in regards to this. We're at a point where we're just starting to see >4c/8t being formally utilized, thanks to Intel having 6c/12t, 8c/16t, and 10c/20t enthusiast-focused processors in the consumer segment for a while now...and they sure aren't going to stop increasing core count with each new generation of X-chipset.

AMD needs to act with some semblance of aggressive competition...and fast.
 
AMD needs to quit fucking around and release a price-conscious industry standard form factor SMP dual socket Ryzen platform into the consumer and enterprise segments a la Athlon MP.

They would see some record CPU sales while putting an actual measurable dent in Intel's market share grip.

That doesn't ring true. Dual sockets for enthusiasts is likely somewhere between 0% and 1% of the market for enthusiast computers. I would bet closer to 0%. It won't result in record sales, it is pretty much irrelevant.
 
That doesn't ring true. Dual sockets for enthusiasts is likely somewhere between 0% and 1% of the market for enthusiast computers. I would bet closer to 0%. It won't result in record sales, it is pretty much irrelevant.

Think back to the P3 Coppermine and Athlon MP days: these were consumer grade processors with SMP capability right out of the box. No need to pay exorbitant price tags on specialty SMP processors carrying namesakes such as Xeon or Opteron reserved for the deepest pockets and wealthiest enterprise customers.

This is something I feel AMD should capitalize on now in order to gain market share in all segments...lower volume consumer and higher volume bread-and-butter business/enterprise (i notice you didn't even retort about the business segment).

It would be a clear advantage.
 
Think back to the P3 Coppermine and Athlon MP days: these were consumer grade processors with SMP capability right out of the box. No need to pay exorbitant price tags on specialty SMP processors carrying namesakes such as Xeon or Opteron reserved for the deepest pockets and wealthiest enterprise customers.

This is something I feel AMD should capitalize on now in order to gain market share in all segments...lower volume consumer and higher volume bread-and-butter business/enterprise (i notice you didn't even retort about the business segment).

It would be a clear advantage.

That is living in the past. This was a thing a decade or so back, because CPUs were essentially single core.

It stopped being a thing when quad cores were common. Even less of an issue with AMD selling 8 core CPUs.
 
This is something I feel AMD should capitalize on now in order to gain market share in all segments...lower volume consumer and higher volume bread-and-butter business/enterprise .

It would be a clear advantage.

That is living in the past. This was a thing a decade or so back, because CPUs were essentially single core.

It stopped being a thing when quad cores were common. Even less of an issue with AMD selling 8 core CPUs.

I dunno, given where we are headed from an IPC perspective (limited changes) and from a frequency perspective (I'm not expecting much) core count is likely all that's left in increasing overall compute resources. Given where AMD has lead and Intel is following along (or at least traveling in parallel) we know it too.

Dejawiz could be correct in that simply replicating that concept of cheap multi socket systems could be something that drives sales for one of the chips vendors.

On the other hand, if I have a modern multi socket system I want the Naples style memory support and connectivity. Dual RAM Channels and 24 lanes of PCIe aren't exactly making my heart flutter over here. Even what this HEDT platform is promising with RAM limited to 128 Gigs isn't doing me a lot of favors.
 
I dunno, given where we are headed from an IPC perspective (limited changes) and from a frequency perspective (I'm not expecting much) core count is likely all that's left in increasing overall compute resources. Given where AMD has lead and Intel is following along (or at least traveling in parallel) we know it too.

Dejawiz could be correct in that simply replicating that concept of cheap multi socket systems could be something that drives sales for one of the chips vendors.

On the other hand, if I have a modern multi socket system I want the Naples style memory support and connectivity. Dual RAM Channels and 24 lanes of PCIe aren't exactly making my heart flutter over here. Even what this HEDT platform is promising with RAM limited to 128 Gigs isn't doing me a lot of favors.

That's precisely my point...if Ryzen came out and completely dominated every mainstream Intel i5 and i7 (as in, at least a 10% performance advantage across the board, even against the 7700K and perhaps even into X-platform offerings), then I'd be singing a different tune. But since real-world performance is what it is (close to Intel in many metrics, but overall a bit lacking, especially when looking at comparative single-threaded IPC), I feel that the opportunity for affordable SMP is a golden one.

In multi-threaded productivity testing, Ryzen hangs with/surpasses even the 8c/16t 6900K...throw a dual socket Ryzen 1700 (16c/32t - $330 cpu1, $330 cpu2, $300+ MoBo = $960+) against an i7-6950X (10c/20t - $1650 cpu, $250+ MoBo = $1900+), and it becomes rather obvious which platform will win the hearts of most buyers.

Heck, we can even compare the pricing to a "high-end" 7700K build (4c/8t - $350 cpu, $300+ MoBo = $650+)...now take the Ryzen dual socket build with only 1 cpu (8c/16t - $330 cpu1, $0 no cpu2, $300+ MoBo = $630+)...Ryzen still wins in price while coming damn close to the overall performance, but has a blatant massive advantage of being able to accept a second cpu. Which platform option has more appeal at this price segment, whether it's in a consumer sense or a business/enterprise one?
 
Some people still use a Horse and Buggy to get around. That doesn't mean Ford and GM should start producing Horse Buggies to boost their sales.

It's a fringe activity in the consumer market, where a 4 core CPU is overkill for most people.

And you're still negating to talk about the most important aspect: high volume business/enterprise segment SMP offerings that are directly available and affordable in the consumer segment; the same that gave SMP-capable success to past P3 Coppermine and Athlon MP offerings in BOTH the business and consumer segments.
 
And you're still negating to talk about the most important aspect: high volume business/enterprise segment SMP offerings that are directly available and affordable in the consumer segment; the same that gave SMP-capable success to past P3 Coppermine and Athlon MP offerings in BOTH the business and consumer segments.

And again, you are referencing the past when CPUs had ONE CORE. So people/business would have to pay to go multi-socket, to do any multi-processing.

Now with 8 core CPUs, the need for this has all but evaporated except in highly specialized markets.
 
Last edited:
And again, you are referencing the past when CPUs had ONE CORE. So people/business would pay to have to go multi-socket to do any multi-processing.

Now with 8 core CPUs, the need for this has all but evaporated except in highly specialized markets.

Just to also add.
Dual socket will not necessarily take-off that quickly for business/enterprise use generally.
Primarily because there are still many general applications that are not NUMA aware and programmed that way, meaning they just use a single processor socket.
But that is more than fine if we have 12C and 16C single socket processors as they would potentially be popular due to AMD's pricing compared to Intels and importantly these SKUs deliver on performance/reasonable TDP - although these will still not be mainstream for business/enterprise compared to 8C/16T.

Cheers
 
And again, you are referencing the past when CPUs had ONE CORE. So people/business would have to pay to go multi-socket, to do any multi-processing.

Now with 8 core CPUs, the need for this has all but evaporated except in highly specialized markets.

I see your point. Reality of it is, laptops with 2-4 cores are the leaders of volume sales right now, regardless of the PC market segment being analyzed.
 
My first 12 core workstation (seven years old) is dual socket as the XEONs of that era were 6 core. My second 12 core workstation (four years old) is single socket as the XEON used is an actual 12 core part. With AMD and Intel offering even larger core configurations, it really just comes down to how many cores do you need and does it make sense that the cores are in the same motherboard instead of being network linked.

I wouldn't say multi-socket configurations are dead but they're less relevant for the average business when multi core CPUs exist in larger configurations than needed for all but the largest workloads.
 
My first 12 core workstation (seven years old) is dual socket as the XEONs of that era were 6 core. My second 12 core workstation (four years old) is single socket as the XEON used is an actual 12 core part. With AMD and Intel offering even larger core configurations, it really just comes down to how many cores do you need and does it make sense that the cores are in the same motherboard instead of being network linked.

I wouldn't say multi-socket configurations are dead but they're less relevant for the average business when multi core CPUs exist in larger configurations than needed for all but the largest workloads.

Quick question: if there was a very cost-effective way (let's say $500-700 total for new dual socket MoBo and 2nd matching CPU) for you to go dual processor with your most recent 12c/24t Xeon for a total of 24c/48t, would you?
 
Quick question: if there was a very cost-effective way (let's say $500-700 total for new dual socket MoBo and 2nd matching CPU) for you to go dual processor with your most recent 12c/24t Xeon for a total of 24c/48t, would you?

It would have to depend upon your applications and whether they are NUMA aware and programmed to be.
Nearly everyone who has ever done a review of 2S XEON have all mentioned they have some applications they need to use that can only use one of the sockects and processors, so half the resources they paid for.
So from a cost/performance perspective, it makes more sense to go with 1 socket solution unless you know all the applications you use can work fully with dual socket architecture.

Cheers
 
Quick question: if there was a very cost-effective way (let's say $500-700 total for new dual socket MoBo and 2nd matching CPU) for you to go dual processor with your most recent 12c/24t Xeon for a total of 24c/48t, would you?

Yes. My workload is highly threaded (custom ffmpeg strings for video encoding). As long as the work is there, all cores will be loaded. I'd love to see a 16c/32t (rumored Naples) x 2 in my future or Intel equivalent. :happy:
 
It would have to depend upon your applications and whether they are NUMA aware and programmed to be.
Nearly everyone who has ever done a review of 2S XEON have all mentioned they have some applications they need to use that can only use one of the sockects and processors, so half the resources they paid for.
So from a cost/performance perspective, it makes more sense to go with 1 socket solution unless you know all the applications you use can work fully with dual socket architecture.

Cheers

I see the same argument being made over single socket 4c being all that's necessary because more than that doesn't get utilized to its extent, if at all. Yet, here Ryzen is...with all three initial release models having 8c, supplementing Intel 6c, 8c, and 10c offerings.

Why release it if only half the cores will be used?

Or is it ultimately better to have it now and see eventual future use, than not have it and suffer a costly upgrade sooner rather than later when it can be used?
 
Yes. My workload is highly threaded (custom ffmpeg strings for video encoding). As long as the work is there, all cores will be loaded. I'd love to see a 16c/32t (rumored Naples) x 2 in my future or Intel equivalent. :happy:

So if you want a rumored Ryzen 16c x dual socket.

Why not a Ryzen 32 Core Single socket. Ryzen 32 core is not even a rumor.

How is
2 socket X 16C,

better than:
1 socket x 32C?
 
AMD-Ryzen-16core-cpu.jpg
 
I see the same argument being made over single socket 4c being all that's necessary because more than that doesn't get utilized to its extent, if at all. Yet, here Ryzen is...with all three initial release models having 8c, supplementing Intel 6c, 8c, and 10c offerings.

Why release it if only half the cores will be used?

Or is it ultimately better to have it now and see eventual future use, than not have it and suffer a costly upgrade sooner rather than later when it can be used?
yeah sure 4 cores are fine, if youre only gaming.
 
I see the same argument being made over single socket 4c being all that's necessary because more than that doesn't get utilized to its extent, if at all. Yet, here Ryzen is...with all three initial release models having 8c, supplementing Intel 6c, 8c, and 10c offerings.

Why release it if only half the cores will be used?

Or is it ultimately better to have it now and see eventual future use, than not have it and suffer a costly upgrade sooner rather than later when it can be used?
Not sure I follow.
How does talking about single core Ryzen and Intel equate to the fact that many general applications out there still cannot utlitise a NUMA dual socket installation?
The NUMA situation is applicable to both Intel and Ryzen and very different to anything discussed to date about Ryzen as a single socket implementation.
As I said it is application dependent, and many reviewers use day-to-day applications that cannot use both processors in a dual setup because of app's lack of awareness.

The average enterprise would never go with dual socket because the level of support required for and also by applications, it has its place but normally more as a server or specific function where the application is aware (and not all rendering apps are as an example).
Most as I said will happily go with the single socket version from both Intel and AMD, which is what I thought the 12C and 16C version was but able to support 2S where needed.
Cheers
 
Quick question: if there was a very cost-effective way (let's say $500-700 total for new dual socket MoBo and 2nd matching CPU) for you to go dual processor with your most recent 12c/24t Xeon for a total of 24c/48t, would you?

Yes I would, swappimg from dual x5670's to something more modern can only be a win, I have everything apart from cheap lga2011 cpu's
 
Back
Top