Ivy Bridge-E or what happens when there's no competitions

[NTMBK]

Arcording to Glenn Hinton. Hyperthreading added less than 5% to the core size.

Example from the P4.

Cool, thanks. Given that an A7 core is ~25% the area of an A15, SMT looks the more efficient way. It'll be interesting to see some benchmarks of the "octacore" when it comes out, I want to see how the two approaches measure up.

 

[colonelciller]

You do know that CPUs have never been so cheap in history as they are now? if you want to see price manipulation and milking, turn the clock back to 2004-2006 and look on K8 prices. Specially X2.

That line of logic is missing the point completely. Market exploitation from 2004-2006 cannot be used to justify market exploitation/manipulation by Intel in 2013.

This isn't about history, this is about what is occurring in the CPU market today, particularly what is going on in the Xeon (Workstation) product line.

In that market Intel is exploiting its monopoly position, and it is exploiting it in a way that is very bad for the consumer.

 

[colonelciller]

Take a look at the CPUs in the above image, compare the specs:

• Xeon E5-2643
• Xeon E5-1620

The specs for the two Xeons are virtually indistinguishable, yet the E5-2643 is nearly quadruple the price of the E5-1620.

• E5-2643: Are users allowed to use the CPU in a dual-socket configuration? YES
• E5-1620: Are users allowed to use the CPU in a dual-socket configuration? NO

Intel allows customers to install 2 CPU's in a single motherboard. For that allowance customers have the "privilege" of paying nearly quadruple the price... times 2. We should all be grateful.

 

[ShintaiDK]

That line of logic is missing the point completely. Market exploitation from 2004-2006 cannot be used to justify market exploitation/manipulation by Intel in 2013.

This isn't about history, this is about what is occurring in the CPU market today, particularly what is going on in the Xeon (Workstation) product line.

In that market Intel is exploiting its monopoly position, and it is exploiting it in a way that is very bad for the consumer.

How can they exploit if its never been so cheap ever in the entire history of man? What should Intel charge for a 3570K for example in your eyes for not being exploitive?

 

[crashtech]

Hey, I want a dual CPU rig on the cheap, too. I also want a new Jag. There's still no point being made. Users that really need that many cores are not average consumers, and have to pay to play. So what if it's what you term an artificial distinction? In my view, the folks that can pay obscene prices for CPUs just help make the cheaper ones even cheaper, because they've already done the heavy lifting of paying for Intel's R&D...

 

[colonelciller]

How can they exploit if its never been so cheap ever in the entire history of man? What should Intel charge for a 3570K for example in your eyes for not being exploitive?

Once again, this is not about the history of mankind and the distraction of history has absolutely nothing to do with intentional actions in the marketplace of 2013.

How could it be exploitative?

Simple. There are users of heavily threaded applications that need to use multi-CPU configurations in order to have minimally tolerable performance of those applications.

Intel understands this very well, and there is nobody other than Intel who can serve this market in 2013.

Please take a look at the specs I just posted and attempt to justify the nearly-quadruple price difference in those two Xeon CPUs.

 

[colonelciller]

Hey, I want a dual CPU rig on the cheap, too. I also want a new Jag. There's still no point being made. Users that really need that many cores are not average consumers, and have to pay to play. So what if it's what you term an artificial distinction? In my view, the folks that can pay obscene prices for CPUs just help make the cheaper ones even cheaper, because they've already done the heavy lifting of paying for Intel's R&D...

please take a look at the specs I just posted for the two Xeon CPUs.
Please let me know if I've left out some important differences.

the quadruple price difference is absurd.
Intel has the market by the b@lls

 

[crashtech]

Once again, this is not about the history of mankind and the distraction of history has absolutely nothing to do with intentional actions in the marketplace of 2013.

How could it be exploitative?

Simple. There are users of heavily threaded applications that need to use multi-CPU configurations in order to have minimally tolerable performance of those applications.

Intel understands this very well, and there is nobody other than Intel who can serve this market in 2013.

Please take a look at the specs I just posted and attempt to justify the nearly-quadruple price difference in those two Xeon CPUs.

You answered your own question.

 

[crashtech]

please take a look at the specs I just posted for the two Xeon CPUs.
Please let me know if I've left out some important differences.

the quadruple price difference is absurd.
Intel has the market by the b@lls

Yes they do. I agree totally with your assessment. Do you have a solution? Is this becoming a political stance?

 

[Charles Kozierok]

The specs for the two Xeons are virtually indistinguishable, yet the E5-2643 is nearly quadruple the price of the E5-1620.

• E5-2643: Are users allowed to use the CPU in a dual-socket configuration? YES
• E5-1620: Are users allowed to use the CPU in a dual-socket configuration? NO

Intel allows customers to install 2 CPU's in a single motherboard. For that allowance customers have the "privilege" of paying nearly quadruple the price... times 2. We should all be grateful.

Oh man, this is taking me back to the "outrage" over the 486SX in the 1990s. What's old is new again indeed.

It's called market segmentation. If it makes you feel better, think of it as the E5-1620 being discounted rather than the E5-2643 being marked up.

Also, the two chips are not really nearly identical. One uses DMI, the other uses QPI. I'd imagine the qualification/verification on the dual chip is much more involved as well.

 

[_Rick_]

IB-E/EP/EX dies comes in 6, 10 and 15 cores.

According to my information, that's not entirely correct. The MDDS states that there is an EP-6 and an EP-10, as well as an EX-15. If those are indeed the core counts, then there will be an EP with 6 cores. Whether this will be used for all desktop chips is not yet clear.
Supposed engineering sample CPU-z screen shots, show 8 cores.

 

[colonelciller]

Yes they do. I agree totally with your assessment. Do you have a solution? Is this becoming a political stance?

I think the solution is for more people to become aware of the issue.
It really does affect many people who aren't using these machines for business purposes. It just something I figured out for myself when attempting to build a decent computer for working with VUE (eon software).

I thought I'd share my thoughts is all.
My hope is a true competitor comes to the market.

 

[colonelciller]

It's called market segmentation. If it makes you feel better, think of it as the E5-1620 being discounted rather than the E5-2643 being marked up.

I suppose I should be happy for the discount then since I'll be installing an E5-1620 as a solo-chip in a dual socket mobo... hoping to be able to afford an upgrade in the future.

 

[crashtech]

My hope is a true competitor comes to the market.

You are certainly not alone in that. Things were much more lively when AMD managed to take the lead while Intel was bogged down in their Netburst fiasco. I'm an admitted opportunist and will jump ship the instant I perceive AMD has a superior product in my price range!

 

[exar333]

They launched Sandy Bridge E ahead of its accompanying Xeons, and what happened? We got a not entirely functional implementation of PCIe3 on first gen boards. If this delay means that IB-E chipsets get the same level of verification and testing that their Xeon counterparts get, I am A-OK with that.

THIS.

Do it right.

 

[ShintaiDK]

According to my information, that's not entirely correct. The MDDS states that there is an EP-6 and an EP-10, as well as an EX-15. If those are indeed the core counts, then there will be an EP with 6 cores. Whether this will be used for all desktop chips is not yet clear.
Supposed engineering sample CPU-z screen shots, show 8 cores.

Intel sells 6 cores today, yet no 6 core dies. It comes from 8 core dies.

8 cores on IB comes from 10 core dies.

 

[ShintaiDK]

Oh man, this is taking me back to the "outrage" over the 486SX in the 1990s. What's old is new again indeed.

It's called market segmentation. If it makes you feel better, think of it as the E5-1620 being discounted rather than the E5-2643 being marked up.

Also, the two chips are not really nearly identical. One uses DMI, the other uses QPI. I'd imagine the qualification/verification on the dual chip is much more involved as well.

Please dont use logic. We need 10000$ BMW M3/M5 as well. Anything else would be exploitation. And 100$ Titan cards too!

 

[Kougar]

Not to mention that this means that Haswell EP also gets pushed back, probably by another year or so.

I doubt it, they are already pretty far behind the curve... Intel appears to be sticking to a predictable 3-year timetable for its enthusiast platform in general.

X58 - Nehalem: Q4'08-Q4'11
X79 - SB-E Q4'11-Q4'13

X79 refresh with fixed southbridge - Q4'13 - Q4'14?
X99 - Haswell-E+? Q4'14-???

So Haswell-E will be 1.5 years behnid Haswell as it stands. If they waited another year they are going to have to skip two generations just to catch back up, and I highly doubt they want to fall THAT far behind with their Xeon architecture versus the mainstream consumer architecture. It'd just look BAD. Or well, worse than it already does

• Xeon E5-2643
• Xeon E5-1620

The specs for the two Xeons are virtually indistinguishable, yet the E5-2643 is nearly quadruple the price of the E5-1620.

• E5-2643: Are users allowed to use the CPU in a dual-socket configuration? YES
• E5-1620: Are users allowed to use the CPU in a dual-socket configuration? NO

Intel allows customers to install 2 CPU's in a single motherboard. For that allowance customers have the "privilege" of paying nearly quadruple the price... times 2. We should all be grateful.

So, AMD is equally guilty of this too then? Remember AMD scaled up prices exactly the same depending on how many sockets the Opterons were designed to share on a board.

​

AMD's margins on their dual core Opteron parts are huge. On average, the second core costs customers over 3x as much as the first core for any of these CPUs. As you will soon see, the performance benefits are definitely worth it, but know that AMD's pricing is not exactly designed to drive dual core into widespread adoption. Source

It was and still is industry standard practice. Not to mention the more interlinks built into a chip the more it is going to cost, and in your example one chip has no QPI interlink whatsoever (Just DMI to the chipset) and the other has dual QPI links. So it is not even the same arch!