Desktop Sandy Bridge E-series (SNB-E) / LGA 2011 CPU Pricing and Naming Speculation

[sethk]

We've probably all seen more details from the Intel roadmap regarding Sandy Bridge-E - i.e. quad channel DDR3, 2 X 16x PCIe 2.0 + 1 X 4x for storage, SATA 6Gbs, etc. The launch processors in Q4 2011 are supposed to be:

A) Core i7 Extreme

6c/12t, 3.3GHz/3.9GHz T, 15MB L3, Fully unlocked​

B) Core i7

6c/12t, 3.2GHz, 3.8GHz, 12MB L3, Fully unlocked​

C) Core i7

4c/8t, 3.6GHz / 3.9GHz, 10MB L3, Partially unlocked​

Any guesses as to pricing and naming?

We went from i7 920 / 950 / 980x to i7 2500 / 2600K as a naming convention, so what will the new processors be named, i7 3xxx? or will they use the space in the 2xxx series to call them 2900X, 2800K, 2700? (X for Extreme, K for enthusiast unlocked, and no-suffix for, well, nothing special.) I could also see them using more of the remaining digits, instead of the xx00 naming.

Pricing wise, I'd start the guessing at $999 for extreme, $500 for the middle one and $300 for the "low end" / 4 core one. This will also mean that the full range of regular Sandy Bridge processors will be bumped down, although I can see 2600K and 2700 sharing the same price.

I'm also puzzled by the L3 cache. Clearly on the 4C processor, I could see this being tied to the 2 missing cores, either laser cut / disabled silicon or a different die, but the middle processor and top processor are 15MB / 12MB? That's just strange. 15MB is an odd number, and plus it's strange to have less cache on the K model - I doubt they will be different dies.

 

[needmorecarnitine]

$349~399 for the entry model, $600 for the middle and $1000 for extreme

 

[ChronoDetector]

I wonder how much faster these perform than the i7 970, 980x, etc.

 

[Bbq]

knowing how intel names things, they'll think of a new naming scheme that is even more convoluted and ridiculous, while providing less information and more confusion.

 

[pxc]

knowing how intel names things,

I bet thinking about the current naming scheme didn't really happen.

From Nehalem to SB, Intel has kept somewhat the same conventions:

Pentium G
Core i3
Core i5
Core i7

SB added a 2 before the 3 digit model number on models besides Pentium G, to signify generation 2 of the "Core i" series (post Ivy Bridge Haswell should get a 3). For example: Core i5-2500. The larger last 3 numbers in any series are faster processors, and each higher i model has more features, cache or cores. It's not exactly difficult to understand that i7 is better than i3 or that an i5-2500 is faster than an i5-2300.

I wouldn't be surprised if the i7 name is finally updated on LGA 2011 to represent a new performance level (Core i9?), at least for the 6 core launch model, and later 8 core models.

 

[x509]

knowing how intel names things, they'll think of a new naming scheme that is even more convoluted and ridiculous, while providing less information and more confusion.

What I find interesting is that with all the leaks in general about Intel's processor roadmap, there seems to be no information, err, rumors and leaks about specific models and pricing.

Nor is there any real info or rumors about motherboards to support LGA 2011.

I'm planning to upgrade when LGA is announced, but it's hard to budget when there is no info at all.

 

[defaultluser]

I wonder how much faster these perform than the i7 970, 980x, etc.

Not much, judging by those clockspeeds. Core i5 750 to Core i5 2500 performance increase was around %30, with a %25 boost in clock speed accounting for most of that. Given the much higher clock speeds of the i7 970, 980x (3.2, 3.33) - I see the difference being almost non-existent (%10 or less), unless these things ship with base clocks of 4 GHz.

 

[Raap]

Wouldn't it be safe to assume that the pricing won't actually be decided until Bulldozer's been released?

 

[Threet3n]

Wouldn't it be safe to assume that the pricing won't actually be decided until Bulldozer's been released?

Fair assumption.

I bet thinking about the current naming scheme didn't really happen.

From Nehalem to SB, Intel has kept somewhat the same conventions:

Pentium G
Core i3
Core i5
Core i7

SB added a 2 before the 3 digit model number on models besides Pentium G, to signify generation 2 of the "Core i" series (post Ivy Bridge Haswell should get a 3). For example: Core i5-2500. The larger last 3 numbers in any series are faster processors, and each higher i model has more features, cache or cores. It's not exactly difficult to understand that i7 is better than i3 or that an i5-2500 is faster than an i5-2300.

I wouldn't be surprised if the i7 name is finally updated on LGA 2011 to represent a new performance level (Core i9?), at least for the 6 core launch model, and later 8 core models.

After reading this, I got to thinking, is there a reason that Intel skipped the even numbers (ie. i3, i5, i7)? Also, It would be pretty sweet if they implement an i9 series.
Prices... I would assume that they would be close to the same as the top of the top 1156 processors, i3 corresponding with i3, i7 corresponding with i7, etc.

 

[Bbq]

I bet thinking about the current naming scheme didn't really happen.

From Nehalem to SB, Intel has kept somewhat the same conventions:

Pentium G
Core i3
Core i5
Core i7

SB added a 2 before the 3 digit model number on models besides Pentium G, to signify generation 2 of the "Core i" series (post Ivy Bridge Haswell should get a 3). For example: Core i5-2500. The larger last 3 numbers in any series are faster processors, and each higher i model has more features, cache or cores. It's not exactly difficult to understand that i7 is better than i3 or that an i5-2500 is faster than an i5-2300.

I wouldn't be surprised if the i7 name is finally updated on LGA 2011 to represent a new performance level (Core i9?), at least for the 6 core launch model, and later 8 core models.

Yes, but don't forget that the 500 part of the 2500k doesn't really give you any information. It doesn't tell you that it's 3.3 ghz, that it's 100mhz "fsb", 33x multiplier, 6mb cache, or 95w. It doesn't give me any information except that it's higher than 400, and lower than 600.

Then, don't forget about that there are dual core i7's, quad core i7's, 6 core i7's, and i7's on socket 1366, 1156, 1155, and pga998

How about dual core + HT i5's and quad core i5's on 1156 and 1155

 

[pxc]

It doesn't tell you that it's 3.3 ghz, that it's 100mhz "fsb", 33x multiplier, 6mb cache, or 95w. It doesn't give me any information except that it's higher than 400, and lower than 600.

Any place you would buy an i5-2500 would have that info on the web page or retail box. It's not reasonable to include all details of a part in the regular (friendly) product name. What I'm getting from your statement is that you would prefer what AMD uses in the ordering part number (e.g. ADAccccxyynzb), which isn't the processor name. And you need a secret decoder ring to decode the information if you choose not to look at the info on the ordering page or retail box. Yeah.

Then, don't forget about that there are dual core i7's, quad core i7's, 6 core i7's, and i7's on socket 1366, 1156, 1155, and pga998

How about dual core + HT i5's and quad core i5's on 1156 and 1155

I am a bit disappointed that Intel blurs the dual/quad core model numbers on the i7 and i5 series. For whatever reasons Intel decided to use the scheme it did, that's all we have. Nothing's perfect.

Additionally, I'm not a fan at all of the crazy splits in micro-segmentation these days. The number of available Sandy Bridge processor models is pretty ridiculous so soon after launch.

 

[Bbq]

Any place you would buy an i5-2500 would have that info on the web page or retail box. It's not reasonable to include all details of a part in the regular (friendly) product name. What I'm getting from your statement is that you would prefer what AMD uses in the ordering part number (e.g. ADAccccxyynzb), which isn't the processor name. And you need a secret decoder ring to decode the information if you choose not to look at the info on the ordering page or retail box. Yeah.

I don't prefer the AMD naming scheme, I think it's worse. With AMD, there's also the whole am2, am2+, am3, am3+, and backwards compatable or not stuff.

I just want something that'll give more relevent details.

For example, i3 for dual cores, i5 for quad cores, i7 for quad core + HT, and i9 for hexa cores. The generation 2 thing I can live with, but how about something that tells you the speed of the processor ? i5 2500k is 3.3ghz and i7 2600k is 3.4ghz, so maybe 33 / 34 in there, HT, and a socket differentiation (currently, 1366 is socket B, 1156 is H, 1155 is H2). i5 233K H2 and i7 234K H2 maybe ? i5 23K / i7 24K ?

Mostly it's just stupid that the "500" part doesn't do anything except provide a number to compare it to "600" or "400". At least make it do something useful.

I am a bit disappointed that Intel blurs the dual/quad core model numbers on the i7 and i5 series. For whatever reasons Intel decided to use the scheme it did, that's all we have. Nothing's perfect.

For sure. You could buy an i7 and have it not fit into 2/3 of the available sockets and also have it either be 2, 4, or 6 cores.

Way to go intel.

 

[PGHammer]

I bet thinking about the current naming scheme didn't really happen.

From Nehalem to SB, Intel has kept somewhat the same conventions:

Pentium G
Core i3
Core i5
Core i7

SB added a 2 before the 3 digit model number on models besides Pentium G, to signify generation 2 of the "Core i" series (post Ivy Bridge Haswell should get a 3). For example: Core i5-2500. The larger last 3 numbers in any series are faster processors, and each higher i model has more features, cache or cores. It's not exactly difficult to understand that i7 is better than i3 or that an i5-2500 is faster than an i5-2300.

I wouldn't be surprised if the i7 name is finally updated on LGA 2011 to represent a new performance level (Core i9?), at least for the 6 core launch model, and later 8 core models.

From what I can see, Pentium-G is (as I thought) the Sandy Bridge equivalent of the LGA775 Pentium DualCore, but with onboard graphics (hence the G) and directly replaces the LGA775 part in Intel's price matrix (naturally, this paves the way for a Celeron-G that replaces the Starbucks-pillaging Celeron DualCore); however, would Intel dare introduce more low-buck flamethrowers simply to avoid eating otherwise-unusable parts? (Remember, the DualCores in LGA775 were gimped Core2Duos, and the Pentium-G, and posited Celeron-G, are gimped i3s; however, the Pentium and Celeron DualCores got reputations as the aforementioned mini-Visigoths and took sales away from the genuine C2D. The question is not will Pentium-G, and possible Celeron-G, take sales away from i3, but how many.)

For LGA 2011, because these CPUs will be *above* i7-2600K, you could have i7-3600 at the bottom, i7-4600K (unlocked, but non-EE) in the middle, with i7-4800X at the top (priced identically with the current i7-990X in LGA1366). There will be *no* i5s or i3s in LGA2011 - this year or any other.

 

[PGHammer]

Any place you would buy an i5-2500 would have that info on the web page or retail box. It's not reasonable to include all details of a part in the regular (friendly) product name. What I'm getting from your statement is that you would prefer what AMD uses in the ordering part number (e.g. ADAccccxyynzb), which isn't the processor name. And you need a secret decoder ring to decode the information if you choose not to look at the info on the ordering page or retail box. Yeah.

I am a bit disappointed that Intel blurs the dual/quad core model numbers on the i7 and i5 series. For whatever reasons Intel decided to use the scheme it did, that's all we have. Nothing's perfect.

Additionally, I'm not a fan at all of the crazy splits in micro-segmentation these days. The number of available Sandy Bridge processor models is pretty ridiculous so soon after launch.

Actually, there is a dual-core/HT i5 in 1155 (in fact, it's the only dual-core i5 in that particular LGA) - the Core i5-2390S. The remaining LGA1155 i5s are quad-cores, but lack HTT support.

 

[plugwash]

Yes, but don't forget that the 500 part of the 2500k doesn't really give you any information. It doesn't tell you that it's 3.3 ghz, that it's 100mhz "fsb", 33x multiplier, 6mb cache, or 95w. It doesn't give me any information except that it's higher than 400, and lower than 600.

Then, don't forget about that there are dual core i7's, quad core i7's, 6 core i7's, and i7's on socket 1366, 1156, 1155, and pga998

How about dual core + HT i5's and quad core i5's on 1156 and 1155

It seems pretty clear to me that the i3,i5,i7 names were designed for marketing. Generally within a series of processors an i7 will be better than an i5 which will be better than an i3 BUT go across different types (ultra mobile VS mobile VS desktop) and/or different generations and this no longer holds. The only reasonable assumption is that Intel marketing wants people to assume that an i7 machine is faster than an i5 machine regardless of whether or not that is actually the case.

 

[wtourist]

I3 2105 is out. Video 3000. I think Intel is making it up as it gioes along.

 

[novadaemon]

I3 2105 is out. Video 3000. I think Intel is making it up as it gioes along.

dont forget, the unlockable i3 2120k will be coming out soon.

 

[insaneshell]

I don't prefer the AMD naming scheme, I think it's worse. With AMD, there's also the whole am2, am2+, am3, am3+, and backwards compatable or not stuff.

I just want something that'll give more relevent details.

For example, i3 for dual cores, i5 for quad cores, i7 for quad core + HT, and i9 for hexa cores. The generation 2 thing I can live with, but how about something that tells you the speed of the processor ? i5 2500k is 3.3ghz and i7 2600k is 3.4ghz, so maybe 33 / 34 in there, HT, and a socket differentiation (currently, 1366 is socket B, 1156 is H, 1155 is H2). i5 233K H2 and i7 234K H2 maybe ? i5 23K / i7 24K ?

Mostly it's just stupid that the "500" part doesn't do anything except provide a number to compare it to "600" or "400". At least make it do something useful.



For sure. You could buy an i7 and have it not fit into 2/3 of the available sockets and also have it either be 2, 4, or 6 cores.

Way to go intel.

AM2 and AM2+ CPU's are discontiunued and the backwards compatability is a major selling point, one reason why I don't jump on the intel bandwagon. I was able to use my ASUS M2N-SLI to install a Phenom 9650 from an Athlon x2 6400, then was able to replace the board for an MSI 770T-C35, then replaced the Phenom later with an Athlon II X3 455 then replace that with an MSI 870-G45 for AM3 and DDR3. I didnt have to buy board/cpu/ram all at once, heck my DDR2-800 I bought in 2006 served me till this christmas when I went AM3 lol.

 

[sethk]

I dont understand why they didn't use the laptop naming scheme for desktops as well, where they would suffix "QM" for the quad core mobile CPUs. That makes it less confusing when you have some dual core i7 processors, but it would also help distinguish hex core processors. i7 960? 970? How do I remember which one is hex core? They could use i7 960Q and i7 970H (or something) to help us keep it straight.

 

[firas]

so will the larger surface area with the SB-E chips mean relatively lower temps?

 

[Brahmzy]

Somethin' tells me my Thermalright Venemous X Black won't completely cover the 2011 heatspreader.

 

[Digital Viper-X-]

pricing will most likely start where SB 1155 stops, so around $300 for the low end SB-Es I'd say

 

[EngrChris]

pricing will most likely start where SB 1155 stops, so around $300 for the low end SB-Es I'd say

I agree. I can imagine the starting price for boards would be around $175.

 

[x509]

Can we all take a deep breath and ask ourselves what we really KNOW, from reliable sources.

I would also like to point out that Intel created the current naming systems because they felt that processor speeds alone were not sufficient to indicate the relative "value" of different parts. This was to "help" consumers.

 

[needmorecarnitine]

so will the larger surface area with the SB-E chips mean relatively lower temps?

They will use more power. The larger surface area may mean lower temps per measured area but overall they will use more power.

 

[plugwash]

Can we all take a deep breath and ask ourselves what we really KNOW, from reliable sources.

It depends what you count as "reliable sources"

Afaict the main thing we have is the image at http://pic.xfastest.com/sxs112/any/1.png (many other copies can probablly be found) showing an intel roadmap. Afaict intel roadmaps are generally reasonable indicators as to what is coming though the i7-980 (NOT the i7-980x) on that roadmap doesn't seem to have actually appeared yet and we are nearly at the end of Q2 2011.

We know that intel traditionally releases extreme editions at ~$1000 (sometimes i've seen them release with a nominal price of $1000 but a slightly higher or lower price in practice) and I doubt they will change this significantly. Assuming the roadmap is accurate we also know that the lowest of the thee LGA2011 desktop processors is both a faster processor and on a higher end platform than the i7-2600 (and is therefore likely to be more expensive that it). What we don't know is which two price points in the gap between the i7-2600 and the extreme edition intel will pick.

 

[German Muscle]

Not much, judging by those clockspeeds. Core i5 750 to Core i5 2500 performance increase was around %30, with a %25 boost in clock speed accounting for most of that. Given the much higher clock speeds of the i7 970, 980x (3.2, 3.33) - I see the difference being almost non-existent (%10 or less), unless these things ship with base clocks of 4 GHz.

Clock speeds mean nothing anymore. Its all about the architecture. With Quad Channel DDR3 and PCI-Express v3.0 and more PCI-E Lanes just adds to it. the 2600K was on par or slightly better at most things except rendering and video editing.

It seems pretty clear to me that the i3,i5,i7 names were designed for marketing. Generally within a series of processors an i7 will be better than an i5 which will be better than an i3 BUT go across different types (ultra mobile VS mobile VS desktop) and/or different generations and this no longer holds. The only reasonable assumption is that Intel marketing wants people to assume that an i7 machine is faster than an i5 machine regardless of whether or not that is actually the case.

Why are you comparing mobile CPUs vs desktop? you cant use one in the other. i7 is better than i5 and i5 is better than i3.

 

[x509]

It depends what you count as "reliable sources"

Exactly. Sometimes when I try to check a "fact" or even a "reliable rumor" I see lots of search hits, but often all the different web posts lead back to the same source. The Internet has been called "a vast echo chamber.

Afaict the main thing we have is the image at http://pic.xfastest.com/sxs112/any/1.png (many other copies can probablly be found) showing an intel roadmap. Afaict intel roadmaps are generally reasonable indicators as to what is coming though the i7-980 (NOT the i7-980x) on that roadmap doesn't seem to have actually appeared yet and we are nearly at the end of Q2 2011.

Agreed. But the Intel roadmap sems as much a technology map as a specific product map, e.g. how many different speeds/bins.

We know that intel traditionally releases extreme editions at ~$1000 (sometimes i've seen them release with a nominal price of $1000 but a slightly higher or lower price in practice) and I doubt they will change this significantly.

Comjpletely agree here. Intel seems to price its parts consistently that way. They are assuming that the people who "need" (have to have) the EE part are not exactly price-sensitive, so they will charge "what the market will bear." And if I'm an Intel shareholder, I'm very happy with this pricing. As a customer, less so.

Assuming the roadmap is accurate we also know that the lowest of the thee LGA2011 desktop processors is both a faster processor and on a higher end platform than the i7-2600 (and is therefore likely to be more expensive that it).

What we don't know is which two price points in the gap between the i7-2600 and the extreme edition intel will pick.

Exactly.